Date: Mon, 28 Jun 2004 17:53:45 -0300 Hector Maletta "SPSSX(r) Discussion" Hector Maletta Re: assign a unique ID number To: "Whitt, Susanne" text/plain; charset="us-ascii"

Susanne, Even if there is actually some systematic bias in the order the cases, that would be only important if you have in mind some analysis of that order, or an analysis of the underlying ordering variable (say by time, by size or whatever criterion was used to order the cases). If you apply any kind of statistical analysis that does not take that order into account, such as regression or tables or ANOVA, the matter is totally immaterial and causes no consequences at all. Most statistical procedures treat all cases as equal members of the data set, and analyze only such variables as you include in each analysis. It is not really overkill but (usually) misplaced kill.

Hector

> -----Original Message----- > From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] > On Behalf Of Whitt, Susanne > Sent: Monday, June 28, 2004 4:53 PM > To: SPSSX-L@LISTSERV.UGA.EDU > Subject: Re: assign a unique ID number > > > I always use the UNIFORM or RV.UNIFORM function in addition > to creating \$casenum, because one can never be certain that > there isn't some underlying systematic bias to the order of > the cases. For example, if cases are entered as they arrive: > is there something different about early responders than > late responders? To make sure there is no possible bias, I > create a variable using UNIFORM or RV.UNIFORM, sort by it, > then compute an ID using \$casenum. Some would say it's > overkill, but it works for me, and besides it's so easy, why not? > > COMPUTE order1 = UNIFORM(111). > EXECUTE. > SORT CASES BY > order1 (A). > COMPUTE id1 = \$casenum. > EXECUTE. > > OR . . . . > > COMPUTE order2 = RV.UNIFORM(1,111). > EXECUTE. > SORT CASES BY > order2 (A). > COMPUTE id2 = \$casenum. > EXECUTE. > > Susanne Whitt, Ph.D. > Research & Evaluation Specialist > Dept. of Defense Education Activity > 4040 N. Fairfax Drive > Arlington, VA 22203 > (703) 588-3162 > (703) 588-3702 (FAX) > swhitt@hq.odedodea.edu > > -----Original Message----- > From: Hector Maletta [mailto:hmaletta@fibertel.com.ar] > Sent: Monday, June 28, 2004 3:12 PM > To: SPSSX-L@LISTSERV.UGA.EDU > Subject: Re: assign a unique ID number > > I was not clear either that the UNIFORM command was still > desired. In fact, it is not necessary at all. \$CASENUM gives > a unique number to each case, and this is all that was > originally needed. Since \$CASENUM has unique values, the > assignment of an ID based on \$CASENUM can be done with > unsorted cases. Later, if desired, cases can be sorted by the > new ID variable. Hector > > > -----Original Message----- > > From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] > On Behalf > > Of Hetter, Rebecca D,,DMDCWEST > > Sent: Monday, June 28, 2004 3:48 PM > > To: SPSSX-L@LISTSERV.UGA.EDU > > Subject: Re: assign a unique ID number > > > > > > I guess I wasn't clear enough when I wrote adding the two > statements. > > I meant following the call to the UNIFORM function. > However, if SORT > > CASES is not used (after the call to UNIFORM), the cases > will remain > > in the original order, which is not random. > > > > HTH > > > > -----Original Message----- > > From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] > On Behalf > > Of Hector Maletta > > Sent: Monday, June 28, 2004 11:28 AM > > To: SPSSX-L@LISTSERV.UGA.EDU > > Subject: Re: assign a unique ID number > > > > > > The SORT CASES command is not required, and furthermore, it is not > > possible to sort by BRAINNO before creating that variable. Hector > > > > > -----Original Message----- > > > From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] > > On Behalf > > > Of Hetter, Rebecca D,,DMDCWEST > > > Sent: Monday, June 28, 2004 3:14 PM > > > To: SPSSX-L@LISTSERV.UGA.EDU > > > Subject: Re: assign a unique ID number > > > > > > > > > The assignments you obtained were random - that's why it's > > possible to > > > have duplicates. The system variable \$casenum qives the sequence > > > numbers of the records within the file, and those are > > unique. Try it > > > adding these > > > statements: > > > > > > sort cases by brainno. > > > compute brainno=\$casenum. > > > > > > HTH > > > > > > > > > -----Original Message----- > > > From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] > > On Behalf > > > Of Jodene Goldenring Fine > > > Sent: Sunday, June 27, 2004 8:40 AM > > > To: SPSSX-L@LISTSERV.UGA.EDU > > > Subject: assign a unique ID number > > > > > > > > > Hello Listservers! > > > > > > I have a simple problem that I just can't seem to get > > around. I have > > > 111 cases to which I would like to assign unique ID numbers from > > > 1-111. This is so I can be "blind" to the data I'm > working with. I > > > tried the following > > > syntax: > > > > > > numeric brainno (F8.0). > > > compute brainno=rv.uniform(1,111). > > > > > > but the assignments were not random. So, I ended up with 3 cases > > > assigned 44 and 2 cases assigned 12, etc. How can I do this? > > > > > > Thanks for the help! > > > > > > Jodene > > > > > > > > > > > > __________________________________ > > > > > > Jodene Goldenring Fine, M.S., N.C.S.P. > > > Doctoral Candidate > > > The University of Texas at Austin > > > Department of Educational Psychology > > > > > > jodene.fine@mail.utexas.edu ________________________________ > > > > > > "Be a nice person and see if it works." > > > Chinese Fortune Cookie, Berkeley, California, circa 1974 > > > > > >

Back to: Top of message | Previous page | Main SPSSX-L page