Hey,
I'm helping a friend with the technical statistics part of their bachelor thesis.
So we sent out 5 different surveys to 5 different groups of people. All surveys contained images of four brands (IKEA, Coca-Cola, Pepsi and McDonald), one survey contained original brand images and the remaining four contained manipulated colors of the brand (either blue, black, red or green). In all surveys below each brand we asked the respondents to state on a scale of 1 (fully disagree) to 6 (fully agree) how much they associated a feeling (healthy, luxury, cheap, high quality, romantic, etc) for the brand image. We also asked the respondents for their age-group and gender.
So far we have done a Kruskal-Wallis test to see if there is variation between groups for each variable and each brand depending on the color manipulation. The results show that there is a variation between groups for 13 of the cases. As an example the association of luxury and IKEA varied depending on the color. We will do follow-up analysis on these results to see where the variations lie.
However, now we want to do more tests. As an example we want to see if there is a difference in associations between the genders for the different color manipulations. We also want to see if a given color actually is more associated with certain feelings (as an example does black increase associations of luxury and green increase associations of healthy?). Can anyone point out which tests are appropriate to do on SPSS? Also, are there other things we could test for that we didn't?
i think , the best way is to use completely randomized block design anova which really considers the external factor which is color so that you can study while adding the factor of color on the results.
find more information on internet and how to use in SPSS.
thanks
As I understand, the 5 different Surveys (original or coloured blue, black, red or green logos) were (randomly) distributed over the 5 different Groups. If so, it seems to me you should analyze the dependent variables with multilevel models; the respondents are nested within the Groups. You can assess the need for a multilevel model comparing a baseline model with only the intercept with a model allowing the intercept to vary across Groups.
In whatever way you want to analyze your data, you could first reduce the number of dependent variable with the PCA (Principal Components Analysis) and the run the analyses on factor scores.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS