双参数模型中系数的置信区间 - ltm
Confidence intervals for coefficients in two-parameter model - ltm
使用ltm package,我计算了这个双参数模型:
TPLMODEL <- ltm(data ~ z1, IRT.param = TRUE)
并通过简单地调用 TPLMODEL
.
来显示系数
如何显示或计算所有系数的置信区间?
您可以使用使用 summary.ltm
得到的标准误差来计算它们。 示例:
library("ltm")
model <- ltm(LSAT ~ z1)
(ms <- summary(model)$coefficients)
# value std.err z.vals
# Dffclt.Item 1 -3.3597341 0.86694625 -3.875366
# Dffclt.Item 2 -1.3696497 0.30733647 -4.456515
# Dffclt.Item 3 -0.2798983 0.09966725 -2.808328
# Dffclt.Item 4 -1.8659189 0.43411993 -4.298165
# Dffclt.Item 5 -3.1235725 0.86998079 -3.590393
# Dscrmn.Item 1 0.8253715 0.25806406 3.198320
# Dscrmn.Item 2 0.7229499 0.18670550 3.872141
# Dscrmn.Item 3 0.8904748 0.23261695 3.828074
# Dscrmn.Item 4 0.6885502 0.18516593 3.718558
# Dscrmn.Item 5 0.6574516 0.21000512 3.130646
ci <- ms[,1] + qt(1-.05/2, Inf)*ms[, 2] %*% cbind(-1, 1)
ms <- `colnames<-`(cbind(ms, ci), c(colnames(ms), paste0(c(2.5, 97.5), "%")))
ms
# value std.err z.vals 2.5% 97.5%
# Dffclt.Item 1 -3.3597341 0.86694625 -3.875366 -5.0589176 -1.66055071
# Dffclt.Item 2 -1.3696497 0.30733647 -4.456515 -1.9720181 -0.76728130
# Dffclt.Item 3 -0.2798983 0.09966725 -2.808328 -0.4752425 -0.08455411
# Dffclt.Item 4 -1.8659189 0.43411993 -4.298165 -2.7167784 -1.01505952
# Dffclt.Item 5 -3.1235725 0.86998079 -3.590393 -4.8287036 -1.41844152
# Dscrmn.Item 1 0.8253715 0.25806406 3.198320 0.3195752 1.33116775
# Dscrmn.Item 2 0.7229499 0.18670550 3.872141 0.3570139 1.08888600
# Dscrmn.Item 3 0.8904748 0.23261695 3.828074 0.4345540 1.34639566
# Dscrmn.Item 4 0.6885502 0.18516593 3.718558 0.3256316 1.05146875
# Dscrmn.Item 5 0.6574516 0.21000512 3.130646 0.2458491 1.06905405
注意:要应用稳健的标准误差,您可以summary(model, robust.se=TRUE)
。
使用ltm package,我计算了这个双参数模型:
TPLMODEL <- ltm(data ~ z1, IRT.param = TRUE)
并通过简单地调用 TPLMODEL
.
如何显示或计算所有系数的置信区间?
您可以使用使用 summary.ltm
得到的标准误差来计算它们。 示例:
library("ltm")
model <- ltm(LSAT ~ z1)
(ms <- summary(model)$coefficients)
# value std.err z.vals
# Dffclt.Item 1 -3.3597341 0.86694625 -3.875366
# Dffclt.Item 2 -1.3696497 0.30733647 -4.456515
# Dffclt.Item 3 -0.2798983 0.09966725 -2.808328
# Dffclt.Item 4 -1.8659189 0.43411993 -4.298165
# Dffclt.Item 5 -3.1235725 0.86998079 -3.590393
# Dscrmn.Item 1 0.8253715 0.25806406 3.198320
# Dscrmn.Item 2 0.7229499 0.18670550 3.872141
# Dscrmn.Item 3 0.8904748 0.23261695 3.828074
# Dscrmn.Item 4 0.6885502 0.18516593 3.718558
# Dscrmn.Item 5 0.6574516 0.21000512 3.130646
ci <- ms[,1] + qt(1-.05/2, Inf)*ms[, 2] %*% cbind(-1, 1)
ms <- `colnames<-`(cbind(ms, ci), c(colnames(ms), paste0(c(2.5, 97.5), "%")))
ms
# value std.err z.vals 2.5% 97.5%
# Dffclt.Item 1 -3.3597341 0.86694625 -3.875366 -5.0589176 -1.66055071
# Dffclt.Item 2 -1.3696497 0.30733647 -4.456515 -1.9720181 -0.76728130
# Dffclt.Item 3 -0.2798983 0.09966725 -2.808328 -0.4752425 -0.08455411
# Dffclt.Item 4 -1.8659189 0.43411993 -4.298165 -2.7167784 -1.01505952
# Dffclt.Item 5 -3.1235725 0.86998079 -3.590393 -4.8287036 -1.41844152
# Dscrmn.Item 1 0.8253715 0.25806406 3.198320 0.3195752 1.33116775
# Dscrmn.Item 2 0.7229499 0.18670550 3.872141 0.3570139 1.08888600
# Dscrmn.Item 3 0.8904748 0.23261695 3.828074 0.4345540 1.34639566
# Dscrmn.Item 4 0.6885502 0.18516593 3.718558 0.3256316 1.05146875
# Dscrmn.Item 5 0.6574516 0.21000512 3.130646 0.2458491 1.06905405
注意:要应用稳健的标准误差,您可以summary(model, robust.se=TRUE)
。