Super junior stats person here. I have a dataset I am looking to analyze properly. I have administered 2 surveys, one pre and one post. They may or may not be the same participants (was distributed via. email to an entire department, and the department was invited to answer the survey both times -- we could have people who only answered pre or just in the post, so I believe that these groups cannot be considered "dependent" in this case). The data I am looking to analyze is a serious of statements and respondents were asked to rank on a Likert scale from Strongly Agree, Agree, Neutral, Disagree, Strongly Disagree. I am looking to see if these opinions changed between the pre period (first survey) to the next period (post survey), and specifically which opinions changed (were they more likely to agree with the statement in the pre vs. post period).
I have done a chi square test of independence by combining Strongly Agree + Agree, and Strongly Disagree + Disagree, so therefore giving 3 categories (Agree, Neutral, and Disagree). My groups are pre and post. However, I want to know if they changed with respect to Agree or Disagree, not whether or not there was just an overall difference in opinion between the 2 time periods (pre/post groups).
I was wondering 3 things: 1. Are these groups considered truly "independent"? 2. Was the chi square test appropriate to run? 3. Which test would help me answer my interest in knowing which specific opinions changed over that time period? i.e. change in % agreement, or change in % disagreement vs. just a general change in the group.