A method appropriate for evaluating biases in criterion statistics attributable to excluding subjects with missing data from an original sample was illustrated using a complex family illness variable. This method incorporates non-criterion information to indirectly reduce the error in bias estimates and appears more accurate than methods which utilize criterion data alone. Since non-criterion data are required, the method is particularly suited to situations in which the original sample is complete but criterion data are missing within forms or subsequently lost through attrition of subjects.
Get full access to this article
View all access options for this article.
References
1.
BonatoR.GuyW.ClearyP.YangK.The information processing system for the ECDEU program. Psychopharmacology Bulletin, 1970, 6, 22–47.
2.
CochranW. G.Sampling techniques. (2nd ed.) New York: Wiley, 1963.
3.
FullerC. H.Weighting to adjust for survey response. Public Opinion Quarterly, 1974, 38, 239–246.
4.
KishL.Survey sampling. New York: Wiley, 1965.
5.
MandellL.When to weight: Determining nonresponse bias in survey data. Public Opinion Quarterly, 1974, 38, 247–252.
6.
McGlashanT. H.The documentation of clinical psychotropic drug trials. Washington, D.C.: U.S. Government Printing Office, 1973. (DHEW Publication No. HSM73–9038).
7.
MeltzoffJ.KornreichM.Research in psychotherapy. New York: McGraw-Hill, 1970.
8.
SiegelS.Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill, 1956.
9.
WittenbornJ. R.The design of clinical trials. In LevineJ.SchieleB. C.BouthiletL. (Eds.), Principles and problems in establishing the efficacy of psychotropic agents. Washington, D. C.: U. S. Government Printing Office, 1971. Pp. 227–264. (PHS Publ. No. 2138).
