sample comparison question
Locked
 
|
Hi all:
I have a subsample of 900 taken from a population of 1700 that I would like to compare on a multiple demographics. In short, I would like to know if this sample of 900 is similar to the popluation of 1700 for demographics such as gender, ethnicity, etc. I am using version 19.0 of SPSS/PASW. Does it do a one-sample z-test of proportions, or any other technique that people think is reasonable to compare a sample proportion to a population proportion? Thanks! J ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Administrator
|
Taking the mean as an example, I think that the only way the mean of the sample of 900 can be the same as the mean of the the entire group (which I hesitate to call a population--see below) is if the means for the groups of 900 and 800 are the same. So in other words, you can do a two-group test comparing those two independent means. (I think this issue came up some time ago in one of the sci.stat.* newsgroups, and I believe Ray Koopman posted this same suggestion there. I just can't find the old thread right now.)
The reason I hesitate to call your 1700 a population is that it is rare to have a case where one truly has no interest in generalizing beyond the data in hand. What are the details of your situation? HTH.
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
In reply to this post by Jordan Kennedy
It is *never* statistically appropriate to compare
a subgroup to the total of which it is a part; though, occasionally, like when there are many subgroups, it can be convenient to carry out the comparison that way, and if the story is simple enough, you can tell it that way. If you want to know whether the 800 are like the 900, you compare the 800 to the 900. -- Rich Ulrich > Date: Mon, 26 Sep 2011 10:43:07 -0700 > From: [hidden email] > Subject: sample comparison question > To: [hidden email] > > Hi all: > > I have a subsample of 900 taken from a population of 1700 that I would > like to compare on a multiple demographics. In short, I would like to > know if this sample of 900 is similar to the popluation of 1700 for > demographics such as gender, ethnicity, etc. I am using version 19.0 > of SPSS/PASW. Does it do a one-sample z-test of proportions, or any > other technique that people think is reasonable to compare a sample > proportion to a population proportion? Thanks! > |
|
I used to do something like this with inexperienced students to demonstrate variation in sample estimates (eg % female or mean age) by taking several sub-samples n from “population” N where n could range from 30 to 300 and N was 3025 cases from the 1989 British Social Attitudes survey. This also demonstrated that size of sample matters for confidence limits. We used to do it the hard way using TEMP. SAMPLE < n> from <N> . a couple of times per student. This didn’t yield many examples and rarely demonstrated the point, but David Marso recently sent me some very neat code to produce 100 or more subsamples from which I was able to make charts to illustrate the point. These will eventually be incorporated into the SPSS tutorials on my site. John F Hall From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rich Ulrich It is *never* statistically appropriate to compare > Date: Mon, 26 Sep 2011 10:43:07 -0700 |
| Free forum by Nabble | Edit this page |