Dear All,

I'm currently investigating the potential effects of transfer of secondary associations between a real parent brand (Schiphol Airport, Amsterdam) and one of its subbrands (See Buy Fly, post-passport shopping centre).

My dependent variable is a customer-based brand equity scale (CBBE). Furthermore, this scale comprises 5 dimensions, value/social image/trustworthiness/attachment/performance. The overall CBBE scale is an aggregate of the 5 dimensions (thus mean average of all dimensions). I'm planning to seperately run a mixed design ANOVA with one DV (CBBE aggregate mean) and a mixed design MANOVA with the five dimensions.

This study is a FIELD study; participants are recruited at the airport, post-passport control, waiting at their departing gates. The stimuli used are the following:

A - See Buy Fly logo, explicitely endorsed by Schiphol --> followed by the CBBE questionnaire about Schiphol
B - See Buy Fly logo, no endorsement --> followed by the CBBE questionnaire about See Buy Fly

Since I'm completely counterbalancing, I'd have one group getting sequence A B, one group getting sequence B A. Notice that so far this is a withing subjects design, with two measurement points (DV is obtained two times, one time after A, one time after B, same scale). Note that I'm interested in possible order effects

However, I'm also adding a between-groups component, the absence or presence of an 'explanatory' link (theoretical background see Bridges, Sood and Keller, 2000). This explanatory link will be introduced between the sequence, e.g. in an AB sequence, the explanatory link will be introduced AFTER A and before B (as to have an effect on B). I'm currently thinking that my design would thus look as follows.

Group 1: AB (no explanatory links, control)
Group 2: BA (no explanatory links, control)
Group 3: AB (explanatory link, treatment)
Group 4: BA (explanatory link, treatment).

Question: HOW MANY FACTORS do I have? I believe that this is a 2 x 2 design --> meaning I need a minimum of N=80. My supervisor thinks is 2x2x2. who's right?

Steven Breij

Dear steven_breij

2x2 makes sense to me, with one within and one between-subjects variable. But what is your supervisor's explanation for his/her interpretation? Does it have to do with the fact that the explanatory link is given between the AB/BA sequence?

I'd be very interested in knowning what would be the third factor.
Thanks

Thanks for the quick reply. If my memory doesn't fail me it had to do with the sequence effects (which are part of my hypotheses: most likely that brand equity (brand associations) will transfer from the primary exposed brand towards the secondary exposed brand) and the fact I use a repeated measure, however these two things are conceptually the same in my opinion.

We did agree on the between-subjects factor by the way! I expect the explanatory links to moderate the potential transfer, I.e. amplify the expected effect.

I have e-mailed my supervisor for feedback on the matter, haven't heard for about 2 weeks so I've decided to stick with my design and am collecting data as we speak.

Came to this forum to look for some reinforcement/critism on my opinion. Will keep you updated though!

Steven

I have yet another question when it comes to terminology of my design. Since I'm conducting my experiment in the field I'm inclined to use an incomplete/quasi-experimental design. As of yet I've named my research a mixed design combined with complete counterbalancing and nonequivalent control group design. It is the latter that I'm concerned about. Since I'm also interested in changes in the "control-groups" (due to sequence/order effects), can I stick to naming them a control group? One could say that, although they do not receive an extra 'treatment', they are exposed to the sequence 'treatment' --> thus are they still a control group or should I name them differently?

Regards,

Steven