sorry , here is some information on my study, am doing a cross sectional study to assess the stress level among students and the common strategies for coping they are using.
among the scles that i am using in my study is the Brief COPE inventory. now i am in analysis stage.
any one knows about this scale?
my objectic is to know the most common coping stratigies my respondants use to cope with stress
i calculate the mean and SD for each of the 14 subscales (which represent 14 copiing stratigies), and based on that , the higher the mean the more common the coping style... is that right?
i cant use the frequency because each subsclea has 4 likert responces (1 not doing.... 4 doing it all the time) and i dont know how to score them or compute them , as in the main paper and in all my litherture no hint about that.
and is it ok to use mean to compare as its more presentitive than median , despite the data not normal depending on that n is more than 30 (mine is 120)?
sorry , here is some information on my study, am doing a cross sectional study to assess the stress level among students and the common strategies for coping they are using.
A little about it. I'm a bit sceptical of how much value it is to know about 14 different coping styles, or whether 2 items per subscale is really a good way to go about things, but c'est la vie.
Based on your description of the response scale, it sounds like the higher the mean, the more often the study participants are using the particular coping style.
There is a lot of debate about whether it is ok to treat Likert scales as interval or only ordinal in nature, and no easy answer to this. If you regard Likert scales as interval, using the mean is ok, but if you regard them as ordinal, using tests based on the mean would not be ok. Here are some articles on the subject:
Likert scales: How to (ab)use them
Resolving the 50-year debate around using and misusing Likert scales
I would not place much stock in the N > 30 rule of thumb.
The fact that each subscale of the brief COPE has only 2 items, and each item only 4 response options makes me hesitant to recommend parametric mean-based tests, if you want to go beyond descriptive analyses. (Because the data will be far from continuous or normally distributed). You may be better served by using a test for medians, or a rank-based differences test such as the Wilcoxon test. But with 14 subscales and therefore 91 pairwise difference comparisons to make, you'd want to be quite careful about why you're comparing the means of any two subscales!
Tweet |