I evaluated schools policy implementation using a Likert scale of 1-4. 1-not yet achieved 2-partially achieved 3 satisfactory 4-outstanding
RQ1: What is the extent of policy implementation in the different schools
There were 74 likert items divided into 10 likert scales / variables. I doesn't make sense to get the mean as these are ordinal data. My sample size is small, so I suppose non parametric methods will be appropriate.
1. How do I get the performancesof schools without using the mean?
2. How do I compare performance across schools on all 10 variables / indicators
3. Which reliability tests should I use?
Thanks
Dear Nothando Andile Mbatha ,
You still might be able to validate your scales despite your small sample size. If you are sure about your 10 scales being independent and measuring different constructs the you can use principal component analysis with them individually. Ideally this needs a sample size of about 100 but it can work with smaller samples according to the strength of the first components in each scale. This technique would effectively turn each of your scales into scale (numerical) data which would simplify your analysis. Please refer to my study guide on scale reliability with small samples.
if you cannot assume your scales are independent then you need to take a step backwards and carry out an exploratory factor analysis. This also needs a sample size of 100+, and at least more than 74 in your case. I have another study guide available on this technique.
https://www.google.com/aclk?sa=l&ai=DChcSEwjLmdzJ86fxAhWQ4XcKHQJgBk8YABAAGgJlZg&ae=2&sig=AOD64_37vmPgg-F67H2LmV3wVlYa0aEvfg&q&nis=1&adurl&ved=2ahUKEwj5xM3J86fxAhUSFRQKHWBuBgwQ0Qx6BAgEEAE
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers