0
$\begingroup$

Good day! Our group conducted a classroom research about Social Goals Role on Student Mindset. For this study, we asked 42 college students to answer 2 sets of questionnaire. The first survey consists of 17 items (YES/NO) while the second set consists of 3 items with 6-point Likert scale.

Please help me on what correlation analysis to use. Thank you in advanced!! :)

$\endgroup$
  • 2
    $\begingroup$ If this is homework please tag it as such. Please also update your title to something that reflects the nature of your problem. Tell us about your 3 items. Do they measure the same construct? Would you like to average them and get one correlation between your items and the YES/NO response? Would you like to look at the correlation between your items and calculate a coefficient of reliability? Please read the link here: meta.stackexchange.com/questions/10811/… What have you already done to try to solve your problem? $\endgroup$ – russellpierce Feb 23 '13 at 15:04
  • $\begingroup$ drknexus our group decided to have a descriptive analysis (mean and percentages) for this study if not correlation. the two test measures different constructs, the Social Goals and the Student Mindset..we already have computed the coefficient of reliability of the tests. our only problem now is how can we correlate the scores from two sets of tests with different number of items. thank you for your comment.sorry for asking vague questions.. :) $\endgroup$ – mai Feb 23 '13 at 15:13
1
$\begingroup$

To get to the answer to your question you need to think about how you can turn a set of X numbers (from each test) into a single number that represents the score on the test (and the underlying construct) as a whole. How do your instructors turn your answers from 17 items on a TRUE/FALSE test into a single number for your grade? Another approach with which you are probably familiar is to find a way to turn those several numbers into a single number that best represents the central tendency of each test.

$\endgroup$
1
$\begingroup$

You first need to decide what question(s) you are trying to answer.

You could analyze your data with anything from running basic summary statistics to doing an unsupervised (or supervised) random forest, the first may not answer the questions of interest (or it may) and the second may be overkill (and still may not answer any questions of interest).

How to combine survey answers into a single score (or even if you should do this) depends on what question you want to answer. Whether to do a single analysis of combined data or do a few different analyses on seperate parts of the data depend on the question(s) of interest.

If you give more information on the actual question(s) of interest and how the data relates to the question(s) then we may be able to give better pointers. Actually, often the process of figuring out a good research question will clarify what techniques to use.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.