Dear Stata users,
Dear colleagues,
In the context of a psychometric properties assessment of a hypothesized scale (12 items, scored on 5-point Likert, two factors), we are following IRT to investigate individual item properties.
Specifically, we want to explore Differential Item Functioning (DIF) within two independent groups in our data set. Based on the Stata manual, I understand that the Stata function for Mantel–Haenszel DIF test (.difmh)is limited to dichotomous items only, whereas the function for the Logistic regression DIF test (.diflogistic) should allow items to be ordinal/continuous. Given the fact that all our items are numeric-ordinal, ranging from 1 to 5, we use the .diflogistic function. Our grouping variable is dichotomous (0=reference and 1=focal), and none of our items have missing values.
However, when we use .diflogistic function for a set of items (in this case, one of the two factors) we always get the same output error:
Code: . diflogistic item2-item8, group(group_X) total(sumscore_items2-8)
Output:
. diflogistic item2-item8, group(group_X) total(sumscore_items2-8)
variable item2 has invalid values;
requires item variables be coded 0, 1, or missing
"
I was wondering whether one of you could help me out here. Does anybody have a suggestion what the problem might be here and what I can do to solve it? Has it something to do with the structure of my data or is there maybe something else that I am missing here?
Thank you very much in advance for your help!
Thomas
Dear colleagues,
In the context of a psychometric properties assessment of a hypothesized scale (12 items, scored on 5-point Likert, two factors), we are following IRT to investigate individual item properties.
Specifically, we want to explore Differential Item Functioning (DIF) within two independent groups in our data set. Based on the Stata manual, I understand that the Stata function for Mantel–Haenszel DIF test (.difmh)is limited to dichotomous items only, whereas the function for the Logistic regression DIF test (.diflogistic) should allow items to be ordinal/continuous. Given the fact that all our items are numeric-ordinal, ranging from 1 to 5, we use the .diflogistic function. Our grouping variable is dichotomous (0=reference and 1=focal), and none of our items have missing values.
However, when we use .diflogistic function for a set of items (in this case, one of the two factors) we always get the same output error:
Code: . diflogistic item2-item8, group(group_X) total(sumscore_items2-8)
Output:
. diflogistic item2-item8, group(group_X) total(sumscore_items2-8)
variable item2 has invalid values;
requires item variables be coded 0, 1, or missing
"
I was wondering whether one of you could help me out here. Does anybody have a suggestion what the problem might be here and what I can do to solve it? Has it something to do with the structure of my data or is there maybe something else that I am missing here?
Thank you very much in advance for your help!
Thomas
