I have an independent variable (related to teachers in schools) measured by a 6-point Likert scale, a probable mediator variable (related to their students) measured by a 5-point Likert scale and a dependent variable (related to their students) measured by a 4-point Likert scale. All scales are standardized.
I actually have 2 questions:
The first is: Can I use Baron's and Kenny's Mediation Analysis that uses simple and linear regression? I am considering my data as interval. I justified this by the fact social researchers do that frequently.
The second is: Can I convert the 5-point scale to a 4-point scale? I'd like to do that since I am targeting 12-year old students and the 5-point scale, measuring an attitude, has a "Not Sure" option that I am afraid students will choose randomly without thinking.
@Salim
In response to you first question, why you still want to follow Baron and Kenny approach when the academia has move way ahead and there are many other robust approaches available for mediation analysis. I suggest you to follow Hayes. Below is a attached file for your reference. For any further assistance, you can also refer to mediation analysis page on David Kenny's website.
In response to you second question, It seems you've not collected the data yet and still in the process of finalizing survey instrument therefore, you can opt any range on the Likert scale seems appropriate for your research. Good luck with your work 👍🏼
@Salim Sorry I forgot to attach reference paper for mediation analysis by Hayes. Below is the attached manuscript.
You cannot consider your scales as interval. They need to have 7 steps for that. Your scales are ordinal.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude