Hello!
I have a set of survey questions I am analysing. Response answers are 5 point [strongly disagree, agree, unsure, disagree, strongly disagree]. I did recode into 3 [disagree, unsure, agree] and run a multinomial regression. That was straightforward in terms of comparing, but I feel recoding to 3 points does lose some interesting information regarding those that felt 'strongly' either way... I dont really want to run a massive multinomial.. So, Im keen to open back to 5 point and run ordinal if this provides more useful information? yet to be determined because the results will look the same for each point along the scale. In other words, Im struggling to see how I might know when a predictor is more inclined to answer 'strongly disagree' vs 'agree' or 'somewhat disagree' vs 'disagree'. I may be making it too complicated or missing something. Any thoughts would be great.
- Badges
- Science topic
- Similar topics
- Psychology
- Emotion