5-item Likert scales ('strongly disagree', 'disagree', 'neutral', 'agree', 'strongly agree') are sometimes converted to 2 levels ('agree' and 'disagree') for statistical analysis as a dichotomous variable.
I heard that sometimes the item 'neutral' is put under 'disagree'.
Is this method explained any published paper?
This is a poor idea. You are throwing a lot of your data away. Develop likert scale(s) and use ordinal logistic regression. David Booth
Many papers explain NOT to this. For example, MacCallum, R. C., Zhang, S., Preacher, K. J., & Rucker, D. D. (2002). On the practice of
dichotomization of quantitative variables. Psychological Methods, 7, 19-40. doi: 10.1037/
1082-989x.7.1.19
I would go further and say that your item measures two variables :
1. Has the person got an opinion on this statement and
2. What is their opinion, IFF they have one.
The factors that predict having an opinion in the first place may be different to the ones that predict the degree of agreement.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey