SPSSX Discussion

correlation for two different samples

Classic

List

Threaded

2 messages Options

abdalla alsmadi-2

correlation for two different samples

Hi everybody

i have a data of two variables

the degree of kids satisfaction about their school

and

their parents degree of participation in school's activities

can i call this as matched pairs?

can i do a correlation coefficient between these two variables? what is the logic?

thanks a lot

Abdalla Alsmadi college of graduate studies Arabian Gulf University - Bahrain Tel: 0097317239999 ext. 676

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

Rich Ulrich

Re: correlation for two different samples

If you have one record per family, you can certainly compute the correlation

between these two variables. It is just an ordinary correlation between two

different variables.

Where school data (with kids and parents) potentially offers a problem is when

you have two kids in the same family, with the same "family" variable used twice;

those two values are not "independent". You can (a) ignore the duplication; (b)

use the average score for kids; (c) chose the older or younger; (d) obtain a version

of r from an ANOVA. My own choice would depend on who my audience is, and how

many "duplicated values" of that sort exist. (d) is the most "proper", though it

probably needs computation by hand from an ANOVA.

The term "matched pairs" makes me think of a paired t-test, which has the same

variable measured for left vs. right, for two individuals, or at two times, or the like.