Hi everyone. I am a student who would need help identifying a suitable static test for my dissertation work. In practice, I should go to evaluate if there is a statistically significant correlation between two parameters. One is a nominal qualitative parameter, specifically the occlusal class(I°,II°,III° class), while the other is quantitative.
I was unable to find adequate statistical tests for this correlation.
I thank those who will answer me.
Looks like an analysis of variance using the occlusal class as the category.
For the nominal variable, you can create a "square" table of Time-1 by Time-2 and do a Chi-sq test.
For the numerical variable, you can do a paired-comparison t-Test (analysis of variance would apply if you had more than two categories, as opposed to only Time- and Time-2)
Hello Dario,
To Paul Golder's answer, I would add: You'll be able to estimate the variance accounted for by occlusal class membership in the quantitative variable via anova (or the equivalent regression model). This will be an R-squared statistic (in ANOVA, the value is sometimes referred to as an eta-squared statistic).
Recall that R-squared is just the square of the multiple R (correlation between scores on a linear composite and a dependent variable), so if you wish to report the value as a "correlation," give the square root of R-squared or eta-squared.
Some authors note that R-squared or eta-squared tend to be biased upward in magnitude as estimators of the population value. For this reason, options such as the "adjusted R-squared" (in regression) or omega-squared (in anova) are frequently used.
Good luck with your work.
David Morse Since the t-Test is formally equivalent to either a dummy variable regression or a two-category ANOVA, it inherently has an R-sq. The only difference is that some statistical packages (SPSS) do not show it in their default output.
David L Morgan , I don't think I claimed otherwise about the t-test.
I hope all are staying safe.
What is the distribution of the quantitative variable like, and what are the ns in the different groups? On the classes, can you explain what these are to someone outside the field, and say if you have a particular model in mind for the relationship between these and your outcome variable that you are evaluating.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis