I'm working on a research proposal paper for my Senior Seminar in Psychology course, and I am unsure which statistical model to use. My research question is whether victims of workplace cyberbullying have higher absenteeism rates than employees who do not experience cyberbullying. My proposed hypothetical study will utilize a cross-sectional self-reported survey methodology. I believe Pearson's r would be appropriate to analyze the data, but I'm not sure.
This will depend on how your variables are calculated - namely, whether they are categorical or continuous.
Pearson correlations are good when both variables are continuous. This would mean that workplace bullying is measured continuously, perhaps based on frequency of bullying occurrences; and absenteeism rates are measured by the number of days missed at work.
Chi-Square is best if you have two categorical variables. A possible way of having this work with your variables would be to have workplace bullying as a dichotomous yes/no answer, or having absenteeism as a yes/no. You can also choose more categories for these variables although a chi-square analysis becomes more difficult to interpret once you have 4-5+ categories.
Based on your variables, I think having these both measured continuously may be best. Your sample size will also be important to consider with regards to the Power needed for these analyses. Alternatively, you can also look into t-tests if you want to have a single continuous (i.e., absenteeism days) and a single categorical (i.e., workplace bullying, yes/no) variable.
it depends on the kinds of variables you have - if they are all categorical (which is what it sounds like they are or could be) then cross-tabulating (ie chi-square) would be the simplest way to go (and the easiest to interpret); eg, 'experienced cyberbullying' (yes/no) crossed with 'absenteeism' (hi, med, low...or whatever you think is most meaningful)
To follow up on the earlier responses, if both bullying and absenteeism are continuous variables with normal-ish distributions, then you could use a correlation test. (But it does sound like the cyberbullying is a dichotomous variable.)
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning