PDA

View Full Version : What kind of statistical analysis to use?!?!



cgrif21
07-20-2011, 07:00 PM
I am a grad student working on a research project and I'm not sure what kind of tests to run in SPSS. I have a set of test scores and a set of 4 grades for a group of 200 students. I am trying to see if there is a relationship between the grades and the test scores where possibly I could predict the grades based on the test score. My boss is hoping to possibly identify a specific test score that would be a good predictor of the four grades for example an 80 on the test would most likely indicate a certain grade requirement. Because I am looking for a predictor I'm thinking regression, but I wasn't sure what kind of regression to use and what to do on SPSS. Thanks!

jpkelley
07-20-2011, 07:15 PM
Could you tell us more about your data structure? So, you have multiple test scores for each of the 200 students? How are the grades different than the test scores? Are the test scores typically averaged to give the grade? Or are these test scores on standardized tests and the grades are what the students earned in school? It sounds like there's a hierarchical structure to the data, because there are multiple test scores and grades for each student.

For SPSS, check out one hierarchical modeling page: HERE (http://www.ats.ucla.edu/stat/paperexamples/singer/default.htm).

It seems that you want to use the resultant model for prediction of future values. I would also check out prediction rather than confidence intervals and see if that works with your design and purpose.

cgrif21
07-21-2011, 12:54 PM
The test scores are standardized tests, I only have multiple test scores for about half of the students. The four grades are course grades from the year following the standardized tests.

Thanks so much for pointing me in the right direction!