Conover (1999) is a good source for the assumptions and/or limitations of nonparametric stats...but Conover says nothing about low sample size being problematic for McNemar tests.

Daniel (1978) in "Applied Nonparametric Statistics" notes that low sample size can cause distorted P values, and mentions that some favor "Gart's Test". I have never used Gart's test, but it appears to be intended to correct the issue of low sample sizes in McNemar.

Evan R. Harrington, Ph.D. Associate Professor Forensic Thesis Track Director The Chicago School of Professional Psychology Department of Forensic Psychology 325 North Wells Street Chicago, IL 60654

Phone: 312 329-6693 Fax: 312 661-1272

-----Original Message----- From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of Marta García-Granero Sent: Sunday, September 20, 2009 12:31 PM To: SPSSX-L@LISTSERV.UGA.EDU Subject: Re: McNemar Test

Hi everybody

AJay Ohri wrote: > T test matched pairs > > Sent from my iPhone

Definitely NOT a solution, since McNemar tests handles BINARY outcomes, not continuous.

Sent from my desktop computer.

MGG > > On Sep 20, 2009, at 12:59 PM, "Handel, Richard W." <HandelRW@EVMS.EDU > <mailto:HandelRW@EVMS.EDU>> wrote: > >> Hi All, >> >> A colleague has asked for assistance in proposing an analysis for a >> 2X2 contigency table for matched pairs. I know that the McNemar test >> is normally appropriate, but at most, 20 matched pairs (40 subjects >> total) will be available for analysis. I am concerned that some >> cells may have very low frequency counts. What are my options? >> >> Thanks! >> >> Rick >> >>

-- For miscellaneous SPSS related statistical stuff, visit: http://gjyp.nl/marta/

===================== To manage your subscription to SPSSX-L, send a message to LISTSERV@LISTSERV.UGA.EDU (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

===================== To manage your subscription to SPSSX-L, send a message to LISTSERV@LISTSERV.UGA.EDU (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