Which test should I use to compare students' mood effects on academic performance? I want to compare mood measured on a likert scale (from point 1= very sad to point= 5 very happy) to three different test scores. They assigned their mood before taking the test. Any help?
Thank you in advance.
Depending on the tests you use to evaluate and measure both variables and at the level (interval, ordinal, nominal ...) at which the scores of the results are obtained and expressed in them: this will allow us to know if statistical correlations can be made , and of what type, "chi square" tests, ANOVAs ... etc ...
1. ANOVA test: For more than two comparison groups and continuous outcome variable (mood score in this case)
2. Chisquare test: For more than two comparison groups and categorical outcome variable (if you have three mood score category)
A psychologist,David Aderman created a system several years ago that may be what you are looking for.I will contact him by monday and see if I can help.
Depending on the tests you use to evaluate and measure both variables and at the level (interval, ordinal, nominal ...) at which the scores of the results are obtained and expressed in them: this will allow us to know if statistical correlations can be made , and of what type, "chi square" tests, ANOVAs ... etc ...
So you have two continuous variables: mood (measured with a single question) and score on an academic test. (Okay, you have 3 such tests, but let that wait.)
There are two obvious answers here.
One is to calculate the zero-order correlation between mood and test score: most likely Pearson's r, unless you want to be finicky about whether Likert-type scales qualify as interval data, in which case you might rank-order the data and use Spearman's rho. Personally, I'd just go with the standard correlation coefficient.
The other is to run a simple regression, using mood to predict grade. Either way, you'll obtain essentially the same results. (In a simple regression, the beta-weight of the predictor is identical to the correlation coefficient between the variables.)
Why pick one or the other? Conceptually, correlations are about relationships between variables (with no causal implication) while regression analyses more or less assume that the 'predictor' (in your case, mood) is primary, either causing or meaningfully preceding the 'outcome' (grade). Your call! As I say, it won't affect the significance or magnitude of the results in any way.
You do need to decide whether to run 3 separate analyses (one for each test) or combine the test scores into one variable. I'd suggest the latter, because it will allow for improved reliability. Just be sure that they are all on the same scale, so they are weighted equally.
- Badges
- Science topic