{"id":44,"date":"2020-09-03T16:38:49","date_gmt":"2020-09-03T13:38:49","guid":{"rendered":"http:\/\/acikkitap.com.tr\/oaek\/chapter\/bolum-3-olcme-ve-ogrenme-analitiginde-kullanimi\/"},"modified":"2020-09-03T16:38:49","modified_gmt":"2020-09-03T13:38:49","slug":"bolum-3-olcme-ve-ogrenme-analitiginde-kullanimi","status":"publish","type":"chapter","link":"https:\/\/acikkitap.com.tr\/oaek\/chapter\/bolum-3-olcme-ve-ogrenme-analitiginde-kullanimi\/","title":{"raw":"B\u00f6l\u00fcm 3 \u00d6l\u00e7me ve \u00d6\u011frenme Analiti\u011finde Kullan\u0131m\u0131","rendered":"B\u00f6l\u00fcm 3 \u00d6l\u00e7me ve \u00d6\u011frenme Analiti\u011finde Kullan\u0131m\u0131"},"content":{"raw":"\n

Yoav Bergner<\/span><\/span><\/p>\n\u00d6\u011frenme Analiti\u011fi Ara\u015ft\u0131rma A\u011f\u0131, New York \u00dcniversitesi, ABD<\/span><\/span>\n\nDOI: 10.18608\/hla17.003<\/span><\/span>\n

\u00d6Z<\/h2>\nPsikolojik \u00f6l\u00e7me, zihinsel durumlar hakk\u0131nda uygunlu\u011fu kan\u0131tlanm\u0131\u015f iddialarda bulunma s\u00fcrecidir. ]Bu haliyle, tipik olarak \u015funlar\u0131 i\u00e7ermektedir: Bir yap\u0131n\u0131n tan\u0131mlanmas\u0131; bir \u00f6l\u00e7me modeli belirlemek ve g\u00fcvenilir bir ara\u00e7 geli\u015ftirmek; \u00e7e\u015fitli hata kaynaklar\u0131n\u0131 analiz etmek (operat\u00f6r hatas\u0131 d\u00e2hil) ve sonucun belirli kullan\u0131mlar\u0131 i\u00e7in ge\u00e7erli bir arg\u00fcman \u00e7er\u00e7evelemek. \u00d6rt\u00fck de\u011fi\u015fkenlerin \u00f6l\u00e7\u00fcm\u00fc, sonu\u00e7ta, bireyler ve gruplar i\u00e7in y\u00fcksek riskli sonu\u00e7lar do\u011furabilecek y\u00fcksek perdeden bir giri\u015fimdir. Bu b\u00f6l\u00fcm, analitik ve e\u011fitsel veri madencili\u011fi \u00f6\u011frenen uygulay\u0131c\u0131lar i\u00e7in e\u011fitsel ve psikolojik \u00f6l\u00e7meye bir giri\u015f niteli\u011findedir. Yap\u0131lar, ara\u00e7lar ve \u00f6l\u00e7me hata kaynaklar\u0131 hakk\u0131ndaki daha kavramsal malzemeden, belirli \u00f6l\u00e7me modelleri ve kullan\u0131mlar\u0131 hakk\u0131nda teknik detaylar\u0131n artt\u0131r\u0131lmas\u0131na y\u00f6nelik olarak, tarihsel olmaktan ziyade tematik olarak d\u00fczenlenmi\u015ftir. A\u00e7\u0131klay\u0131c\u0131 ve kestirimci modelleme aras\u0131ndaki felsefi farkl\u0131l\u0131klar\u0131n baz\u0131lar\u0131 sona do\u011fru incelenmi\u015ftir.<\/span>\n\nAnahtar Kelimeler<\/span>: \u00d6l\u00e7me, \u00f6rt\u00fck s\u0131n\u0131f modelleri, model uyumu<\/span>\n

\u00d6\u011frencilerin ne bildi\u011finin ve -duyu\u015fsal \u00f6l\u00e7\u00fctlere giderek daha fazla \u00f6nem verildi\u011fi g\u00f6z \u00f6n\u00fcne al\u0131nd\u0131\u011f\u0131nda- nas\u0131l hissettiklerini bilmek, \u00f6\u011frenmeye ili\u015fkin \u00e7o\u011fu sohbetin \u00f6z\u00fcn\u00fc olu\u015fturur. Bununla birlikte, bir \u00f6\u011frencinin bilgi, becerilerini, tutumlar\u0131n\u0131\/ istidat \/ yeteneklerini (BB\u0130) ve \/ veya duygular\u0131n\u0131 \u00f6l\u00e7mek, boy veya kilosunu \u00f6l\u00e7mekten daha karma\u015f\u0131k bir i\u015ftir. Psikolojik \u00f6l\u00e7me, \u00f6zel bir programa tahsis edilme (ileri d\u00fczey veya telafi), bir \u00fcniversiteye kabul, i\u015fe al\u0131m, hastaneye yat\u0131\u015f veya tutuklanma gibi y\u00fcksek riskli sonu\u00e7lar do\u011furabilecek rahats\u0131z edici bir u\u011fra\u015ft\u0131r. Bireysel seviyedeki k\u00fc\u00e7\u00fck \u00f6l\u00e7me yan\u0131lg\u0131lar\u0131 bile bulgular\u0131 gruplar i\u00e7in birle\u015ftirildi\u011finde b\u00fcy\u00fck sonu\u00e7lar do\u011furabilir. (Kane, 2010). Bu sonu\u00e7lardaki hassasiyet, E\u011fitsel ve Psikolojik \u00d6l\u00e7me Standartlar\u0131nda<\/i><\/span> yer alan bir y\u00fczy\u0131ldan fazla s\u00fcren y\u00f6ntem bilim ara\u015ft\u0131rmas\u0131yla ortaya \u00e7\u0131km\u0131\u015ft\u0131r.(AERA, APA ve NCME, 2014). \u00d6l\u00e7me bu d\u00fczeyde, \u00f6\u011frenme ve \u00f6\u011frenme ortamlar\u0131n\u0131 anlamak ve en iyi hale getirmek amac\u0131yla \u00f6\u011frenme analiti\u011fi ve e\u011fitsel veri madencili\u011finde kullan\u0131labiliyorsa (Siemens ve Baker, 2012), \u00f6l\u00e7mede kabul edilebilir hatalar neler olacakt\u0131r? Ne de olsa verinin \u201cdijital okyanusundan faydalanman\u0131n\u201d nihayetinde ayr\u0131 de\u011ferlendirmelere duyulan ihtiyac\u0131n yerini alabilece\u011fi iddia edilmi\u015ftir (Behrens ve DiCerbo, 2014). Bir taraftan da ki\u015fi en az\u0131ndan \u00f6\u011frenmeyi yanl\u0131\u015f anlamaktan<\/i><\/span> veya eksilen<\/i><\/span> \u00f6\u011frenen deneyimlerinden ka\u00e7\u0131nmak isteyecektir.<\/p>\n\n

\u00d6L\u00c7ME NED\u0130R? FELSEFE VE TEMEL F\u0130K\u0130RLER<\/h2>\n

Psikolojik \u00f6l\u00e7me tart\u0131\u015fmalar\u0131, genellikle fiziksel \u00f6l\u00e7me ile z\u0131tl\u0131klar \u00e7izerek ba\u015flamaktad\u0131r (\u00f6r. Armstrong, 1967; Borsboom, 2008 DeVellis, 2003; Lord ve Novick, 1968; Maul, Irribarra ve Wilson, 2016; Michell, 1999; Sijtsma, 2011). S\u00fcre\u00e7te, ara\u00e7salla\u015ft\u0131rma veya i\u015flemselle\u015ftirme, \u00f6l\u00e7\u00fcmlerin tekrarlanabilirli\u011fi veya kesinli\u011fi, hata kaynaklar\u0131 ve \u00f6nlemin kendisinin yorumlanmas\u0131 gibi bir dizi \u00f6nemli psikolojik \u00f6l\u00e7me fakt\u00f6r\u00fc ortaya \u00e7\u0131kar. Psikolojik \u00f6l\u00e7menin a\u015fa\u011f\u0131dakileri i\u00e7erdi\u011fi s\u00f6ylenebilir; bir yap\u0131y\u0131 tan\u0131mlamak, bir \u00f6l\u00e7me y\u00f6ntemi belirlemek ve g\u00fcvenilir bir ara\u00e7 (geli\u015ftirmek); \u00e7e\u015fitli hata kaynaklar\u0131n\u0131 analiz etmek ve nedenlerini a\u00e7\u0131klamak (uygulay\u0131c\u0131 hatas\u0131 d\u00e2hil) ve sonucun belirli kullan\u0131mlar\u0131 i\u00e7in ge\u00e7erli bir arg\u00fcman \u00e7er\u00e7evelemek.<\/p>\n\n

Yap\u0131lar<\/h3>\n

Psikolojik yap\u0131lar ger\u00e7ekten var m\u0131? Hangi anlamda \u00f6\u011frencinin halet-i ruhiyesini ger\u00e7ekten bilebiliriz? Bir nesnenin fiziksel uzunlu\u011fu gibi de\u011fi\u015fkenlerin do\u011frudan g\u00f6zlendi\u011fini veya tezah\u00fcr etti\u011fini s\u00f6ylerken, bireyin zihinsel durumlar\u0131n\u0131 veya psikolojik \u00f6zelliklerini yaln\u0131zca dolayl\u0131 olarak g\u00f6zlemlendi\u011fi veya gizlendi\u011fini s\u00f6yl\u00fcyoruz. Yap\u0131 terimi, \u00f6rt\u00fck de\u011fi\u015fkenle<\/i><\/span> de\u011fi\u015fmeli<\/i><\/span> olarak kullan\u0131l\u0131rken \u00f6zellik, yap\u0131n\u0131n zamana g\u00f6re sabit olu\u015funu ima etmek i\u00e7in kullan\u0131l\u0131r (Lord ve Novick, 1968). Asl\u0131nda, fiziksel \u00f6l\u00e7me bile dolayl\u0131 olarak ger\u00e7ekle\u015ftirilir. Uzunlu\u011fu do\u011frudan duyular\u0131m\u0131zla alg\u0131layabilmemize ra\u011fmen, uzunlu\u011fun \u00f6l\u00e7\u00fclmesi<\/i><\/span>, bir mezura gibi bir referans nesnesi veya alet ile bir kar\u015f\u0131la\u015ft\u0131rma i\u015flemini i\u00e7erir. Mezura, uzunluk kar\u015f\u0131la\u015ft\u0131rmalar\u0131n\u0131 resmile\u015ftiren in\u00e7 veya santimetre gibi bir \u00f6l\u00e7ek sa\u011flar. \u00d6rne\u011fin, iki uzunluk aras\u0131ndaki fark\u0131 bir \u00f6l\u00e7\u00fcm\u00fc di\u011ferinden \u00e7\u0131kartarak inceleyebiliriz.<\/p>\n

Yirminci y\u00fczy\u0131l\u0131n ilk yar\u0131s\u0131nda, \u00f6l\u00e7menin felsefi \u00f6l\u00e7me meselelerini \u00e7\u00f6zme \u00e7abalar\u0131 Bridgman'\u0131 (1927) ve di\u011ferlerini i\u015flemselcili\u011fe y\u00f6nlendirdi; burada uzunluk, k\u00fctle ve yo\u011funluk gibi fiziksel kavramlar ile bunlar\u0131 \u00f6l\u00e7mek i\u00e7in kullan\u0131lan i\u015flemlerin \u201ce\u015f anlaml\u0131\u201d oldu\u011fu anla\u015f\u0131ld\u0131. Yani, uzunluk (muhtemelen faraz\u00ee) bir uzunluk \u00f6l\u00e7\u00fcm y\u00f6nteminin \u00fcr\u00fcn\u00fc olarak anla\u015f\u0131lmaktad\u0131r. Bu fikir, yap\u0131lar\u0131 onlar\u0131 \u00f6l\u00e7mek i\u00e7in kullan\u0131lan ara\u00e7lardaki puanlarla e\u015fle\u015ftirme yoluyla matematik yetene\u011fi ve d\u0131\u015fa d\u00f6n\u00fckl\u00fck gibi psikolojik yap\u0131lara aktar\u0131labilir. B\u00f6ylelikle matematik yetene\u011fi daha sonra bir matematik testindeki bir puana ve d\u0131\u015fa d\u00f6n\u00fckl\u00fck, Likert madde anketinde verilen bir puana e\u015f de\u011fer olur. Bu pozitivist tutum, Stevens'\u0131n \u201cnesnelere ya da olaylara kurallara g\u00f6re say\u0131lar\u0131n atanmas\u0131\u201d olarak yapt\u0131\u011f\u0131 \u00f6l\u00e7me tan\u0131m\u0131nda yans\u0131t\u0131lmaktad\u0131r (1946, s. 677). Yap\u0131lara ili\u015fkin i\u015flemselci g\u00f6r\u00fc\u015f ge\u00e7mi\u015fte olduk\u00e7a etkiliydi ancak bir\u00e7ok nedenden dolay\u0131 \u00f6zellikle de i\u015flemselcilik yap\u0131n\u0131n onu \u00f6l\u00e7mek i\u00e7in var olan her ara\u00e7 i\u00e7in yeniden tan\u0131mlamay\u0131 gerektirmesi nedeniyle reddedildi (Maul, Irribarra ve Wilson, 2016; Michell, 1999).<\/p>\n

\u0130\u015flemselci bir yorum reddedildi\u011finde \u00f6rt\u00fck de\u011fi\u015fkenlere dair epistemolojik ve ontolojik sorular\u0131 a\u00e7\u0131kta b\u0131rakt\u0131\u011f\u0131 g\u00f6r\u00fclmektedir. Mislevy (2009, 2012), yap\u0131land\u0131rmac\u0131-ger\u00e7ek\u00e7i bir konumu a\u00e7\u0131k\u00e7a belirtir; yani,<\/i><\/span> model temelli bir ak\u0131l y\u00fcr\u00fctmeyi taahh\u00fct ederek, kat\u0131 ger\u00e7ek\u00e7ili\u011fe ba\u011fl\u0131 olmadan bir yap\u0131 varm\u0131\u015f gibi konu\u015fabiliriz. Model temelli ak\u0131l y\u00fcr\u00fctme, bir sistemin -\u00f6rne\u011fin, ki\u015filer ve cevaplar aras\u0131ndaki yap\u0131 arac\u0131l\u0131 ili\u015fki- g\u00f6ze \u00e7arpan y\u00f6nleri (\u00f6r. \u00f6r\u00fcnt\u00fcler) yakalayan basitle\u015ftirilmi\u015f bir temsilini, kabul etmek anlam\u0131na gelir ve olgular\u0131 a\u00e7\u0131klamam\u0131z\u0131 veya tahmin etmemizi sa\u011flar (Mislevy, 2009; a\u00e7\u0131klay\u0131c\u0131 \/ kestirimci modelleri bu b\u00f6l\u00fcm\u00fcn ilerleyen k\u0131s\u0131mlar\u0131nda ele alaca\u011f\u0131z). George Box'\u0131n \u00fcnl\u00fc s\u00f6z\u00fcnde dedi\u011fi gibi, \u201ct\u00fcm modeller yanl\u0131\u015f ancak baz\u0131lar\u0131 yararl\u0131d\u0131r\u201d (Box, 1979). Zorluk, faydal\u0131 modeller veya Stevens'\u0131n tan\u0131m\u0131 ile ifade edildi\u011finde, yararl\u0131 \u00f6l\u00e7me kurallar\u0131 ile ortaya \u00e7\u0131kmaya devam etmektedir.<\/p>\n

Fiziksel teoriler say\u0131ca az ve daha kapsaml\u0131 olma e\u011filimindeyken, psikolojik teoriler \u00e7ok say\u0131da ve s\u0131n\u0131rl\u0131 bir \u015fekilde tan\u0131ml\u0131d\u0131rlar (DeVellis, 2003). Yap\u0131lar uydurulmu\u015f\/icat edilen \u015feyler oldu\u011fu i\u00e7in, say\u0131lar\u0131 i\u00e7in deneysel bir s\u0131n\u0131r yoktur. Bir yap\u0131 hakk\u0131nda bir \u00f6l\u00e7\u00fcm arac\u0131n\u0131n<\/i><\/span> yoklu\u011funda konu\u015fmak m\u00fcmk\u00fcnd\u00fcr ancak bir \u00f6l\u00e7\u00fcm arac\u0131 her zaman bir \u015feyi \u00f6l\u00e7mek i\u00e7in tasarlanm\u0131\u015ft\u0131r. Bu nedenle, kendilerini \u00f6l\u00e7mek i\u00e7in \u00f6nceden geli\u015ftirilen ara\u00e7lara uygun ve son derece k\u0131smi bir \u00f6\u011frenme analiti\u011fine ili\u015fkin yap\u0131lar listesi \u00e7\u0131karsamas\u0131 yapabiliriz. \u00d6rnekler aras\u0131nda zek\u00e2 (\u00f6r. Stanford-Binet Zek\u00e2 \u00d6l\u00e7e\u011fi), akademik yatk\u0131nl\u0131k (\u00f6r. bu SAT1<\/sup><\/a> testi), akademik ba\u015far\u0131 (hem b\u00fcy\u00fck \u00f6l\u00e7ekli s\u0131navlar hem de ders ba\u015far\u0131 s\u0131navlar\u0131 d\u00e2hil say\u0131s\u0131z \u00f6rnek), ki\u015filik (\u00f6r. \u201cb\u00fcy\u00fck be\u015f\u201d fakt\u00f6r modeli; Digman, 1990), ba\u015far\u0131 hedef oryantasyonu (\u00f6r. Midgley vd., 2000), tatmin duygular\u0131 (Pekrun, Goetz, Frenzel, Barchfeld ve Perry, 2011), sab\u0131r (Duckworth, Peterson, Matthews ve Kelly, 2007), \u00f6z yeterlilik teorileri ve sabit \/ b\u00fcy\u00fcme zihniyeti teorileri (Dweck, 2000; Yeager ve Dweck, 2012), i\u00e7sel motivasyon (Deci ve Ryan, 1985; Guay, Vallerand ve Blanchard, 2000), \u00f6z y\u00f6netimli \u00f6\u011frenme ve \u00f6z yeterlik (\u00f6r. Pintrich ve De Groot, 1990), \u00f6\u011frenme g\u00fcc\u00fc (Buckingham Shum ve Deakin Crick, 2012; Crick, Broadfoot ve Claxton, 2004) ve kitle kaynakl\u0131 \u00f6\u011frenme yetene\u011fi (Milligan ve Griffin, 2016) vard\u0131r.<\/p>\n

Yukar\u0131da listelenen yap\u0131lar\u0131n bir\u00e7o\u011fu \u00e7ok boyutludur, yani birden \u00e7ok fakt\u00f6r i\u00e7erirler. \u0130li\u015fkili yap\u0131lar\u0131 ayr\u0131\u015ft\u0131rman\u0131n ya da birle\u015ftirmenin de\u011feri bir tart\u0131\u015fma konusudur (Edwards, 2001; Schwartz, 2007).<\/p>\n\n

\u00d6l\u00e7me Ara\u00e7lar\u0131<\/h3>\n

Psikolojik \u00f6l\u00e7me ara\u00e7lar\u0131na genellikle test veya soru formlar\u0131 (ayr\u0131ca anketler ve envanterler) denir ve maddelerden veya g\u00f6stergelerden olu\u015furlar. Test kelimesi daha \u00e7ok zek\u00e2, bili\u015fsel yetenek ve psikomotor becerileri gibi yap\u0131lar i\u00e7in kullan\u0131l\u0131r; burada derse veya s\u0131nava giren ki\u015finin performans\u0131n\u0131 en \u00fcst seviyeye \u00e7\u0131karmaya \u00e7al\u0131\u015fmas\u0131 istenir (Sijtsma, 2011). Soru formu kat\u0131l\u0131mc\u0131lar\u0131ndan, aksine, d\u00fc\u015f\u00fcnceleri, duygular\u0131 ve davran\u0131\u015flar\u0131 ile ilgili d\u00fcr\u00fcst\u00e7e cevaplar vermeleri istenir. (Tepki yanl\u0131l\u0131k de\u011feri, ge\u00e7erlili\u011fe geldi\u011fimizde tan\u0131mlayaca\u011f\u0131m\u0131z gibi, bu ayr\u0131m\u0131 bulan\u0131kla\u015ft\u0131rabilir). Deneklerin ara\u00e7larla nas\u0131l etkile\u015fime girmesi beklendi\u011fine ili\u015fkin bu tan\u0131mlaman\u0131n bir \u00f6l\u00e7me modelinin<\/i><\/span> temel ilkelerini ortaya \u00e7\u0131kard\u0131\u011f\u0131na dikkat ediniz. Daha yetenekli bir s\u0131nav kat\u0131l\u0131mc\u0131s\u0131n\u0131n bir yetenek s\u0131nav\u0131nda daha y\u00fcksek puan alaca\u011f\u0131n\u0131 ve daha endi\u015feli bireyin kayg\u0131 anketinde daha y\u00fcksek puan alaca\u011f\u0131n\u0131 varsay\u0131yoruz.<\/p>\n

Bazen \u00f6l\u00e7me \u00f6l\u00e7e\u011fi terimi, enstr\u00fcmanla de\u011fi\u015fimli olarak kullan\u0131l\u0131r (DeVellis, 2003). \u00d6l\u00e7ek test veya anketin puanland\u0131\u011f\u0131n\u0131 g\u00f6sterir. Do\u011fru ve yanl\u0131\u015f cevaplar\u0131 olan ve evet \/ hay\u0131r sorular\u0131na sahip olan ikili maddeler genellikle {0, 1} 'de yer alan de\u011ferlerle ikili bir \u015fekilde puanlan\u0131r. Likert \u00f6l\u00e7e\u011fi, puanlama \u00f6l\u00e7e\u011fi ve g\u00f6rsel-analog \u00f6l\u00e7ekler (Luria, 1975), kesikli veya s\u00fcrekli say\u0131sal de\u011ferler alabilen di\u011fer madde t\u00fcrleridir. Bireysel maddelerin puanlar\u0131n\u0131n toplanarak bir toplam puana (ayr\u0131ca ham puan olarak) d\u00f6n\u00fc\u015ft\u00fcr\u00fclmesi, bir arac\u0131n puanlanmas\u0131 i\u00e7in bir y\u00f6ntemdir ancak tek veya zorunlu olarak en iyi y\u00f6ntem de\u011fildir (Lord ve Novick, 1968; Millsap, 2012). A\u011f\u0131rl\u0131kl\u0131 toplam puanlar ve madde tepki teorisi (MTK; Baker ve Kim, 2004) bir dizi alternatif sunar.<\/p>\n

Testlerin ve soru formlar\u0131n\u0131n kullan\u0131lmas\u0131, insanlar\u0131 ger\u00e7ek hayatta g\u00f6zlemlemenin ve kendili\u011finden d\u00fc\u015f\u00fcnceleri ifade etmelerini veya ilgilenilen davran\u0131\u015flar\u0131 sergilemelerini beklemenin alternatifine k\u0131yasla hem verimlilik hem de standardizasyon meselesidir (Sijtsma, 2011). \u00d6\u011frenme analiti\u011finde, verilerin verimli bir \u015fekilde toplanmas\u0131 genellikle sorun de\u011fildir ancak standardizasyon eksikli\u011fi \u00f6l\u00e7me hatas\u0131na bir a\u00e7\u0131klama getirmeyi zorla\u015ft\u0131rabilir.<\/p>\n\n

\u00d6l\u00e7mede Hata Kayna\u011f\u0131<\/h3>\n

Tecr\u00fcbelerden biliyoruz ki psikolojik \u00f6l\u00e7meler fiziksel \u00f6l\u00e7meler kadar tutarl\u0131 bir \u015fekilde tekrar edilebilir de\u011fildir. \u0130nsanlar\u0131n bir araca verdi\u011fi cevaplar\u0131n yeteneklerini, tutumlar\u0131n\u0131 veya di\u011fer ilgi alanlar\u0131n\u0131 g\u00fcvenilir bir \u015fekilde yans\u0131tmayabilece\u011fini de biliyoruz. \u0130statistiksel modeller, \u00f6geleri, g\u00f6stergeleri veya testleri \u00f6rt\u00fck bir de\u011fi\u015fkenin rastgele \u00f6rnekleri olarak d\u00fc\u015f\u00fcnmemize izin verir. \u00d6rt\u00fck de\u011fi\u015fken rastgele bir de\u011fi\u015fken olabilir veya ger\u00e7ek puan teorisinde oldu\u011fu gibi sabitlenebilir (Lord ve Novick, 1968). Her iki durumda da \u00f6l\u00e7me numuneleri bazen rastgele hata olarak adland\u0131r\u0131lan ve \u00f6z\u00fcnde i\u00e7sel tekrarlanamazl\u0131ktan kaynaklanan ve yans\u0131z olan hataya sahip olacakt\u0131r (tekrarlanan \u00f6l\u00e7\u00fcmlerin bir miktar\u0131n\u0131n da\u011f\u0131t\u0131m\u0131 \u00fczerine s\u0131f\u0131r beklentisine sahip olma anlam\u0131nda). \u00d6n yarg\u0131l\u0131 sistematik, yanl\u0131 olan bir hata da bulunabilir.<\/p>\n

Bir \u00f6l\u00e7me \u00e7er\u00e7evesi veya modeli benimsedi\u011fimizde hatayla ilgili daha kesin veya bi\u00e7imsel ifadeler ortaya \u00e7\u0131kar. \u00d6rne\u011fin, ger\u00e7ek puan teorisi ve fakt\u00f6r analizinde, bir arac\u0131n g\u00fcvenilirli\u011fine ili\u015fkin tahminler t\u00fcretmek i\u00e7in paralel testler veya e\u015fde\u011fer formlar a\u00e7\u0131s\u0131ndan ak\u0131l y\u00fcr\u00fctebiliriz. \u00d6l\u00e7\u00fcm hatas\u0131, modelde a\u00e7\u0131kland\u0131\u011f\u0131 gibi verilerdeki yap\u0131ya atfedilmemi\u015f herhangi bir de\u011fi\u015fiklik olarak da tan\u0131mlanabilir (AERA, APA ve NCME, 2014). \u00d6l\u00e7me modelleri konusundaki tart\u0131\u015fmam\u0131z\u0131 bitirdikten sonra hata kaynaklar\u0131n\u0131 tekrar g\u00f6zden ge\u00e7irece\u011fiz.<\/p>\n\n

G\u00fcvenilirlik<\/h3>\n

G\u00fcvenilirlik, bir araca atfedilir ve puanlar\u0131n tutarl\u0131l\u0131\u011f\u0131n\u0131n (AERA, APA ve NCME, 2014), \u00f6zellikle de toplam de\u011fi\u015fkenli\u011fin \u00f6rt\u00fck de\u011fi\u015fkene atfedilen puanlardaki oran\u0131n\u0131n bir \u00f6l\u00e7\u00fcs\u00fcd\u00fcr (DeVellis, 2003). \u00d6rnekleme (ger\u00e7ek puan teorisinde) ve modele ba\u011fl\u0131 (daha karma\u015f\u0131k modellerde) olabilir. Bu kelime bazen, yayg\u0131n olarak Cronbach's (1951) alfa a olan, [0, 1] aras\u0131nda de\u011fi\u015fen belirli bir g\u00fcvenilirlik katsay\u0131s\u0131 anlam\u0131na gelir. Bununla birlikte, g\u00fcvenilirlik terimi, asl\u0131nda bir korelasyon olan ve test- tekrar test g\u00fcvenilirli\u011fi ve puanlay\u0131c\u0131lar aras\u0131 g\u00fcvenirlik anlam\u0131nda da kullan\u0131lmaktad\u0131r (\u00f6r. Cohen'in kappa, k<\/i><\/span>; Cohen, 1968). Uygulay\u0131c\u0131lar bazen, \u00f6l\u00e7eklerin kullanmak i\u00e7in yeterince iyi oldu\u011funa karar vermek i\u00e7in .70 alt s\u0131n\u0131r a<\/i><\/span> olarak kabul edilebilir (Cortina, 1993) de\u011ferlere dair y\u00f6nergelere sorgulamadan ba\u011fl\u0131 kal\u0131rlar. Ancak istatistiksel g\u00fcc\u00fcn a<\/i><\/span>'n\u0131n daha y\u00fcksek de\u011ferlerle artt\u0131\u011f\u0131na dikkat edilmelidir (DeVellis, 2003). Bu nedenle, bir \u00f6l\u00e7e\u011fin g\u00fcvenilirli\u011fini geli\u015ftirme \u00e7abas\u0131, daha b\u00fcy\u00fck \u00f6rneklemeler al\u0131nmas\u0131n\u0131n faydalar\u0131ndan a\u011f\u0131r basabilir.<\/p>\n\n

Ge\u00e7erlik<\/h3>\n

Ge\u00e7erlilik, Standartlar\u0131n\u0131n<\/i><\/span> ilk b\u00f6l\u00fcm\u00fc \u201cGe\u00e7erlilik, kan\u0131tlar\u0131n ve teorinin, testlerin \u00f6nerilen kullan\u0131m\u0131 i\u00e7in test puanlar\u0131n\u0131n yorumlanmas\u0131n\u0131 destekleme derecesini belirtir. ....testin ge\u00e7erli\u011fi\" \u015feklindeki niteliksiz ifadeyi kullanmak do\u011fru de\u011fildir. \" (s.11) olarak ba\u015flayan standartlar'\u0131n en \u00f6nemli konusudur. Daha geni\u015f olan \u201c\u00f6l\u00e7\u00fc\u201d terimini daha dar olan \u201ctest\u201din yerine ge\u00e7irme sayesinde, ge\u00e7erlili\u011fin \u00f6\u011frenme analiti\u011fi i\u00e7in ne b\u00fcy\u00fck \u00f6nem ta\u015f\u0131d\u0131\u011f\u0131 a\u00e7\u0131k\u00e7a g\u00f6r\u00fclmelidir. Standartlar<\/i><\/span>'da do\u011frulama arg\u00fcmanlar\u0131nda kullan\u0131lan dili \u015fekillendirmeye dair Messick'in (1995), Cronbach ve Meehl'i (1955) etkili bir \u015fekilde elden ge\u00e7irmesinde de belirgin olan bir yakla\u015f\u0131m olarak (bk. Ayr\u0131ca Kane, 2001) dili \u015fekillendirmeye dair somut bir odaklanma vard\u0131r. Ge\u00e7erli\u011fe ili\u015fkin kan\u0131t t\u00fcrleri (\u201cge\u00e7erlilik t\u00fcrleri\u201dnden ziyade), tepki s\u00fcre\u00e7leri hakk\u0131ndaki kan\u0131t, arac\u0131n i\u00e7 yap\u0131s\u0131 hakk\u0131ndaki kan\u0131t, yak\u0131nsak ve ay\u0131r\u0131c\u0131 kan\u0131t, kriter referanslar\u0131 (\u00f6ng\u00f6r\u00fclen kriterler d\u00e2hil) ve genellenebilirlik hakk\u0131nda kan\u0131t i\u00e7erir.<\/p>\n

Bu b\u00f6l\u00fcm\u00fcn ba\u015flar\u0131nda, anketlere verilen cevaplar\u0131n d\u00fcr\u00fcst d\u00fc\u015f\u00fcncelere ve duygulara kar\u015f\u0131l\u0131k geldi\u011fi varsay\u0131m\u0131na de\u011finmi\u015ftik. Bununla birlikte, tepki yanl\u0131l\u0131\u011f\u0131 t\u00fcrleri hakk\u0131nda kabul edilebilme yanl\u0131l\u0131\u011f\u0131ndan (evet diyerek; Messick ve Jackson, 1961) sosyal istenirlik yanl\u0131l\u0131\u011f\u0131na (ayr\u0131ca, iyiyi oynama; Nederhof, 1985) a\u015f\u0131r\u0131 ve \u0131l\u0131ml\u0131 cevaplay\u0131c\u0131 yanl\u0131l\u0131\u011f\u0131na (yani, Likert-skalalar\u0131n\u0131n a\u015f\u0131r\u0131 u\u00e7lar\u0131n\u0131 se\u00e7me e\u011filiminde olan insanlar) geni\u015f bir literat\u00fcr bulunmaktad\u0131r. (Bachman ve O'Malley, 1984). Hile yapmaya istekli olma, cinsel fanteziler veya \u0131rkla ilgili tutumlar gibi hassas konular hakk\u0131ndaki soru formlar\u0131 ve anketler i\u00e7in daha s\u0131k belgelenmesine ra\u011fmen, Newtoncu d\u00fc\u015f\u00fcnceyi de\u011ferlendirmek i\u00e7in kullan\u0131lan kuvvet kavram\u0131 envanteri gibi (KKE; Hestenes, Wells ve Swackhamer, 1992) cevaplar\u0131n \u00f6z uyumlanma ve oto sans\u00fcr\u00fc de e\u011fitsel testlerde ger\u00e7ekle\u015febilir. Mazur (2007), \u00f6zellikle \u201cBu sorulara nas\u0131l cevap vermeliyim? diye soran bir \u00f6\u011frenci oldu\u011funu bildirmi\u015ftir. Bize \u00f6\u011frettiklerinize g\u00f6re, ya da bu \u015feylerle ilgili olarak benim d\u00fc\u015f\u00fcnd\u00fcklerim gibi mi?\u201d sorusunu soran bir \u00f6\u011freneni bildirmi\u015ftir. Son olarak, kas\u0131tl\u0131 h\u0131zl\u0131 tahmin etme davran\u0131\u015f\u0131 bir cevap yanl\u0131l\u0131\u011f\u0131 bi\u00e7imi olarak d\u00fc\u015f\u00fcn\u00fclebilir (Wise ve Kong, 2005). T\u00fcm bu cevap yanl\u0131l\u0131\u011f\u0131n\u0131n kaynaklar\u0131 \u00f6l\u00e7ek puanlar\u0131n\u0131n ele\u015ftirel olmayan yorumlar\u0131na meydan okudu\u011fu bilinmelidir.<\/p>\n\n

\u00d6l\u00e7me Modelleri<\/h3>\n

Bu s\u00fcrecin en zorlu k\u0131sm\u0131 \u00f6l\u00e7me modellerinin teknik detaylar\u0131ndad\u0131r. \u00d6l\u00e7me modeli, \u00f6rt\u00fck bir de\u011fi\u015fken veya de\u011fi\u015fken k\u00fcmesi ile g\u00f6zlemlenebilir bir de\u011fi\u015fken veya de\u011fi\u015fken k\u00fcmesi aras\u0131ndaki resm\u00ee bir matematiksel ili\u015fkidir. Tamamen istatistiksel bir \u00f6l\u00e7me modeli \u00f6rt\u00fck de\u011fi\u015fken(ler) i\u00e7in bir da\u011f\u0131l\u0131m, g\u00f6zlenen de\u011fi\u015fken(ler) i\u00e7in bir da\u011f\u0131l\u0131m ve aralar\u0131ndaki fonksiyonel bir ili\u015fkiyi belirtebilir. \u00d6rt\u00fck de\u011fi\u015fkenler \u00e7o\u011fu zaman hatalara tabi olan g\u00f6zlemleri nedensel olarak<\/i><\/span> a\u00e7\u0131klayanlar \u015feklinde anla\u015f\u0131lmaktad\u0131r. Rasgele de\u011fi\u015fkenlerin varyanslar\u0131 ve kovaryanslar\u0131, modelde a\u00e7\u0131k\u00e7a veya \u00f6rt\u00fck olarak a\u00e7\u0131klanm\u0131\u015ft\u0131r. Modeller, \u00f6rne\u011fin yap\u0131 ile \u00f6l\u00e7\u00fc aras\u0131ndaki ili\u015fkinin monotonluk (veya daha kat\u0131, do\u011frusall\u0131k) varsay\u0131m\u0131 ya da tekil \u00f6gelerin hata terimleri aras\u0131nda s\u0131f\u0131r kovaryans varsay\u0131m\u0131 yapar. Bir modelin varsay\u0131mlar\u0131 ihlal edilirse, model kullan\u0131larak yap\u0131lan \u00e7\u0131kar\u0131mlar yanl\u0131\u015f olabilir (Lord ve Novick, 1968).<\/p>\n

Kategorik ve s\u00fcrekli de\u011fi\u015fkenler farkl\u0131 istatistiksel y\u00f6ntemler i\u00e7erdi\u011finden, \u00f6l\u00e7me modelleri t\u00fcrleri bazen Tablo 3.1'de g\u00f6sterildi\u011fi gibi \u00f6rt\u00fck ve g\u00f6zlenen de\u011fi\u015fkenlerin t\u00fcr\u00fcne g\u00f6re aileler olarak tasnif edilir. Bu tasnif ayr\u0131nt\u0131l\u0131 de\u011fildir, \u00e7\u00fcnk\u00fc hibrit modellerin yan\u0131 s\u0131ra bu model ailelerin \u00f6zel durumlar haline geldi\u011fi genelle\u015ftirilmi\u015f \u00e7at\u0131lar (Skrondal ve Rabe-Hesketh, 2004) da vard\u0131r. B\u00fcy\u00fcme modelleri, \u00f6l\u00e7me modellerinin tekrarlanan \u00f6l\u00e7\u00fcmlere geni\u015fletilmesidir ve hem s\u00fcrekli hem de kategorik \u00f6rt\u00fck de\u011fi\u015fkenlere uygulanabilir (\u00f6r. Meredith ve Tisak, 1990; Rabiner, 1989; Raudenbush ve Bryk, 2002).<\/p>\n

<\/a><\/a><\/a><\/a><\/a> Tablo 3.1. Gizli De\u011fi\u015fken Modellerin Aileleri<\/span><\/i><\/span><\/p>\n\n<\/colgroup>\n\n\n\n\n
\n

\u00d6rt\u00fck\/ G\u00f6zlenen<\/b><\/p>\n<\/td>\n

\n

G\u00f6zlenen s\u00fcrekli<\/b><\/p>\n<\/td>\n

\n

G\u00f6zlenen kategorik<\/b><\/p>\n<\/td>\n<\/tr>\n

\u00d6rt\u00fck s\u00fcrekli<\/span><\/td>\nFakt\u00f6r modelleri (Bollen, 1989; Mulaik, 2009)<\/span><\/td>\nMadde tepki modelleri (Lord ve Novick, 1968; Baker ve Kim, 2004)<\/span><\/td>\n<\/tr>\n
\u00d6rt\u00fck kategorik<\/span><\/td>\n\u00d6rt\u00fck kar\u0131\u015f\u0131m modelleri (McLachlan ve Peel, 2004)<\/span><\/td>\n\u00d6rt\u00fck s\u0131n\u0131f modelleri (Goodman, 2002)<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

\u00d6\u011eRENME ANAL\u0130T\u0130\u011e\u0130NDE \u00d6L\u00c7ME MODELLER\u0130N\u0130N \u00d6ZEL KULLANIMI<\/h2>\n

Daha \u00f6nce, psikolojik ve e\u011fitsel \u00f6l\u00e7menin, s\u0131n\u0131fland\u0131rma, tan\u0131lama, s\u0131ralama, yerle\u015ftirme ve bireylerin belgelendirilmesinin yan\u0131 s\u0131ra gruplara dair uygun \u00e7\u0131kar\u0131mlar d\u00e2hil olmak \u00fczere \u00e7e\u015fitli ama\u00e7lar i\u00e7in kullan\u0131ld\u0131\u011f\u0131n\u0131 belirtmi\u015ftik. \u00d6\u011frenme analiti\u011fi ve e\u011fitsel veri madencili\u011fi alan\u0131ndaki \u00e7al\u0131\u015fmalar, dijital \u00f6\u011frenme ortamlar\u0131ndaki psikolojik \u00f6l\u00e7ekler, davran\u0131\u015flar ve performans aras\u0131ndaki karma\u015f\u0131k ili\u015fki a\u011f\u0131n\u0131 da ara\u015ft\u0131rmaktad\u0131r (Tempelaar, Rienties ve Giesbers, 2015). Bu teman\u0131n amac\u0131, modeller ve bunlar\u0131n analitik ve e\u011fitsel veri madencili\u011fini \u00f6\u011frenmedeki kullan\u0131mlar\u0131 hakk\u0131nda biraz daha derinlik sa\u011flamakt\u0131r. T\u00fcm konular e\u015fit \u00f6l\u00e7\u00fcde ele al\u0131nmaz, bu da hem alan k\u0131s\u0131tlamalar\u0131n\u0131 hem de se\u00e7im yanl\u0131l\u0131\u011f\u0131n\u0131 yans\u0131t\u0131r.<\/p>\n\n

Fakt\u00f6r analizi<\/h3>\n

Fakt\u00f6r analizi (Mulaik, 2009), g\u00f6zlenen de\u011fi\u015fkenler aras\u0131ndaki korelasyonlar\u0131, fakt\u00f6r olarak bilinen bir dizi \u00f6rt\u00fck de\u011fi\u015fkenle do\u011frusal bir ili\u015fki yoluyla modellemektedir. Orijinal tek fakt\u00f6rl\u00fc model Spearman'\u0131n (1904) genel zek\u00e2 g<\/i><\/span> modelidir, ilgisiz konu testlerindeki puanlar aras\u0131ndaki ili\u015fkileri a\u00e7\u0131klamak i\u00e7in kullan\u0131l\u0131r. Klasik test teorisi (Lord ve Novick, 1968) olarak da bilinen ger\u00e7ek puan teorisi, t\u00fcm madde fakt\u00f6r\u00fc y\u00fcklerinin ayn\u0131 oldu\u011fu tek fakt\u00f6r modelinin \u00f6zel bir hali olarak elde edilebilir. Thurstone (1947), \u00e7oklu (yedi) fakt\u00f6r zek\u00e2 modelini geli\u015ftirdi.<\/p>\n

Fakt\u00f6r analizi, genellikle iki te\u015febb\u00fcse b\u00f6l\u00fcnm\u00fc\u015ft\u00fcr. A\u00e7\u0131mlay\u0131c\u0131 fakt\u00f6r analizi (AFA), g\u00fc\u00e7l\u00fc teorik varsay\u0131mlar olmadan verilerdeki \u00f6rt\u00fck fakt\u00f6rlerin say\u0131s\u0131n\u0131 belirlemek i\u00e7in kullan\u0131l\u0131r ve genellikle \u00f6l\u00e7ek geli\u015ftirmenin bir par\u00e7as\u0131d\u0131r. Bununla birlikte, AFA, e\u011fer zay\u0131f yap\u0131l\u0131rsa sorunlu sonu\u00e7lara yol a\u00e7abilecek birka\u00e7 \u00f6nemli metodolojik karar gerektirir (Fabrigar, Wegener, MacCallum ve Strahan, 1999). Fabrigar vd. (1999), AFA'n\u0131n, ger\u00e7ek fakt\u00f6r yap\u0131s\u0131 hakk\u0131nda hatal\u0131 \u00e7\u0131kar\u0131mlara yol a\u00e7abilecek, bir boyutsall\u0131k azaltma tekni\u011fi olan temel bile\u015fenler analizi (TBA) ile kar\u0131\u015ft\u0131r\u0131lmamas\u0131 konusunda uyar\u0131larda bulundu. Do\u011frulay\u0131c\u0131 fakt\u00f6r analizi (DFA), beklenen ve g\u00f6zlemlenen korelasyonlar aras\u0131ndaki kal\u0131nt\u0131lar\u0131 inceleyerek teorik olarak \u00f6nerilen bir fakt\u00f6r modelini test etmek i\u00e7in yap\u0131lm\u0131\u015f tamamlay\u0131c\u0131 teknikler setidir. B\u00f6ylece, bir modeli reddetmek i\u00e7in DFA kullan\u0131labilir. DFA, yol \u00e7\u00f6z\u00fcmlemesi ve gizli b\u00fcy\u00fcme modelleri ile birlikte, yap\u0131sal e\u015fitlik modellemesi ile g\u00fcvence alt\u0131na al\u0131nm\u0131\u015ft\u0131r (SEM; Bollen, 1989; Kline, 2010). Do\u011frulay\u0131c\u0131 fakt\u00f6r analizi, durum ikincisini ger\u00e7ekle\u015ftirmek i\u00e7in yap\u0131lm\u0131\u015f olmas\u0131na ra\u011fmen AFA'n\u0131n farkl\u0131 pop\u00fclasyon \u00f6rnekleriyle birden fazla kez \u00e7al\u0131\u015ft\u0131r\u0131lmas\u0131 ile ayn\u0131 \u015fey de\u011fildir. (DeVellis, 2003).<\/p>\n

Baz\u0131 \u00f6\u011frenme analiti\u011fi ara\u015ft\u0131rmalar\u0131, \u00f6l\u00e7ek geli\u015ftirme ve bunun \u00f6\u011frenme y\u00f6netimi sistemlerinden toplanan verilerle birle\u015ftirilmesi ile do\u011frudan ilgilidir (\u00f6r. Buckingham Shum ve Deakin Crick, 2012; Milligan ve Griffin, 2016). Di\u011fer \u00e7al\u0131\u015fmalar, ba\u015far\u0131 \u00f6l\u00e7ekleri (Pekrun vd., 2011) ile y\u00fcz y\u00fcze ve \u00e7evrimi\u00e7i e\u011fitim (Tempelaar, Niculescu, Rienties, Giesbers ve Gijselaers, 2012) aras\u0131ndaki ili\u015fki gibi mevcut \u00f6l\u00e7ekler ve sonu\u00e7 \u00f6l\u00e7\u00fcmleri veya motivasyon \u00f6nlemleri ile kitlesel a\u00e7\u0131k \u00e7evrimi\u00e7i dersin tamamlanmas\u0131 aras\u0131ndaki ili\u015fkilere odaklanmaktad\u0131r (Wang ve Baker, 2015). Bir arac\u0131 veya \u00f6zellikle bir arac\u0131n bir b\u00f6l\u00fcm\u00fcn\u00fc yeni ama\u00e7lar i\u00e7in uyarlarken, uygulay\u0131c\u0131lar bu yeni kullan\u0131mlar\u0131n yeni do\u011frulama arg\u00fcmanlar\u0131na de\u011fip de\u011fmedi\u011fi konusunda dikkatli olmal\u0131d\u0131r.<\/p>\n\n

\u00d6rt\u00fck S\u0131n\u0131f ve \u00d6rt\u00fck Kar\u0131\u015f\u0131m Modelleri<\/h3>\n

Dedic, Rosenfeld ve Lasry (2010), \u00f6\u011frencilerin bir fizik kavram\u0131 testindeki yanl\u0131\u015f cevaplar\u0131na dayanarak fizik kavram yan\u0131lg\u0131lar\u0131n\u0131n da\u011f\u0131l\u0131m\u0131n\u0131 anlamak i\u00e7in \u00f6rt\u00fck s\u0131n\u0131f analizini kullanm\u0131\u015ft\u0131r. Veriler, bir fizik kursu \u00f6ncesi ve sonras\u0131ndaki uygulamalardan gelmi\u015ftir. (\u00f6n ve son test). Yazarlar, kesikli bask\u0131nl\u0131k yan\u0131lma s\u0131n\u0131flar\u0131 arac\u0131l\u0131\u011f\u0131yla, Aristotelist' ten Newtoncu d\u00fc\u015f\u00fcnceye kadar bariz bir ilerleme tespit etmi\u015ftir. Belgelerin konu modellemesi i\u00e7in yayg\u0131n olarak kullan\u0131lan bir y\u00f6ntem olan gizli Dirichlet tahsisi (GDT; Blei, Ng ve Jordan, 2003; ayr\u0131ca, bu ciltte birka\u00e7 b\u00f6l\u00fcme bak\u0131n\u0131z) \u00f6rt\u00fck bir kar\u0131\u015f\u0131m modelidir. Kar\u0131\u015f\u0131k \u00fcyelik modelleri (Erosheva, Fienberg ve Lafferty, 2004), bir bireyin birden fazla s\u0131n\u0131fa \u201cbelirsiz\u201d veya a\u011f\u0131rl\u0131kl\u0131 olarak atanmas\u0131na izin vererek \u00f6rt\u00fck kar\u0131\u015f\u0131mlar\u0131 daha da genellemektedir. Gauss kar\u0131\u015f\u0131m modeli, KA\u00c7D \u00f6\u011frenenlerin performans g\u00fczerg\u00e2hlar\u0131na uygulanan model tabanl\u0131 k\u00fcmeleme analizi (Fraley ve Raftery, 1998) i\u00e7in temel olu\u015fturmaktad\u0131r (Bergner, Kerr ve Pritchard, 2015). Bununla birlikte, k\u00fcmeleme algoritmalar\u0131n\u0131n hepsinin \u00f6rt\u00fck kar\u0131\u015f\u0131m modeli olmad\u0131\u011f\u0131 unutulmamal\u0131d\u0131r.<\/p>\n\n

Madde Tepki Kuram\u0131 (MTK)<\/h3>\n

Madde tepki kuram\u0131, klasik test teorisinde oldu\u011fu gibi, toplam test puanlar\u0131ndan ziyade bireysel ki\u015filik-madde etkile\u015fimlerini modelleyerek, test teorisinin tarihsel geli\u015fiminde kendisine ayr\u0131 bir yer edinmi\u015ftir. Kavramsal olarak, MTK'nin amac\u0131 \u201cmaddeleri, madde parametrelerine g\u00f6re ve s\u0131nava girenleri, inceleme parametrelerine g\u00f6re; benzer s\u0131nava girenler daha \u00f6nce benzer maddeleri hi\u00e7 cevaplamad\u0131ysa bile herhangi bir s\u0131nava girenin herhangi bir maddeye cevab\u0131n\u0131 olas\u0131l\u0131\u011fa dayal\u0131 olarak tahmin edebilecek \u015fekilde tan\u0131mlamakt\u0131r\" (Lord, 1980, s. 11). \u0130ki bile\u015fenli bir madde i\u00e7in (\u00f6r. do\u011fru \/ yanl\u0131\u015f, ayn\u0131 fikirde \/ kat\u0131lm\u0131yorum, vb.) bir \u00f6rnek madde karakteristik e\u011frisi (MKE) veya e\u015fde\u011ferde madde tepki fonksiyonu (MTF), \u015eekil 3.1'de g\u00f6sterilmektedir.<\/p>\n

\"\"<\/p>\n<\/a> \u015eekil 3.1. Bir \u00f6rnek madde karakteristik e\u011frisi (MKE). Noktal\u0131 \u00e7izgiler P = 0.5 kesi\u015fimini g\u00f6sterir.<\/i><\/span>\n

\u015eekil 3.1'in belirgin \u00f6zellikleri a\u015fa\u011f\u0131daki gibidir:<\/p>\n\n

    \n \t
  1. \n

    \u00d6zellik (\u00f6r. yetenek) s\u00fcrekli rastgele bir de\u011fi\u015fken olarak \u00f6l\u00e7\u00fcl\u00fcr ve yatay eksende <\/span>\u04e8<\/i><\/span> ile temsil edilir. De\u011fi\u015fken, ilgilenilen pop\u00fclasyonda ortalama s\u0131f\u0131ra ve 1 varyans\u0131na sahip olacak \u015fekilde standardize edilmi\u015ftir. Daha y\u00fcksek bir \u04e8<\/i><\/span> de\u011ferine kar\u015f\u0131l\u0131k gelen \u00f6zellikten daha fazlas\u0131n\u0131n, pozitif (veya do\u011fru) bir cevab\u0131n P<\/i><\/span> olas\u0131l\u0131\u011f\u0131n\u0131 artt\u0131rmas\u0131 beklenir. G\u00f6zlenen bir monotonluk ihlali<\/i><\/span>, temel \u015fah\u0131s-madde ili\u015fkisinin yanl\u0131\u015f oldu\u011fu ve teste maddenin d\u00e2hil edilmesinin k\u00f6t\u00fc bir uyum olaca\u011f\u0131 ve g\u00fcvenilmez \u00e7\u0131kar\u0131mlara yol a\u00e7aca\u011f\u0131 anlam\u0131na gelir.<\/p>\n<\/li>\n \t

  2. \n

    Bu e\u011frileri yorumlaman\u0131n iki yolu Holland (1990) taraf\u0131ndan tan\u0131mlanm\u0131\u015ft\u0131r. Stokastik denek yorumunda, ki\u015fi bu e\u011frinin, performans\u0131 \u00f6ng\u00f6r\u00fclemeyen bir bireye uyguland\u0131\u011f\u0131n\u0131 d\u00fc\u015f\u00fcn\u00fcr. Holland'\u0131 anlamsal olarak al\u0131nt\u0131layacak olursak, stokastik denek a\u00e7\u0131klamas\u0131 sezgiseldir ancak tamamen tatmin edici de\u011fildir; \u00f6znenin stokastik do\u011fas\u0131 i\u00e7in mekanik bir a\u00e7\u0131klamam\u0131z yoktur. \u00d6te yandan, rastgele \u00f6rneklem yorumunda, bu e\u011fri, s\u0131nava girenlerin \u00f6rneklem pop\u00fclasyonuna uyguland\u0131\u011f\u0131nda anlaml\u0131d\u0131r. \u00d6rne\u011fin, belirli bir yetenek aral\u0131\u011f\u0131ndaki s\u0131nava girenler aras\u0131nda, baz\u0131 oranlar do\u011fru cevap verecektir. \u015eekildeki noktalar ve hata \u00e7ubuklar\u0131 bu g\u00f6zlemi yans\u0131tmaktad\u0131r.2<\/sup><\/a><\/span><\/p>\n<\/li>\n \t

  3. \n

    P<\/i><\/span> = 0.5 olan \u04e8<\/i><\/span> de\u011feri, bir bili\u015fsel yetenek test maddesi i\u00e7in zorluk olarak adland\u0131r\u0131lan bir referans kesi\u015fimidir. Zorlu\u011fun, yetenek ile ipso facto<\/i><\/span> (kendili\u011finden) ayn\u0131 \u00f6l\u00e7ekte oldu\u011funu ve b\u00f6ylece bir ki\u015finin yetene\u011fi ile bir maddenin zorlu\u011fu aras\u0131ndaki fark hakk\u0131nda konu\u015fman\u0131n mant\u0131kl\u0131 olabilece\u011fi unutulmamal\u0131d\u0131r.<\/p>\n<\/li>\n \t

  4. \n

    Olas\u0131l\u0131k ba\u011flant\u0131s\u0131n\u0131n \u015fekli, bireyin <\/span>\u04e8<\/i><\/span>i<\/sub> \u00f6zelli\u011fi i<\/i><\/span> ve j maddesi i\u00e7in bir dizi B<\/i><\/span><\/span>j<\/sub>, madde parametreleri y\u00f6n\u00fcnden genellikle parametriktir,\nP<\/i><\/span>ij<\/sub> = <\/span>P<\/i><\/span>(<\/span>X<\/i><\/span>ij<\/sub> = 1|<\/span>\u03b8<\/i><\/span><\/span>i<\/sub>, <\/span>\u03b2<\/i><\/span><\/span>j<\/sub>) = <\/span>f<\/i><\/span>(<\/span>\u03b8<\/i><\/span><\/span>i<\/sub>, <\/span>\u03b2<\/i><\/span><\/span>j<\/sub>), (1)\n<\/span>Rasch modelinde (bir tekil zorluk parametresi) veya iki parametreli lojistik (2PL) modelinde oldu\u011fu gibi. 2PL modeli, \u015eekil 3.2'de g\u00f6sterilmektedir; verilere uygunluk g\u00f6zle g\u00f6r\u00fcl\u00fcr derecede iyi ve \u00f6rt\u00fc\u015fme d\u00fczeyi testi G<\/i><\/span>2<\/span><\/sup> ayn\u0131 miktardad\u0131r. Parametrik olmayan MTK y\u00f6ntemlerinin var oldu\u011fu belirtilmelidir (Sijtsma, 1998).<\/p>\n<\/li>\n<\/ol>\n

    Bir ki\u015fi bir \u00f6l\u00e7me arac\u0131nda birka\u00e7 maddeye cevap verdi\u011finde, buradaki fikir, \u00f6zelli\u011fin sonsal tahminlerini yapmak i\u00e7in cevap bilgilerini birle\u015ftirmektir. Bir cevap vekt\u00f6r\u00fcn\u00fcn, bireysel madde seviyesindeki olas\u0131l\u0131klar\u0131n bir \u00fcr\u00fcn\u00fcne \u00e7arpan olabilirlik durumu i\u00e7in, cevaplar, niteli\u011fe ba\u011fl\u0131 olarak aksi takdirde ba\u011f\u0131ms\u0131z olmal\u0131d\u0131r. Bu ko\u015fullu ba\u011f\u0131ms\u0131zl\u0131k varsay\u0131m\u0131<\/i><\/span> maddeler aras\u0131 ba\u011f\u0131ml\u0131l\u0131\u011f\u0131 (\u00f6r. Rijmen, 2010) a\u00e7\u0131klayan ek fakt\u00f6rlerin takdimini gerektirebilir.<\/p>\n

    MTK'nin y\u00fcksek riskli test uygulamalar\u0131 d\u0131\u015f\u0131nda e\u011fitimde bir miktar \u00e7ekim g\u00fcc\u00fcne sahip oldu\u011funa dair kan\u0131tlar fizik e\u011fitimi ara\u015ft\u0131rma uygulamalar\u0131nda kuvvet kavram\u0131 envanterine (KKE; Hestenes vd., 1992) ve temel mekanik bilgi testine bak\u0131larak bulunabilir (MBT; Hestenes ve Wells, 1992). Bu ara\u00e7lar yirmi be\u015f y\u0131ld\u0131r kullan\u0131lmakta iken, madde tepki modeli analizleri daha yak\u0131n zamanda ortaya \u00e7\u0131kmaya ba\u015flam\u0131\u015ft\u0131r (Morris vd., 2006; Wang ve Bao, 2010). KKE i\u00e7in model-veri uygunlu\u011fu genel olarak kabul edilebilir durumdayd\u0131. Bununla birlikte Cardamone vd. (2011), MBT'de, madde tepki fonksiyonlar\u0131n\u0131 inceleyerek, k\u00f6t\u00fc \u00e7al\u0131\u015fan iki maddeyi ke\u015ffetmi\u015ftir. \u015eekil 3.2'de g\u00f6sterilmi\u015ftir.\"\"<\/p>\n<\/a> \u015eekil 3.2. Mekanik bilgi testinden (MBT) k\u00f6t\u00fc uyum sa\u011flayan bir madde.<\/span><\/i><\/span>\n

    D\u00fc\u015f\u00fck yetenekli \u00f6\u011frencilerin ortalama yetenekli \u00f6\u011frencilere g\u00f6re bir maddeye do\u011fru cevap vermeleri daha muhtemel ise, burada \u015f\u00fcpheli bir durum vard\u0131r. Daha detayl\u0131 bir inceleme ile bu test maddesindeki mu\u011flak ifadelerin \u00f6\u011frencilerin alg\u0131lar\u0131n\u0131 hatal\u0131 y\u00f6nlendirdi\u011fi ve yanl\u0131\u015fl\u0131kla do\u011fru cevab\u0131 bulmalar\u0131n\u0131 sa\u011flad\u0131\u011f\u0131 tespit edilmi\u015ftir. Bu durumda, iki yanl\u0131\u015f bir do\u011fru yapm\u0131\u015f oldu.<\/p>\n

    Birden fazla boyut tan\u0131mlayan KKE'nin a\u00e7\u0131mlay\u0131c\u0131 fakt\u00f6r analizlerini takiben (Ding ve Beichner, 2009; Scott, Schumayer ve Gray, 2012), MBT'ye \u00e7ok boyutlu MTK'nin bir varyasyonu uyguland\u0131 (Bergner, Rayyan, Seaton ve Pritchard, 2013). Madde tepki kuram\u0131 modelleri de \u00e7evrimi\u00e7i \u00f6devlerde s\u0131k\u00e7a g\u00f6r\u00fclen bir kolayl\u0131k olan, birden fazla cevap verme giri\u015fimlerinin (do\u011fru olana dek cevaplama) ard\u0131ndaki kendinden s\u0131ral\u0131 s\u00fcrece geni\u015fletildi (Attali, 2011; Bergner, Colvin ve Pritchard, 2015; 2014).<\/p>\n\n

    B\u00fcy\u00fcme Modelleri<\/h3>\n

    B\u00fcy\u00fcme modelleri, \u00f6rt\u00fck bir \u00f6zelli\u011fin \u00f6l\u00e7\u00fcmler aras\u0131nda sistematik olarak de\u011fi\u015fti\u011fi herhangi bir anda uygulan\u0131r. De\u011fi\u015fen tutumlara uygulanabilirler (\u00f6r. George, 2000), fakat biz burada bili\u015fsel yetenek alanlar\u0131 uygulamas\u0131na odaklan\u0131yoruz. \u00d6\u011fretim program\u0131 d\u00fczenleme \u00f6\u011freticilerinden3<\/sup><\/a> ay\u0131rt edilen ak\u0131ll\u0131 problem \u00e7\u00f6zme \u00f6\u011freticileri i\u00e7in \u00f6\u011frenci modellerine dair e\u011fitsel veri madencili\u011finde kapsaml\u0131 bir literat\u00fcr bulunmaktad\u0131r (Desmarais ve Baker, 2011).<\/p>\n

    Matematik i\u00e7in bili\u015fsel \u00f6\u011freticilerde (Anderson, Corbett, Koedinger ve Pelletier, 1995), uygulama madde dizilimleri, bili\u015fsel bir modele g\u00f6re ince taneli bilgi bile\u015fenlerinin tam \u00f6\u011frenilmesini desteklemek i\u00e7in tasarlanm\u0131\u015ft\u0131r. Bu sistemlerde verilerde ustal\u0131\u011fa do\u011fru b\u00fcy\u00fcmeyi modelleme amac\u0131 olan iki yakla\u015f\u0131mdan biri Bayesci bilgi takibi (BBT; Corbett ve Anderson, 1995) ve toplamsal fakt\u00f6r modelleridir; Cen, Koedinger ve Junker, 2008; Draney, Pirolli ve Wilson, 1995). \u00d6\u011frenme e\u011frileri analizi (Kaser, Koedinger ve Gross, 2014; Martin'in, Mitrovic, Mathan ve Koedinger, 2010), verilerle \u00f6\u011freticinin temelindeki bili\u015fsel model ve veri aras\u0131ndaki uyu\u015fmazl\u0131klar\u0131 kontrol etmek i\u00e7in de kullan\u0131lm\u0131\u015ft\u0131r.<\/p>\n

    \u201cUygulama Yasas\u0131\u201d na g\u00f6re (Newell ve Rosenbloom, 1981), B ve a<\/i><\/span>' n\u0131n deneysel olarak belirlendi\u011fi \"T = Bn<\/sub>-a<\/sup>, \"g\u00fc\u00e7 yasas\u0131na g\u00f6re\", n<\/i><\/span> uygulama f\u0131rsat\u0131n\u0131n bir fonksiyonu olarak toplam hata oran\u0131 T azalmal\u0131d\u0131r. Veri ve model aras\u0131ndaki uyum, \u00f6rne\u011fin r kare \u00f6l\u00e7\u00fcmleri kullanmak, bilgi e\u015flemedeki geli\u015fmeleri motive edebilir. Bu hatal\u0131 bir maddenin tespit edildi\u011fi \u015eekil 3.2'deki madde analizine benze\u015fik olarak g\u00f6r\u00fclebilir. Bununla birlikte, bu durumda, bir bilgi bile\u015fenine bir madde diziliminin atanmas\u0131 hatal\u0131 olarak g\u00f6r\u00fclmektedir.<\/p>\n

    BBT'de \u00f6rt\u00fck de\u011fi\u015fken, bir i\u015flemsel bilgi bile\u015feninin ustal\u0131\u011f\u0131d\u0131r ve ikili de\u011fere sahiptir, M<\/i><\/span> \u2208<\/span> {0, 1}. Herhangi bir f\u0131rsatta ustal\u0131k ve do\u011fruluk X<\/i><\/span> \u2208<\/span> {0, 1} aras\u0131ndaki ba\u011flant\u0131s\u0131 ancak Eq ile benze\u015fik olarak bir 2x2 ko\u015fullu olas\u0131l\u0131k tablosudur. (1) tahmin (g<\/i><\/span>) ve kayma (s<\/i><\/span>) parametreleri y\u00f6n\u00fcnden \u015f\u00f6yle yaz\u0131labilir,<\/p>\n

    P(X = 1|M) = (1 - s)M<\/sup> g(1-M)<\/sup> (2)<\/p>\n

    \u00d6nemli bi\u00e7imde, giri\u015fimler ba\u011f\u0131ms\u0131z olarak g\u00f6r\u00fclmemektedir. Aksine, BBT'deki kilit fikir, \u00f6\u011frencilerin kurallara g\u00f6re her bir uygulama durumunda \u00f6n bir ustal\u0131k olas\u0131l\u0131\u011f\u0131 ile ba\u015flamalar\u0131 ve ustal\u0131\u011fa do\u011fru hareket ediyor (\u00f6\u011frenirler) olmalar\u0131d\u0131r,\nP<\/i><\/span>(M<\/i><\/span>n<\/span><\/sub>) = P(M<\/i><\/span>n-1<\/span><\/sub>) + t(1 - P(M<\/i><\/span>n-1<\/span><\/sub>))<\/i><\/span> (3)\nBurada t<\/i><\/span> bir b\u00fcy\u00fcme parametresidir. Son zamanlarda, van de Sande (2013) BBT'nin uygulama giri\u015fimleri ve hata oranlar\u0131 aras\u0131nda bir g\u00fc\u00e7 yasas\u0131 ili\u015fkisinden ziyade bir \u00fcstellik belirtti\u011fini g\u00f6stermi\u015ftir. Bu BBT'yi uygulaman\u0131n g\u00fc\u00e7 yasas\u0131n\u0131 sa\u011flayan veriler i\u00e7in yanl\u0131\u015f tan\u0131mlanm\u0131\u015f bir model yapar. Aksine, toplamsal fakt\u00f6r modeli, uygulama paradigmas\u0131n\u0131n g\u00fc\u00e7 yasas\u0131na uyacak \u015fekilde tasarlanm\u0131\u015ft\u0131r. Kaser vd. (2014) BBT'nin kestirimsel keskinli\u011finin TFM'den ay\u0131rt edilemez oldu\u011funu g\u00f6sterdi. \u0130kincisinin uyumu ile ilgili olarak, toplam kal\u0131nt\u0131 analizlerinde sistematik yanl\u0131l\u0131\u011fa dikkat \u00e7ektiler.<\/p>\n

    TFM, MTK'nin bir uzant\u0131s\u0131 olarak adland\u0131r\u0131lm\u0131\u015ft\u0131r (Koedinger, McLaughlin ve Stamper, 2012) ve asl\u0131nda do\u011frusal lojistik test modeliyle olan (DLTM, Fischer, 1973) ili\u015fki bu modelin \u00f6nc\u00fcl\u00fcnde a\u00e7\u0131kt\u0131 (Draney vd., 1995). Bununla birlikte, mevcut \u015fekline ge\u00e7erken, model kritik bir \u015fekilde de\u011fi\u015ftirildi. DLTM, bir maddenin zorlu\u011funun, maddenin potansiyel \u00f6zellikleri \u00fczerinde bir toplam olarak ayr\u0131\u015ft\u0131r\u0131ld\u0131\u011f\u0131 Rasch tipi bir MTK modelidir. Rasch modelini \u015f\u00f6yle yazabiliriz,<\/p>\n

    logit(P<\/span>ij<\/span><\/sub>) = ln(P<\/span>ij<\/span><\/sub>\/(1-P<\/span>ij<\/span><\/sub>)) = \u03b8<\/span>i<\/span><\/sub> - \u03b2<\/span>j<\/span><\/sub>,<\/span> <\/i>(4)\n\u03b2<\/span> \u00f6gesinin zorlu\u011fu <\/span><\/sub>j ayr\u0131ca ayr\u0131\u015ft\u0131r\u0131l\u0131r,\n\u03b2<\/span>j<\/span><\/sub>=c<\/span>j <\/span><\/sub>+ \u03a3<\/span>k<\/span><\/sub> w<\/span>jk<\/span><\/sub>\u03b1<\/span>k<\/span><\/sub>, <\/span>(5)\na<\/i><\/span>k<\/span><\/sub>, \u201ctemel\u201d i\u015flemlerin (Fischer\u2019in terimi) zorluklar\u0131d\u0131r ve w<\/i><\/span>ik<\/span><\/sub> g\u00f6stergeleri, bu i\u015flemlerin j maddesinde gerekip gerekmedi\u011fine ba\u011fl\u0131 olarak 0 veya 1\u2019dir. T\u00fcm \u00f6geler ayn\u0131 i\u015flemleri kullan\u0131yorsa model basit bir kayma ile Rasch modeline a\u00e7\u0131k\u00e7a indirgenir,\n\u03b2<\/span>j<\/span><\/sub>=c<\/span>j <\/span><\/sub>+\u03b1. <\/span>(6)\nDraney vd. (1995)'n\u0131n modeli madde seviyesinde bir zorluk parametresi i\u00e7ermekte olsa da TFM'de sadece bile\u015fen becerilerinin zorluklar\u0131 korunmaktad\u0131r. Ek olarak, bir uygulama terimi getirilir,4<\/sup><\/a><\/p>\n

    \u03b2<\/span>j<\/span><\/sub>AFM<\/span><\/sup> = \u03a3<\/span>k<\/span><\/sub>w<\/span>jk<\/span><\/sub>\u03b1<\/span>k<\/span><\/sub> - \u03a3<\/span>k<\/span><\/sub>wj<\/span>k<\/span><\/sub>\u03b3<\/span>k<\/span><\/sub>T<\/span>ik<\/span><\/sub> , <\/span>(7)\nburadaki y<\/span>k<\/span><\/sub>, bir b\u00fcy\u00fcme parametresidir ve Tik<\/sub>, \u00f6\u011frenen i<\/i><\/span>' nin beceri k <\/i><\/span>\u00fczerindeki \u00f6nceki uygulama giri\u015fimlerinin bir say\u0131s\u0131d\u0131r. Bir uygulama problemi diziliminin t\u00fcm\u00fc, \u00f6\u011fretici uygulamalar\u0131 i\u00e7in ortak olan ayn\u0131 becerileri i\u00e7eriyorsa, o zaman her bir dizi i\u00e7in, bu parametre,\n\u03b2<\/i><\/span>j<\/i><\/span><\/sub>AFM<\/i><\/span><\/sup> = \u03b1\u2013\u03b3T<\/i><\/span>i <\/i><\/span><\/sub>.<\/i><\/span> (8)\n<\/span>\u00d6nemli olarak, bu asl\u0131nda, sa\u011f taraftaki alt simgelerden de a\u00e7\u0131k\u00e7a anla\u015f\u0131ld\u0131\u011f\u0131 gibi, maddenin hi\u00e7bir \u00f6zelli\u011fine de\u011fil yaln\u0131zca \u00f6\u011frenmeye ba\u011fl\u0131d\u0131r. c<\/i><\/span>j<\/span><\/sub> parametresini Denklem(7) - (8) 'de b\u0131rakma sayesinde, TFM asl\u0131nda sabit bir etki b\u00fcy\u00fcme modeli haline gelmi\u015ftir.<\/p>\n

    Modelleme a\u00e7\u0131s\u0131ndan bak\u0131ld\u0131\u011f\u0131nda hem zorluk hem de b\u00fcy\u00fcme parametrelerinin saklanmas\u0131 tan\u0131mlanabilirlik i\u00e7in bir sorun olu\u015fturdu\u011fundan madde d\u00fczeyinde zorluk parametresinin kald\u0131r\u0131lmas\u0131 \u015fa\u015f\u0131rt\u0131c\u0131 de\u011fildir. Bir model, parametreleri yeterli veri g\u00f6z \u00f6n\u00fcne al\u0131nd\u0131\u011f\u0131nda a\u00e7\u0131k\u00e7a \u00f6\u011frenilebiliyorsa tan\u0131mlanabilir. Ancak sabit bir madde dizilim \u00fczerinde \u00e7al\u0131\u015fan \u00f6\u011frenciler i\u00e7in, \u00f6\u011frenme \/ b\u00fcy\u00fcme nedeniyle artan ba\u015far\u0131 oran\u0131, azalan madde zorlu\u011funa ba\u011flanabilir. B\u00fcy\u00fcme olmayan ko\u015fullar alt\u0131nda madde zorluklar\u0131 ayr\u0131 ayr\u0131 kalibre edilmedik\u00e7e, iki etki ay\u0131rt edilemez.<\/p>\n\n

    Bili\u015fsel Te\u015fhis Modelleri<\/h3>\n

    Bili\u015fsel g\u00f6rev analizi kullan\u0131larak yap\u0131lan kar\u0131\u015f\u0131k say\u0131daki \u00e7\u0131karma \u00e7al\u0131\u015fmas\u0131na dair \u00e7\u0131\u011f\u0131r a\u00e7\u0131c\u0131 bir \u00e7al\u0131\u015fma, Tatsuoka'y\u0131 (1983), bir e\u011fitim testinde Q-matris y\u00f6ntemini ve belirli alt becerilerin te\u015fhisi i\u00e7in bir model (\u00f6r. bir say\u0131n\u0131n tamam\u0131n\u0131 kesire d\u00f6n\u00fc\u015ft\u00fcrme) geli\u015ftirmesine yol a\u00e7t\u0131. Q-matrisi, alt becerileri gerektirecek maddelerin kesikli bir haritas\u0131d\u0131r ve de\u011ferlendirme modelinde geleneksel olarak belirtilir. Bili\u015fsel tan\u0131sal modeller o zamandan beri olduk\u00e7a yayg\u0131nla\u015ft\u0131r\u0131lm\u0131\u015ft\u0131r (Rupp ve Templin, 2008; von Davier, 2005) ve Q matrisini verilerden \u00f6\u011frenme \u00e7abalar\u0131, e\u011fitsel veri madencili\u011fi ara\u015ft\u0131rmalar\u0131nda ortaya \u00e7\u0131km\u0131\u015ft\u0131r (Barnes, 2005; Desmarais, 2012; Koedinger vd., 2012).<\/p>\n\n

    HATA KAYNAKLARI, TEKRAR G\u00d6ZDEN GE\u00c7\u0130R\u0130LMES\u0130<\/h2>\n

    Motivasyon, duygu ve bili\u015f \u00e7al\u0131\u015fmalar\u0131na d\u00e2hil olan baz\u0131 \u00f6l\u00e7me modellerini ara\u015ft\u0131rd\u0131ktan sonra, \u00f6nemli olan hata konusu tekrar g\u00f6zden ge\u00e7irmeye de\u011fer. Uygulay\u0131c\u0131lar, yanl\u0131\u015f parametreli modeller kullanarak, yanl\u0131\u015f modeller kullanarak veya modelleri yanl\u0131\u015f kullanarak ek hata kaynaklar\u0131n\u0131n ortaya \u00e7\u0131kabilece\u011fine dikkat etmelidir.<\/p>\n

    Bir modelin kullan\u0131m\u0131, tahmini hataya tabi olan parametrelere ba\u011fl\u0131 olabilir. Bu belirsizlikler kabul edilmelidir ancak model g\u00f6zlenen veriler i\u00e7in veri \u00fcreten bir model<\/i><\/span> olarak tutarl\u0131ysa, bunlar ille de ciddi belirsizlikler de\u011fildir. Yani, istatistiksel modeli veri \u00fcretmek i\u00e7in de kullan\u0131labilecek stokastik bir s\u00fcre\u00e7 olarak g\u00f6r\u00fcyoruz (ayr\u0131ca, \u00f6rnekleme veya benzetme) (Breiman, 2001). \u00d6rne\u011fin, ger\u00e7ek madeni paran\u0131n hilesiz olup olmad\u0131\u011f\u0131ndan emin olmasak bile, bir Bernoulli i\u015flemi kullanarak madeni para atma deneyi verilerini sim\u00fcle edebiliriz. Prensip olarak, modelimizdeki tura olas\u0131l\u0131klar\u0131 parametresi, ger\u00e7ek madeni paradan daha fazla veri ile geli\u015ftirilebilir. Bu modelin kendisinin ya \u00f6rt\u00fck de\u011fi\u015fkenler ya da ba\u011flant\u0131 i\u015flevleri a\u00e7\u0131s\u0131ndan, ger\u00e7ek \u00fcretici modelle tutars\u0131z oldu\u011fu durumdan farkl\u0131d\u0131r. \u0130kinci vaka modelin yanl\u0131\u015f tan\u0131mlanmas\u0131 olarak adland\u0131r\u0131l\u0131r (White, 1996). Uyumluluk testleri, modeli korumak veya reddetmek i\u00e7in g\u00f6zlenen veriler ile \u00fcretici model aras\u0131ndaki tutarl\u0131l\u0131\u011f\u0131 de\u011ferlendirir (White, 1996; Haberman, 2009; Ames ve Penfield, 2015).<\/p>\n\n

    A\u00c7IKLAMA VE YORDAMA<\/h3>\n

    Kestirimci modelleme, e\u011fitsel veri madencili\u011finde en \u00f6nemli metodolojik yakla\u015f\u0131mlardan biridir (Baker ve Siemens, 2014; Baker ve Yacef, 2009). \u00d6l\u00e7me teorisi, aksine, sosyal bilimlerde geleneksel olarak kullan\u0131lan istatistiksel y\u00f6ntemlerin \u00e7o\u011funda oldu\u011fu gibi, tamamen a\u00e7\u0131klay\u0131c\u0131d\u0131r (Breiman, 2001; Shmueli, 2010). A\u00e7\u0131klay\u0131c\u0131 bir model, \u00f6ng\u00f6r\u00fclerde bulunmak i\u00e7in kullan\u0131labilirken -ve hatas\u0131z- bir a\u00e7\u0131klay\u0131c\u0131 model, kusursuz tahminlerde bulunabilir; kestirimci bir model muhakkak a\u00e7\u0131klay\u0131c\u0131 olmak zorunda de\u011fildir. Breiman (2001) iki k\u00fclt\u00fcr a\u00e7\u0131s\u0131ndan ayr\u0131m\u0131 ifade etmi\u015ftir: veri modelleme k\u00fclt\u00fcr\u00fc (Breiman'a g\u00f6re gayriresm\u00ee olarak istatistiklerin %98'i) ve algoritmik modelleme k\u00fclt\u00fcr\u00fc (Breiman'\u0131n kendisini i\u00e7erdi\u011fi %2).5<\/sup><\/a> Shmueli (2010), bir tahmin veya a\u00e7\u0131klama merce\u011finden bak\u0131ld\u0131\u011f\u0131nda istatistiksel modelleme i\u00e7in t\u00fcm tasar\u0131m s\u00fcrecinin kar\u015f\u0131la\u015ft\u0131rmas\u0131n\u0131 yapm\u0131\u015ft\u0131r. Yorday\u0131c\u0131lar\u0131n karma\u015f\u0131k bir tahmin modelinde yorumlanabilirli\u011fi veya yorumlanamazl\u0131\u011f\u0131, ayr\u0131m\u0131n yaln\u0131zca bir y\u00f6n\u00fcd\u00fcr (ayr\u0131ca bk. Liu ve Koedinger, bu say\u0131). Farkl\u0131 bak\u0131\u015f a\u00e7\u0131lar\u0131, ara\u015ft\u0131rmac\u0131lar\u0131n hata ve belirsizlikle nas\u0131l ba\u015fa \u00e7\u0131kt\u0131klar\u0131 hakk\u0131nda temel olarak bilgilendirmektedir.<\/p>\n

    Kestirimci g\u00f6r\u00fc\u015f, \u00f6rne\u011fin, e\u011fitsel veri madencili\u011fi konferans\u0131ndaki en son ve en iyi makalede a\u00e7\u0131klanm\u0131\u015ft\u0131r. Yazarlar, \u201cmodel varsay\u0131mlar\u0131n\u0131n do\u011fru olup olmad\u0131\u011f\u0131n\u0131 belirlemenin tek yolu, farkl\u0131 varsay\u0131mlar yapan alternatif bir model olu\u015fturmak ve alternatifin [tahmin d\u0131\u015f\u0131] BBT' den daha iyi performans g\u00f6sterip g\u00f6stermedi\u011fini belirlemektir\u201d iddias\u0131ndad\u0131r (Khajah, Lindsey ve Mozer, 2016, 95, edit\u00f6r notu eklendi). A\u00e7\u0131k\u00e7as\u0131 model tahmin performans\u0131 model varsay\u0131mlar\u0131n\u0131n ihlal edilip edilmedi\u011finin belirlemesi i\u00e7in bir yolu de\u011fildir. Aksine hem gayriresm\u00ee kontroller hem de uyumlulu\u011fa y\u00f6nelik resm\u00ee testler yukar\u0131da tart\u0131\u015f\u0131lm\u0131\u015ft\u0131r. Bununla birlikte, al\u0131nt\u0131, modellerin \u00f6ng\u00f6r\u00fcc\u00fc do\u011frulukla onayland\u0131\u011f\u0131 algoritmik modelleme k\u00fclt\u00fcr\u00fcn\u00fcn bir yans\u0131mas\u0131d\u0131r (Breiman, 2001). Daha problematik olarak, bu yorday\u0131c\u0131 g\u00fcc\u00fcn daha ger\u00e7ek\u00e7i bir modele i\u015faret etti\u011fi varsay\u0131m\u0131n\u0131 ta\u015f\u0131r. Asl\u0131nda, bu rol\u00fc oynayan a\u00e7\u0131klay\u0131c\u0131 bir g\u00fc\u00e7t\u00fcr. Varyans bile\u015fenleri a\u00e7\u0131s\u0131ndan, \u201ca\u00e7\u0131klay\u0131c\u0131 modellemede odak, temel teorinin en do\u011fru temsilini elde etmek i\u00e7in yanl\u0131l\u0131\u011f\u0131 en aza indirmektir. Buna kar\u015f\u0131l\u0131k, kestirimci modelleme \u00f6nyarg\u0131 ve varyans kombinasyonunu en aza indirmeyi, zaman zamansa geli\u015fmi\u015f deneysel kesinlik i\u00e7in teorik do\u011frulu\u011fu feda etmeyi ama\u00e7lamaktad\u0131r \u201d(Shmueli, 2010, s. 293). A\u00e7\u0131klay\u0131c\u0131 g\u00fcc\u00fcn ve yorday\u0131c\u0131 g\u00fcc\u00fcn her zaman ayn\u0131 y\u00f6ne i\u015faret etmedi\u011fi vurgulanmal\u0131d\u0131r. Nitekim, Hagerty ve Srinivasan (1991), karma\u015f\u0131k durumlarda, yetersiz tan\u0131mlanm\u0131\u015f \u00e7oklu regresyon modellerinin do\u011fru (ger\u00e7ek) modelden daha fazla yorday\u0131c\u0131 g\u00fcce sahip oldu\u011funu kan\u0131tlam\u0131\u015ft\u0131r.<\/p>\n

    Suthers ve Verbert (2013) \u00f6\u011frenme analiti\u011fini, \u00f6\u011frenme bilimi ve analitik aras\u0131nda \u201corta alan\u201d olarak tan\u0131mlam\u0131\u015ft\u0131r. Belki de a\u00e7\u0131klay\u0131c\u0131 ve kestirimci yakla\u015f\u0131mlar aras\u0131nda metodolojik bir orta alan\u0131 i\u015fgal etti\u011fi d\u00fc\u015f\u00fcn\u00fclebilir. Bu durumda, alan her iki bak\u0131\u015f\u0131n n\u00fcanslar\u0131n\u0131 anlamakta fayda elde edebilir.<\/p>\n\n

    DAHA FAZLA OKUMA<\/h3>\n

    Psikolojik \u00f6l\u00e7meler neredeyse psikolojinin kendisi ve istatistikler kadar eskidir. G\u00fcvenilir, teknik ve bir nevi ansiklopedik kaynaklar, \u0130statistik El Kitab\u0131<\/i><\/span> serisindeki psikometri antolojisi(Rao ve Sinharay, 2006) ve \u015fu an da d\u00f6rd\u00fcnc\u00fc bask\u0131s\u0131nda olan E\u011fitsel \u00d6l\u00e7menin<\/i><\/span> \u201c\u0130ncil'i\u201ddir (Brennan, 2006). Belirli say\u0131lar\u0131n g\u00fcvenilirlik, ge\u00e7erlilik, genelle\u015ftirilebilirlik, kar\u015f\u0131la\u015ft\u0131r\u0131labilirlik ve do\u011fruluk oldu\u011fu e\u011fitsel \u00f6l\u00e7me sorunlar\u0131 ve Standartlar<\/i><\/span> s\u0131navlara vurgu yapar (AERA, APA ve NCME, 2014). DeVellis'in (2003) \u00f6l\u00e7ek geli\u015ftirmede \u00f6zl\u00fc hacim makalesi, psikolojik \u00f6l\u00e7meye teknik olmayan bir giri\u015f sunar ve paralel test formlar\u0131ndan al\u0131nan puanlar\u0131 birbirine ba\u011flamak gibi b\u00fcy\u00fck \u00f6l\u00e7ekli testlere \u00f6zg\u00fc konular\u0131 g\u00f6z ard\u0131 etmektedir.<\/p>\nKAYNAK\u00c7A\n\nAERA, APA, & NCME (American Educational Research Association, American Psychological Association, & National Council on Measurement in Education). (2014). Standards for educational and psychological testing<\/i>. Washington, DC: AERA. <\/span>\n\nAmes, A. J., & Penfield, R. D. (2015). An NCME instructional module on polytomous item response theory models. Educational Measurement: Issues and Practice, 34<\/i>(3), 39\u201348. doi:10.1111\/emip.12023 <\/span>\n\nAnderson, J. R., Corbett, A. T., Koedinger, K. R., & Pelletier, R. (1995). Cognitive tutors: Lessons learned. The Journal of the Learning Sciences, 4<\/i>(2), 167\u2013207. <\/span>\n\nArmstrong, J. S. (1967). Derivation of theory by means of factor analysis or Tom Swift and his electric factor analysis machine. The American Statistician, 21<\/i>, 17\u201321. <\/span>\n\nAttali, Y. (2011). Immediate feedback and opportunity to revise answers: Application of a graded response IRT model. Applied Psychological Measurement, 35<\/i>(6), 472\u2013479. <\/span>\n\nBaker, F. B., & Kim, S.-H. (Eds.). (2004). Item response theory: Parameter estimation techniques<\/i>. Boca Raton, FL: CRC Press. <\/span>\n\nBaker, R. S., & Siemens, G. (2014). Educational data mining and learning analytics. In R. Sawyer (Ed), The Cambridge handbook of the learning sciences <\/i>(pp. 253\u2013272). Cambridge University Press. <\/span>\n\nBaker, R. S., & Yacef, K. (2009). The state of educational data mining in 2009: A review and future visions. Journal of Educational Data Mining, 1<\/i>(1), 3\u201317. <\/span>\n\nBarnes, T. (2005). The Q-matrix method: Mining student response data for knowledge. In the Technical Report (WS-05-02) of the AAAI-05 Workshop on Educational Data Mining. <\/span>\n\nBehrens, J. T., & DiCerbo, K. E. (2014). Harnessing the currents of the digital ocean. In J. A. Larusson & B. White (Eds.), Learning analytics: From research to practice <\/i>(pp. 39\u201360). New York: Springer. <\/span>\n\nBachman, J. G., & O\u2019Malley, P.M. (1984). Yea-saying, nay-saying, and going to extremes: Black-white differences in response styles. Public Opinion Quarterly, 48<\/i>, 491\u2013509. <\/span>\n\nBergner, Y., Colvin, K., & Pritchard, D. E. (2015). Estimation of ability from homework items when there are missing and\/or multiple attempts. Proceedings of the 5th International Conference on Learning Analytics and Knowledge <\/i>(LAK\u201915), 16\u201320 March 2015, Poughkeepsie, NY, USA (pp. 118\u2013125). New York: ACM. <\/span>\n\nBergner, Y., Kerr, D., & Pritchard, D. E. (2015). Methodological challenges in the analysis of MOOC data for exploring the relationship between discussion forum views and learning outcomes. In O. C. Santos et al. (Eds.), Proceedings of the 8th International Conference on Educational Data Mining <\/i>(EDM2015), 26\u201329 June 2015, Madrid, Spain (pp. 234\u2013241). International Educational Data Mining Society. <\/span>\n\nBergner, Y., Rayyan, S., Seaton, D., & Pritchard, D. E. (2013). Multidimensional student skills with collaborative filtering. AIP Conference Proceedings, 1513<\/i>(1), 74\u201377. doi:10.1063\/1.4789655 <\/span>\n\nBlei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent Dirichlet allocation. Journal of Machine Learning Research, 3<\/i>(Jan.), 993\u20131022. <\/span>\n\nBollen, K. A. (1989). Structural equations with latent variables<\/i>. John Wiley & Sons. <\/span>\n\nBorsboom, D. (2008). Latent variable theory. Measurement: Interdisciplinary Research & Perspective, 6<\/i>(1\u20132), 25\u201353. http:\/\/doi.org\/10.1080\/15366360802035497 <\/span>\n\nBox, G. E. (1979). Robustness in the strategy of scientific model building. Robustness in Statistics, 1<\/i>, 201\u2013236. <\/span>\n\nBreiman, L. (2001). Statistical modeling: The two cultures. Statistical Science, 16<\/i>(3), 199\u2013215. http:\/\/doi.org\/10.2307\/2676681 <\/span>\n\nBrennan, R. L. (Ed.). (2006). Educational measurement<\/i>. Praeger Publishers.<\/span>\n\nBridgman, P. W. (1927). The logic of modern physics<\/i>. New York: Macmillan. <\/span>\n\nBuckingham Shum, S., & Deakin Crick, R. (2012). Learning dispositions and transferable competencies: Pedagogy, modeling and learning analytics. Proceedings of the 2nd International Conference on Learning Analytics and Knowledge <\/i>(LAK\u201912), 29 April\u20132 May 2012, Vancouver, BC, Canada (pp. 92\u2013101). New York: ACM. <\/span>\n\nCardamone, C. N., Abbott, J. E., Rayyan, S., Seaton, D. T., Pawl, A., & Pritchard, D. E. (2011). Item response theory analysis of the mechanics baseline test. Proceedings of the 2011 Physics Education Research Conference <\/i>(PERC 2011), 3\u20134 August 2011, Omaha, NE, USA (pp. 135\u2013138). doi:10.1063\/1.3680012 <\/span>\n\nCen, H., Koedinger, K. R., & Junker, B. (2008). Comparing two IRT models for conjunctive skills. In B. Woolf, E. A\u00efmeur, R. Nkambou, & S. Lajoie (Eds.), Proceedings of the 9th International Conference on Intelligent Tutoring Systems <\/i>(ITS 2008), 23\u201327 June 2008, Montreal, PQ, Canada (pp. 796\u2013798). Springer. <\/span>\n\nCohen, J. (1968). Weighted kappa: Nominal scale agreement provision for scaled disagreement or partial credit. Psychological Bulletin, 70<\/i>(4), 213\u2013220. <\/span>\n\nCorbett, A. T., & Anderson, J. R. (1995). Knowledge tracing: Modeling the acquisition of procedural knowledge. User Modeling and User-Adapted Interaction, 4<\/i>, 253\u2013278. <\/span>\n\nCortina, J.M. (1993). What is coefficient alpha? An examination of theory and applications. Journal of Applied Psychology, 78<\/i>(1), 98. <\/span>\n\nCrick, R. D., Broadfoot, P., & Claxton, G. (2004). Developing an effective lifelong learning inventory: The ELLI project. Assessment in Education: Principles, Policy & Practice, 11<\/i>(3), 247\u2013272. <\/span>\n\nCronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16<\/i>(3), 297\u2013334. <\/span>\n\nCronbach, L. J., & Meehl, P. E. (1955). Construct validity in psychological tests. Psychological Bulletin, 52<\/i>(4), 281\u2013302. <\/span>\n\nCulpepper, S. A. (2014). If at first you don\u2019t succeed, try, try again: Applications of sequential IRT models to cognitive assessments. Applied Psychological Measurement, 38<\/i>(8), 632\u2013644. doi:10.1177\/0146621614536464 <\/span>\n\nDeci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behaviour. New York: Plenum. <\/span>\n\nDedic, H., Rosenfield, S., & Lasry, N. (2010). Are all wrong FCI answers equivalent? AIP Conference Proceedings, 1289<\/i>, 125\u2013128. doi.org\/10.1063\/1.3515177 <\/span>\n\nDesmarais, M.C. (2012). Mapping question items to skills with non-negative matrix factorization. ACM SIGKDD Explorations Newsletter, 13<\/i>(2), 30\u201336. <\/span>\n\nDesmarais, M. C., & Baker, R. S. (2011). A review of recent advances in learner and skill modeling in intelligent learning environments. User Modeling and User-Adapted Interaction, 22<\/i>(1\u20132), 9\u201338. doi:10.1007\/s11257-011- 9106-8 <\/span>\n\nDeVellis, R. F. (2003). Scale development: Theory and applications<\/i>. Applied Social Research Methods Series (Vol. 26). Thousand Oaks, CA: Sage Publications. <\/span>\n\nDigman, J.M. (1990). Personality structure: Emergence of the five-factor model. Annual Review of Psychology, 41<\/i>(1), 417\u2013440. <\/span>\n\nDing, L., & Beichner, R. (2009). Approaches to data analysis of multiple-choice questions. Physical Review Special Topics: Physics Education Research, 5<\/i>(2), 1\u201317. doi:10.1103\/PhysRevSTPER.5.020103 <\/span>\n\nDraney, K., Pirolli, P., & Wilson, M. R. (1995). A measurement model for a complex cognitive skill. In P. Nichols, S. Chipman, & R. Brennan (Eds.), Cognitively diagnostic assessment<\/i>. Hillsdale, NJ: Lawrence Erlbaum Associates. <\/span>\n\nDuckworth, A. L., Peterson, C., Matthews, M. D., & Kelly, D. R. (2007). Grit: Perseverance and passion for long-term goals. Journal of Personality and Social Psychology, 9<\/i>, 1087\u20131101. <\/span>\n\nDweck, C. S. (2000). Self-theories: Their role in motivation, personality and development. Philadelphia, PA: Taylor & Francis.<\/span>\n\nEdwards, J. R. (2001). Multidimensional constructs in organizational behavior research: An integrative analytical framework. Organizational Research Methods, 4<\/i>(2), 144\u2013192. <\/span>\n\nErosheva, E., Fienberg, S., & Lafferty, J. (2004). Mixed-membership models of scientific publications. Proceedings of the National Academy of Sciences, 101<\/i>(suppl 1), 5220\u20135227. <\/span>\n\nFabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. (1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods, 4<\/i>(3), 272. <\/span>\n\nFischer, G.H. (1973). The linear logistic test model as an instrument in educational research. Acta Psychologica, 37<\/i>(6), 359\u2013374. <\/span>\n\nFraley, C., & Raftery, A. E. (1998). How many clusters? Which clustering method? Answers via model-based cluster analysis. The Computer Journal, 41<\/i>(8), 578\u2013588. <\/span>\n\nGeorge, R. (2000). Measuring change in students\u2019 attitudes toward science over time: An application of latent variable growth modeling. Journal of Science Education and Technology, 9<\/i>(3), 213\u2013225. <\/span>\n\nGoodman, L. (2002) Latent class analysis: The empirical study of latent types, latent variables, and latent structures. In J. A. Hagenaars & A. L. McCutcheon (Eds.), Applied latent class analysis <\/i>(pp. 3\u201355). Cambridge, UK: Cambridge University Press. <\/span>\n\nGuay, F., Vallerand, R. J., & Blanchard, C. (2000). On the assessment of situational intrinsic and extrinsic motivation: The situational motivation scale (SIMS). Motivation and Emotion, 24<\/i>(3), 175\u2013213. <\/span>\n\nHaberman, S. J. (2009). Use of generalized residuals to examine goodness of fit of item response models. ETS Research Report RR-09-15<\/i>. <\/span>\n\nHagerty, M. R., & Srinivasan, V. (1991). Comparing the predictive powers of alternative multiple regression models. Psychometrika, 56<\/i>(1), 77\u201385. <\/span>\n\nHestenes, D., & Wells, M. (1992). A mechanics baseline test. The Physics Teacher, 30<\/i>(3), 159\u2013166. <\/span>\n\nHestenes, D., Wells, M., & Swackhamer, G. (1992). Force concept inventory. The Physics Teacher, 30<\/i>(3), 141. doi:10.1119\/1.2343497 <\/span>\n\nHolland, P.W. (1990). On the sampling theory roundations of item response theory models. Psychometrika, 55<\/i>(4), 577\u2013601. http:\/\/doi.org\/10.1007\/BF02294609 <\/span>\n\nKane, M.T. (2001). Current concerns in validity theory. Journal of Educational Measurement, 38<\/i>(4), 319\u2013342. <\/span>\n\nKane, M. (2010). Errors of measurement, theory, and public policy. William H. Angoff Memorial Lecture Series. Educational Testing Service<\/i>. https:\/\/www.ets.org\/Media\/Research\/pdf\/PICANG12.pdf <\/span>\n\nK\u00e4ser, T., Koedinger, K. R., & Gross, M. (2014). Different parameters \u2014 same prediction: An analysis of learning curves. In S. K. D\u2019Mello, R. A. Calvo, & A. Olney (Eds.), Proceedings of the 6th International Conference on Educational Data Mining <\/i>(EDM2013), 6\u20139 July 2013, Memphis, TN, USA (pp. 52\u201359). International Educational Data Mining Society\/Springer. <\/span>\n\nKhajah, M., Lindsey, R. V., & Mozer, M. C. (2016). How deep is knowledge tracing? In T. Barnes, M. Chi, & M. Feng (Eds.), Proceedings of the 9th International Conference on Educational Data Mining <\/i>(EDM2016), 29 June\u20132 July 2016, Raleigh, NC, USA (pp. 94\u2013101). International Educational Data Mining Society. <\/span>\n\nKline, R. B. (2010). Principles and practice of structural equation modeling. New York: Guilford. <\/span>\n\nKoedinger, K. R., McLaughlin, E. A., & Stamper, J. (2012). Automated student model improvement. In K. Yacef et al. (Eds.), Proceedings of the 5th International Conference on Educational Data Mining <\/i>(EDM2012), 19\u201321 June 2012, Chania, Greece. International Educational Data Mining Society. http:\/\/www.learnlab.org\/research\/ wiki\/images\/e\/e1\/KoedingerMcLaughlinStamperEDM12.pdf <\/span>\n\nLord, F. M. (1980). Applications of item response theory to practical testing problems. Routledge. <\/span>\n\nLord, F. M., & Novick, M. R. (1968). Statistical theories of mental test scores<\/i>. Addison-Wesley.<\/span>\n\nLuria, R. E. (1975). The validity and reliability of the visual analogue mood scale. Journal of Psychiatric Research, 12<\/i>(1), 51\u201357. <\/span>\n\nMartin, B., Mitrovic, T., Mathan, S., & Koedinger, K. R. (2010). Evaluating and improving adaptive educational systems with learning curves. User Modeling and User-Adapted Interaction: The Journal of Personalization Research, 21<\/i>, 249\u2013283. <\/span>\n\nMaul, A., Irribarra, D. T., & Wilson, M. (2016). On the philosophical foundations of psychological measurement. Measurement, 79<\/i>, 311\u2013320. http:\/\/doi.org\/10.1016\/j.measurement.2015.11.001 <\/span>\n\nMazur, E. (2007). Confessions of a converted lecturer. https:\/\/www.math.upenn.edu\/~pemantle\/active-papers\/Mazurpubs_605.pdf <\/span>\n\nMcLachlan, G., & Peel, D. (2004). Finite mixture models<\/i>. John Wiley & Sons. <\/span>\n\nMeredith, W., & Tisak, J. (1990). Latent curve analysis. Psychometrika, 55<\/i>(1), 107\u2013122. <\/span>\n\nMessick, S. (1995). Validity of psychological assessment: Validation of inferences from persons\u2019 responses and performances as scientific inquiry into score meaning. American Psychologist, 50<\/i>(9), 741\u2013749. <\/span>\n\nMessick, S., & Jackson, D. (1961). Acquiescence and the factorial interpretation of the MMPI. Psychological Bulletin, 58<\/i>(4), 299\u2013304 <\/span>\n\nMichell, J. (1999). Measurement in psychology: A critical history of a methodological concept (Vol. 53). Cambridge University Press. <\/span>\n\nMidgley, C., Maehr, M. L., Hruda, L., Anderman, E. M., Anderman, L., Freeman, K. E., et al. (2000). Manual for the patterns of adaptive learning scales (PALS)<\/i>. Ann Arbor, MI: University of Michigan. <\/span>\n\nMilligan, S. K., & Griffin, P. (2016). Understanding learning and learning design in MOOCs: A measurement-based interpretation. Journal of Learning Analytics, 3<\/i>(2), 88\u2013115. <\/span>\n\nMillsap, R.E. (2012). Statistical approaches to measurement invariance. Routledge. <\/span>\n\nMislevy, R. J. (2009). Validity from the perspective of model-based reasoning. In R. L. Lissitz (Ed.), The concept of validity: Revisions, new directions and applications <\/i>(pp. 83\u2013108). Charlotte, NC: Information Age Publishing. <\/span>\n\nMislevy, R. J. (2012). Four metaphors we need to understand assessment. Draft paper commissioned by the Gordon Commission. http:\/\/www.gordoncommission.com\/rsc\/pdfs\/mislevy_four_metaphors_understand_assessment.pdf <\/span>\n\nMorris, G. A., Branum-Martin, L., Harshman, N., Baker, S. D., Mazur, E., Dutta, S., \u2026 McCauley, V. (2006). Testing the test: Item response curves and test quality. American Journal of Physics, 74<\/i>(5), 449. doi:10.1119\/1.2174053 <\/span>\n\nMulaik, S. A. (2009). Foundations of factor analysis<\/i>. Boca Raton, FL: CRC Press. <\/span>\n\nNederhof, A. J. (1985). Methods of coping with social desirability bias: A review. European Journal of Social Psychology, 15<\/i>(3), 263\u2013280. http:\/\/doi.org\/10.1002\/ejsp.2420150303 <\/span>\n\nNewell, A., & Rosenbloom, P. S. (1981). Mechanisms of skill acquisition and the law of practice. Cognitive Skills and their Acquisition, 6<\/i>, 1\u201355. <\/span>\n\nPekrun, R., Goetz, T., Frenzel, A. C., Barchfeld, P., & Perry, R. P. (2011). Measuring emotions in students\u2019 learning and performance: The achievement emotions questionnaire (AEQ). Contemporary Educational Psychology, 36<\/i>(1), 36\u201348. http:\/\/doi.org\/10.1016\/j.cedpsych.2010.10.002 <\/span>\n\nPintrich, P. R., & De Groot, E. V. (1990). Motivational and self-regulated learning components of classroom academic performance. Journal of Educational Psychology, 82<\/i>(1), 33. <\/span>\n\nRabiner, L. R. (1989). A tutorial on hidden Markov models and selected applications in speech recognition. Proceedings of the IEEE, 77<\/i>(2), 257\u2013286.<\/span>\n\nRao, C. R., & Sinharay, S. (Eds.). (2006). Handbook of statistics 26: Psychometrics<\/i>. Elsevier. doi:10.1016\/S0169- 7161(06)26037-1 <\/span>\n\nRaudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (Vol. 1). Sage. <\/span>\n\nRijmen, F. (2010). Formal relations and an empirical comparison among the bi-factor, the testlet, and a second-order multidimensional IRT model. Journal of Educational Measurement, 47<\/i>(3), 361\u2013372. doi:10.1111\/ j.1745-3984.2010.00118.x <\/span>\n\nRupp, A., & Templin, J. L. (2008). Unique characteristics of diagnostic classification models: A comprehensive review of the current state-of-the-art. Measurement: Interdisciplinary Research & Perspective, 6<\/i>(4), 219\u2013 262. doi:10.1080\/15366360802490866 <\/span>\n\nSchwartz, S. (2007). The structure of identity consolidation: Multiple correlated constructs or one superordinate construct? Identity, 7<\/i>(1), 27\u201349. <\/span>\n\nScott, T. F., Schumayer, D., & Gray, A. R. (2012). Exploratory factor analysis of a force concept inventory data set. Physical Review Special Topics: Physics Education Research, 8<\/i>(2). doi:10.1103\/PhysRevSTPER.8.020105 <\/span>\n\nShmueli, G. (2010). To explain or to predict? Statistical Science, 25<\/i>(3), 289\u2013310. http:\/\/doi.org\/10.1214\/10- STS330 <\/span>\n\nSiemens, G., & Baker, R. S. (2012). Learning analytics and educational data mining: Towards communication and collaboration. Proceedings of the 2nd International Conference on Learning Analytics and Knowledge <\/i>(LAK\u201912), 29 April\u20132 May 2012, Vancouver, BC, Canada (pp. 252\u2013254). New York: ACM. <\/span>\n\nSijtsma, K. (2011). Introduction to the measurement of psychological attributes. Measurement, 44<\/i>(7), 1209\u20131219. doi: 10.1016 \/ j.measurement.2011.03.019 <\/span>\n\nSijtsma, K. (1998). Methodology review: Nonparametric IRT approaches to the analysis of dichotomous item scores. Applied Psychological Measurement, 22<\/i>(1), 3\u201331. doi:10.1177\/01466216980221001 <\/span>\n\nSkrondal, A., & Rabe-Hesketh, S. (2004). Generalized latent variable modeling: Multilevel, longitudinal and structural equation models. Boca Raton, FL: Chapman & Hall\/CRC Press. <\/span>\n\nSpearman, C. (1904). \u201cGeneral intelligence,\u201d objectively determined and measured. The American Journal of Psychology, 15<\/i>(2), 201\u2013292. <\/span>\n\nSpray, J. A. (1997). Multiple-attempt, single-item response models. In W. J. van der Linden & R. K. Hambleton (Eds.), Handbook of modern item response theory <\/i>(pp. 209\u2013220). New York: Springer. <\/span>\n\nStevens, S.S. (1946). On the theory of scales of measurement. Science, 103<\/i>(2684), 677\u2013680. <\/span>\n\nSuthers, D., & Verbert, K. (2013). Learning analytics as a middle space. Proceedings of the 3rd International Conference on Learning Analytics and Knowledge <\/i>(LAK\u201913), 8\u201312 April 2013, Leuven, Belgium (pp. 1\u20134). New York: ACM. <\/span>\n\nTatsuoka, K.K. (1983). Rule space: An approach for dealing with misconceptions based on item response theory. Journal of Educational Measurement, 20<\/i>, 345\u2013354. <\/span>\n\nTempelaar, D. T., Niculescu, A., Rienties, B., Giesbers, B., & Gijselaers, W. H. (2012). How achievement emotions impact students\u2019 decisions for online learning, and what precedes those emotions. Internet and Higher Education, 15<\/i>(3), 161\u2013169. doi: 10.1016 \/ j.iheduc.2011.10.003 <\/span>\n\nTempelaar, D. T., Rienties, B., & Giesbers, B. (2015). In search for the most informative data for feedback generation: Learning analytics in a data-rich context. Computers in Human Behavior, 47<\/i>, 157\u2013167. doi:10.1016\/j. chb.2014.05.038 <\/span>\n\nThurstone, L.L. (1947). Multiple factor analysis<\/i>. Chicago, IL: University of Chicago Press. <\/span>\n\nvan de Sande, B. (2013). Properties of the Bayesian knowledge tracing model. Journal of Educational Data Mining, 5<\/i>(2), 1\u201310.<\/span>\n\nvon Davier, M. (2005). A general diagnostic model applied to language testing data. The British Journal of Mathematical and Statistical Psychology, 61<\/i>(Pt 2), 287\u2013307. doi:10.1348\/000711007X193957 <\/span>\n\nWang, Y., & Baker, R. S. (2015). Content or platform: Why do students complete MOOCs? Journal of Online Learning and Teaching, 11<\/i>(1), 17. <\/span>\n\nWang, J., & Bao, L. (2010). Analyzing force concept inventory with item response theory. American Journal of Physics, 78<\/i>(10), 1064. doi:10.1119\/1.3443565 <\/span>\n\nWhite, H. (1996). Estimation, inference and specification analysis <\/i>(No. 22). Cambridge University Press. <\/span>\n\nWise, S., & Kong, X. (2005). Response time effort: A new measure of examinee motivation in computer-based tests. Applied Measurement in Education, 18<\/i>(2), 163\u2013183. <\/span>\n\nYeager, D. S., & Dweck, C. S. (2012). Mindsets that promote resilience: When students belie<\/span>\n\n

    \n\n
    \n

    1<\/a> Amerika\u2019daki y\u00fcksekokul ve \u00fcniversitelere giri\u015fte hem amerikan vatanda\u015flar\u0131 hem de yabanc\u0131 uyruklu \u00f6\u011frenciler taraf\u0131ndan kullan\u0131lan ayn\u0131 zamanda T\u00fcrkiye\u2019deki yabanc\u0131 uyruklu \u00f6\u011frencilerin T\u00fcrk \u00fcniversitelerine yerle\u015ftirilmesi s\u00fcrecinde de bir\u00e7ok \u00fcniversite taraf\u0131ndan kabul edilen bir s\u0131navd\u0131r.<\/span><\/span><\/p>\n\n<\/div>\n

    \n\n2<\/a> Stokastik konu i\u00e7in, bu \u00f6rnek de\u011ferler rep di\u011fer \u00e7al\u0131\u015fmalarda hi\u00e7bir bellek ile ayn\u0131 konuyu ayn\u0131 deneyler bir dizi k\u0131rg\u0131n olurdu. Bu bili\u015fsel test \u00f6gesi garip g\u00f6r\u00fcnse de, psikomotor bir ba\u011flamda b\u00fcy\u00fck bir ihtimaldir. Bknz Sprey (1997).<\/span>\n\n<\/div>\n
    \n\n3<\/a> orj. curriculum sequencing<\/span>\n\n<\/div>\n
    \n\n4<\/a> Cen vd.nin i\u015faret uzla\u015f\u0131m\u0131 kural\u0131 (2008), modeli al\u0131\u015f\u0131lm\u0131\u015f Rasch modeli ile tutarl\u0131 hale getirmek i\u00e7in kolayl\u0131ktan \u00e7ok bir zorluk parametresi olarak de\u011fi\u015ftirildi.<\/span>\n\n<\/div>\n
    \n\n5<\/a> Breiman a\u00e7\u0131klama ve tahmin i\u00e7in incontrast yerine d\u00f6nem bilgileri kullan\u0131r.<\/span>\n\n<\/div>\n","rendered":"

    Yoav Bergner<\/span><\/span><\/p>\n

    \u00d6\u011frenme Analiti\u011fi Ara\u015ft\u0131rma A\u011f\u0131, New York \u00dcniversitesi, ABD<\/span><\/span><\/p>\n

    DOI: 10.18608\/hla17.003<\/span><\/span><\/p>\n

    \u00d6Z<\/h2>\n

    Psikolojik \u00f6l\u00e7me, zihinsel durumlar hakk\u0131nda uygunlu\u011fu kan\u0131tlanm\u0131\u015f iddialarda bulunma s\u00fcrecidir. ]Bu haliyle, tipik olarak \u015funlar\u0131 i\u00e7ermektedir: Bir yap\u0131n\u0131n tan\u0131mlanmas\u0131; bir \u00f6l\u00e7me modeli belirlemek ve g\u00fcvenilir bir ara\u00e7 geli\u015ftirmek; \u00e7e\u015fitli hata kaynaklar\u0131n\u0131 analiz etmek (operat\u00f6r hatas\u0131 d\u00e2hil) ve sonucun belirli kullan\u0131mlar\u0131 i\u00e7in ge\u00e7erli bir arg\u00fcman \u00e7er\u00e7evelemek. \u00d6rt\u00fck de\u011fi\u015fkenlerin \u00f6l\u00e7\u00fcm\u00fc, sonu\u00e7ta, bireyler ve gruplar i\u00e7in y\u00fcksek riskli sonu\u00e7lar do\u011furabilecek y\u00fcksek perdeden bir giri\u015fimdir. Bu b\u00f6l\u00fcm, analitik ve e\u011fitsel veri madencili\u011fi \u00f6\u011frenen uygulay\u0131c\u0131lar i\u00e7in e\u011fitsel ve psikolojik \u00f6l\u00e7meye bir giri\u015f niteli\u011findedir. Yap\u0131lar, ara\u00e7lar ve \u00f6l\u00e7me hata kaynaklar\u0131 hakk\u0131ndaki daha kavramsal malzemeden, belirli \u00f6l\u00e7me modelleri ve kullan\u0131mlar\u0131 hakk\u0131nda teknik detaylar\u0131n artt\u0131r\u0131lmas\u0131na y\u00f6nelik olarak, tarihsel olmaktan ziyade tematik olarak d\u00fczenlenmi\u015ftir. A\u00e7\u0131klay\u0131c\u0131 ve kestirimci modelleme aras\u0131ndaki felsefi farkl\u0131l\u0131klar\u0131n baz\u0131lar\u0131 sona do\u011fru incelenmi\u015ftir.<\/span><\/p>\n

    Anahtar Kelimeler<\/span>: \u00d6l\u00e7me, \u00f6rt\u00fck s\u0131n\u0131f modelleri, model uyumu<\/span><\/p>\n

    \u00d6\u011frencilerin ne bildi\u011finin ve -duyu\u015fsal \u00f6l\u00e7\u00fctlere giderek daha fazla \u00f6nem verildi\u011fi g\u00f6z \u00f6n\u00fcne al\u0131nd\u0131\u011f\u0131nda- nas\u0131l hissettiklerini bilmek, \u00f6\u011frenmeye ili\u015fkin \u00e7o\u011fu sohbetin \u00f6z\u00fcn\u00fc olu\u015fturur. Bununla birlikte, bir \u00f6\u011frencinin bilgi, becerilerini, tutumlar\u0131n\u0131\/ istidat \/ yeteneklerini (BB\u0130) ve \/ veya duygular\u0131n\u0131 \u00f6l\u00e7mek, boy veya kilosunu \u00f6l\u00e7mekten daha karma\u015f\u0131k bir i\u015ftir. Psikolojik \u00f6l\u00e7me, \u00f6zel bir programa tahsis edilme (ileri d\u00fczey veya telafi), bir \u00fcniversiteye kabul, i\u015fe al\u0131m, hastaneye yat\u0131\u015f veya tutuklanma gibi y\u00fcksek riskli sonu\u00e7lar do\u011furabilecek rahats\u0131z edici bir u\u011fra\u015ft\u0131r. Bireysel seviyedeki k\u00fc\u00e7\u00fck \u00f6l\u00e7me yan\u0131lg\u0131lar\u0131 bile bulgular\u0131 gruplar i\u00e7in birle\u015ftirildi\u011finde b\u00fcy\u00fck sonu\u00e7lar do\u011furabilir. (Kane, 2010). Bu sonu\u00e7lardaki hassasiyet, E\u011fitsel ve Psikolojik \u00d6l\u00e7me Standartlar\u0131nda<\/i><\/span> yer alan bir y\u00fczy\u0131ldan fazla s\u00fcren y\u00f6ntem bilim ara\u015ft\u0131rmas\u0131yla ortaya \u00e7\u0131km\u0131\u015ft\u0131r.(AERA, APA ve NCME, 2014). \u00d6l\u00e7me bu d\u00fczeyde, \u00f6\u011frenme ve \u00f6\u011frenme ortamlar\u0131n\u0131 anlamak ve en iyi hale getirmek amac\u0131yla \u00f6\u011frenme analiti\u011fi ve e\u011fitsel veri madencili\u011finde kullan\u0131labiliyorsa (Siemens ve Baker, 2012), \u00f6l\u00e7mede kabul edilebilir hatalar neler olacakt\u0131r? Ne de olsa verinin \u201cdijital okyanusundan faydalanman\u0131n\u201d nihayetinde ayr\u0131 de\u011ferlendirmelere duyulan ihtiyac\u0131n yerini alabilece\u011fi iddia edilmi\u015ftir (Behrens ve DiCerbo, 2014). Bir taraftan da ki\u015fi en az\u0131ndan \u00f6\u011frenmeyi yanl\u0131\u015f anlamaktan<\/i><\/span> veya eksilen<\/i><\/span> \u00f6\u011frenen deneyimlerinden ka\u00e7\u0131nmak isteyecektir.<\/p>\n

    \u00d6L\u00c7ME NED\u0130R? FELSEFE VE TEMEL F\u0130K\u0130RLER<\/h2>\n

    Psikolojik \u00f6l\u00e7me tart\u0131\u015fmalar\u0131, genellikle fiziksel \u00f6l\u00e7me ile z\u0131tl\u0131klar \u00e7izerek ba\u015flamaktad\u0131r (\u00f6r. Armstrong, 1967; Borsboom, 2008 DeVellis, 2003; Lord ve Novick, 1968; Maul, Irribarra ve Wilson, 2016; Michell, 1999; Sijtsma, 2011). S\u00fcre\u00e7te, ara\u00e7salla\u015ft\u0131rma veya i\u015flemselle\u015ftirme, \u00f6l\u00e7\u00fcmlerin tekrarlanabilirli\u011fi veya kesinli\u011fi, hata kaynaklar\u0131 ve \u00f6nlemin kendisinin yorumlanmas\u0131 gibi bir dizi \u00f6nemli psikolojik \u00f6l\u00e7me fakt\u00f6r\u00fc ortaya \u00e7\u0131kar. Psikolojik \u00f6l\u00e7menin a\u015fa\u011f\u0131dakileri i\u00e7erdi\u011fi s\u00f6ylenebilir; bir yap\u0131y\u0131 tan\u0131mlamak, bir \u00f6l\u00e7me y\u00f6ntemi belirlemek ve g\u00fcvenilir bir ara\u00e7 (geli\u015ftirmek); \u00e7e\u015fitli hata kaynaklar\u0131n\u0131 analiz etmek ve nedenlerini a\u00e7\u0131klamak (uygulay\u0131c\u0131 hatas\u0131 d\u00e2hil) ve sonucun belirli kullan\u0131mlar\u0131 i\u00e7in ge\u00e7erli bir arg\u00fcman \u00e7er\u00e7evelemek.<\/p>\n

    Yap\u0131lar<\/h3>\n

    Psikolojik yap\u0131lar ger\u00e7ekten var m\u0131? Hangi anlamda \u00f6\u011frencinin halet-i ruhiyesini ger\u00e7ekten bilebiliriz? Bir nesnenin fiziksel uzunlu\u011fu gibi de\u011fi\u015fkenlerin do\u011frudan g\u00f6zlendi\u011fini veya tezah\u00fcr etti\u011fini s\u00f6ylerken, bireyin zihinsel durumlar\u0131n\u0131 veya psikolojik \u00f6zelliklerini yaln\u0131zca dolayl\u0131 olarak g\u00f6zlemlendi\u011fi veya gizlendi\u011fini s\u00f6yl\u00fcyoruz. Yap\u0131 terimi, \u00f6rt\u00fck de\u011fi\u015fkenle<\/i><\/span> de\u011fi\u015fmeli<\/i><\/span> olarak kullan\u0131l\u0131rken \u00f6zellik, yap\u0131n\u0131n zamana g\u00f6re sabit olu\u015funu ima etmek i\u00e7in kullan\u0131l\u0131r (Lord ve Novick, 1968). Asl\u0131nda, fiziksel \u00f6l\u00e7me bile dolayl\u0131 olarak ger\u00e7ekle\u015ftirilir. Uzunlu\u011fu do\u011frudan duyular\u0131m\u0131zla alg\u0131layabilmemize ra\u011fmen, uzunlu\u011fun \u00f6l\u00e7\u00fclmesi<\/i><\/span>, bir mezura gibi bir referans nesnesi veya alet ile bir kar\u015f\u0131la\u015ft\u0131rma i\u015flemini i\u00e7erir. Mezura, uzunluk kar\u015f\u0131la\u015ft\u0131rmalar\u0131n\u0131 resmile\u015ftiren in\u00e7 veya santimetre gibi bir \u00f6l\u00e7ek sa\u011flar. \u00d6rne\u011fin, iki uzunluk aras\u0131ndaki fark\u0131 bir \u00f6l\u00e7\u00fcm\u00fc di\u011ferinden \u00e7\u0131kartarak inceleyebiliriz.<\/p>\n

    Yirminci y\u00fczy\u0131l\u0131n ilk yar\u0131s\u0131nda, \u00f6l\u00e7menin felsefi \u00f6l\u00e7me meselelerini \u00e7\u00f6zme \u00e7abalar\u0131 Bridgman’\u0131 (1927) ve di\u011ferlerini i\u015flemselcili\u011fe y\u00f6nlendirdi; burada uzunluk, k\u00fctle ve yo\u011funluk gibi fiziksel kavramlar ile bunlar\u0131 \u00f6l\u00e7mek i\u00e7in kullan\u0131lan i\u015flemlerin \u201ce\u015f anlaml\u0131\u201d oldu\u011fu anla\u015f\u0131ld\u0131. Yani, uzunluk (muhtemelen faraz\u00ee) bir uzunluk \u00f6l\u00e7\u00fcm y\u00f6nteminin \u00fcr\u00fcn\u00fc olarak anla\u015f\u0131lmaktad\u0131r. Bu fikir, yap\u0131lar\u0131 onlar\u0131 \u00f6l\u00e7mek i\u00e7in kullan\u0131lan ara\u00e7lardaki puanlarla e\u015fle\u015ftirme yoluyla matematik yetene\u011fi ve d\u0131\u015fa d\u00f6n\u00fckl\u00fck gibi psikolojik yap\u0131lara aktar\u0131labilir. B\u00f6ylelikle matematik yetene\u011fi daha sonra bir matematik testindeki bir puana ve d\u0131\u015fa d\u00f6n\u00fckl\u00fck, Likert madde anketinde verilen bir puana e\u015f de\u011fer olur. Bu pozitivist tutum, Stevens’\u0131n \u201cnesnelere ya da olaylara kurallara g\u00f6re say\u0131lar\u0131n atanmas\u0131\u201d olarak yapt\u0131\u011f\u0131 \u00f6l\u00e7me tan\u0131m\u0131nda yans\u0131t\u0131lmaktad\u0131r (1946, s. 677). Yap\u0131lara ili\u015fkin i\u015flemselci g\u00f6r\u00fc\u015f ge\u00e7mi\u015fte olduk\u00e7a etkiliydi ancak bir\u00e7ok nedenden dolay\u0131 \u00f6zellikle de i\u015flemselcilik yap\u0131n\u0131n onu \u00f6l\u00e7mek i\u00e7in var olan her ara\u00e7 i\u00e7in yeniden tan\u0131mlamay\u0131 gerektirmesi nedeniyle reddedildi (Maul, Irribarra ve Wilson, 2016; Michell, 1999).<\/p>\n

    \u0130\u015flemselci bir yorum reddedildi\u011finde \u00f6rt\u00fck de\u011fi\u015fkenlere dair epistemolojik ve ontolojik sorular\u0131 a\u00e7\u0131kta b\u0131rakt\u0131\u011f\u0131 g\u00f6r\u00fclmektedir. Mislevy (2009, 2012), yap\u0131land\u0131rmac\u0131-ger\u00e7ek\u00e7i bir konumu a\u00e7\u0131k\u00e7a belirtir; yani,<\/i><\/span> model temelli bir ak\u0131l y\u00fcr\u00fctmeyi taahh\u00fct ederek, kat\u0131 ger\u00e7ek\u00e7ili\u011fe ba\u011fl\u0131 olmadan bir yap\u0131 varm\u0131\u015f gibi konu\u015fabiliriz. Model temelli ak\u0131l y\u00fcr\u00fctme, bir sistemin -\u00f6rne\u011fin, ki\u015filer ve cevaplar aras\u0131ndaki yap\u0131 arac\u0131l\u0131 ili\u015fki- g\u00f6ze \u00e7arpan y\u00f6nleri (\u00f6r. \u00f6r\u00fcnt\u00fcler) yakalayan basitle\u015ftirilmi\u015f bir temsilini, kabul etmek anlam\u0131na gelir ve olgular\u0131 a\u00e7\u0131klamam\u0131z\u0131 veya tahmin etmemizi sa\u011flar (Mislevy, 2009; a\u00e7\u0131klay\u0131c\u0131 \/ kestirimci modelleri bu b\u00f6l\u00fcm\u00fcn ilerleyen k\u0131s\u0131mlar\u0131nda ele alaca\u011f\u0131z). George Box’\u0131n \u00fcnl\u00fc s\u00f6z\u00fcnde dedi\u011fi gibi, \u201ct\u00fcm modeller yanl\u0131\u015f ancak baz\u0131lar\u0131 yararl\u0131d\u0131r\u201d (Box, 1979). Zorluk, faydal\u0131 modeller veya Stevens’\u0131n tan\u0131m\u0131 ile ifade edildi\u011finde, yararl\u0131 \u00f6l\u00e7me kurallar\u0131 ile ortaya \u00e7\u0131kmaya devam etmektedir.<\/p>\n

    Fiziksel teoriler say\u0131ca az ve daha kapsaml\u0131 olma e\u011filimindeyken, psikolojik teoriler \u00e7ok say\u0131da ve s\u0131n\u0131rl\u0131 bir \u015fekilde tan\u0131ml\u0131d\u0131rlar (DeVellis, 2003). Yap\u0131lar uydurulmu\u015f\/icat edilen \u015feyler oldu\u011fu i\u00e7in, say\u0131lar\u0131 i\u00e7in deneysel bir s\u0131n\u0131r yoktur. Bir yap\u0131 hakk\u0131nda bir \u00f6l\u00e7\u00fcm arac\u0131n\u0131n<\/i><\/span> yoklu\u011funda konu\u015fmak m\u00fcmk\u00fcnd\u00fcr ancak bir \u00f6l\u00e7\u00fcm arac\u0131 her zaman bir \u015feyi \u00f6l\u00e7mek i\u00e7in tasarlanm\u0131\u015ft\u0131r. Bu nedenle, kendilerini \u00f6l\u00e7mek i\u00e7in \u00f6nceden geli\u015ftirilen ara\u00e7lara uygun ve son derece k\u0131smi bir \u00f6\u011frenme analiti\u011fine ili\u015fkin yap\u0131lar listesi \u00e7\u0131karsamas\u0131 yapabiliriz. \u00d6rnekler aras\u0131nda zek\u00e2 (\u00f6r. Stanford-Binet Zek\u00e2 \u00d6l\u00e7e\u011fi), akademik yatk\u0131nl\u0131k (\u00f6r. bu SAT1<\/sup><\/a> testi), akademik ba\u015far\u0131 (hem b\u00fcy\u00fck \u00f6l\u00e7ekli s\u0131navlar hem de ders ba\u015far\u0131 s\u0131navlar\u0131 d\u00e2hil say\u0131s\u0131z \u00f6rnek), ki\u015filik (\u00f6r. \u201cb\u00fcy\u00fck be\u015f\u201d fakt\u00f6r modeli; Digman, 1990), ba\u015far\u0131 hedef oryantasyonu (\u00f6r. Midgley vd., 2000), tatmin duygular\u0131 (Pekrun, Goetz, Frenzel, Barchfeld ve Perry, 2011), sab\u0131r (Duckworth, Peterson, Matthews ve Kelly, 2007), \u00f6z yeterlilik teorileri ve sabit \/ b\u00fcy\u00fcme zihniyeti teorileri (Dweck, 2000; Yeager ve Dweck, 2012), i\u00e7sel motivasyon (Deci ve Ryan, 1985; Guay, Vallerand ve Blanchard, 2000), \u00f6z y\u00f6netimli \u00f6\u011frenme ve \u00f6z yeterlik (\u00f6r. Pintrich ve De Groot, 1990), \u00f6\u011frenme g\u00fcc\u00fc (Buckingham Shum ve Deakin Crick, 2012; Crick, Broadfoot ve Claxton, 2004) ve kitle kaynakl\u0131 \u00f6\u011frenme yetene\u011fi (Milligan ve Griffin, 2016) vard\u0131r.<\/p>\n

    Yukar\u0131da listelenen yap\u0131lar\u0131n bir\u00e7o\u011fu \u00e7ok boyutludur, yani birden \u00e7ok fakt\u00f6r i\u00e7erirler. \u0130li\u015fkili yap\u0131lar\u0131 ayr\u0131\u015ft\u0131rman\u0131n ya da birle\u015ftirmenin de\u011feri bir tart\u0131\u015fma konusudur (Edwards, 2001; Schwartz, 2007).<\/p>\n

    \u00d6l\u00e7me Ara\u00e7lar\u0131<\/h3>\n

    Psikolojik \u00f6l\u00e7me ara\u00e7lar\u0131na genellikle test veya soru formlar\u0131 (ayr\u0131ca anketler ve envanterler) denir ve maddelerden veya g\u00f6stergelerden olu\u015furlar. Test kelimesi daha \u00e7ok zek\u00e2, bili\u015fsel yetenek ve psikomotor becerileri gibi yap\u0131lar i\u00e7in kullan\u0131l\u0131r; burada derse veya s\u0131nava giren ki\u015finin performans\u0131n\u0131 en \u00fcst seviyeye \u00e7\u0131karmaya \u00e7al\u0131\u015fmas\u0131 istenir (Sijtsma, 2011). Soru formu kat\u0131l\u0131mc\u0131lar\u0131ndan, aksine, d\u00fc\u015f\u00fcnceleri, duygular\u0131 ve davran\u0131\u015flar\u0131 ile ilgili d\u00fcr\u00fcst\u00e7e cevaplar vermeleri istenir. (Tepki yanl\u0131l\u0131k de\u011feri, ge\u00e7erlili\u011fe geldi\u011fimizde tan\u0131mlayaca\u011f\u0131m\u0131z gibi, bu ayr\u0131m\u0131 bulan\u0131kla\u015ft\u0131rabilir). Deneklerin ara\u00e7larla nas\u0131l etkile\u015fime girmesi beklendi\u011fine ili\u015fkin bu tan\u0131mlaman\u0131n bir \u00f6l\u00e7me modelinin<\/i><\/span> temel ilkelerini ortaya \u00e7\u0131kard\u0131\u011f\u0131na dikkat ediniz. Daha yetenekli bir s\u0131nav kat\u0131l\u0131mc\u0131s\u0131n\u0131n bir yetenek s\u0131nav\u0131nda daha y\u00fcksek puan alaca\u011f\u0131n\u0131 ve daha endi\u015feli bireyin kayg\u0131 anketinde daha y\u00fcksek puan alaca\u011f\u0131n\u0131 varsay\u0131yoruz.<\/p>\n

    Bazen \u00f6l\u00e7me \u00f6l\u00e7e\u011fi terimi, enstr\u00fcmanla de\u011fi\u015fimli olarak kullan\u0131l\u0131r (DeVellis, 2003). \u00d6l\u00e7ek test veya anketin puanland\u0131\u011f\u0131n\u0131 g\u00f6sterir. Do\u011fru ve yanl\u0131\u015f cevaplar\u0131 olan ve evet \/ hay\u0131r sorular\u0131na sahip olan ikili maddeler genellikle {0, 1} ‘de yer alan de\u011ferlerle ikili bir \u015fekilde puanlan\u0131r. Likert \u00f6l\u00e7e\u011fi, puanlama \u00f6l\u00e7e\u011fi ve g\u00f6rsel-analog \u00f6l\u00e7ekler (Luria, 1975), kesikli veya s\u00fcrekli say\u0131sal de\u011ferler alabilen di\u011fer madde t\u00fcrleridir. Bireysel maddelerin puanlar\u0131n\u0131n toplanarak bir toplam puana (ayr\u0131ca ham puan olarak) d\u00f6n\u00fc\u015ft\u00fcr\u00fclmesi, bir arac\u0131n puanlanmas\u0131 i\u00e7in bir y\u00f6ntemdir ancak tek veya zorunlu olarak en iyi y\u00f6ntem de\u011fildir (Lord ve Novick, 1968; Millsap, 2012). A\u011f\u0131rl\u0131kl\u0131 toplam puanlar ve madde tepki teorisi (MTK; Baker ve Kim, 2004) bir dizi alternatif sunar.<\/p>\n

    Testlerin ve soru formlar\u0131n\u0131n kullan\u0131lmas\u0131, insanlar\u0131 ger\u00e7ek hayatta g\u00f6zlemlemenin ve kendili\u011finden d\u00fc\u015f\u00fcnceleri ifade etmelerini veya ilgilenilen davran\u0131\u015flar\u0131 sergilemelerini beklemenin alternatifine k\u0131yasla hem verimlilik hem de standardizasyon meselesidir (Sijtsma, 2011). \u00d6\u011frenme analiti\u011finde, verilerin verimli bir \u015fekilde toplanmas\u0131 genellikle sorun de\u011fildir ancak standardizasyon eksikli\u011fi \u00f6l\u00e7me hatas\u0131na bir a\u00e7\u0131klama getirmeyi zorla\u015ft\u0131rabilir.<\/p>\n

    \u00d6l\u00e7mede Hata Kayna\u011f\u0131<\/h3>\n

    Tecr\u00fcbelerden biliyoruz ki psikolojik \u00f6l\u00e7meler fiziksel \u00f6l\u00e7meler kadar tutarl\u0131 bir \u015fekilde tekrar edilebilir de\u011fildir. \u0130nsanlar\u0131n bir araca verdi\u011fi cevaplar\u0131n yeteneklerini, tutumlar\u0131n\u0131 veya di\u011fer ilgi alanlar\u0131n\u0131 g\u00fcvenilir bir \u015fekilde yans\u0131tmayabilece\u011fini de biliyoruz. \u0130statistiksel modeller, \u00f6geleri, g\u00f6stergeleri veya testleri \u00f6rt\u00fck bir de\u011fi\u015fkenin rastgele \u00f6rnekleri olarak d\u00fc\u015f\u00fcnmemize izin verir. \u00d6rt\u00fck de\u011fi\u015fken rastgele bir de\u011fi\u015fken olabilir veya ger\u00e7ek puan teorisinde oldu\u011fu gibi sabitlenebilir (Lord ve Novick, 1968). Her iki durumda da \u00f6l\u00e7me numuneleri bazen rastgele hata olarak adland\u0131r\u0131lan ve \u00f6z\u00fcnde i\u00e7sel tekrarlanamazl\u0131ktan kaynaklanan ve yans\u0131z olan hataya sahip olacakt\u0131r (tekrarlanan \u00f6l\u00e7\u00fcmlerin bir miktar\u0131n\u0131n da\u011f\u0131t\u0131m\u0131 \u00fczerine s\u0131f\u0131r beklentisine sahip olma anlam\u0131nda). \u00d6n yarg\u0131l\u0131 sistematik, yanl\u0131 olan bir hata da bulunabilir.<\/p>\n

    Bir \u00f6l\u00e7me \u00e7er\u00e7evesi veya modeli benimsedi\u011fimizde hatayla ilgili daha kesin veya bi\u00e7imsel ifadeler ortaya \u00e7\u0131kar. \u00d6rne\u011fin, ger\u00e7ek puan teorisi ve fakt\u00f6r analizinde, bir arac\u0131n g\u00fcvenilirli\u011fine ili\u015fkin tahminler t\u00fcretmek i\u00e7in paralel testler veya e\u015fde\u011fer formlar a\u00e7\u0131s\u0131ndan ak\u0131l y\u00fcr\u00fctebiliriz. \u00d6l\u00e7\u00fcm hatas\u0131, modelde a\u00e7\u0131kland\u0131\u011f\u0131 gibi verilerdeki yap\u0131ya atfedilmemi\u015f herhangi bir de\u011fi\u015fiklik olarak da tan\u0131mlanabilir (AERA, APA ve NCME, 2014). \u00d6l\u00e7me modelleri konusundaki tart\u0131\u015fmam\u0131z\u0131 bitirdikten sonra hata kaynaklar\u0131n\u0131 tekrar g\u00f6zden ge\u00e7irece\u011fiz.<\/p>\n

    G\u00fcvenilirlik<\/h3>\n

    G\u00fcvenilirlik, bir araca atfedilir ve puanlar\u0131n tutarl\u0131l\u0131\u011f\u0131n\u0131n (AERA, APA ve NCME, 2014), \u00f6zellikle de toplam de\u011fi\u015fkenli\u011fin \u00f6rt\u00fck de\u011fi\u015fkene atfedilen puanlardaki oran\u0131n\u0131n bir \u00f6l\u00e7\u00fcs\u00fcd\u00fcr (DeVellis, 2003). \u00d6rnekleme (ger\u00e7ek puan teorisinde) ve modele ba\u011fl\u0131 (daha karma\u015f\u0131k modellerde) olabilir. Bu kelime bazen, yayg\u0131n olarak Cronbach’s (1951) alfa a olan, [0, 1] aras\u0131nda de\u011fi\u015fen belirli bir g\u00fcvenilirlik katsay\u0131s\u0131 anlam\u0131na gelir. Bununla birlikte, g\u00fcvenilirlik terimi, asl\u0131nda bir korelasyon olan ve test- tekrar test g\u00fcvenilirli\u011fi ve puanlay\u0131c\u0131lar aras\u0131 g\u00fcvenirlik anlam\u0131nda da kullan\u0131lmaktad\u0131r (\u00f6r. Cohen’in kappa, k<\/i><\/span>; Cohen, 1968). Uygulay\u0131c\u0131lar bazen, \u00f6l\u00e7eklerin kullanmak i\u00e7in yeterince iyi oldu\u011funa karar vermek i\u00e7in .70 alt s\u0131n\u0131r a<\/i><\/span> olarak kabul edilebilir (Cortina, 1993) de\u011ferlere dair y\u00f6nergelere sorgulamadan ba\u011fl\u0131 kal\u0131rlar. Ancak istatistiksel g\u00fcc\u00fcn a<\/i><\/span>‘n\u0131n daha y\u00fcksek de\u011ferlerle artt\u0131\u011f\u0131na dikkat edilmelidir (DeVellis, 2003). Bu nedenle, bir \u00f6l\u00e7e\u011fin g\u00fcvenilirli\u011fini geli\u015ftirme \u00e7abas\u0131, daha b\u00fcy\u00fck \u00f6rneklemeler al\u0131nmas\u0131n\u0131n faydalar\u0131ndan a\u011f\u0131r basabilir.<\/p>\n

    Ge\u00e7erlik<\/h3>\n

    Ge\u00e7erlilik, Standartlar\u0131n\u0131n<\/i><\/span> ilk b\u00f6l\u00fcm\u00fc \u201cGe\u00e7erlilik, kan\u0131tlar\u0131n ve teorinin, testlerin \u00f6nerilen kullan\u0131m\u0131 i\u00e7in test puanlar\u0131n\u0131n yorumlanmas\u0131n\u0131 destekleme derecesini belirtir. ….testin ge\u00e7erli\u011fi” \u015feklindeki niteliksiz ifadeyi kullanmak do\u011fru de\u011fildir. ” (s.11) olarak ba\u015flayan standartlar’\u0131n en \u00f6nemli konusudur. Daha geni\u015f olan \u201c\u00f6l\u00e7\u00fc\u201d terimini daha dar olan \u201ctest\u201din yerine ge\u00e7irme sayesinde, ge\u00e7erlili\u011fin \u00f6\u011frenme analiti\u011fi i\u00e7in ne b\u00fcy\u00fck \u00f6nem ta\u015f\u0131d\u0131\u011f\u0131 a\u00e7\u0131k\u00e7a g\u00f6r\u00fclmelidir. Standartlar<\/i><\/span>‘da do\u011frulama arg\u00fcmanlar\u0131nda kullan\u0131lan dili \u015fekillendirmeye dair Messick’in (1995), Cronbach ve Meehl’i (1955) etkili bir \u015fekilde elden ge\u00e7irmesinde de belirgin olan bir yakla\u015f\u0131m olarak (bk. Ayr\u0131ca Kane, 2001) dili \u015fekillendirmeye dair somut bir odaklanma vard\u0131r. Ge\u00e7erli\u011fe ili\u015fkin kan\u0131t t\u00fcrleri (\u201cge\u00e7erlilik t\u00fcrleri\u201dnden ziyade), tepki s\u00fcre\u00e7leri hakk\u0131ndaki kan\u0131t, arac\u0131n i\u00e7 yap\u0131s\u0131 hakk\u0131ndaki kan\u0131t, yak\u0131nsak ve ay\u0131r\u0131c\u0131 kan\u0131t, kriter referanslar\u0131 (\u00f6ng\u00f6r\u00fclen kriterler d\u00e2hil) ve genellenebilirlik hakk\u0131nda kan\u0131t i\u00e7erir.<\/p>\n

    Bu b\u00f6l\u00fcm\u00fcn ba\u015flar\u0131nda, anketlere verilen cevaplar\u0131n d\u00fcr\u00fcst d\u00fc\u015f\u00fcncelere ve duygulara kar\u015f\u0131l\u0131k geldi\u011fi varsay\u0131m\u0131na de\u011finmi\u015ftik. Bununla birlikte, tepki yanl\u0131l\u0131\u011f\u0131 t\u00fcrleri hakk\u0131nda kabul edilebilme yanl\u0131l\u0131\u011f\u0131ndan (evet diyerek; Messick ve Jackson, 1961) sosyal istenirlik yanl\u0131l\u0131\u011f\u0131na (ayr\u0131ca, iyiyi oynama; Nederhof, 1985) a\u015f\u0131r\u0131 ve \u0131l\u0131ml\u0131 cevaplay\u0131c\u0131 yanl\u0131l\u0131\u011f\u0131na (yani, Likert-skalalar\u0131n\u0131n a\u015f\u0131r\u0131 u\u00e7lar\u0131n\u0131 se\u00e7me e\u011filiminde olan insanlar) geni\u015f bir literat\u00fcr bulunmaktad\u0131r. (Bachman ve O’Malley, 1984). Hile yapmaya istekli olma, cinsel fanteziler veya \u0131rkla ilgili tutumlar gibi hassas konular hakk\u0131ndaki soru formlar\u0131 ve anketler i\u00e7in daha s\u0131k belgelenmesine ra\u011fmen, Newtoncu d\u00fc\u015f\u00fcnceyi de\u011ferlendirmek i\u00e7in kullan\u0131lan kuvvet kavram\u0131 envanteri gibi (KKE; Hestenes, Wells ve Swackhamer, 1992) cevaplar\u0131n \u00f6z uyumlanma ve oto sans\u00fcr\u00fc de e\u011fitsel testlerde ger\u00e7ekle\u015febilir. Mazur (2007), \u00f6zellikle \u201cBu sorulara nas\u0131l cevap vermeliyim? diye soran bir \u00f6\u011frenci oldu\u011funu bildirmi\u015ftir. Bize \u00f6\u011frettiklerinize g\u00f6re, ya da bu \u015feylerle ilgili olarak benim d\u00fc\u015f\u00fcnd\u00fcklerim gibi mi?\u201d sorusunu soran bir \u00f6\u011freneni bildirmi\u015ftir. Son olarak, kas\u0131tl\u0131 h\u0131zl\u0131 tahmin etme davran\u0131\u015f\u0131 bir cevap yanl\u0131l\u0131\u011f\u0131 bi\u00e7imi olarak d\u00fc\u015f\u00fcn\u00fclebilir (Wise ve Kong, 2005). T\u00fcm bu cevap yanl\u0131l\u0131\u011f\u0131n\u0131n kaynaklar\u0131 \u00f6l\u00e7ek puanlar\u0131n\u0131n ele\u015ftirel olmayan yorumlar\u0131na meydan okudu\u011fu bilinmelidir.<\/p>\n

    \u00d6l\u00e7me Modelleri<\/h3>\n

    Bu s\u00fcrecin en zorlu k\u0131sm\u0131 \u00f6l\u00e7me modellerinin teknik detaylar\u0131ndad\u0131r. \u00d6l\u00e7me modeli, \u00f6rt\u00fck bir de\u011fi\u015fken veya de\u011fi\u015fken k\u00fcmesi ile g\u00f6zlemlenebilir bir de\u011fi\u015fken veya de\u011fi\u015fken k\u00fcmesi aras\u0131ndaki resm\u00ee bir matematiksel ili\u015fkidir. Tamamen istatistiksel bir \u00f6l\u00e7me modeli \u00f6rt\u00fck de\u011fi\u015fken(ler) i\u00e7in bir da\u011f\u0131l\u0131m, g\u00f6zlenen de\u011fi\u015fken(ler) i\u00e7in bir da\u011f\u0131l\u0131m ve aralar\u0131ndaki fonksiyonel bir ili\u015fkiyi belirtebilir. \u00d6rt\u00fck de\u011fi\u015fkenler \u00e7o\u011fu zaman hatalara tabi olan g\u00f6zlemleri nedensel olarak<\/i><\/span> a\u00e7\u0131klayanlar \u015feklinde anla\u015f\u0131lmaktad\u0131r. Rasgele de\u011fi\u015fkenlerin varyanslar\u0131 ve kovaryanslar\u0131, modelde a\u00e7\u0131k\u00e7a veya \u00f6rt\u00fck olarak a\u00e7\u0131klanm\u0131\u015ft\u0131r. Modeller, \u00f6rne\u011fin yap\u0131 ile \u00f6l\u00e7\u00fc aras\u0131ndaki ili\u015fkinin monotonluk (veya daha kat\u0131, do\u011frusall\u0131k) varsay\u0131m\u0131 ya da tekil \u00f6gelerin hata terimleri aras\u0131nda s\u0131f\u0131r kovaryans varsay\u0131m\u0131 yapar. Bir modelin varsay\u0131mlar\u0131 ihlal edilirse, model kullan\u0131larak yap\u0131lan \u00e7\u0131kar\u0131mlar yanl\u0131\u015f olabilir (Lord ve Novick, 1968).<\/p>\n

    Kategorik ve s\u00fcrekli de\u011fi\u015fkenler farkl\u0131 istatistiksel y\u00f6ntemler i\u00e7erdi\u011finden, \u00f6l\u00e7me modelleri t\u00fcrleri bazen Tablo 3.1’de g\u00f6sterildi\u011fi gibi \u00f6rt\u00fck ve g\u00f6zlenen de\u011fi\u015fkenlerin t\u00fcr\u00fcne g\u00f6re aileler olarak tasnif edilir. Bu tasnif ayr\u0131nt\u0131l\u0131 de\u011fildir, \u00e7\u00fcnk\u00fc hibrit modellerin yan\u0131 s\u0131ra bu model ailelerin \u00f6zel durumlar haline geldi\u011fi genelle\u015ftirilmi\u015f \u00e7at\u0131lar (Skrondal ve Rabe-Hesketh, 2004) da vard\u0131r. B\u00fcy\u00fcme modelleri, \u00f6l\u00e7me modellerinin tekrarlanan \u00f6l\u00e7\u00fcmlere geni\u015fletilmesidir ve hem s\u00fcrekli hem de kategorik \u00f6rt\u00fck de\u011fi\u015fkenlere uygulanabilir (\u00f6r. Meredith ve Tisak, 1990; Rabiner, 1989; Raudenbush ve Bryk, 2002).<\/p>\n

    <\/a><\/a><\/a><\/a><\/a> Tablo 3.1. Gizli De\u011fi\u015fken Modellerin Aileleri<\/span><\/i><\/span><\/p>\n\n\n\n\n<\/colgroup>\n\n\n\n\n
    \n

    \u00d6rt\u00fck\/ G\u00f6zlenen<\/b><\/p>\n<\/td>\n

    		\n

    G\u00f6zlenen s\u00fcrekli<\/b><\/p>\n<\/td>\n

    		\n

    G\u00f6zlenen kategorik<\/b><\/p>\n<\/td>\n<\/tr>\n

    \u00d6rt\u00fck s\u00fcrekli<\/span><\/td>\nFakt\u00f6r modelleri (Bollen, 1989; Mulaik, 2009)<\/span><\/td>\nMadde tepki modelleri (Lord ve Novick, 1968; Baker ve Kim, 2004)<\/span><\/td>\n<\/tr>\n
    \u00d6rt\u00fck kategorik<\/span><\/td>\n\u00d6rt\u00fck kar\u0131\u015f\u0131m modelleri (McLachlan ve Peel, 2004)<\/span><\/td>\n\u00d6rt\u00fck s\u0131n\u0131f modelleri (Goodman, 2002)<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    \u00d6\u011eRENME ANAL\u0130T\u0130\u011e\u0130NDE \u00d6L\u00c7ME MODELLER\u0130N\u0130N \u00d6ZEL KULLANIMI<\/h2>\n

    Daha \u00f6nce, psikolojik ve e\u011fitsel \u00f6l\u00e7menin, s\u0131n\u0131fland\u0131rma, tan\u0131lama, s\u0131ralama, yerle\u015ftirme ve bireylerin belgelendirilmesinin yan\u0131 s\u0131ra gruplara dair uygun \u00e7\u0131kar\u0131mlar d\u00e2hil olmak \u00fczere \u00e7e\u015fitli ama\u00e7lar i\u00e7in kullan\u0131ld\u0131\u011f\u0131n\u0131 belirtmi\u015ftik. \u00d6\u011frenme analiti\u011fi ve e\u011fitsel veri madencili\u011fi alan\u0131ndaki \u00e7al\u0131\u015fmalar, dijital \u00f6\u011frenme ortamlar\u0131ndaki psikolojik \u00f6l\u00e7ekler, davran\u0131\u015flar ve performans aras\u0131ndaki karma\u015f\u0131k ili\u015fki a\u011f\u0131n\u0131 da ara\u015ft\u0131rmaktad\u0131r (Tempelaar, Rienties ve Giesbers, 2015). Bu teman\u0131n amac\u0131, modeller ve bunlar\u0131n analitik ve e\u011fitsel veri madencili\u011fini \u00f6\u011frenmedeki kullan\u0131mlar\u0131 hakk\u0131nda biraz daha derinlik sa\u011flamakt\u0131r. T\u00fcm konular e\u015fit \u00f6l\u00e7\u00fcde ele al\u0131nmaz, bu da hem alan k\u0131s\u0131tlamalar\u0131n\u0131 hem de se\u00e7im yanl\u0131l\u0131\u011f\u0131n\u0131 yans\u0131t\u0131r.<\/p>\n

    Fakt\u00f6r analizi<\/h3>\n

    Fakt\u00f6r analizi (Mulaik, 2009), g\u00f6zlenen de\u011fi\u015fkenler aras\u0131ndaki korelasyonlar\u0131, fakt\u00f6r olarak bilinen bir dizi \u00f6rt\u00fck de\u011fi\u015fkenle do\u011frusal bir ili\u015fki yoluyla modellemektedir. Orijinal tek fakt\u00f6rl\u00fc model Spearman’\u0131n (1904) genel zek\u00e2 g<\/i><\/span> modelidir, ilgisiz konu testlerindeki puanlar aras\u0131ndaki ili\u015fkileri a\u00e7\u0131klamak i\u00e7in kullan\u0131l\u0131r. Klasik test teorisi (Lord ve Novick, 1968) olarak da bilinen ger\u00e7ek puan teorisi, t\u00fcm madde fakt\u00f6r\u00fc y\u00fcklerinin ayn\u0131 oldu\u011fu tek fakt\u00f6r modelinin \u00f6zel bir hali olarak elde edilebilir. Thurstone (1947), \u00e7oklu (yedi) fakt\u00f6r zek\u00e2 modelini geli\u015ftirdi.<\/p>\n

    Fakt\u00f6r analizi, genellikle iki te\u015febb\u00fcse b\u00f6l\u00fcnm\u00fc\u015ft\u00fcr. A\u00e7\u0131mlay\u0131c\u0131 fakt\u00f6r analizi (AFA), g\u00fc\u00e7l\u00fc teorik varsay\u0131mlar olmadan verilerdeki \u00f6rt\u00fck fakt\u00f6rlerin say\u0131s\u0131n\u0131 belirlemek i\u00e7in kullan\u0131l\u0131r ve genellikle \u00f6l\u00e7ek geli\u015ftirmenin bir par\u00e7as\u0131d\u0131r. Bununla birlikte, AFA, e\u011fer zay\u0131f yap\u0131l\u0131rsa sorunlu sonu\u00e7lara yol a\u00e7abilecek birka\u00e7 \u00f6nemli metodolojik karar gerektirir (Fabrigar, Wegener, MacCallum ve Strahan, 1999). Fabrigar vd. (1999), AFA’n\u0131n, ger\u00e7ek fakt\u00f6r yap\u0131s\u0131 hakk\u0131nda hatal\u0131 \u00e7\u0131kar\u0131mlara yol a\u00e7abilecek, bir boyutsall\u0131k azaltma tekni\u011fi olan temel bile\u015fenler analizi (TBA) ile kar\u0131\u015ft\u0131r\u0131lmamas\u0131 konusunda uyar\u0131larda bulundu. Do\u011frulay\u0131c\u0131 fakt\u00f6r analizi (DFA), beklenen ve g\u00f6zlemlenen korelasyonlar aras\u0131ndaki kal\u0131nt\u0131lar\u0131 inceleyerek teorik olarak \u00f6nerilen bir fakt\u00f6r modelini test etmek i\u00e7in yap\u0131lm\u0131\u015f tamamlay\u0131c\u0131 teknikler setidir. B\u00f6ylece, bir modeli reddetmek i\u00e7in DFA kullan\u0131labilir. DFA, yol \u00e7\u00f6z\u00fcmlemesi ve gizli b\u00fcy\u00fcme modelleri ile birlikte, yap\u0131sal e\u015fitlik modellemesi ile g\u00fcvence alt\u0131na al\u0131nm\u0131\u015ft\u0131r (SEM; Bollen, 1989; Kline, 2010). Do\u011frulay\u0131c\u0131 fakt\u00f6r analizi, durum ikincisini ger\u00e7ekle\u015ftirmek i\u00e7in yap\u0131lm\u0131\u015f olmas\u0131na ra\u011fmen AFA’n\u0131n farkl\u0131 pop\u00fclasyon \u00f6rnekleriyle birden fazla kez \u00e7al\u0131\u015ft\u0131r\u0131lmas\u0131 ile ayn\u0131 \u015fey de\u011fildir. (DeVellis, 2003).<\/p>\n

    Baz\u0131 \u00f6\u011frenme analiti\u011fi ara\u015ft\u0131rmalar\u0131, \u00f6l\u00e7ek geli\u015ftirme ve bunun \u00f6\u011frenme y\u00f6netimi sistemlerinden toplanan verilerle birle\u015ftirilmesi ile do\u011frudan ilgilidir (\u00f6r. Buckingham Shum ve Deakin Crick, 2012; Milligan ve Griffin, 2016). Di\u011fer \u00e7al\u0131\u015fmalar, ba\u015far\u0131 \u00f6l\u00e7ekleri (Pekrun vd., 2011) ile y\u00fcz y\u00fcze ve \u00e7evrimi\u00e7i e\u011fitim (Tempelaar, Niculescu, Rienties, Giesbers ve Gijselaers, 2012) aras\u0131ndaki ili\u015fki gibi mevcut \u00f6l\u00e7ekler ve sonu\u00e7 \u00f6l\u00e7\u00fcmleri veya motivasyon \u00f6nlemleri ile kitlesel a\u00e7\u0131k \u00e7evrimi\u00e7i dersin tamamlanmas\u0131 aras\u0131ndaki ili\u015fkilere odaklanmaktad\u0131r (Wang ve Baker, 2015). Bir arac\u0131 veya \u00f6zellikle bir arac\u0131n bir b\u00f6l\u00fcm\u00fcn\u00fc yeni ama\u00e7lar i\u00e7in uyarlarken, uygulay\u0131c\u0131lar bu yeni kullan\u0131mlar\u0131n yeni do\u011frulama arg\u00fcmanlar\u0131na de\u011fip de\u011fmedi\u011fi konusunda dikkatli olmal\u0131d\u0131r.<\/p>\n

    \u00d6rt\u00fck S\u0131n\u0131f ve \u00d6rt\u00fck Kar\u0131\u015f\u0131m Modelleri<\/h3>\n

    Dedic, Rosenfeld ve Lasry (2010), \u00f6\u011frencilerin bir fizik kavram\u0131 testindeki yanl\u0131\u015f cevaplar\u0131na dayanarak fizik kavram yan\u0131lg\u0131lar\u0131n\u0131n da\u011f\u0131l\u0131m\u0131n\u0131 anlamak i\u00e7in \u00f6rt\u00fck s\u0131n\u0131f analizini kullanm\u0131\u015ft\u0131r. Veriler, bir fizik kursu \u00f6ncesi ve sonras\u0131ndaki uygulamalardan gelmi\u015ftir. (\u00f6n ve son test). Yazarlar, kesikli bask\u0131nl\u0131k yan\u0131lma s\u0131n\u0131flar\u0131 arac\u0131l\u0131\u011f\u0131yla, Aristotelist’ ten Newtoncu d\u00fc\u015f\u00fcnceye kadar bariz bir ilerleme tespit etmi\u015ftir. Belgelerin konu modellemesi i\u00e7in yayg\u0131n olarak kullan\u0131lan bir y\u00f6ntem olan gizli Dirichlet tahsisi (GDT; Blei, Ng ve Jordan, 2003; ayr\u0131ca, bu ciltte birka\u00e7 b\u00f6l\u00fcme bak\u0131n\u0131z) \u00f6rt\u00fck bir kar\u0131\u015f\u0131m modelidir. Kar\u0131\u015f\u0131k \u00fcyelik modelleri (Erosheva, Fienberg ve Lafferty, 2004), bir bireyin birden fazla s\u0131n\u0131fa \u201cbelirsiz\u201d veya a\u011f\u0131rl\u0131kl\u0131 olarak atanmas\u0131na izin vererek \u00f6rt\u00fck kar\u0131\u015f\u0131mlar\u0131 daha da genellemektedir. Gauss kar\u0131\u015f\u0131m modeli, KA\u00c7D \u00f6\u011frenenlerin performans g\u00fczerg\u00e2hlar\u0131na uygulanan model tabanl\u0131 k\u00fcmeleme analizi (Fraley ve Raftery, 1998) i\u00e7in temel olu\u015fturmaktad\u0131r (Bergner, Kerr ve Pritchard, 2015). Bununla birlikte, k\u00fcmeleme algoritmalar\u0131n\u0131n hepsinin \u00f6rt\u00fck kar\u0131\u015f\u0131m modeli olmad\u0131\u011f\u0131 unutulmamal\u0131d\u0131r.<\/p>\n

    Madde Tepki Kuram\u0131 (MTK)<\/h3>\n

    Madde tepki kuram\u0131, klasik test teorisinde oldu\u011fu gibi, toplam test puanlar\u0131ndan ziyade bireysel ki\u015filik-madde etkile\u015fimlerini modelleyerek, test teorisinin tarihsel geli\u015fiminde kendisine ayr\u0131 bir yer edinmi\u015ftir. Kavramsal olarak, MTK’nin amac\u0131 \u201cmaddeleri, madde parametrelerine g\u00f6re ve s\u0131nava girenleri, inceleme parametrelerine g\u00f6re; benzer s\u0131nava girenler daha \u00f6nce benzer maddeleri hi\u00e7 cevaplamad\u0131ysa bile herhangi bir s\u0131nava girenin herhangi bir maddeye cevab\u0131n\u0131 olas\u0131l\u0131\u011fa dayal\u0131 olarak tahmin edebilecek \u015fekilde tan\u0131mlamakt\u0131r” (Lord, 1980, s. 11). \u0130ki bile\u015fenli bir madde i\u00e7in (\u00f6r. do\u011fru \/ yanl\u0131\u015f, ayn\u0131 fikirde \/ kat\u0131lm\u0131yorum, vb.) bir \u00f6rnek madde karakteristik e\u011frisi (MKE) veya e\u015fde\u011ferde madde tepki fonksiyonu (MTF), \u015eekil 3.1’de g\u00f6sterilmektedir.<\/p>\n

    \"\"<\/p>\n

    <\/a> \u015eekil 3.1. Bir \u00f6rnek madde karakteristik e\u011frisi (MKE). Noktal\u0131 \u00e7izgiler P = 0.5 kesi\u015fimini g\u00f6sterir.<\/i><\/span><\/p>\n

    \u015eekil 3.1’in belirgin \u00f6zellikleri a\u015fa\u011f\u0131daki gibidir:<\/p>\n

      \n
    1. \n

      \u00d6zellik (\u00f6r. yetenek) s\u00fcrekli rastgele bir de\u011fi\u015fken olarak \u00f6l\u00e7\u00fcl\u00fcr ve yatay eksende <\/span>\u04e8<\/i><\/span> ile temsil edilir. De\u011fi\u015fken, ilgilenilen pop\u00fclasyonda ortalama s\u0131f\u0131ra ve 1 varyans\u0131na sahip olacak \u015fekilde standardize edilmi\u015ftir. Daha y\u00fcksek bir \u04e8<\/i><\/span> de\u011ferine kar\u015f\u0131l\u0131k gelen \u00f6zellikten daha fazlas\u0131n\u0131n, pozitif (veya do\u011fru) bir cevab\u0131n P<\/i><\/span> olas\u0131l\u0131\u011f\u0131n\u0131 artt\u0131rmas\u0131 beklenir. G\u00f6zlenen bir monotonluk ihlali<\/i><\/span>, temel \u015fah\u0131s-madde ili\u015fkisinin yanl\u0131\u015f oldu\u011fu ve teste maddenin d\u00e2hil edilmesinin k\u00f6t\u00fc bir uyum olaca\u011f\u0131 ve g\u00fcvenilmez \u00e7\u0131kar\u0131mlara yol a\u00e7aca\u011f\u0131 anlam\u0131na gelir.<\/p>\n<\/li>\n

    2. \n

      Bu e\u011frileri yorumlaman\u0131n iki yolu Holland (1990) taraf\u0131ndan tan\u0131mlanm\u0131\u015ft\u0131r. Stokastik denek yorumunda, ki\u015fi bu e\u011frinin, performans\u0131 \u00f6ng\u00f6r\u00fclemeyen bir bireye uyguland\u0131\u011f\u0131n\u0131 d\u00fc\u015f\u00fcn\u00fcr. Holland’\u0131 anlamsal olarak al\u0131nt\u0131layacak olursak, stokastik denek a\u00e7\u0131klamas\u0131 sezgiseldir ancak tamamen tatmin edici de\u011fildir; \u00f6znenin stokastik do\u011fas\u0131 i\u00e7in mekanik bir a\u00e7\u0131klamam\u0131z yoktur. \u00d6te yandan, rastgele \u00f6rneklem yorumunda, bu e\u011fri, s\u0131nava girenlerin \u00f6rneklem pop\u00fclasyonuna uyguland\u0131\u011f\u0131nda anlaml\u0131d\u0131r. \u00d6rne\u011fin, belirli bir yetenek aral\u0131\u011f\u0131ndaki s\u0131nava girenler aras\u0131nda, baz\u0131 oranlar do\u011fru cevap verecektir. \u015eekildeki noktalar ve hata \u00e7ubuklar\u0131 bu g\u00f6zlemi yans\u0131tmaktad\u0131r.2<\/sup><\/a><\/span><\/p>\n<\/li>\n

    3. \n

      P<\/i><\/span> = 0.5 olan \u04e8<\/i><\/span> de\u011feri, bir bili\u015fsel yetenek test maddesi i\u00e7in zorluk olarak adland\u0131r\u0131lan bir referans kesi\u015fimidir. Zorlu\u011fun, yetenek ile ipso facto<\/i><\/span> (kendili\u011finden) ayn\u0131 \u00f6l\u00e7ekte oldu\u011funu ve b\u00f6ylece bir ki\u015finin yetene\u011fi ile bir maddenin zorlu\u011fu aras\u0131ndaki fark hakk\u0131nda konu\u015fman\u0131n mant\u0131kl\u0131 olabilece\u011fi unutulmamal\u0131d\u0131r.<\/p>\n<\/li>\n

    4. \n

      Olas\u0131l\u0131k ba\u011flant\u0131s\u0131n\u0131n \u015fekli, bireyin <\/span>\u04e8<\/i><\/span>i<\/sub> \u00f6zelli\u011fi i<\/i><\/span> ve j maddesi i\u00e7in bir dizi B<\/i><\/span><\/span>j<\/sub>, madde parametreleri y\u00f6n\u00fcnden genellikle parametriktir,
      \nP<\/i><\/span>ij<\/sub> = <\/span>P<\/i><\/span>(<\/span>X<\/i><\/span>ij<\/sub> = 1|<\/span>\u03b8<\/i><\/span><\/span>i<\/sub>, <\/span>\u03b2<\/i><\/span><\/span>j<\/sub>) = <\/span>f<\/i><\/span>(<\/span>\u03b8<\/i><\/span><\/span>i<\/sub>, <\/span>\u03b2<\/i><\/span><\/span>j<\/sub>), (1)
      \n<\/span>Rasch modelinde (bir tekil zorluk parametresi) veya iki parametreli lojistik (2PL) modelinde oldu\u011fu gibi. 2PL modeli, \u015eekil 3.2’de g\u00f6sterilmektedir; verilere uygunluk g\u00f6zle g\u00f6r\u00fcl\u00fcr derecede iyi ve \u00f6rt\u00fc\u015fme d\u00fczeyi testi G<\/i><\/span>2<\/span><\/sup> ayn\u0131 miktardad\u0131r. Parametrik olmayan MTK y\u00f6ntemlerinin var oldu\u011fu belirtilmelidir (Sijtsma, 1998).<\/p>\n<\/li>\n<\/ol>\n

      Bir ki\u015fi bir \u00f6l\u00e7me arac\u0131nda birka\u00e7 maddeye cevap verdi\u011finde, buradaki fikir, \u00f6zelli\u011fin sonsal tahminlerini yapmak i\u00e7in cevap bilgilerini birle\u015ftirmektir. Bir cevap vekt\u00f6r\u00fcn\u00fcn, bireysel madde seviyesindeki olas\u0131l\u0131klar\u0131n bir \u00fcr\u00fcn\u00fcne \u00e7arpan olabilirlik durumu i\u00e7in, cevaplar, niteli\u011fe ba\u011fl\u0131 olarak aksi takdirde ba\u011f\u0131ms\u0131z olmal\u0131d\u0131r. Bu ko\u015fullu ba\u011f\u0131ms\u0131zl\u0131k varsay\u0131m\u0131<\/i><\/span> maddeler aras\u0131 ba\u011f\u0131ml\u0131l\u0131\u011f\u0131 (\u00f6r. Rijmen, 2010) a\u00e7\u0131klayan ek fakt\u00f6rlerin takdimini gerektirebilir.<\/p>\n

      MTK’nin y\u00fcksek riskli test uygulamalar\u0131 d\u0131\u015f\u0131nda e\u011fitimde bir miktar \u00e7ekim g\u00fcc\u00fcne sahip oldu\u011funa dair kan\u0131tlar fizik e\u011fitimi ara\u015ft\u0131rma uygulamalar\u0131nda kuvvet kavram\u0131 envanterine (KKE; Hestenes vd., 1992) ve temel mekanik bilgi testine bak\u0131larak bulunabilir (MBT; Hestenes ve Wells, 1992). Bu ara\u00e7lar yirmi be\u015f y\u0131ld\u0131r kullan\u0131lmakta iken, madde tepki modeli analizleri daha yak\u0131n zamanda ortaya \u00e7\u0131kmaya ba\u015flam\u0131\u015ft\u0131r (Morris vd., 2006; Wang ve Bao, 2010). KKE i\u00e7in model-veri uygunlu\u011fu genel olarak kabul edilebilir durumdayd\u0131. Bununla birlikte Cardamone vd. (2011), MBT’de, madde tepki fonksiyonlar\u0131n\u0131 inceleyerek, k\u00f6t\u00fc \u00e7al\u0131\u015fan iki maddeyi ke\u015ffetmi\u015ftir. \u015eekil 3.2’de g\u00f6sterilmi\u015ftir.\"\"<\/p>\n

      <\/a> \u015eekil 3.2. Mekanik bilgi testinden (MBT) k\u00f6t\u00fc uyum sa\u011flayan bir madde.<\/span><\/i><\/span><\/p>\n

      D\u00fc\u015f\u00fck yetenekli \u00f6\u011frencilerin ortalama yetenekli \u00f6\u011frencilere g\u00f6re bir maddeye do\u011fru cevap vermeleri daha muhtemel ise, burada \u015f\u00fcpheli bir durum vard\u0131r. Daha detayl\u0131 bir inceleme ile bu test maddesindeki mu\u011flak ifadelerin \u00f6\u011frencilerin alg\u0131lar\u0131n\u0131 hatal\u0131 y\u00f6nlendirdi\u011fi ve yanl\u0131\u015fl\u0131kla do\u011fru cevab\u0131 bulmalar\u0131n\u0131 sa\u011flad\u0131\u011f\u0131 tespit edilmi\u015ftir. Bu durumda, iki yanl\u0131\u015f bir do\u011fru yapm\u0131\u015f oldu.<\/p>\n

      Birden fazla boyut tan\u0131mlayan KKE’nin a\u00e7\u0131mlay\u0131c\u0131 fakt\u00f6r analizlerini takiben (Ding ve Beichner, 2009; Scott, Schumayer ve Gray, 2012), MBT’ye \u00e7ok boyutlu MTK’nin bir varyasyonu uyguland\u0131 (Bergner, Rayyan, Seaton ve Pritchard, 2013). Madde tepki kuram\u0131 modelleri de \u00e7evrimi\u00e7i \u00f6devlerde s\u0131k\u00e7a g\u00f6r\u00fclen bir kolayl\u0131k olan, birden fazla cevap verme giri\u015fimlerinin (do\u011fru olana dek cevaplama) ard\u0131ndaki kendinden s\u0131ral\u0131 s\u00fcrece geni\u015fletildi (Attali, 2011; Bergner, Colvin ve Pritchard, 2015; 2014).<\/p>\n

      B\u00fcy\u00fcme Modelleri<\/h3>\n

      B\u00fcy\u00fcme modelleri, \u00f6rt\u00fck bir \u00f6zelli\u011fin \u00f6l\u00e7\u00fcmler aras\u0131nda sistematik olarak de\u011fi\u015fti\u011fi herhangi bir anda uygulan\u0131r. De\u011fi\u015fen tutumlara uygulanabilirler (\u00f6r. George, 2000), fakat biz burada bili\u015fsel yetenek alanlar\u0131 uygulamas\u0131na odaklan\u0131yoruz. \u00d6\u011fretim program\u0131 d\u00fczenleme \u00f6\u011freticilerinden3<\/sup><\/a> ay\u0131rt edilen ak\u0131ll\u0131 problem \u00e7\u00f6zme \u00f6\u011freticileri i\u00e7in \u00f6\u011frenci modellerine dair e\u011fitsel veri madencili\u011finde kapsaml\u0131 bir literat\u00fcr bulunmaktad\u0131r (Desmarais ve Baker, 2011).<\/p>\n

      Matematik i\u00e7in bili\u015fsel \u00f6\u011freticilerde (Anderson, Corbett, Koedinger ve Pelletier, 1995), uygulama madde dizilimleri, bili\u015fsel bir modele g\u00f6re ince taneli bilgi bile\u015fenlerinin tam \u00f6\u011frenilmesini desteklemek i\u00e7in tasarlanm\u0131\u015ft\u0131r. Bu sistemlerde verilerde ustal\u0131\u011fa do\u011fru b\u00fcy\u00fcmeyi modelleme amac\u0131 olan iki yakla\u015f\u0131mdan biri Bayesci bilgi takibi (BBT; Corbett ve Anderson, 1995) ve toplamsal fakt\u00f6r modelleridir; Cen, Koedinger ve Junker, 2008; Draney, Pirolli ve Wilson, 1995). \u00d6\u011frenme e\u011frileri analizi (Kaser, Koedinger ve Gross, 2014; Martin’in, Mitrovic, Mathan ve Koedinger, 2010), verilerle \u00f6\u011freticinin temelindeki bili\u015fsel model ve veri aras\u0131ndaki uyu\u015fmazl\u0131klar\u0131 kontrol etmek i\u00e7in de kullan\u0131lm\u0131\u015ft\u0131r.<\/p>\n

      \u201cUygulama Yasas\u0131\u201d na g\u00f6re (Newell ve Rosenbloom, 1981), B ve a<\/i><\/span>‘ n\u0131n deneysel olarak belirlendi\u011fi “T = Bn<\/sub>-a<\/sup>, “g\u00fc\u00e7 yasas\u0131na g\u00f6re”, n<\/i><\/span> uygulama f\u0131rsat\u0131n\u0131n bir fonksiyonu olarak toplam hata oran\u0131 T azalmal\u0131d\u0131r. Veri ve model aras\u0131ndaki uyum, \u00f6rne\u011fin r kare \u00f6l\u00e7\u00fcmleri kullanmak, bilgi e\u015flemedeki geli\u015fmeleri motive edebilir. Bu hatal\u0131 bir maddenin tespit edildi\u011fi \u015eekil 3.2’deki madde analizine benze\u015fik olarak g\u00f6r\u00fclebilir. Bununla birlikte, bu durumda, bir bilgi bile\u015fenine bir madde diziliminin atanmas\u0131 hatal\u0131 olarak g\u00f6r\u00fclmektedir.<\/p>\n

      BBT’de \u00f6rt\u00fck de\u011fi\u015fken, bir i\u015flemsel bilgi bile\u015feninin ustal\u0131\u011f\u0131d\u0131r ve ikili de\u011fere sahiptir, M<\/i><\/span> \u2208<\/span> {0, 1}. Herhangi bir f\u0131rsatta ustal\u0131k ve do\u011fruluk X<\/i><\/span> \u2208<\/span> {0, 1} aras\u0131ndaki ba\u011flant\u0131s\u0131 ancak Eq ile benze\u015fik olarak bir 2×2 ko\u015fullu olas\u0131l\u0131k tablosudur. (1) tahmin (g<\/i><\/span>) ve kayma (s<\/i><\/span>) parametreleri y\u00f6n\u00fcnden \u015f\u00f6yle yaz\u0131labilir,<\/p>\n

      P(X = 1|M) = (1 – s)M<\/sup> g(1-M)<\/sup> (2)<\/p>\n

      \u00d6nemli bi\u00e7imde, giri\u015fimler ba\u011f\u0131ms\u0131z olarak g\u00f6r\u00fclmemektedir. Aksine, BBT’deki kilit fikir, \u00f6\u011frencilerin kurallara g\u00f6re her bir uygulama durumunda \u00f6n bir ustal\u0131k olas\u0131l\u0131\u011f\u0131 ile ba\u015flamalar\u0131 ve ustal\u0131\u011fa do\u011fru hareket ediyor (\u00f6\u011frenirler) olmalar\u0131d\u0131r,
      \nP<\/i><\/span>(M<\/i><\/span>n<\/span><\/sub>) = P(M<\/i><\/span>n-1<\/span><\/sub>) + t(1 – P(M<\/i><\/span>n-1<\/span><\/sub>))<\/i><\/span> (3)
      \nBurada t<\/i><\/span> bir b\u00fcy\u00fcme parametresidir. Son zamanlarda, van de Sande (2013) BBT’nin uygulama giri\u015fimleri ve hata oranlar\u0131 aras\u0131nda bir g\u00fc\u00e7 yasas\u0131 ili\u015fkisinden ziyade bir \u00fcstellik belirtti\u011fini g\u00f6stermi\u015ftir. Bu BBT’yi uygulaman\u0131n g\u00fc\u00e7 yasas\u0131n\u0131 sa\u011flayan veriler i\u00e7in yanl\u0131\u015f tan\u0131mlanm\u0131\u015f bir model yapar. Aksine, toplamsal fakt\u00f6r modeli, uygulama paradigmas\u0131n\u0131n g\u00fc\u00e7 yasas\u0131na uyacak \u015fekilde tasarlanm\u0131\u015ft\u0131r. Kaser vd. (2014) BBT’nin kestirimsel keskinli\u011finin TFM’den ay\u0131rt edilemez oldu\u011funu g\u00f6sterdi. \u0130kincisinin uyumu ile ilgili olarak, toplam kal\u0131nt\u0131 analizlerinde sistematik yanl\u0131l\u0131\u011fa dikkat \u00e7ektiler.<\/p>\n

      TFM, MTK’nin bir uzant\u0131s\u0131 olarak adland\u0131r\u0131lm\u0131\u015ft\u0131r (Koedinger, McLaughlin ve Stamper, 2012) ve asl\u0131nda do\u011frusal lojistik test modeliyle olan (DLTM, Fischer, 1973) ili\u015fki bu modelin \u00f6nc\u00fcl\u00fcnde a\u00e7\u0131kt\u0131 (Draney vd., 1995). Bununla birlikte, mevcut \u015fekline ge\u00e7erken, model kritik bir \u015fekilde de\u011fi\u015ftirildi. DLTM, bir maddenin zorlu\u011funun, maddenin potansiyel \u00f6zellikleri \u00fczerinde bir toplam olarak ayr\u0131\u015ft\u0131r\u0131ld\u0131\u011f\u0131 Rasch tipi bir MTK modelidir. Rasch modelini \u015f\u00f6yle yazabiliriz,<\/p>\n

      logit(P<\/span>ij<\/span><\/sub>) = ln(P<\/span>ij<\/span><\/sub>\/(1-P<\/span>ij<\/span><\/sub>)) = \u03b8<\/span>i<\/span><\/sub> – \u03b2<\/span>j<\/span><\/sub>,<\/span> <\/i>(4)
      \n\u03b2<\/span> \u00f6gesinin zorlu\u011fu <\/span><\/sub>j ayr\u0131ca ayr\u0131\u015ft\u0131r\u0131l\u0131r,
      \n\u03b2<\/span>j<\/span><\/sub>=c<\/span>j <\/span><\/sub>+ \u03a3<\/span>k<\/span><\/sub> w<\/span>jk<\/span><\/sub>\u03b1<\/span>k<\/span><\/sub>, <\/span>(5)
      \na<\/i><\/span>k<\/span><\/sub>, \u201ctemel\u201d i\u015flemlerin (Fischer\u2019in terimi) zorluklar\u0131d\u0131r ve w<\/i><\/span>ik<\/span><\/sub> g\u00f6stergeleri, bu i\u015flemlerin j maddesinde gerekip gerekmedi\u011fine ba\u011fl\u0131 olarak 0 veya 1\u2019dir. T\u00fcm \u00f6geler ayn\u0131 i\u015flemleri kullan\u0131yorsa model basit bir kayma ile Rasch modeline a\u00e7\u0131k\u00e7a indirgenir,
      \n\u03b2<\/span>j<\/span><\/sub>=c<\/span>j <\/span><\/sub>+\u03b1. <\/span>(6)
      \nDraney vd. (1995)’n\u0131n modeli madde seviyesinde bir zorluk parametresi i\u00e7ermekte olsa da TFM’de sadece bile\u015fen becerilerinin zorluklar\u0131 korunmaktad\u0131r. Ek olarak, bir uygulama terimi getirilir,      4<\/sup><\/a><\/p>\n

      \u03b2<\/span>j<\/span><\/sub>AFM<\/span><\/sup> = \u03a3<\/span>k<\/span><\/sub>w<\/span>jk<\/span><\/sub>\u03b1<\/span>k<\/span><\/sub> – \u03a3<\/span>k<\/span><\/sub>wj<\/span>k<\/span><\/sub>\u03b3<\/span>k<\/span><\/sub>T<\/span>ik<\/span><\/sub> , <\/span>(7)
      \nburadaki y<\/span>k<\/span><\/sub>, bir b\u00fcy\u00fcme parametresidir ve Tik<\/sub>, \u00f6\u011frenen i<\/i><\/span>‘ nin beceri k <\/i><\/span>\u00fczerindeki \u00f6nceki uygulama giri\u015fimlerinin bir say\u0131s\u0131d\u0131r. Bir uygulama problemi diziliminin t\u00fcm\u00fc, \u00f6\u011fretici uygulamalar\u0131 i\u00e7in ortak olan ayn\u0131 becerileri i\u00e7eriyorsa, o zaman her bir dizi i\u00e7in, bu parametre,
      \n\u03b2<\/i><\/span>j<\/i><\/span><\/sub>AFM<\/i><\/span><\/sup> = \u03b1\u2013\u03b3T<\/i><\/span>i <\/i><\/span><\/sub>.<\/i><\/span> (8)
      \n<\/span>\u00d6nemli olarak, bu asl\u0131nda, sa\u011f taraftaki alt simgelerden de a\u00e7\u0131k\u00e7a anla\u015f\u0131ld\u0131\u011f\u0131 gibi, maddenin hi\u00e7bir \u00f6zelli\u011fine de\u011fil yaln\u0131zca \u00f6\u011frenmeye ba\u011fl\u0131d\u0131r. c<\/i><\/span>j<\/span><\/sub> parametresini Denklem(7) – (8) ‘de b\u0131rakma sayesinde, TFM asl\u0131nda sabit bir etki b\u00fcy\u00fcme modeli haline gelmi\u015ftir.<\/p>\n

      Modelleme a\u00e7\u0131s\u0131ndan bak\u0131ld\u0131\u011f\u0131nda hem zorluk hem de b\u00fcy\u00fcme parametrelerinin saklanmas\u0131 tan\u0131mlanabilirlik i\u00e7in bir sorun olu\u015fturdu\u011fundan madde d\u00fczeyinde zorluk parametresinin kald\u0131r\u0131lmas\u0131 \u015fa\u015f\u0131rt\u0131c\u0131 de\u011fildir. Bir model, parametreleri yeterli veri g\u00f6z \u00f6n\u00fcne al\u0131nd\u0131\u011f\u0131nda a\u00e7\u0131k\u00e7a \u00f6\u011frenilebiliyorsa tan\u0131mlanabilir. Ancak sabit bir madde dizilim \u00fczerinde \u00e7al\u0131\u015fan \u00f6\u011frenciler i\u00e7in, \u00f6\u011frenme \/ b\u00fcy\u00fcme nedeniyle artan ba\u015far\u0131 oran\u0131, azalan madde zorlu\u011funa ba\u011flanabilir. B\u00fcy\u00fcme olmayan ko\u015fullar alt\u0131nda madde zorluklar\u0131 ayr\u0131 ayr\u0131 kalibre edilmedik\u00e7e, iki etki ay\u0131rt edilemez.<\/p>\n

      Bili\u015fsel Te\u015fhis Modelleri<\/h3>\n

      Bili\u015fsel g\u00f6rev analizi kullan\u0131larak yap\u0131lan kar\u0131\u015f\u0131k say\u0131daki \u00e7\u0131karma \u00e7al\u0131\u015fmas\u0131na dair \u00e7\u0131\u011f\u0131r a\u00e7\u0131c\u0131 bir \u00e7al\u0131\u015fma, Tatsuoka’y\u0131 (1983), bir e\u011fitim testinde Q-matris y\u00f6ntemini ve belirli alt becerilerin te\u015fhisi i\u00e7in bir model (\u00f6r. bir say\u0131n\u0131n tamam\u0131n\u0131 kesire d\u00f6n\u00fc\u015ft\u00fcrme) geli\u015ftirmesine yol a\u00e7t\u0131. Q-matrisi, alt becerileri gerektirecek maddelerin kesikli bir haritas\u0131d\u0131r ve de\u011ferlendirme modelinde geleneksel olarak belirtilir. Bili\u015fsel tan\u0131sal modeller o zamandan beri olduk\u00e7a yayg\u0131nla\u015ft\u0131r\u0131lm\u0131\u015ft\u0131r (Rupp ve Templin, 2008; von Davier, 2005) ve Q matrisini verilerden \u00f6\u011frenme \u00e7abalar\u0131, e\u011fitsel veri madencili\u011fi ara\u015ft\u0131rmalar\u0131nda ortaya \u00e7\u0131km\u0131\u015ft\u0131r (Barnes, 2005; Desmarais, 2012; Koedinger vd., 2012).<\/p>\n

      HATA KAYNAKLARI, TEKRAR G\u00d6ZDEN GE\u00c7\u0130R\u0130LMES\u0130<\/h2>\n

      Motivasyon, duygu ve bili\u015f \u00e7al\u0131\u015fmalar\u0131na d\u00e2hil olan baz\u0131 \u00f6l\u00e7me modellerini ara\u015ft\u0131rd\u0131ktan sonra, \u00f6nemli olan hata konusu tekrar g\u00f6zden ge\u00e7irmeye de\u011fer. Uygulay\u0131c\u0131lar, yanl\u0131\u015f parametreli modeller kullanarak, yanl\u0131\u015f modeller kullanarak veya modelleri yanl\u0131\u015f kullanarak ek hata kaynaklar\u0131n\u0131n ortaya \u00e7\u0131kabilece\u011fine dikkat etmelidir.<\/p>\n

      Bir modelin kullan\u0131m\u0131, tahmini hataya tabi olan parametrelere ba\u011fl\u0131 olabilir. Bu belirsizlikler kabul edilmelidir ancak model g\u00f6zlenen veriler i\u00e7in veri \u00fcreten bir model<\/i><\/span> olarak tutarl\u0131ysa, bunlar ille de ciddi belirsizlikler de\u011fildir. Yani, istatistiksel modeli veri \u00fcretmek i\u00e7in de kullan\u0131labilecek stokastik bir s\u00fcre\u00e7 olarak g\u00f6r\u00fcyoruz (ayr\u0131ca, \u00f6rnekleme veya benzetme) (Breiman, 2001). \u00d6rne\u011fin, ger\u00e7ek madeni paran\u0131n hilesiz olup olmad\u0131\u011f\u0131ndan emin olmasak bile, bir Bernoulli i\u015flemi kullanarak madeni para atma deneyi verilerini sim\u00fcle edebiliriz. Prensip olarak, modelimizdeki tura olas\u0131l\u0131klar\u0131 parametresi, ger\u00e7ek madeni paradan daha fazla veri ile geli\u015ftirilebilir. Bu modelin kendisinin ya \u00f6rt\u00fck de\u011fi\u015fkenler ya da ba\u011flant\u0131 i\u015flevleri a\u00e7\u0131s\u0131ndan, ger\u00e7ek \u00fcretici modelle tutars\u0131z oldu\u011fu durumdan farkl\u0131d\u0131r. \u0130kinci vaka modelin yanl\u0131\u015f tan\u0131mlanmas\u0131 olarak adland\u0131r\u0131l\u0131r (White, 1996). Uyumluluk testleri, modeli korumak veya reddetmek i\u00e7in g\u00f6zlenen veriler ile \u00fcretici model aras\u0131ndaki tutarl\u0131l\u0131\u011f\u0131 de\u011ferlendirir (White, 1996; Haberman, 2009; Ames ve Penfield, 2015).<\/p>\n

      A\u00c7IKLAMA VE YORDAMA<\/h3>\n

      Kestirimci modelleme, e\u011fitsel veri madencili\u011finde en \u00f6nemli metodolojik yakla\u015f\u0131mlardan biridir (Baker ve Siemens, 2014; Baker ve Yacef, 2009). \u00d6l\u00e7me teorisi, aksine, sosyal bilimlerde geleneksel olarak kullan\u0131lan istatistiksel y\u00f6ntemlerin \u00e7o\u011funda oldu\u011fu gibi, tamamen a\u00e7\u0131klay\u0131c\u0131d\u0131r (Breiman, 2001; Shmueli, 2010). A\u00e7\u0131klay\u0131c\u0131 bir model, \u00f6ng\u00f6r\u00fclerde bulunmak i\u00e7in kullan\u0131labilirken -ve hatas\u0131z- bir a\u00e7\u0131klay\u0131c\u0131 model, kusursuz tahminlerde bulunabilir; kestirimci bir model muhakkak a\u00e7\u0131klay\u0131c\u0131 olmak zorunda de\u011fildir. Breiman (2001) iki k\u00fclt\u00fcr a\u00e7\u0131s\u0131ndan ayr\u0131m\u0131 ifade etmi\u015ftir: veri modelleme k\u00fclt\u00fcr\u00fc (Breiman’a g\u00f6re gayriresm\u00ee olarak istatistiklerin %98’i) ve algoritmik modelleme k\u00fclt\u00fcr\u00fc (Breiman’\u0131n kendisini i\u00e7erdi\u011fi %2).5<\/sup><\/a> Shmueli (2010), bir tahmin veya a\u00e7\u0131klama merce\u011finden bak\u0131ld\u0131\u011f\u0131nda istatistiksel modelleme i\u00e7in t\u00fcm tasar\u0131m s\u00fcrecinin kar\u015f\u0131la\u015ft\u0131rmas\u0131n\u0131 yapm\u0131\u015ft\u0131r. Yorday\u0131c\u0131lar\u0131n karma\u015f\u0131k bir tahmin modelinde yorumlanabilirli\u011fi veya yorumlanamazl\u0131\u011f\u0131, ayr\u0131m\u0131n yaln\u0131zca bir y\u00f6n\u00fcd\u00fcr (ayr\u0131ca bk. Liu ve Koedinger, bu say\u0131). Farkl\u0131 bak\u0131\u015f a\u00e7\u0131lar\u0131, ara\u015ft\u0131rmac\u0131lar\u0131n hata ve belirsizlikle nas\u0131l ba\u015fa \u00e7\u0131kt\u0131klar\u0131 hakk\u0131nda temel olarak bilgilendirmektedir.<\/p>\n

      Kestirimci g\u00f6r\u00fc\u015f, \u00f6rne\u011fin, e\u011fitsel veri madencili\u011fi konferans\u0131ndaki en son ve en iyi makalede a\u00e7\u0131klanm\u0131\u015ft\u0131r. Yazarlar, \u201cmodel varsay\u0131mlar\u0131n\u0131n do\u011fru olup olmad\u0131\u011f\u0131n\u0131 belirlemenin tek yolu, farkl\u0131 varsay\u0131mlar yapan alternatif bir model olu\u015fturmak ve alternatifin [tahmin d\u0131\u015f\u0131] BBT’ den daha iyi performans g\u00f6sterip g\u00f6stermedi\u011fini belirlemektir\u201d iddias\u0131ndad\u0131r (Khajah, Lindsey ve Mozer, 2016, 95, edit\u00f6r notu eklendi). A\u00e7\u0131k\u00e7as\u0131 model tahmin performans\u0131 model varsay\u0131mlar\u0131n\u0131n ihlal edilip edilmedi\u011finin belirlemesi i\u00e7in bir yolu de\u011fildir. Aksine hem gayriresm\u00ee kontroller hem de uyumlulu\u011fa y\u00f6nelik resm\u00ee testler yukar\u0131da tart\u0131\u015f\u0131lm\u0131\u015ft\u0131r. Bununla birlikte, al\u0131nt\u0131, modellerin \u00f6ng\u00f6r\u00fcc\u00fc do\u011frulukla onayland\u0131\u011f\u0131 algoritmik modelleme k\u00fclt\u00fcr\u00fcn\u00fcn bir yans\u0131mas\u0131d\u0131r (Breiman, 2001). Daha problematik olarak, bu yorday\u0131c\u0131 g\u00fcc\u00fcn daha ger\u00e7ek\u00e7i bir modele i\u015faret etti\u011fi varsay\u0131m\u0131n\u0131 ta\u015f\u0131r. Asl\u0131nda, bu rol\u00fc oynayan a\u00e7\u0131klay\u0131c\u0131 bir g\u00fc\u00e7t\u00fcr. Varyans bile\u015fenleri a\u00e7\u0131s\u0131ndan, \u201ca\u00e7\u0131klay\u0131c\u0131 modellemede odak, temel teorinin en do\u011fru temsilini elde etmek i\u00e7in yanl\u0131l\u0131\u011f\u0131 en aza indirmektir. Buna kar\u015f\u0131l\u0131k, kestirimci modelleme \u00f6nyarg\u0131 ve varyans kombinasyonunu en aza indirmeyi, zaman zamansa geli\u015fmi\u015f deneysel kesinlik i\u00e7in teorik do\u011frulu\u011fu feda etmeyi ama\u00e7lamaktad\u0131r \u201d(Shmueli, 2010, s. 293). A\u00e7\u0131klay\u0131c\u0131 g\u00fcc\u00fcn ve yorday\u0131c\u0131 g\u00fcc\u00fcn her zaman ayn\u0131 y\u00f6ne i\u015faret etmedi\u011fi vurgulanmal\u0131d\u0131r. Nitekim, Hagerty ve Srinivasan (1991), karma\u015f\u0131k durumlarda, yetersiz tan\u0131mlanm\u0131\u015f \u00e7oklu regresyon modellerinin do\u011fru (ger\u00e7ek) modelden daha fazla yorday\u0131c\u0131 g\u00fcce sahip oldu\u011funu kan\u0131tlam\u0131\u015ft\u0131r.<\/p>\n

      Suthers ve Verbert (2013) \u00f6\u011frenme analiti\u011fini, \u00f6\u011frenme bilimi ve analitik aras\u0131nda \u201corta alan\u201d olarak tan\u0131mlam\u0131\u015ft\u0131r. Belki de a\u00e7\u0131klay\u0131c\u0131 ve kestirimci yakla\u015f\u0131mlar aras\u0131nda metodolojik bir orta alan\u0131 i\u015fgal etti\u011fi d\u00fc\u015f\u00fcn\u00fclebilir. Bu durumda, alan her iki bak\u0131\u015f\u0131n n\u00fcanslar\u0131n\u0131 anlamakta fayda elde edebilir.<\/p>\n

      DAHA FAZLA OKUMA<\/h3>\n

      Psikolojik \u00f6l\u00e7meler neredeyse psikolojinin kendisi ve istatistikler kadar eskidir. G\u00fcvenilir, teknik ve bir nevi ansiklopedik kaynaklar, \u0130statistik El Kitab\u0131<\/i><\/span> serisindeki psikometri antolojisi(Rao ve Sinharay, 2006) ve \u015fu an da d\u00f6rd\u00fcnc\u00fc bask\u0131s\u0131nda olan E\u011fitsel \u00d6l\u00e7menin<\/i><\/span> \u201c\u0130ncil’i\u201ddir (Brennan, 2006). Belirli say\u0131lar\u0131n g\u00fcvenilirlik, ge\u00e7erlilik, genelle\u015ftirilebilirlik, kar\u015f\u0131la\u015ft\u0131r\u0131labilirlik ve do\u011fruluk oldu\u011fu e\u011fitsel \u00f6l\u00e7me sorunlar\u0131 ve Standartlar<\/i><\/span> s\u0131navlara vurgu yapar (AERA, APA ve NCME, 2014). DeVellis’in (2003) \u00f6l\u00e7ek geli\u015ftirmede \u00f6zl\u00fc hacim makalesi, psikolojik \u00f6l\u00e7meye teknik olmayan bir giri\u015f sunar ve paralel test formlar\u0131ndan al\u0131nan puanlar\u0131 birbirine ba\u011flamak gibi b\u00fcy\u00fck \u00f6l\u00e7ekli testlere \u00f6zg\u00fc konular\u0131 g\u00f6z ard\u0131 etmektedir.<\/p>\n

      KAYNAK\u00c7A<\/p>\n

      AERA, APA, & NCME (American Educational Research Association, American Psychological Association, & National Council on Measurement in Education). (2014). Standards for educational and psychological testing<\/i>. Washington, DC: AERA. <\/span><\/p>\n

      Ames, A. J., & Penfield, R. D. (2015). An NCME instructional module on polytomous item response theory models. Educational Measurement: Issues and Practice, 34<\/i>(3), 39\u201348. doi:10.1111\/emip.12023 <\/span><\/p>\n

      Anderson, J. R., Corbett, A. T., Koedinger, K. R., & Pelletier, R. (1995). Cognitive tutors: Lessons learned. The Journal of the Learning Sciences, 4<\/i>(2), 167\u2013207. <\/span><\/p>\n

      Armstrong, J. S. (1967). Derivation of theory by means of factor analysis or Tom Swift and his electric factor analysis machine. The American Statistician, 21<\/i>, 17\u201321. <\/span><\/p>\n

      Attali, Y. (2011). Immediate feedback and opportunity to revise answers: Application of a graded response IRT model. Applied Psychological Measurement, 35<\/i>(6), 472\u2013479. <\/span><\/p>\n

      Baker, F. B., & Kim, S.-H. (Eds.). (2004). Item response theory: Parameter estimation techniques<\/i>. Boca Raton, FL: CRC Press. <\/span><\/p>\n

      Baker, R. S., & Siemens, G. (2014). Educational data mining and learning analytics. In R. Sawyer (Ed), The Cambridge handbook of the learning sciences <\/i>(pp. 253\u2013272). Cambridge University Press. <\/span><\/p>\n

      Baker, R. S., & Yacef, K. (2009). The state of educational data mining in 2009: A review and future visions. Journal of Educational Data Mining, 1<\/i>(1), 3\u201317. <\/span><\/p>\n

      Barnes, T. (2005). The Q-matrix method: Mining student response data for knowledge. In the Technical Report (WS-05-02) of the AAAI-05 Workshop on Educational Data Mining. <\/span><\/p>\n

      Behrens, J. T., & DiCerbo, K. E. (2014). Harnessing the currents of the digital ocean. In J. A. Larusson & B. White (Eds.), Learning analytics: From research to practice <\/i>(pp. 39\u201360). New York: Springer. <\/span><\/p>\n

      Bachman, J. G., & O\u2019Malley, P.M. (1984). Yea-saying, nay-saying, and going to extremes: Black-white differences in response styles. Public Opinion Quarterly, 48<\/i>, 491\u2013509. <\/span><\/p>\n

      Bergner, Y., Colvin, K., & Pritchard, D. E. (2015). Estimation of ability from homework items when there are missing and\/or multiple attempts. Proceedings of the 5th International Conference on Learning Analytics and Knowledge <\/i>(LAK\u201915), 16\u201320 March 2015, Poughkeepsie, NY, USA (pp. 118\u2013125). New York: ACM. <\/span><\/p>\n

      Bergner, Y., Kerr, D., & Pritchard, D. E. (2015). Methodological challenges in the analysis of MOOC data for exploring the relationship between discussion forum views and learning outcomes. In O. C. Santos et al. (Eds.), Proceedings of the 8th International Conference on Educational Data Mining <\/i>(EDM2015), 26\u201329 June 2015, Madrid, Spain (pp. 234\u2013241). International Educational Data Mining Society. <\/span><\/p>\n

      Bergner, Y., Rayyan, S., Seaton, D., & Pritchard, D. E. (2013). Multidimensional student skills with collaborative filtering. AIP Conference Proceedings, 1513<\/i>(1), 74\u201377. doi:10.1063\/1.4789655 <\/span><\/p>\n

      Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent Dirichlet allocation. Journal of Machine Learning Research, 3<\/i>(Jan.), 993\u20131022. <\/span><\/p>\n

      Bollen, K. A. (1989). Structural equations with latent variables<\/i>. John Wiley & Sons. <\/span><\/p>\n

      Borsboom, D. (2008). Latent variable theory. Measurement: Interdisciplinary Research & Perspective, 6<\/i>(1\u20132), 25\u201353. http:\/\/doi.org\/10.1080\/15366360802035497 <\/span><\/p>\n

      Box, G. E. (1979). Robustness in the strategy of scientific model building. Robustness in Statistics, 1<\/i>, 201\u2013236. <\/span><\/p>\n

      Breiman, L. (2001). Statistical modeling: The two cultures. Statistical Science, 16<\/i>(3), 199\u2013215. http:\/\/doi.org\/10.2307\/2676681 <\/span><\/p>\n

      Brennan, R. L. (Ed.). (2006). Educational measurement<\/i>. Praeger Publishers.<\/span><\/p>\n

      Bridgman, P. W. (1927). The logic of modern physics<\/i>. New York: Macmillan. <\/span><\/p>\n

      Buckingham Shum, S., & Deakin Crick, R. (2012). Learning dispositions and transferable competencies: Pedagogy, modeling and learning analytics. Proceedings of the 2nd International Conference on Learning Analytics and Knowledge <\/i>(LAK\u201912), 29 April\u20132 May 2012, Vancouver, BC, Canada (pp. 92\u2013101). New York: ACM. <\/span><\/p>\n

      Cardamone, C. N., Abbott, J. E., Rayyan, S., Seaton, D. T., Pawl, A., & Pritchard, D. E. (2011). Item response theory analysis of the mechanics baseline test. Proceedings of the 2011 Physics Education Research Conference <\/i>(PERC 2011), 3\u20134 August 2011, Omaha, NE, USA (pp. 135\u2013138). doi:10.1063\/1.3680012 <\/span><\/p>\n

      Cen, H., Koedinger, K. R., & Junker, B. (2008). Comparing two IRT models for conjunctive skills. In B. Woolf, E. A\u00efmeur, R. Nkambou, & S. Lajoie (Eds.), Proceedings of the 9th International Conference on Intelligent Tutoring Systems <\/i>(ITS 2008), 23\u201327 June 2008, Montreal, PQ, Canada (pp. 796\u2013798). Springer. <\/span><\/p>\n

      Cohen, J. (1968). Weighted kappa: Nominal scale agreement provision for scaled disagreement or partial credit. Psychological Bulletin, 70<\/i>(4), 213\u2013220. <\/span><\/p>\n

      Corbett, A. T., & Anderson, J. R. (1995). Knowledge tracing: Modeling the acquisition of procedural knowledge. User Modeling and User-Adapted Interaction, 4<\/i>, 253\u2013278. <\/span><\/p>\n

      Cortina, J.M. (1993). What is coefficient alpha? An examination of theory and applications. Journal of Applied Psychology, 78<\/i>(1), 98. <\/span><\/p>\n

      Crick, R. D., Broadfoot, P., & Claxton, G. (2004). Developing an effective lifelong learning inventory: The ELLI project. Assessment in Education: Principles, Policy & Practice, 11<\/i>(3), 247\u2013272. <\/span><\/p>\n

      Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16<\/i>(3), 297\u2013334. <\/span><\/p>\n

      Cronbach, L. J., & Meehl, P. E. (1955). Construct validity in psychological tests. Psychological Bulletin, 52<\/i>(4), 281\u2013302. <\/span><\/p>\n

      Culpepper, S. A. (2014). If at first you don\u2019t succeed, try, try again: Applications of sequential IRT models to cognitive assessments. Applied Psychological Measurement, 38<\/i>(8), 632\u2013644. doi:10.1177\/0146621614536464 <\/span><\/p>\n

      Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behaviour. New York: Plenum. <\/span><\/p>\n

      Dedic, H., Rosenfield, S., & Lasry, N. (2010). Are all wrong FCI answers equivalent? AIP Conference Proceedings, 1289<\/i>, 125\u2013128. doi.org\/10.1063\/1.3515177 <\/span><\/p>\n

      Desmarais, M.C. (2012). Mapping question items to skills with non-negative matrix factorization. ACM SIGKDD Explorations Newsletter, 13<\/i>(2), 30\u201336. <\/span><\/p>\n

      Desmarais, M. C., & Baker, R. S. (2011). A review of recent advances in learner and skill modeling in intelligent learning environments. User Modeling and User-Adapted Interaction, 22<\/i>(1\u20132), 9\u201338. doi:10.1007\/s11257-011- 9106-8 <\/span><\/p>\n

      DeVellis, R. F. (2003). Scale development: Theory and applications<\/i>. Applied Social Research Methods Series (Vol. 26). Thousand Oaks, CA: Sage Publications. <\/span><\/p>\n

      Digman, J.M. (1990). Personality structure: Emergence of the five-factor model. Annual Review of Psychology, 41<\/i>(1), 417\u2013440. <\/span><\/p>\n

      Ding, L., & Beichner, R. (2009). Approaches to data analysis of multiple-choice questions. Physical Review Special Topics: Physics Education Research, 5<\/i>(2), 1\u201317. doi:10.1103\/PhysRevSTPER.5.020103 <\/span><\/p>\n

      Draney, K., Pirolli, P., & Wilson, M. R. (1995). A measurement model for a complex cognitive skill. In P. Nichols, S. Chipman, & R. Brennan (Eds.), Cognitively diagnostic assessment<\/i>. Hillsdale, NJ: Lawrence Erlbaum Associates. <\/span><\/p>\n

      Duckworth, A. L., Peterson, C., Matthews, M. D., & Kelly, D. R. (2007). Grit: Perseverance and passion for long-term goals. Journal of Personality and Social Psychology, 9<\/i>, 1087\u20131101. <\/span><\/p>\n

      Dweck, C. S. (2000). Self-theories: Their role in motivation, personality and development. Philadelphia, PA: Taylor & Francis.<\/span><\/p>\n

      Edwards, J. R. (2001). Multidimensional constructs in organizational behavior research: An integrative analytical framework. Organizational Research Methods, 4<\/i>(2), 144\u2013192. <\/span><\/p>\n

      Erosheva, E., Fienberg, S., & Lafferty, J. (2004). Mixed-membership models of scientific publications. Proceedings of the National Academy of Sciences, 101<\/i>(suppl 1), 5220\u20135227. <\/span><\/p>\n

      Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. (1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods, 4<\/i>(3), 272. <\/span><\/p>\n

      Fischer, G.H. (1973). The linear logistic test model as an instrument in educational research. Acta Psychologica, 37<\/i>(6), 359\u2013374. <\/span><\/p>\n

      Fraley, C., & Raftery, A. E. (1998). How many clusters? Which clustering method? Answers via model-based cluster analysis. The Computer Journal, 41<\/i>(8), 578\u2013588. <\/span><\/p>\n

      George, R. (2000). Measuring change in students\u2019 attitudes toward science over time: An application of latent variable growth modeling. Journal of Science Education and Technology, 9<\/i>(3), 213\u2013225. <\/span><\/p>\n

      Goodman, L. (2002) Latent class analysis: The empirical study of latent types, latent variables, and latent structures. In J. A. Hagenaars & A. L. McCutcheon (Eds.), Applied latent class analysis <\/i>(pp. 3\u201355). Cambridge, UK: Cambridge University Press. <\/span><\/p>\n

      Guay, F., Vallerand, R. J., & Blanchard, C. (2000). On the assessment of situational intrinsic and extrinsic motivation: The situational motivation scale (SIMS). Motivation and Emotion, 24<\/i>(3), 175\u2013213. <\/span><\/p>\n

      Haberman, S. J. (2009). Use of generalized residuals to examine goodness of fit of item response models. ETS Research Report RR-09-15<\/i>. <\/span><\/p>\n

      Hagerty, M. R., & Srinivasan, V. (1991). Comparing the predictive powers of alternative multiple regression models. Psychometrika, 56<\/i>(1), 77\u201385. <\/span><\/p>\n

      Hestenes, D., & Wells, M. (1992). A mechanics baseline test. The Physics Teacher, 30<\/i>(3), 159\u2013166. <\/span><\/p>\n

      Hestenes, D., Wells, M., & Swackhamer, G. (1992). Force concept inventory. The Physics Teacher, 30<\/i>(3), 141. doi:10.1119\/1.2343497 <\/span><\/p>\n

      Holland, P.W. (1990). On the sampling theory roundations of item response theory models. Psychometrika, 55<\/i>(4), 577\u2013601. http:\/\/doi.org\/10.1007\/BF02294609 <\/span><\/p>\n

      Kane, M.T. (2001). Current concerns in validity theory. Journal of Educational Measurement, 38<\/i>(4), 319\u2013342. <\/span><\/p>\n

      Kane, M. (2010). Errors of measurement, theory, and public policy. William H. Angoff Memorial Lecture Series. Educational Testing Service<\/i>. https:\/\/www.ets.org\/Media\/Research\/pdf\/PICANG12.pdf <\/span><\/p>\n

      K\u00e4ser, T., Koedinger, K. R., & Gross, M. (2014). Different parameters \u2014 same prediction: An analysis of learning curves. In S. K. D\u2019Mello, R. A. Calvo, & A. Olney (Eds.), Proceedings of the 6th International Conference on Educational Data Mining <\/i>(EDM2013), 6\u20139 July 2013, Memphis, TN, USA (pp. 52\u201359). International Educational Data Mining Society\/Springer. <\/span><\/p>\n

      Khajah, M., Lindsey, R. V., & Mozer, M. C. (2016). How deep is knowledge tracing? In T. Barnes, M. Chi, & M. Feng (Eds.), Proceedings of the 9th International Conference on Educational Data Mining <\/i>(EDM2016), 29 June\u20132 July 2016, Raleigh, NC, USA (pp. 94\u2013101). International Educational Data Mining Society. <\/span><\/p>\n

      Kline, R. B. (2010). Principles and practice of structural equation modeling. New York: Guilford. <\/span><\/p>\n

      Koedinger, K. R., McLaughlin, E. A., & Stamper, J. (2012). Automated student model improvement. In K. Yacef et al. (Eds.), Proceedings of the 5th International Conference on Educational Data Mining <\/i>(EDM2012), 19\u201321 June 2012, Chania, Greece. International Educational Data Mining Society. http:\/\/www.learnlab.org\/research\/ wiki\/images\/e\/e1\/KoedingerMcLaughlinStamperEDM12.pdf <\/span><\/p>\n

      Lord, F. M. (1980). Applications of item response theory to practical testing problems. Routledge. <\/span><\/p>\n

      Lord, F. M., & Novick, M. R. (1968). Statistical theories of mental test scores<\/i>. Addison-Wesley.<\/span><\/p>\n

      Luria, R. E. (1975). The validity and reliability of the visual analogue mood scale. Journal of Psychiatric Research, 12<\/i>(1), 51\u201357. <\/span><\/p>\n

      Martin, B., Mitrovic, T., Mathan, S., & Koedinger, K. R. (2010). Evaluating and improving adaptive educational systems with learning curves. User Modeling and User-Adapted Interaction: The Journal of Personalization Research, 21<\/i>, 249\u2013283. <\/span><\/p>\n

      Maul, A., Irribarra, D. T., & Wilson, M. (2016). On the philosophical foundations of psychological measurement. Measurement, 79<\/i>, 311\u2013320. http:\/\/doi.org\/10.1016\/j.measurement.2015.11.001 <\/span><\/p>\n

      Mazur, E. (2007). Confessions of a converted lecturer. https:\/\/www.math.upenn.edu\/~pemantle\/active-papers\/Mazurpubs_605.pdf <\/span><\/p>\n

      McLachlan, G., & Peel, D. (2004). Finite mixture models<\/i>. John Wiley & Sons. <\/span><\/p>\n

      Meredith, W., & Tisak, J. (1990). Latent curve analysis. Psychometrika, 55<\/i>(1), 107\u2013122. <\/span><\/p>\n

      Messick, S. (1995). Validity of psychological assessment: Validation of inferences from persons\u2019 responses and performances as scientific inquiry into score meaning. American Psychologist, 50<\/i>(9), 741\u2013749. <\/span><\/p>\n

      Messick, S., & Jackson, D. (1961). Acquiescence and the factorial interpretation of the MMPI. Psychological Bulletin, 58<\/i>(4), 299\u2013304 <\/span><\/p>\n

      Michell, J. (1999). Measurement in psychology: A critical history of a methodological concept (Vol. 53). Cambridge University Press. <\/span><\/p>\n

      Midgley, C., Maehr, M. L., Hruda, L., Anderman, E. M., Anderman, L., Freeman, K. E., et al. (2000). Manual for the patterns of adaptive learning scales (PALS)<\/i>. Ann Arbor, MI: University of Michigan. <\/span><\/p>\n

      Milligan, S. K., & Griffin, P. (2016). Understanding learning and learning design in MOOCs: A measurement-based interpretation. Journal of Learning Analytics, 3<\/i>(2), 88\u2013115. <\/span><\/p>\n

      Millsap, R.E. (2012). Statistical approaches to measurement invariance. Routledge. <\/span><\/p>\n

      Mislevy, R. J. (2009). Validity from the perspective of model-based reasoning. In R. L. Lissitz (Ed.), The concept of validity: Revisions, new directions and applications <\/i>(pp. 83\u2013108). Charlotte, NC: Information Age Publishing. <\/span><\/p>\n

      Mislevy, R. J. (2012). Four metaphors we need to understand assessment. Draft paper commissioned by the Gordon Commission. http:\/\/www.gordoncommission.com\/rsc\/pdfs\/mislevy_four_metaphors_understand_assessment.pdf <\/span><\/p>\n

      Morris, G. A., Branum-Martin, L., Harshman, N., Baker, S. D., Mazur, E., Dutta, S., \u2026 McCauley, V. (2006). Testing the test: Item response curves and test quality. American Journal of Physics, 74<\/i>(5), 449. doi:10.1119\/1.2174053 <\/span><\/p>\n

      Mulaik, S. A. (2009). Foundations of factor analysis<\/i>. Boca Raton, FL: CRC Press. <\/span><\/p>\n

      Nederhof, A. J. (1985). Methods of coping with social desirability bias: A review. European Journal of Social Psychology, 15<\/i>(3), 263\u2013280. http:\/\/doi.org\/10.1002\/ejsp.2420150303 <\/span><\/p>\n

      Newell, A., & Rosenbloom, P. S. (1981). Mechanisms of skill acquisition and the law of practice. Cognitive Skills and their Acquisition, 6<\/i>, 1\u201355. <\/span><\/p>\n

      Pekrun, R., Goetz, T., Frenzel, A. C., Barchfeld, P., & Perry, R. P. (2011). Measuring emotions in students\u2019 learning and performance: The achievement emotions questionnaire (AEQ). Contemporary Educational Psychology, 36<\/i>(1), 36\u201348. http:\/\/doi.org\/10.1016\/j.cedpsych.2010.10.002 <\/span><\/p>\n

      Pintrich, P. R., & De Groot, E. V. (1990). Motivational and self-regulated learning components of classroom academic performance. Journal of Educational Psychology, 82<\/i>(1), 33. <\/span><\/p>\n

      Rabiner, L. R. (1989). A tutorial on hidden Markov models and selected applications in speech recognition. Proceedings of the IEEE, 77<\/i>(2), 257\u2013286.<\/span><\/p>\n

      Rao, C. R., & Sinharay, S. (Eds.). (2006). Handbook of statistics 26: Psychometrics<\/i>. Elsevier. doi:10.1016\/S0169- 7161(06)26037-1 <\/span><\/p>\n

      Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (Vol. 1). Sage. <\/span><\/p>\n

      Rijmen, F. (2010). Formal relations and an empirical comparison among the bi-factor, the testlet, and a second-order multidimensional IRT model. Journal of Educational Measurement, 47<\/i>(3), 361\u2013372. doi:10.1111\/ j.1745-3984.2010.00118.x <\/span><\/p>\n

      Rupp, A., & Templin, J. L. (2008). Unique characteristics of diagnostic classification models: A comprehensive review of the current state-of-the-art. Measurement: Interdisciplinary Research & Perspective, 6<\/i>(4), 219\u2013 262. doi:10.1080\/15366360802490866 <\/span><\/p>\n

      Schwartz, S. (2007). The structure of identity consolidation: Multiple correlated constructs or one superordinate construct? Identity, 7<\/i>(1), 27\u201349. <\/span><\/p>\n

      Scott, T. F., Schumayer, D., & Gray, A. R. (2012). Exploratory factor analysis of a force concept inventory data set. Physical Review Special Topics: Physics Education Research, 8<\/i>(2). doi:10.1103\/PhysRevSTPER.8.020105 <\/span><\/p>\n

      Shmueli, G. (2010). To explain or to predict? Statistical Science, 25<\/i>(3), 289\u2013310. http:\/\/doi.org\/10.1214\/10- STS330 <\/span><\/p>\n

      Siemens, G., & Baker, R. S. (2012). Learning analytics and educational data mining: Towards communication and collaboration. Proceedings of the 2nd International Conference on Learning Analytics and Knowledge <\/i>(LAK\u201912), 29 April\u20132 May 2012, Vancouver, BC, Canada (pp. 252\u2013254). New York: ACM. <\/span><\/p>\n

      Sijtsma, K. (2011). Introduction to the measurement of psychological attributes. Measurement, 44<\/i>(7), 1209\u20131219. doi: 10.1016 \/ j.measurement.2011.03.019 <\/span><\/p>\n

      Sijtsma, K. (1998). Methodology review: Nonparametric IRT approaches to the analysis of dichotomous item scores. Applied Psychological Measurement, 22<\/i>(1), 3\u201331. doi:10.1177\/01466216980221001 <\/span><\/p>\n

      Skrondal, A., & Rabe-Hesketh, S. (2004). Generalized latent variable modeling: Multilevel, longitudinal and structural equation models. Boca Raton, FL: Chapman & Hall\/CRC Press. <\/span><\/p>\n

      Spearman, C. (1904). \u201cGeneral intelligence,\u201d objectively determined and measured. The American Journal of Psychology, 15<\/i>(2), 201\u2013292. <\/span><\/p>\n

      Spray, J. A. (1997). Multiple-attempt, single-item response models. In W. J. van der Linden & R. K. Hambleton (Eds.), Handbook of modern item response theory <\/i>(pp. 209\u2013220). New York: Springer. <\/span><\/p>\n

      Stevens, S.S. (1946). On the theory of scales of measurement. Science, 103<\/i>(2684), 677\u2013680. <\/span><\/p>\n

      Suthers, D., & Verbert, K. (2013). Learning analytics as a middle space. Proceedings of the 3rd International Conference on Learning Analytics and Knowledge <\/i>(LAK\u201913), 8\u201312 April 2013, Leuven, Belgium (pp. 1\u20134). New York: ACM. <\/span><\/p>\n

      Tatsuoka, K.K. (1983). Rule space: An approach for dealing with misconceptions based on item response theory. Journal of Educational Measurement, 20<\/i>, 345\u2013354. <\/span><\/p>\n

      Tempelaar, D. T., Niculescu, A., Rienties, B., Giesbers, B., & Gijselaers, W. H. (2012). How achievement emotions impact students\u2019 decisions for online learning, and what precedes those emotions. Internet and Higher Education, 15<\/i>(3), 161\u2013169. doi: 10.1016 \/ j.iheduc.2011.10.003 <\/span><\/p>\n

      Tempelaar, D. T., Rienties, B., & Giesbers, B. (2015). In search for the most informative data for feedback generation: Learning analytics in a data-rich context. Computers in Human Behavior, 47<\/i>, 157\u2013167. doi:10.1016\/j. chb.2014.05.038 <\/span><\/p>\n

      Thurstone, L.L. (1947). Multiple factor analysis<\/i>. Chicago, IL: University of Chicago Press. <\/span><\/p>\n

      van de Sande, B. (2013). Properties of the Bayesian knowledge tracing model. Journal of Educational Data Mining, 5<\/i>(2), 1\u201310.<\/span><\/p>\n

      von Davier, M. (2005). A general diagnostic model applied to language testing data. The British Journal of Mathematical and Statistical Psychology, 61<\/i>(Pt 2), 287\u2013307. doi:10.1348\/000711007X193957 <\/span><\/p>\n

      Wang, Y., & Baker, R. S. (2015). Content or platform: Why do students complete MOOCs? Journal of Online Learning and Teaching, 11<\/i>(1), 17. <\/span><\/p>\n

      Wang, J., & Bao, L. (2010). Analyzing force concept inventory with item response theory. American Journal of Physics, 78<\/i>(10), 1064. doi:10.1119\/1.3443565 <\/span><\/p>\n

      White, H. (1996). Estimation, inference and specification analysis <\/i>(No. 22). Cambridge University Press. <\/span><\/p>\n

      Wise, S., & Kong, X. (2005). Response time effort: A new measure of examinee motivation in computer-based tests. Applied Measurement in Education, 18<\/i>(2), 163\u2013183. <\/span><\/p>\n

      Yeager, D. S., & Dweck, C. S. (2012). Mindsets that promote resilience: When students belie<\/span><\/p>\n

      \n
      \n

      1<\/a> Amerika\u2019daki y\u00fcksekokul ve \u00fcniversitelere giri\u015fte hem amerikan vatanda\u015flar\u0131 hem de yabanc\u0131 uyruklu \u00f6\u011frenciler taraf\u0131ndan kullan\u0131lan ayn\u0131 zamanda T\u00fcrkiye\u2019deki yabanc\u0131 uyruklu \u00f6\u011frencilerin T\u00fcrk \u00fcniversitelerine yerle\u015ftirilmesi s\u00fcrecinde de bir\u00e7ok \u00fcniversite taraf\u0131ndan kabul edilen bir s\u0131navd\u0131r.<\/span><\/span><\/p>\n<\/div>\n

      \n

      2<\/a> Stokastik konu i\u00e7in, bu \u00f6rnek de\u011ferler rep di\u011fer \u00e7al\u0131\u015fmalarda hi\u00e7bir bellek ile ayn\u0131 konuyu ayn\u0131 deneyler bir dizi k\u0131rg\u0131n olurdu. Bu bili\u015fsel test \u00f6gesi garip g\u00f6r\u00fcnse de, psikomotor bir ba\u011flamda b\u00fcy\u00fck bir ihtimaldir. Bknz Sprey (1997).<\/span><\/p>\n<\/div>\n

      \n

      3<\/a> orj. curriculum sequencing<\/span><\/p>\n<\/div>\n

      \n

      4<\/a> Cen vd.nin i\u015faret uzla\u015f\u0131m\u0131 kural\u0131 (2008), modeli al\u0131\u015f\u0131lm\u0131\u015f Rasch modeli ile tutarl\u0131 hale getirmek i\u00e7in kolayl\u0131ktan \u00e7ok bir zorluk parametresi olarak de\u011fi\u015ftirildi.<\/span><\/p>\n<\/div>\n

      \n

      5<\/a> Breiman a\u00e7\u0131klama ve tahmin i\u00e7in incontrast yerine d\u00f6nem bilgileri kullan\u0131r.<\/span><\/p>\n<\/div>\n","protected":false},"author":1,"menu_order":3,"template":"","meta":{"pb_show_title":"on","pb_short_title":"","pb_subtitle":"","pb_authors":["yoav-bergner"],"pb_section_license":""},"chapter-type":[48],"contributor":[76],"license":[],"class_list":["post-44","chapter","type-chapter","status-publish","hentry","chapter-type-numberless","contributor-yoav-bergner"],"part":31,"_links":{"self":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters\/44","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters"}],"about":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/types\/chapter"}],"author":[{"embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/users\/1"}],"version-history":[{"count":0,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters\/44\/revisions"}],"part":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/parts\/31"}],"metadata":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters\/44\/metadata\/"}],"wp:attachment":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/media?parent=44"}],"wp:term":[{"taxonomy":"chapter-type","embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapter-type?post=44"},{"taxonomy":"contributor","embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/contributor?post=44"},{"taxonomy":"license","embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/license?post=44"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}