What is the best method to establish psychometric properties for a questionnaire based on a 3 point rating scale?
Is it possible to use Exploratory Factor Analysis to reduce the items in the form of factors?
Yes you can test the psychometric property for an instrument with items at ORDINAL level {assuming your scale has the 3 step in order, and not one of them as not applicable or don’t know}
Also, you need to have number of cases {preferred} 10 cases per item
If the scale validity is already established in other setting or other language, then it is preferred to do the confirmatory factor analysis. If the result support the original instrument structure, then no need to do the Exploratory factor analysis. However if the CFA did not support the original tool validity, then you have to do EFA.
Hi,
It depends of number of observations and items. If you have enough data, you can used a two steps method: Exploratory Factor Analysis then Confirmatory Factor Analysis.
In few words, Exploratory Factor Analysis aims to extract latent variables in a data-driven approach. And Confirmatory Factor Analysis aims to evaluate measurement and structural models on your data.
Dear Alishba
I would like to add: If possible, try to establish convergent validity by also selecting an instrument that comes close to test whatever it is that you want to measure.
Yes you can test the psychometric property for an instrument with items at ORDINAL level {assuming your scale has the 3 step in order, and not one of them as not applicable or don’t know}
Also, you need to have number of cases {preferred} 10 cases per item
If the scale validity is already established in other setting or other language, then it is preferred to do the confirmatory factor analysis. If the result support the original instrument structure, then no need to do the Exploratory factor analysis. However if the CFA did not support the original tool validity, then you have to do EFA.
You can read about the methods in one of theses books:
Brown, T.: (2006). Confirmatory factory analysis for applied research. New York: Guilford.
This book shows examples for the PC-programs LISREL, EQS, Amos, SPSS, SAS and Mplus.
The follwing books uses the best of these programs, Mplus:
Byrne, B.M. (2012). Structural equation modeling with Mplus. Basic concepts, applications and programming. New York: Routledge.
Geiser, C. (2013). Data analysis with Mplus. New York: Guilford.
You can also learn a lot at the web site for Mplus: www.statmodel.com. Here are questions and answers, as well as videos of teaching: Look at the section: "Mplus Web Training and Handouts"
Best,
I'd also recommend you analyze the data with item response theory, using models like Andrich's rating scale model or Samejima's graded response model. These are in some ways a repackaging of factor analysis but are very useful for other reasons. I'd recommend you determine your scales/subscales with factor analysis but then evaluate the psychometric characteristics of them with IRT.
I would like to add that, nowadays, according to some authors, it is preferable to use a polychoric correlation matrix instead of a Pearson one for the Exploratory Factor Analysis of ordinal variables (like Likert-type items).
For example:
Article Polychoric versus Pearson correlations in Exploratory and Co...
The IRT (item response theory can be utilized to establish the hproperties of a questionnaire. Free and paid softwares such as R and Excaliber etc. can do the needful. They provide good comprehensive outcomes. As Thompson suggested Samejimaa's graded model is better one.
Dear Alishba,
Using IRT is appropriate if you focus on the questionnaire items, but if you want to focus on the questionnaire and the dimensions of the questionnaire (or its components) it is better to use factor analysis. When using factor analysis, it is necessary to use the tetrachoric correlation matrix. I will not focus on factor analysis because there is a lot of information about factor analysis in previous answers. If you have the time, knowledge and motivation you can apply both. So you will have quite detailed information.
- Badges
- Science topic
- Similar topics
- Psychology
- Psychometrics