1

I want to extract the variance that is common in 3 variables from my items. I believe I could perform a CFA and fix the loadings to be equal and then extract the "measurement error," which I expect would be scores per item for every datapoint, such as you can extract the factor scores for every data point.

It seems that this is not really common and I can't find code explaining how to do this.

Anyone got an idea?

hamagust
  • 728
  • 2
  • 10
  • 28
BrentHendrickx
  • 11
  • 2

1 Answers1

0

You can define an orthogonal single-indicator construct per indicator to capture its unique-factor variance, by fixing its residual variance to 0.

HS.model <- ' visual  =~ x1 + x2 + x3
   e1 =~ 1*x1  ;  x1 ~~ 0*x1  ;  e1 ~~ NA*e1
   e2 =~ 1*x2  ;  x2 ~~ 0*x2  ;  e2 ~~ NA*e2
   e3 =~ 1*x3  ;  x3 ~~ 0*x3  ;  e3 ~~ NA*e3 '
fit <- cfa(HS.model, data = HolzingerSwineford1939,
           orthogonal = TRUE, std.lv = TRUE)
head(lavPredict(fit))

Then you can get factor scores for the unique factors as well as common factors.

Terrence
  • 780
  • 4
  • 7