Hi, I'm having a problem in the analysis of my research data, where my dependent variable is continuous. I want to check the association of the continuous dependent variable with socio-demographics factors that are categorical.
1. What is the best test to use? Is it ANOVA or Chi square? Why?
2. If I'm using Chi-square, How I'm to decide on the cut-off value?
Appreciate your support. Thanks in advance
Hi Dilini
Of the two, ANOVA. A chi-squared test uses categorical dependent and independent variables. That does not describe your data. Your second question indicates categorising a continuous variable, but this is generally not advised unless you have a good reason. Some of the problems are justifying your cut-off and losing statistical power.
However, ANOVA may not be ideal either. It assumes a normally distributed dependent variable and homogeneity of variances. It also only tells you if there is a significant difference, not which groups are different from one another if you have more than two groups. You need to run a Tukey's post-hoc test for that.
It's pretty much par for the course on this questions board to suggest a different test entirely, but I think a linear regression would be better. It relaxes some of the assumptions around normality of distribution and if you want, you can include all variables in a single multivariable regression model, if you have sufficient observations.
Hope that helps!
Tyler
use ANOVA to describe your dataset. Chi-square only describes the differences of observed counts.
Since your dependent variable is continuous, I recommend ANOVA if Normality, Homogeneity of variance and independence assumptions are fulfilled. Otherwise, nonparametric tests (e.g. Kruskal-Wallis test) are recommended.
Chi-2 is appropriate when the dependent variable is discrete too. ANOVA is the right option. Check out the normality of errors and equality of variance of errors in any socioeconomic group.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA