0

The research is about comparison of different education levels on self-efficacy.

The education level samples size are: Bachelor degree , 18 Diploma, 21 Pre-U ,4 High school/secondaryschool, 17

Self-efficacy is measured in a questionnaire with Likert scale.

Can I analyse these data with one-way Anova using SPSS ?

kjetil b halvorsen
  • 77,844
user320021
  • 3

1 Answers1

2

There's no minimum number required to perform an ANOVA, but when you have very small datasets it is less likely to determine a difference in the means. You might actually be better off making a qualitative assessment by plotting these and describing the differences. I'm specifically looking at the group of 4, which is a really small dataset to make conclusions with. Also, a Likert scale can suffer from the effects of binning, depending on the number of available responses. Basically, give it a try and see what you come up with, and if you don't see differences but suspect there truly are population differences, dig deeper.

If you suspect this should lead to a trend, try plotting like I suggested above and also include a regression line.

Porter
  • 201
  • the results come out as significant differences where higher scores for Bachelor and Diploma than high school. However, can I remove the group of 4 in the analysis to generate the result to make better conclusion? Thank you for your commenting again! – user320021 Apr 29 '21 at 02:36
  • Removing data is always a big debate, in general I'm not a fan of removing any data points unless I can show they don't actually represent the population. I prefer to pool data where relevant. Can you show the relationship for pre- and post- high school graduation? Or can you show a trend, where even those 4 data points help paint a picture? Additionally you could look into a Tukey's HSD test to look at pairwise comparisons. – Porter Apr 29 '21 at 12:49