I am looking to measure attitudes of respondents based on a 7-point Likert scale with 8 items. However, I need to summarize these attitudes into a single score.
Originally, I performed a factor analysis, only keeping a single dimension (Kaiser's Rule), which explained over 70% of total variance.
However, I was told by my colleague that I need to use another method, namely computing a new variable "Attitude" which is simply a mean of the eight items. Apparently this method is supposedly better (I wouldn't know why) and it is dominantly used in marketing research.
Could anyone enlighten me on this subject? Thanks in advance!
