Scale analysis with R
邱浩恩
2016年11月25日
題項分析
載入套件
library(psychometric)進行題項分析
item<-read.csv("D:/104/ML_R/scale_test.csv",
header=TRUE, sep=",")
head(item)## self_1 self_2 self_3 self_4 self_5 self_6 self_7 self_8 self_9 self_10
## 1 4 2 4 4 1 2 4 3 1 4
## 2 4 2 4 4 2 1 4 3 2 4
## 3 4 2 5 5 2 2 4 4 1 4
## 4 4 2 4 4 1 1 4 4 1 4
## 5 4 2 3 3 2 3 4 4 2 3
## 6 4 2 4 3 2 1 4 3 1 4item1<- item.exam(item[,1:10], discrim = TRUE)
head(item1)## Sample.SD Item.total Item.Tot.woi Difficulty Discrimination
## self_1 0.7559976 0.3412282 0.11353359 3.477901 0.5666667
## self_2 0.8914767 0.3682440 0.09903319 2.129834 0.6166667
## self_3 0.7225804 0.4887244 0.29088131 3.574586 0.7583333
## self_4 0.7302545 0.5104715 0.31422734 3.604972 0.8166667
## self_5 0.8950874 0.2573135 -0.02019437 2.187845 0.4583333
## self_6 0.9513181 0.3621139 0.07264161 2.160221 0.5666667
## Item.Criterion Item.Reliab Item.Rel.woi Item.Validity
## self_1 NA 0.2576111 0.08571249 NA
## self_2 NA 0.3278272 0.08816376 NA
## self_3 NA 0.3526546 0.20989461 NA
## self_4 NA 0.3722588 0.22914876 NA
## self_5 NA 0.2299997 -0.01805074 NA
## self_6 NA 0.3440093 0.06900976 NASample.SD–> 題項標準差
Item.total–>題項和量表總分的相關
Item.Tot.woi–>題項和量表總分的相關(扣掉本身題項分數)
Difficulty–>均值
Discrimination–>前25%-33%(U)和後25%-33%(L)之差異
U和L都答對---->0
U全對L都錯---->1
U全錯L都對---->-1
Item.Reliab Item reliability index–>題項一致性信度
Item.Rel.woi Item reliability index–>題項一致性信度(扣掉本身題項分數)
題目信度
item2 <-item1$Item.Rel.woi先卸下套件
題目刪除後全量表信度變化
detach("package:psychometric", unload = TRUE)
library(psych)
item3 <- alpha(item[,1:10],check.keys=TRUE)$alpha.drop[,'raw_alpha']把分析結果合起來
item4 <- as.data.frame(t(rbind(item3, item2)))
names(item4) <- c('總量表信度（刪題）', '題目信度')
head(item4)## 總量表信度（刪題） 題目信度
## 1 0.8384603 0.08571249
## 2 0.8453470 0.08816376
## 3 0.8408557 0.20989461
## 4 0.8462150 0.22914876
## 5 0.8384205 -0.01805074
## 6 0.8455503 0.06900976加上題目名字、顯示三位
row.names(item4) <- names(item[,1:10])
round(item4, 3)## 總量表信度（刪題） 題目信度
## self_1 0.838 0.086
## self_2 0.845 0.088
## self_3 0.841 0.210
## self_4 0.846 0.229
## self_5 0.838 -0.018
## self_6 0.846 0.069
## self_7 0.849 0.204
## self_8 0.876 0.275
## self_9 0.843 0.014
## self_10 0.835 0.067因素分析
利用PCA探索因素數量
fa.parallel(item[, 1:10], fa = "pc", show.legend = FALSE)
## Parallel analysis suggests that the number of factors = NA and the number of components = 2建議因素數是2，看看因素結構
print.psych(fa(item[, 1:10], nfactor = 2, fm = "pa", rotate = "promax"), cut = .3)## Factor Analysis using method = pa
## Call: fa(r = item[, 1:10], nfactors = 2, rotate = "promax", fm = "pa")
## Standardized loadings (pattern matrix) based upon correlation matrix
## PA2 PA1 h2 u2 com
## self_1 0.60 0.55 0.45 1.3
## self_2 0.84 0.64 0.36 1.0
## self_3 0.81 0.65 0.35 1.0
## self_4 0.79 0.56 0.44 1.0
## self_5 0.72 0.61 0.39 1.0
## self_6 0.83 0.64 0.36 1.0
## self_7 0.65 0.42 0.58 1.0
## self_8 0.50 0.18 0.82 1.4
## self_9 0.66 0.51 0.49 1.0
## self_10 0.56 0.56 0.44 1.4
##
## PA2 PA1
## SS loadings 2.73 2.59
## Proportion Var 0.27 0.26
## Cumulative Var 0.27 0.53
## Proportion Explained 0.51 0.49
## Cumulative Proportion 0.51 1.00
##
## With factor correlations of
## PA2 PA1
## PA2 1.00 -0.54
## PA1 -0.54 1.00
##
## Mean item complexity = 1.1
## Test of the hypothesis that 2 factors are sufficient.
##
## The degrees of freedom for the null model are 45 and the objective function was 4.46 with Chi Square of 1589.85
## The degrees of freedom for the model are 26 and the objective function was 0.21
##
## The root mean square of the residuals (RMSR) is 0.03
## The df corrected root mean square of the residuals is 0.04
##
## The harmonic number of observations is 362 with the empirical chi square 37.67 with prob < 0.065
## The total number of observations was 362 with MLE Chi Square = 74.71 with prob < 1.3e-06
##
## Tucker Lewis Index of factoring reliability = 0.945
## RMSEA index = 0.073 and the 90 % confidence intervals are 0.053 0.091
## BIC = -78.47
## Fit based upon off diagonal values = 0.99
## Measures of factor score adequacy
## PA2 PA1
## Correlation of scores with factors 0.93 0.93
## Multiple R square of scores with factors 0.86 0.87
## Minimum correlation of possible factor scores 0.72 0.74