My research question is to find if there is a relation among body image dissatisfaction, disordered eating and suicidal ideation among adolescents. The sampling technique I am using is Purposive sampling. Thus, the data obtained will not be normally distributed. For the first hypothesis, to find if there is a relation between the three variables , I am thinking of using Scheirer Ray Hare test. Is this correct? What are some alternatives? I also found multinomial logistic regression can be used for this. But, I could find it's non parametric equivalent. Kindly guide further.
You are confused about regression. Start by searching on what regression should I use. Best wishes David Booth
I don't understand what you about sampling and distributions. The sampling procedure can affect the distribution of different variables, but unless you tell use more about your sampling I don't see why, from this, you'd assume they are not drawn from normal distributions (and of course they aren't, "all assumptions are wrong" Tukey, but that is a different issue. The technique I suggest is a scatter plot.
BTW purposive sampling is non probability sampling so it won't generalize anyway. David Booth
WHY? Why "I can only use non random (non probability) sampling technique for drawing the sample. Thereby, the data so obtained will not be normally distributed." Yes, no real distribution is likely exactly normal, but this isn't what causes that. And why "so I can use only a non parametric test"? If you want to estimate a parameter, do it. If you mean a test that makes less assumptions about the distributions, this is different. I think from David Eugene Booth 's and my comments, it is probably best to consult a statistician at your university. I'd recommend doing this before any further data collection.
A little tricky to understand. Which do you consider to be independent variables and which are dependent. Are there some variables that are independent and some (more than one) that are dependent? Or are he variables just a cluster of variables yet not differentiated into independent or dependent?
How large is the sample, on what level are the variables measured etc.
- Badges
- Science topic
- Similar topics
- Computer Science
- Research Topics