The principles of the Ederer-Myers-Mantel procedure for seeking evidence of disease clustering are reviewed. The procedure is based on cumulative empirical clusters, i.e., the largest frequency in a single time period or in two successive time periods, and comparing that cumulation with the expected cumulation of largest frequencies under random occurrence. Original tabulations covered totals of up to 15 cases distributed among three, four or five time periods. Present tabulations of expectations and variances cover up to 500 cases distributed among two or three time periods and 200 cases distributed among four or five time periods. Asymptotic formulas are provided for the expectation and variance of the largest frequency in a single period when arbitrarily many cases are distributed at random among two, three, four or five time periods.
|Number of pages||9|
|Journal||American Journal of Epidemiology|
|State||Published - Nov 1976|
Bibliographical noteFunding Information:
Received for publication December 29, 1975 and in final form April 14, 1976. Abbreviation: E-M-M, Ederer-Myers-Mantel (reference 1). From the Biometry Branch, National Cancer Institute, Bethesda, MD 20014. 1Presently with Biostatistics Center, George Washington University, 7979 Old Georgetown Road, Bethesda, MD 20014. Supported in part by USPHS Grant CA-15686 from the National Cancer Institute.
- Disease models
ASJC Scopus subject areas