Dear Listers. How it can be done in spss if I have a spss data set that have 10000 records In my data set I have a question ( Do you have some disease( Yes=1 and No=2 ). That question we are asking 10000 times. If suppose out of 10000, 150 cases are Yes=1. Ofcourse that cases would have some ids. How can I make a new data set from 10000 cases dataset that would have only 300 cases ( 150 disease=Yes ) and rest 150 would be just next case of that case where disease=Yes ) Example Sno. Question of Disease ( Yes=1/No=2 ) 1 1 ( Here is Yes so I need this case in my new data file ) 2 2 and I need this case also as just next to above. 3 2 4 1 5 2 6 2 7 2 8 2 9 2 10 2 11 2 12 1 13 2 14 2 . . 10000 So out of these 10000 cases, I should have a new data file that has only 1…………. 2 4 5 12 13 I hope u understand my query. Please help in this regard. Thanks in advance. Warm Regards Manoj India. |
1) Do you want
to select the first 150 records where Disease=Yes and the first 150
records where Disease=No,
OR
2) do you want to
randomly select 150 records where Disease=Yes and randomly
select 150 records where Disease=No??
If 1)
then
sort cases by
disease.
compute
rec=1.
if (disease eq
lag(disease)) rec=lag(rec)+1.
select if (rec le
150).
if 2) Look at the
Sample command and see if that will work for you. You will have to split the
current file into a Disease=Yes file and a Disease=No file, apply sample
to each and then put the two resulting files back together.
An alternate way
is this.
compute
rnum=uniform(1).
sort cases by
disease rnum.
compute
rec=1.
if (disease eq
lag(disease)) rec=lag(rec)+1. select if (rec le
150). From: SPSSX(r) Discussion [mailto:[hidden email]]
On Behalf Of Manoj Soni
Sent: Friday, September 09, 2011 7:28 AM To: [hidden email] Subject: A Query in Spss Dear Listers. How it can be done in spss if I have a spss data set that
have 10000 records In my data set I have a question ( Do you have some disease(
Yes=1 and No=2 ). That question we are asking 10000 times. If suppose out of 10000, 150 cases are Yes=1. Ofcourse that cases would have some ids. How can I make a new data set from 10000 cases dataset that
would have only 300 cases ( 150 disease=Yes ) and rest 150 would be just
next case of that case where disease=Yes ) Example
Sno.
Question of Disease ( Yes=1/No=2 ) 1
1 ( Here is Yes so I need this case in my new data file
) 2
2 and I need this case also as just next to
above. 3
2 4
1 5
2 6
2 7
2 8
2 9
2 10
2 11
2 12
1 13
2 14
2 . . 10000 So out of these 10000 cases, I should have a new data file
that has only 1…………. 2 4 5 12 13 I hope u understand my query. Please help in this
regard. Thanks in advance. Warm Regards Manoj
India. |
Administrator
|
In reply to this post by Manoj Soni
Something like the following?
*BUT* what if 2 consecutive cases in the file have disease=1? COMPUTE KeepCase= LAG(disease) EQ 1 OR disease EQ 1. SELECT IF KeepCase.
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
Dear SPSS listers,
This is not a query about SPSS; but... I intent to make a meta-analyse of studies where correlations are computed between several measures taken upon the same individuals. For instance, the same participants are evaluated using the A-version, B-version and C-version of a task (for instance, a segment bisection task) and then their performance on those three tasks is correlated with another variable (for instance, anxiety). One of my goals is to evaluate if the version of the task is a moderator for the relation between the task and the other variable (for instance, is the version of the task a moderator of the correlation between anxiety and segment bisection performance?). The three correlation coefficients are not independent (because they evaluate the same skill and are based in the same sample), so I believe I cannot use them at the same time in the analysis. Is this right? For each study, should I choose only one correlation and discard the other two (because they are not independent) or there is any kind of "within-subject design" for meta-analytic studies involving correlations that allow me to use the three correlations (some kind of repeated measures ANOVA for correlation coefficients)? I hope I have put my query clearly enough for your understanding. Any suggestion or bibliographic reference on this topic will be very helpful. Thank you in advance. Louis ===================== 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
|
It sounds like your problem is similar to the one discussed here (mind the line-wrap):
http://books.google.ca/books?hl=en&lr=&id=Nr11DicNrkUC&oi=fnd&pg=PA179&dq=Gleser+%26+Olkin+meta-analysis+multiple+treatment+groups&ots=Choy0nDHUL&sig=VtpwwW3iALbi0Ao21lkTFm7bIwM#v=onepage&q&f=false http://statistics.stanford.edu/~ckirby/techreports/GEN/2007/2007-2.pdf 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 Luís Faísca
I think I may understand what you are asking, but I have
to translate your terminology, which I think you are using in an unconventional way. See below, for interjected questions and comments. > Date: Sat, 10 Sep 2011 02:36:29 +0100 > From: [hidden email] > Subject: Meta-analysis for corelation between repeated measures > To: [hidden email] > > Dear SPSS listers, > > This is not a query about SPSS; but... > I intent to make a meta-analyse of studies where correlations are computed > between several measures taken upon the same individuals. I think of a meta-analysis as doing what you can do, to try to emulate a full ANOVA design, when you are handed the summary statistics from a number of studies. However, you don't seem to have a "number of studies" and you don't seem confined to previous summary statistics. > For instance, the same participants are evaluated using the A-version, > B-version and C-version of a task (for instance, a segment bisection task) > and then their performance on those three tasks is correlated with another > variable (for instance, anxiety). One of my goals is to evaluate if the > version of the task is a moderator for the relation between the task and the > other variable (for instance, is the version of the task a moderator of the > correlation between anxiety and segment bisection performance?). This seems to parse badly, "if the version of the test is a moderator..." when you are referring to several correlations. - A moderator variable measures a characteristic of a person; it is not a choice of predictors. In trying to grasp this, I draw a parallel : Suppose that you had 3 measures of depression, and you wanted to know if they measured the same aspect of (your variable) anxiety. I would not say, "Does the choice of depression-score moderate the measured relation between depression and anxiety?" It *seems* like you want to know if one of the test-versions predicts better (or worse) than the others, and that would be the more direct way to say it. If that is so, then the first *new* concern is to clarify *that* hypothesis. Do you really want to know if one is better, or do you want to know, as well, how they are *different*? If you put all three into a prediction equation by multiple regression, then, if they are highly similar, the whole prediction, with 3 d.f., will be strong, and the tests on each of the three separate coefficients will be Not Significant (or, much less significant). (With a large enough sample, you could find significant added prediction simply by lengthening a test; that is the justification for using longer scales.) If any of the three tests is Significant, that is a result that says, "This test shows some *unique* prediction, above and beyond the prediction of the other two (apparently similar) tests." However, I think you may want a more direct comparison of "correlated correlations." That would be the pairwise comparisons of the three tests, x1, x2 and x3 -- x1 with z versus x2 with z; x1 with z versus x3 with z; x2 with z versus x3 with z. The top citation that Google gives me looks pretty good - http://psych.unl.edu/psycrs/statpage/biv_corr_comp_eg.pdf He describes both Meng's test and Hotelling's test. I can add something to this, from a previous note that I once had online in a Stats-FAQ for usenet stats-groups. === Communication, 2002, from Paul von Hippel " ... if you read the appendix and related articles you realize that they're confining themselves to the case where the regressors are random variables. If the regressors are fixed, as in an experimental design, then Hotelling's test is appropriate. Hotelling (1940) was quite explicit about this, so what Meng, Rosenthal, & Rubin are really criticizing is the mistaken practice of using Hotelling's test with random regressors. "Williams (1959) adapted Hotelling's test to the case of random regressors. In simulation studies Williams' test has held up quite well against the alternatives described by Meng, Rosenthal, & Rubin. This is all in the papers cited in MR&R's bibliography." === end of communication ]. > The three correlation coefficients are not independent (because they > evaluate the same skill and are based in the same sample), so I believe I > cannot use them at the same time in the analysis. Is this right? For each > study, should I choose only one correlation and discard the other two > (because they are not independent) or there is any kind of "within-subject > design" for meta-analytic studies involving correlations that allow me to > use the three correlations (some kind of repeated measures ANOVA for > correlation coefficients)? > > I hope I have put my query clearly enough for your understanding. Any > suggestion or bibliographic reference on this topic will be very helpful. > Thank you in advance. > > Louis -- Rich Ulrich |
Free forum by Nabble | Edit this page |