I am conducting research to investigate changes in negative relating and personality traits following completion of a therapeutic community. The research involves comparing the pre and post therapy psychometric scores for the same participants. The measures I am using are the Eysenck Personality Questionnaire (EPQR), the International Personality Disorder Examination Screening (IPDE-SQ) the Person's Related to Others Questionnaire (PROQ3).
For each psychometric, I have the individual subscale score and the total score for the measure for each participant, pre and post therapy. The IPDE has 10 sub-scales, the EPQR has 6 and the PROQ3 has 8.
How do I go about analysing the data? Published research into this area seems to have used paired t-tests to compare each pre and post subscale and the total scores. However, I have read about the disadvantage of increasing Type I errors with this methodology. Someone has also suggested the use of an MANOVA? But this baffles if I'm honest!
Any help would be greatly appreciated :)
You may benefit from Latent Growth Curve Analysis. This analysis is pretty similar than a multigroup factor analysis, but LGC considers time as an index. Mplus can deal with that, and its youtube channel has some lectures about this method.
https://www.researchgate.net/publication/325723337_Screening_for_Social_and_Emotional_Delays_in_Young_Children_Who_Live_in_Poverty_A_Brazilian_Example has an example of what I'm saying.
As you seem to have two measurement occasions, a possible solution may be Latent Change Score Models, which can easily accommodate multiple DVs.
Gain scores (pre/post) are notoriously difficult in CTT framework. IRT is recommended for proceeding.
- Badges
- Science topic
- Similar topics
- Medicine
- Counseling Psychology