I am working on a research article about the effects of infant sleep problems on mothers’ well-being and how self-efficacy moderates this relationship. I have three independent variables, which are actually three subscales of the BISQ questionnaire. This questionnaire examines the sleep patterns of infants from 0 to 36 months. Its three subscales are infant sleep, parental perception, and parental behavior. My moderating variable is maternal self-efficacy. I also have six dependent variables: mother’s depression, anxiety, stress, life satisfaction, happiness, and sleep quality, which were measured with 4 questionnaires. For the statistical analysis of my findings, I performed 18 moderator analyses using the Process macro in SPSS, considering three independent variables, one moderating variable, and six dependent variables. However, I am not sure if this is a common or appropriate approach. I have not found any similar articles that do this. Are there any other multivariate techniques that I could use? How can I choose the best data analysis strategy for my research problem? Thank you for your help.
Hello Amir Ali Mazandarani I think you can find the answer to your question in the link below.
https://haskayne.ucalgary.ca/CCRAM/resource-hub
I have sent a help file and a Extension file for further guidance.
It is an interesting topic.
Did you conduct Confirmatory Factor Analysis to determine whether: 1) your three IVs could all be represented by a single "unmeasured" variable (factor), and 2) whether there was any factor structure to your DVs?
In your situation, MANOVA could be applied to test the overall effect of the three independent variables (subscales of the BISQ questionnaire) on the six dependent variables (maternal outcomes) while considering the moderating effect of maternal self-efficacy. First you can employ factor analyses (evaluative) to explore the underlying structure of your dependent variables and reduce them to a smaller set of latent factors, which could simplify subsequent analyses. Then you could apply SEM analysis to assess the relationships between infant sleep problems (latent variable represented by the three BISQ subscales), maternal self-efficacy, and the six maternal outcomes in a comprehensive way.
Anthony Bagherian We have been discussing Structural Equation Modeling and I would recommend that over MANOVA because it makes fewer assumptions. Plus, you do not have to convert the factors into single item indicators.
David L Morgan MANOVA is primarily deployed for comparing means across multiple DVs in different groups, while SEM is a more comprehensive approach that allows for the testing of complex links among variables, including latent constructs. MANOVA focuses on differences in group means, while SEM focuses on the underlying structure and relationships within a set of variables. In this stance, agree we don´t have groups but three independent variables, so SEM would be appropriate.
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning