{"id":1616,"date":"2023-07-25T15:47:14","date_gmt":"2023-07-25T15:47:14","guid":{"rendered":"https:\/\/statorials.org\/tr\/rdeki-welchs-anova\/"},"modified":"2023-07-25T15:47:14","modified_gmt":"2023-07-25T15:47:14","slug":"rdeki-welchs-anova","status":"publish","type":"post","link":"https:\/\/statorials.org\/tr\/rdeki-welchs-anova\/","title":{"rendered":"R'de welch'in anova's\u0131 nas\u0131l ger\u00e7ekle\u015ftirilir (ad\u0131m ad\u0131m)"},"content":{"rendered":"

<\/p>\n

\n

Welch’in ANOVA’s\u0131,<\/strong> varyanslar\u0131n e\u015fitli\u011fi varsay\u0131m\u0131<\/a> kar\u015f\u0131lanmad\u0131\u011f\u0131nda tipik tek y\u00f6nl\u00fc ANOVA’ya<\/a> bir alternatiftir.<\/span><\/p>\n

A\u015fa\u011f\u0131daki ad\u0131m ad\u0131m \u00f6rnek, Welch’in ANOVA’s\u0131n\u0131n R’de nas\u0131l ger\u00e7ekle\u015ftirilece\u011fini g\u00f6sterir.<\/span><\/p>\n

1. Ad\u0131m: Verileri olu\u015fturun<\/strong><\/span><\/h3>\n

\u00dc\u00e7 farkl\u0131 \u00e7al\u0131\u015fma tekni\u011finin farkl\u0131 s\u0131nav sonu\u00e7lar\u0131na yol a\u00e7\u0131p a\u00e7mad\u0131\u011f\u0131n\u0131 belirlemek i\u00e7in, bir profes\u00f6r rastgele 10 \u00f6\u011frenciye her tekni\u011fi (Teknik A, B veya C) bir hafta boyunca kullanmalar\u0131 i\u00e7in atar ve ard\u0131ndan her \u00f6\u011frenciye e\u015fit zorlukta bir test verir.<\/span><\/p>\n

30 \u00f6\u011frencinin s\u0131nav sonu\u00e7lar\u0131 a\u015fa\u011f\u0131da sunulmu\u015ftur:<\/span><\/p>\n

 #create data frame<\/span>\ndf <-data. frame<\/span> (group = rep<\/span> (c(' A<\/span> ', ' B<\/span> ', ' C<\/span> '), each<\/span> =10),\n                score = c(64, 66, 68, 75, 78, 94, 98, 79, 71, 80,\n                          91, 92, 93, 85, 87, 84, 82, 88, 95, 96,\n                          79, 78, 88, 94, 92, 85, 83, 85, 82, 81))\n\n#view first six rows of data frame\n<\/span>head(df)\n\n   group score\n1 to 64\n2 to 66\n3 to 68\n4 to 75\n5 to 78\n6 to 94<\/strong><\/pre>\n
 Ad\u0131m 2: E\u015fit Farklar Testi<\/strong><\/span><\/h3>\n
 Daha sonra, her grup aras\u0131ndaki varyanslar\u0131n e\u015fit olup olmad\u0131\u011f\u0131n\u0131 belirlemek i\u00e7in Bartlett testi<\/a> yapabiliriz.<\/span><\/p>\n
 Test istatisti\u011finin p de\u011feri<\/a> belirli bir anlaml\u0131l\u0131k d\u00fczeyinin alt\u0131ndaysa (\u03b1 = 0,05 gibi), o zaman s\u0131f\u0131r hipotezini reddedebilir ve t\u00fcm gruplar\u0131n ayn\u0131 varyansa sahip olmad\u0131\u011f\u0131 sonucuna varabiliriz.<\/span><\/p>\n
 Bartlett testini ger\u00e7ekle\u015ftirmek i\u00e7in R taban\u0131ndaki a\u015fa\u011f\u0131daki s\u00f6zdizimini kullanan bartlett.test<\/strong> i\u015flevini kullanabiliriz:<\/span><\/p>\n
 bartlett.test(form\u00fcl, veri)<\/strong><\/span><\/p>\n
 \u00d6rne\u011fimizde bu fonksiyonun nas\u0131l kullan\u0131laca\u011f\u0131 a\u015fa\u011f\u0131da a\u00e7\u0131klanm\u0131\u015ft\u0131r:<\/span><\/p>\n
 #perform Bartlett's test\n<\/span>bartlett. test<\/span> (score ~ group, data = df)\n\n\tBartlett test of homogeneity of variances\n\ndata: score by group\nBartlett's K-squared = 8.1066, df = 2, p-value = 0.01737\n<\/strong><\/pre>\n
 Bartlett testinin p de\u011feri ( .01737<\/strong> ) \u03b1 = 0.05’ten k\u00fc\u00e7\u00fckt\u00fcr; bu, her grubun ayn\u0131 varyansa sahip oldu\u011fu bo\u015f hipotezini reddedebilece\u011fimiz anlam\u0131na gelir.<\/span><\/p>\n
 B\u00f6ylece varyanslar\u0131n e\u015fitli\u011fi varsay\u0131m\u0131 ihlal edilmi\u015f olur ve Welch’in ANOVA’s\u0131na ge\u00e7ebiliriz.<\/span><\/p>\n
 Ad\u0131m 3: Welch’in ANOVA’s\u0131n\u0131 ger\u00e7ekle\u015ftirin<\/strong><\/span><\/h3>\n
 Welch’in ANOVA’s\u0131n\u0131 R’de ger\u00e7ekle\u015ftirmek i\u00e7in R base oneway.test()<\/strong> i\u015flevini a\u015fa\u011f\u0131daki gibi kullanabiliriz:<\/span><\/p>\n
 #perform Welch's ANOVA\n<\/span>oneway. test<\/span> (score ~ group, data = df, var. equal<\/span> = FALSE<\/span> )\n\n\tOne-way analysis of means (not assuming equal variances)\n\ndata: score and group\nF = 5.3492, num df = 2.00, denom df = 16.83, p-value = 0.01591\n<\/strong><\/pre>\n
 ANOVA tablosunun genel p de\u011feri ( 0,01591<\/strong> ) \u03b1 = 0,05’ten k\u00fc\u00e7\u00fckt\u00fcr; bu, s\u0131nav sonu\u00e7lar\u0131n\u0131n \u00fc\u00e7 \u00e7al\u0131\u015fma tekni\u011fi aras\u0131nda e\u015fit oldu\u011funa ili\u015fkin bo\u015f hipotezi reddedebilece\u011fimiz anlam\u0131na gelir.<\/span><\/p>\n
 Daha sonra hangi grup ortalamalar\u0131n\u0131n farkl\u0131 oldu\u011funu belirlemek i\u00e7in post hoc testi yapabiliriz. R’de \u00e7e\u015fitli post-hoc testlerin nas\u0131l ger\u00e7ekle\u015ftirilece\u011fini g\u00f6rmek i\u00e7in a\u015fa\u011f\u0131daki e\u011fitimlere bak\u0131n:<\/span><\/p>\n