# Testing differences in survey category responses

Through a market research survey to consumers I'm trying to infer if emotional associations with retail brands differ in some meaningful way. The goal of this analysis is to understand differences in how consumers associate various themes with different brands.

I'm analyzing survey data about brands to determine if the observed differences in the number of respondants answering in a particular way is significant at some level for different brands. Survey respondants are asked

Carrying a [BRAND] bag makes me feel ...

Choose as many of the following as apply.

- fashionable
- confident
- luxurious
- glamorous


Each respondant is asked about at most 3 brands. There are more than 3 brands total and therefore not all respondants will answer questions about all brands.

I'm wondering about approaches to comparing differences across responses, e.g. if 700/1000 respondants say carrying a brand1 bag makes them feel confident vs 100/1000 respondants say carrying a brand2 bag makes them feel confident. Note that these two groups of 1000 will have some overlap but will not be the exact same group of respondants.