Hi
I have run an exploratory factor analysis on 300 questionnaire responses. We have 54 re-test responses. Our weighted kappa are not that good (0.37-0.47 for factor 1 and 0.45-0.69 for factor 2). My supervisor has suggested, since I've done a factor analysis, that I should do ICC on the aggregated data for test-retest reliability. I have this (below) in my do-file for test-retest reliability, but the ICC's there are for the single items. I am not quite sure how to do what my supervisor is asking. I've searched STATA help and googled without finding a useful answer. Can anyone help please?
use "efa.dta", clear
bysort record_id: keep if _N==2
tempname fil
postfile `fil' str30 item double icc double icc_bias double bias using "retest.dta", replace
foreach v of varlist in_charge - caretaker_involve_lgvgt {
preserve
keep if !mi(`v')
bysort record_id: keep if _N==2
local bias = .
cap noi mixed `v' || record_id:
local icc = .
if _rc==0 {
local bias = r(table)[1,2]
estat icc
local icc = r(icc2)
}
cap noi mixed `v' i.retest || record_id:
local icc_bias = .
if _rc==0 {
estat icc
local icc_bias = r(icc2)
}
post `fil' ("`v'") (`icc') (`icc_bias') (`bias')
restore
}
postclose `fil'
I have run an exploratory factor analysis on 300 questionnaire responses. We have 54 re-test responses. Our weighted kappa are not that good (0.37-0.47 for factor 1 and 0.45-0.69 for factor 2). My supervisor has suggested, since I've done a factor analysis, that I should do ICC on the aggregated data for test-retest reliability. I have this (below) in my do-file for test-retest reliability, but the ICC's there are for the single items. I am not quite sure how to do what my supervisor is asking. I've searched STATA help and googled without finding a useful answer. Can anyone help please?
use "efa.dta", clear
bysort record_id: keep if _N==2
tempname fil
postfile `fil' str30 item double icc double icc_bias double bias using "retest.dta", replace
foreach v of varlist in_charge - caretaker_involve_lgvgt {
preserve
keep if !mi(`v')
bysort record_id: keep if _N==2
local bias = .
cap noi mixed `v' || record_id:
local icc = .
if _rc==0 {
local bias = r(table)[1,2]
estat icc
local icc = r(icc2)
}
cap noi mixed `v' i.retest || record_id:
local icc_bias = .
if _rc==0 {
estat icc
local icc_bias = r(icc2)
}
post `fil' ("`v'") (`icc') (`icc_bias') (`bias')
restore
}
postclose `fil'
