What are you trying to accomplish in winnowing down your product sets from seven to three within each? How do you intend to use the live study ratings on your final sets? What specific research hypothesis/es are you trying to address in your study?
Would it matter that each category could end up having a different sift of products?
Without knowing more information, it's impossible to offer constructive recommendations. Can you elaborate?
Good afternoon, One of the technic used to select variables reducing the dimensions is the Principal Component Analysis. However, your number of products 14 seems not too large. Why do you decide to keep only 3 hedonic and 3 utilitarian products?
For my research I want to find out how product type and message framing (intrinsic and extrinsic goal framing to be precisely) affect consumers willingness to buy.
The goal of this preliminary test was to find out whether respondents view utilitarian products as utilitarian and hedonic as hedonic to not include products that are seen differently. Now I need to figure out which products I should select from my pre-test to include in advertisements that I am going to create for my main survey (advertisements consist: products with intrinsic goal framed message, extrinsic goal framed message and no goal frame).
My initial idea was to select 3-5 utilitarian products and 3-5 hedonic products, For this question I said 3 but this is not yet fully determined.
I already conducted analyses based on the means for each dimension for utilitarian and hedonic products based on the scale from Voss et al. (2003) but I was wondering if there are other ways and tests to make the product selection more substantiated.
DId you have your pilot study respondents rate each of the purported utilitarian products on the hedonic dimensions/attributes, and vice versa?
Otherwise, how could you be certain that an out-of-category product wouldn't rate higher than the in-category products? (Even if you thought that such an outcome was unlikely.)
If you must reduce each set, then likely you would opt for utilitarian products which had highest average ratings across the five utilitarian dimensions _and_ (ideally) lowest average ratings across the five hedonic dimensions; the opposite would be the target for hedonic products.
I don't think that PCA would be productive here, since any observed coalition of products might be due to some completely different attribute, and not necessarily the 5 (utilitarian) and 5 (hedonic) you've identified.