I'm using full profile conjoint analysis for my master dissertation and have encountered some problems regarding interaction of independent variables and statistical significance tests.
About my research: I conducted two experiments, one with 3 attributes (respectively 2, 2 and 5 levels) and one with six attributes (respectively 2, 2, 2, 2, 2, and 3 levels). Full factorial plans would result in 20 respectively 96 cards, but I used a fractional factorial plan.
Both experiments use rating (SCORE) data from 7 point Likert scales. Participants rated 20 'cards' with information on multiple variables presented on each card. In both experiments, the 20 cards encompassed 4 holdouts.
I have used SPSS's CONJOINT PLAN syntax like this:
CONJOINT PLAN='C:....sav' (plan file with card numbers and corresponding attribute levels)
/DATA='C:....sav' (participant responses)
/SCORE=SCORE1 TO SCORE20
This worked nicely, but leaves me with two problems:
1. It doesn't allow for interaction between independent variables, something that is key to my research.
2. It doesn't show statistical significance tests for the respondents' utilities.
Looking into how the CONJOINT PLAN syntax works, I found out it uses a (two-way/factorial) MANOVA. MANOVA outputs seem to give a lot of data on significance and also allow for interactions. This tells me that SPSS can do what I need it to, but I have no idea how to do this.
Could someone explain to me a) how I could calculate utilities, allowing for interaction between two of the independent variables, and b) how I can generate significance tests for these conjoint utilities?
Is a factorial MANOVA the way to go? If so, please explain how.
Lastly, I am a fairly inexperienced user so I would ask that any explanations be extremely simple.
Thank you very much.