I am clueless and need help fast. I am new to this forum so I hope I'm in the right place!
Anyway, my research study is about determining whether or not there is a relationship between Parental Behaviors and self-esteem in adolescence. Each subject completed two measures, one for perceived parental behaviors and the other to assess the respondent’s self-esteem. Each measure has four scales and the items use a five point lickert scale of strongly agree, agree, not sure, disagree and strongly disagree. These were scored from 5-1 for positive statements and 1-5 for negative statements.
How do I correlate the two measures to establish whether there is a relationship or not?
This thread might help you!
How many items are in each of the four sub-scales? Generally, I would sum the scores on the scales, convert them to rank order (Likert scales are ordinal data), and use a Spearman Correlation to examine the relationship between the variables. Tied ranks could be a problem depending on the range of the summed scale scores.
thanks for the thread Katie.. very helpful! Jacque, some sub scales have more items than others so I summed the scores of each scale and eventually, each questionnaire and generated percentages to make them equal and easy for me to compare. I hope I did the right thing. The minimum score for each scale was 20% and the maximum 100% so I considered scores falling below average (70%) low and those above average high. Then proceeded with conducting a spearman correlational analysis.
It might be justified in light of your research goal, but you lose quite some information by only distinguishing between high and low scores.
Thanks for that observation Ingo. However, how do you suggest I go about it so as to capture as much info as possible?
Honestly, I am not exactly sure how you aggregated the scales, but a first step would be not to impose a threshold at 70%. As I understand it, you then have a summative index comprising multiple scales. Am I right?
*sorry, the average is actually 60% and not 70%*
well the measures have subscales with individual scores whose totals make up the scores for each measure. I ran correlations across subscales to determine relationships among scales before I proceeded to correlate the overall scores. I ranked scores above 60% 'high' with a value of 1 and those below 'low' with a value of 2. I used those values in my analysis.
hmmm... hope it makes sense. If I am not right, please let me know!
I would not say that it doesn't make sense, but it depends on what you find out. If a dichotomous measure fits your research aim, you should go forward with this. You just should be prepared that other people will also ask why give up information by dichotomizing your scales.
But: if you expect the subscales to measure the same construct, you may also assess their reliability and assess their validity by a principal component/factor analysis. A good text to start still is Carmines, E.G. and Zeller, R.A. (1979): Reliability and validity assessment, Beverly Hills, Calif.: Sage Publications.
You have me thinking maybe what I have done is too simplistic... I could get more from my data.
Thank you for your patience!