1

I am trying to do statistical analysis for a user survey. The total number of survey to be analysed is 59. The variables in the survey can be defined as a categorical variable. Some of the contingency tables for the survey looks as follows:

2   1
11  13
0   32

11  9   1
14  23  1

13  7
23  15

I am confused regarding whether to use fisher test or chi square test? As not all the tables are of 2*2 matrix, as well as some of the tables have expected values of less than 5. (The total number of responses vary in each table).

If I have to use fisher test, I use F-Test Two-Sample for Variances in excel. The solution by that method gives the following terms. What does these term mean?

F
P(F<=f) one-tail
F Critical one-tail

At last, when can I reject null hypothesis in excel.

gung - Reinstate Monica
  • 132,789
  • 81
  • 357
  • 650
Punit Gandhi
  • 11
  • 1
  • 2
  • 3
    "If I have to use fisher test, I use F-Test Two-Sample for Variances in excel." -- I'm sorry, this makes no sense at all. Why would you do a two sample test for variances? You said you wanted to do a Fisher test (i.e. a Fisher exact test for contingency tables). They're not at all the same thing. – Glen_b Dec 20 '15 at 10:14
  • There are several posts about Fisher vs $\chi^2$, such as [this one](http://stats.stackexchange.com/questions/14226/given-the-power-of-computers-these-days-is-there-ever-a-reason-to-do-a-chi-squa). According to @FrankHarrell, the "$\geq 5$" requirement is a myth. There are other ones in the list of related posts at right of your post. – Stéphane Laurent Dec 20 '15 at 12:58

1 Answers1

1

To solve yours problems with performing independence test in Excel I recommend you this two links:

http://www.real-statistics.com/chi-square-and-f-distributions/independence-testing/

http://www.real-statistics.com/chi-square-and-f-distributions/fishers-exact-test/

If you're using Fisher's Exact Test, you should define whether to use one-tailed or two-tailed test. The clear explanation (with online calc for FET) is here: http://quantpsy.org/fisher/fisher.htm

To get to know the theory behind this test, I would rather recommend readinng at least some chapters from http://www.biostathandbook.com/ , particularly:

  1. http://www.biostathandbook.com/chiind.html
  2. http://www.biostathandbook.com/fishers.html
  3. http://www.biostathandbook.com/small.html
Adam Przedniczek
  • 1,082
  • 2
  • 10
  • 24
  • @Punit Gandhi Does this solve your problem? Do you manage to perform Fisher's Exact Test in Excel or do you have further issues with that? – Adam Przedniczek Dec 22 '15 at 15:31