Help with intervals for dependent variables

#1
Hello,

I am needing some quick help. I have searched and searched and can't find a solution. I have some dependent variables that are a number of times a student has gotten into trouble at school. About 1/2 of the students have 0(none) and then the other 1/2 range from 1 to over 30. There are very few students in the upper limits so when I put them into intervals, I have some very large cell counts and then some that are too small. I don't want to change the interval spread because I will lose meaning (there is a lot of difference in interpretation of why a child is in trouble 1-5 times versus 20 times.) Anyway, this data is obviously skewed because of this problem. Can I treat some of the data as outliers? If so, can you explain what to do? I am using some ANOVA and some regression. If you need more info, just ask and hopefully I can explain. THANK YOU!!!!

Dawn
 
#3
hypothesis

The hypotheses that concern these dependent variables have to do with whether a student's family structure is a factor of how many times the student got into trouble.
 
#4
How many independent variables do you have? As I understand you want to predict how family structure influences "getting in trouble".
You need at least 60 students in yout research. As you said, 50% do not get in trouble at all. I would look what family structure "0" trouble have most. The same with the second group (50% who get in trouble). If there is no difference ( I don't know your criterias:both parents, divorced parents, ect). than family structure has no influence.
...
this is not a professional advise:p use at your own risk:wave:
 
#5
thanks

I have over 5000 student records. I guess I'm just not going to be able to get as detailed results as I wanted since no one has any ideas on the interval issue. Thanks!