Can anyone advise on splitting continuous data in to high and low groups for analysis with spss ?
I have some continuous data (variable 1 ) and I want to look for differences on (variable 2) which is also continuous data. However I want to examine if scores are significantly different on V2 from those scoring high on V1 from those scoring low on V1. My first thought was to to use a median split to form high and low groups, but I have found several sources saying this is a bad idea, but they do not suggest alternatives.
I am sure this type of analysis must be required often, but surprised the answer is difficult to find. If you have the solution to this problem I would be very grateful and also appreciate any reference for your suggested approach.
I think the alternative is to treat the variables as continuous variables, and not try to split them into groups.
Plotting the two variables. Perhaps looking at simple Pearson correlation or Spearman correlation. Or perhaps fitting a more complex curved model.
In order to group the variables, it might make sense to use cluster analysis.
You might share a plot of your data and see what people think.
After scatterplots you might consider appropriate regression models.. best wishes David Booth
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication