Comparing 2 percentages - how?

laura marie

New Member
Hi again guys,
I need to compare two percentages.

I have the percentage 12% and the percentage 38.5%

I need to state if they are significantly different.

I've been told I need to do a chi-square test but I'm not sure how exactly I'm meant to do that.

Any advice would be much appreciated! Many thanks in advance

Laura

TheEcologist

R purist
Hi again guys,
I need to compare two percentages.

I have the percentage 12% and the percentage 38.5%

I need to state if they are significantly different.

I've been told I need to do a chi-square test but I'm not sure how exactly I'm meant to do that.

Any advice would be much appreciated! Many thanks in advance

Laura
You can use an exact Binomial test;

Online reference:

http://en.wikipedia.org/wiki/Binomial_test.

SPSS;
http://academic.udayton.edu/gregelvers/psy216/SPSS/nominaldata.htm

Other references:

Clopper, C. J. & Pearson, E. S. (1934). The use of confidence or fiducial limits illustrated in the case of the binomial. Biometrika, 26, 404–413.

William J. Conover (1971), Practical nonparametric statistics. New York: John Wiley & Sons. Pages 97–104.

Myles Hollander & Douglas A. Wolfe (1973), Nonparametric statistical inference. New York: John Wiley & Sons. Pages 15–22.

good luck,

laura marie

New Member
Thank you for your reply, unfortunately I have read and re-read the links thoroughly and still don't know how I am meant to compute this.

I only have these four numbers.

The 38.5% is out of 52 participants.
The 12% is out of 114 participants.

I am using SPSS but I don't know how to arrange the numbers to run the test. It is very very frustrating!

TheEcologist

R purist
Thank you for your reply, unfortunately I have read and re-read the links thoroughly and still don't know how I am meant to compute this.

I only have these four numbers.

The 38.5% is out of 52 participants.
The 12% is out of 114 participants.

I am using SPSS but I don't know how to arrange the numbers to run the test. It is very very frustrating!
Lets see if this helps:

You're sample size is probably sufficient (this is subjective) to allow the use of a Chi-squared and/or G-test.

These test might be less difficult to compute and are explained in great detail on the web.

Heres a resource:
http://udel.edu/~mcdonald/statgtestind.html

Mean Joe

TS Contributor
Thank you for your reply, unfortunately I have read and re-read the links thoroughly and still don't know how I am meant to compute this.

I only have these four numbers.

The 38.5% is out of 52 participants.
The 12% is out of 114 participants.

I am using SPSS but I don't know how to arrange the numbers to run the test. It is very very frustrating!
You can do this in Excel (sorry I don't know SPSS). A very simple-minded Excel worksheet is attached, pardon its look.

You know that 38.5% of 52 participants = 20 are "successes" in row 1
You know that 12% of 114 participants = 14 are "successes" in row 2

(See the ACTUAL table, rows 1-3)

Thus you know that 34 of 166 participants (20.5%) are successes overall. So using this overall % of successes, you can determine the EXPECTED table (see rows 5-7)
eg 20.5% of 52 participants = expected 10.6506 "successes"
and 20.5% of 114 participants = expected 23.3494 "successes"

Chi-square test compares the actual successes vs expected successes in row 1, and in row 2 (alternatively, you can say it compares the actual non-successes vs expected non-successes, since successes + non-successes = participants).

Excel has a CHITEST function (see cell A12).