0
Age  12-17  18-24   25-34   35-44   45-54   55-64   65+
Category
1       5   16  14  44  13  12  0
2       0   1   3   0   0   0   0
3       3   2   6   1   0   1   0
4       27  3   3   13  0   3   2
5       0   4   4   9   6   7   4

The above is results of questionnaire data, I was going to use the chi^2 test on this but as it has very small values and values of zero, it is not appropriate. I would like to test the if there is a relationship between age and category (which is about where the shop, i.e five 12-17 year olds shop at category 1). What test if any could I use? thanks

whuber
  • 322,774
debbie scott
  • 1
  • You misunderstand the limitations of chi-squared tests. (This is a FAQ). The expected counts need to be reasonably large, not the counts themselves. See some of our other related questions for examples and details. I have linked to one of the more directly useful ones as a duplicate. – whuber Apr 23 '15 at 13:29
  • Thanks for your reply, but the expected counts are also very low in this data, I have looked in the FAQ but cant find an answer for an alternative, any help is greatly appreciated – debbie scott Apr 23 '15 at 13:33
  • Your total is about 200, covering 7*5 = 35 categories, for a mean expected count of 6. That's not too bad. The duplicate I linked to specifically addresses the issue of low expected counts--look at its title! – whuber Apr 23 '15 at 14:00

0 Answers0