Complex samples, with replacement PPS, why only one stage?

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

davirix
Posts: 4
Joined: Mon Sep 10, 2012 3:41 am

Complex samples, with replacement PPS, why only one stage?

Postby davirix » Mon Sep 10, 2012 5:39 am

Hi All,

First post, new SPSS user, struggling with Complex Samples functions.

SPSS ver. 16 on windows 7.

I'm working on a survey, stratified (in provinces), 40 cluster (County) selected with systematic PPS in every Province, then one Village selected (simple random selection) in every cluster/County and 20 cases selected (simple random selection) in every Village. Cluster could be selected twice Some clusters have been selected twice, due to their large size, so I suppose that I should consider this a WOR method for PPS.

However when I want to calculate the weighing factors with SPSS complex samples, and I chose the PPS systematic option I think SPSS consider it as WOR sampling. In fact at the end of the procedure I got this message: ""the maximum Measure of Size (MOS) value across population units cannot be greater than the total MOS in the population divided by the samples size." which I think means that one or more clusters are so large that they would be reselected, which is not compatible with the systematic PPS WOR method, and so the error message.

If on the other side a select a 'simple' PPS and then chose a WR method, SPSS would not allow a second stage of sampling.

So my questions are:
(i) is systematic PPS always WOR?
(ii) why SPSS does not allow multiple stage sampling when WR sampling is selected at first?
(iii) what would be the best procedure to estimate weighing factors in this survey?


Thanks,

David
statman
Administrator
Posts: 2757
Joined: Tue Jun 12, 2007 12:08 pm
Location: Florida, USA

Re: Complex samples, with replacement PPS, why only one stag

Postby statman » Mon Sep 10, 2012 1:02 pm

A lot has changed between V16 & at least V18 wrt the CS module. If you want to post or email me your dataset and output viewer files with take a look (running V19-V21 so can see what might be happening)

FYI, V16 could be unstable in W7 (not certified beyond XP) so make sure you have all the patches applied as well as saving your data frequently
See the note below

NOTE: Please read the Posting Guidelines and always tell us your OS, the SPSS version and information about your study and data!

Statman
Statistical Services
davirix
Posts: 4
Joined: Mon Sep 10, 2012 3:41 am

Re: Complex samples, with replacement PPS, why only one stag

Postby davirix » Tue Sep 11, 2012 3:33 am

Thanks Statman. Please find attached a dummy database that I am created and am using for pre-testing data analysis, the final data are not yet ready but variables will be the same. I am also the .csplan file for calculating the weighing factors which is giving the error message (also attached as output file).

But in any case, do you know why with SPSS (in both 'select a sample' and 'prepare for analysis' wizard) a WR sampling does not allow for a further sampling stage?

Best,
You do not have the required permissions to view the files attached to this post.
statman
Administrator
Posts: 2757
Joined: Tue Jun 12, 2007 12:08 pm
Location: Florida, USA

Re: Complex samples, with replacement PPS, why only one stag

Postby statman » Tue Sep 11, 2012 12:50 pm

Yes, PPS systematic is WOR but could use straight PPS and then WR

If I am clear as to your need, CS needs the sample selection variables (Stage 2 cluster variable) to continue if you have multiple stages as it seems you do but I don't see Village in the dataset

Are we on the same page?
davirix
Posts: 4
Joined: Mon Sep 10, 2012 3:41 am

Re: Complex samples, with replacement PPS, why only one stag

Postby davirix » Wed Sep 12, 2012 12:37 am

Thanks statman. My first post was simplified for better understanding. What I call village in the post is labelled as WG in the database. I am just realising how bad I did write my last post... sorry about that, I was too tired yesterday!

D
statman
Administrator
Posts: 2757
Joined: Tue Jun 12, 2007 12:08 pm
Location: Florida, USA

Re: Complex samples, with replacement PPS, why only one stag

Postby statman » Thu Sep 13, 2012 2:13 pm

D, I am still looking at this and can't overcome the MOS error either even if I let it scan the data for a viable sample under S1 and then do a % select for S2

Will keep in touch if figure anything out

Update:
Might have a start, seems to be in the definitions of your data. Here is a link to the Output in my Box account. Reply via email once download

Code: Select all

https://www.box.com/s/4xagl2b5q7u5mtbgbdnz

Who is online

Users browsing this forum: No registered users and 2 guests

cron