Same sample; different variables-paired sample t-test?

[mistermagoo]

I am very confused.
Say for example I have one sample of subjects, and they both do a maths exam and english exam. Does this make them a related sample?
If I wanted to see if there was a significant difference between the scores they got on the exams, I'm pretty sure I wouldnt want to use a paired samples t-test because the scores being compared individually wouldnt make sense.
Have I lost the plot here...? Is it just a correlation to be used, or would you use a one sample t-test and enter in the value of the other variables mean...?
Heeeeeeeelllp!!!!!!!!!!!!!!!!!

[Dragan]

You would use a paired-sample t-test, which is logically equivanlent to a one-sample t-test on the difference scores. Note: I'm assuming that your tests are scaled appropriately so that you can make an adaquate comparison between the means of the math and english test scores - otherwise you could resort the the Pearson correlation as an option.

[mistermagoo]

Ok, so say next I wanted to predict the score on the maths exam via the score on the english exam, I would use a simple linear regression using the english scores as the independent or predictor variable. However if the scales used were different, say the english exam was marked between 0-10, and the maths exam 0-8, and they showed a significant pearsons correlation, would that matter in the regression with SPSS? Or would I need to transform the data so they were both in the same range?

[Karabiner]

You can regress yearly income on age. You can regress number of
crimes in an area on average temperature. Etc. pp.

With kind regards

K.

[mistermagoo]

Great! but what if the two variables don't correlate to begin with? I ran a regression but its come up with an adjusted R square of -.057 which I'm guessing is bad news.