a-paramter in bifactor model

[Yutaro Sakamoto]

Hi,

I'm trying to do bifactor model and estimate a-parameter in G ,F1, F2, F3 factor in mirt.

data: 25 items (N=9,000)

But two a-parameters I estimeted are negative.

Before I questioned this problem, Phil said that this often occer because slopes are 'partial regression coefficients'.

this means multicollinearity in regression analysis?

When a-parameter is negative, must I change estimetation method(eg. MHRM→EM) ?

thanks

[Phil Chalmers]

It has nothing to do with the estimator. Consider the same phenomenon as followed with simple linear regression. Note that although all correlations in the population generating model are positive the regression slopes for the second predictor is negative. 

library(mvtnorm)

C <- matrix(c(1, .5, .1,

              .5, 1, .3,

              .1, .3, 1), 3, 3)

set.seed(1)

dat <- data.frame(rmvnorm(9000, sigma=C))

colnames(dat) <- c('y', 'x1', 'x2')

lm(y ~ x1 + x2, dat)

Phil

--
You received this message because you are subscribed to the Google Groups "mirt-package" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mirt-package...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Yutaro Sakamoto]

Hi,Phil

thank you for your advice.

OK...,

I doubt that if a parameter in group factor(in bifactor model) is negative  , that means the item scarcely do not reflect group factor?

I think if item reflect group factor, that a parameter in the item(s) is positive.

thanks

6年5月26日木曜日 11時27分30秒 UTC+9 Yutaro Sakamoto: