I am conducting a study on the relationship between age (IV) and motivation (DV):
- I am using 5 age groups, with 20 participants in each age group
- My survey has a 5 point Likert scale with 2 statements per DV (12 statements in total for 6 DV's)
- I am also asking participants for their gender, which I want to analyse differences in
What are the best tests to use in SPSS in order to discover significant relationships between age groups and motivation types, as well as gender, age groups and motivation types?
Thanks
Dear Joe,
Two statements per DV is a problem. You need at least three to construct a scale, unless you can combine all your motivation questions together into a single scale. You would first need to run a scale reliability analysis in order to validate that scale. If it was acceptable then the best technique might be Jonckheere-Terpstra, as you might be looking for a trend in motivation scores based on age category.
See my guide on scale reliability analysis: https://www.researchgate.net/publication/280936182_Advice_on_Reliability_Analysis_with_Small_Samples
Technical Report Advice on Reliability Analysis with Small Samples
Joe,
You have several options. If all six DVs are elements of motivation, those scores will likely be correlated, and a multivariate analysis would be a logical choice. A manova with main effects of age (5 levels) and sex (2 levels) and 6 measured variables would allow direct answers to age influence, sex influence, and age x sex interactive influence. If you find significant effects, you could run a discriminant analysis to determine which DVs are most influential in separating groups.
"Plan B" could be a canonical correlation analysis, in which you'd dummy code cell memberships as one set of variables [4 dummy variates for age level; 1 for sex; and 4 for the interaction terms; 9 in all] and use the 6 motivation scores as the other set. That would allow for more direct inspection of which elements on each side contribute most to maximized correlations across sets.
For future reference, the DV scores would be more stable, more reliable, and behave more like continuous variables if they were comprised of more than two Likert-type stimuli each.
Good luck with your work!
peter and david, I think you are on point. I support your suggestions
- Badges
- Science topic
- Similar topics
- Medicine
- Epidemiology
- Age Groups