|
|
Various types of indicators |
|
Message/Author |
|
Hongbo Wang posted on Wednesday, April 02, 2008 - 12:51 pm
|
|
|
Hi, I'm creating a scale of self-control based on a big set of selected survey items. They scatter across different questionnaires and, as a result, are measured quite differently. I have responses coded using a 5-point scale, say, "Never" "Once a month" "Once a week" "Once a day and more" while others scaled from "Very important" to "Important" "Indifferent" to "unimportant" to "Not important at all". There are continuous/count responses as well. So what would you suggest me to do with these indicators for a 1 or 2-factor CFA? I'm assuming Mplus allows for indicators measured differently. I'm currently thinking of dichotomizing them first but worry that it can lead to artificial results... Thanks, Hongbo |
|
|
In Mplus factor indicators can be continuous, censored, binary, ordered categorical (ordinal), counts, or combinations of these variable types. |
|
Hongbo Wang posted on Wednesday, April 02, 2008 - 3:59 pm
|
|
|
Hi Linda: Thank you! But on what conditions would you suggest dichotomization of all indicators? I roughly recall that it could be a problem when indicators have too many categories (>4?)... Best, Hongbo |
|
|
No need to dichotomize 5-category variables; they can be treated as categorical (i.e. ordinal). |
|
Hongbo Wang posted on Thursday, April 03, 2008 - 12:58 pm
|
|
|
Dear Dr. Muthen: Thanks. Best, Hongbo |
|
Back to top |
|
|