Comparing 2 percentages - how?

#1
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
 
#2
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,
 
#3
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!
 
#4
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
#5
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).