×

Nearly exact tests of conditional independence and marginal homogeneity for sparse contingency tables. (English) Zbl 0900.62292

Summary: We discuss a variety of test statistics for the hypothesis of conditional independence in three-way contingency tables. Statistics that are designed to detect association between nominal variables, between ordinal variables, and between nominal and ordinal variables are presented. Tests previously presented by Birch (1965) and Landis et al. (1978) are efficient score statistics for alternative hypotheses corresponding to loglinear models that assume homogeneous associations across levels of the control variable. Additional score statistics are presented for loglinear model alternatives that permit interaction, in the form of heterogeneous associations. Ordinary asymptotic chi-squared inference is well established for both types of alternatives. For small samples or sparse data, however, software for exact conditional inference is currently unavailable for most cases having multiple levels of response variables. Monte-Carlo simulation of exact distributions is computationally simple and quick, even for large sparse tables. It results in precise estimates of P-values for exact conditional tests, and hence ‘nearly exact’ tests, for cases lacking software or cases that are likely to be computationally infeasible in the indefinite future. The tests with homogeneous association alternatives also can be applied to nearly exact testing of marginal homogeneity for multivariate responses having the same categorical scale for each component.

MSC:

62H17 Contingency tables
62G10 Nonparametric hypothesis testing

Software:

AS 159; DIGRAM; StatXact
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] Agresti, A.: A survey of exact inference for contingency tables. Statist. sci. 7, 131-177 (1992) · Zbl 0955.62587
[2] Agresti, A.: Computing conditional maximum likelihood estimates for generalized rasch models using simple loglinear models with diagonals parameters. Scand. J. Statist. 20, 63-72 (1993) · Zbl 0770.62095
[3] Agresti, A.; Kezouh, A.: Association models for multidimensional cross-classifications of ordinal variables. Comm. statist. 12, 1261-1276 (1983)
[4] Agresti, A.; Lang, J. B.; Mehta, C. R.: Some empirical comparisons of exact, modified exact, and higher-order asymptotic tests of independence for ordered categorical variables. Comm. statist. Simulation comput. 22, 1-18 (1993)
[5] Agresti, A.; Lipsitz, S.; Lang, J. B.: Comparing marginal distributions of large, sparse contingency tables. Comput. statist. Data anal. 14, 55-73 (1992) · Zbl 0875.62240
[6] Agresti, A.; Mehta, C. R.; Patel, N. R.: Exact inference for contingency tables with ordered categories. J. amer. Statist. assoc. 85, 453-458 (1990) · Zbl 0731.62112
[7] Agresti, A.; Wackerly, D.; Boyett, J.: Exact conditional tests for cross-classifications: approximation of attained significance levels. Psychometrika 44, 75-83 (1979) · Zbl 0419.62046
[8] Andersen, A. H.: Multidimensional contingency tables. Scand. J. Statist. 1, 115-127 (1974) · Zbl 0307.62039
[9] Baglivo, J.; Olivier, D.; Pagano, M.: Methods for the analysis of contingency tables with large and small counts. J. amer. Statist. assoc. 83, 1006-1013 (1988)
[10] Baglivo, J.; Olivier, D.; Pagano, M.: Methods for exact goodness-of-fit tests. J. amer. Statist. assoc. 87, 464-469 (1992) · Zbl 0850.62392
[11] Baglivo, J.; Olivier, D.; Pagano, M.: Analysis of discrete data: rerandomization methods and complexity. Comput. statist. Data anal. 16, 175-184 (1993) · Zbl 0937.62541
[12] Birch, M. W.: The detection of partial association II: The general case. J. roy. Statist. soc. Ser. B 27, 111-124 (1965) · Zbl 0151.23704
[13] Caussinus, H.: Contribution à l’analyse statistique des tableaux de correlation. Ann. fac. Sci. univ. Toulouse 29, 77-182 (1966) · Zbl 0168.39904
[14] Clogg, C. C.: Some models for the analysis of association in multiway cross-classifications having ordered categories. J. amer. Statist. assoc. 77, 803-815 (1982)
[15] Cohen, A.; Sackrowitz, H. B.: An evaluation of some tests of trend in contingency tables. J. amer. Statist. assoc. 87, 470-475 (1992)
[16] Conaway, M.: Analysis of repeated categorical measurements with conditional likelihood methods. J. amer. Statist. assoc. 84, 53-62 (1989)
[17] Darroch, J. N.: The mantel-haenszel test and tests of marginal symmetry; fixed effects and mixed models for a categorical response. Internat. statist. Rev. 49, 285-307 (1981) · Zbl 0484.62067
[18] Diaconis, P.; Sturmfels, B.: Algebraic algorithms for sampling from conditional distributions. Tech. report no. 430 (1993) · Zbl 0952.62088
[19] Fienberg, S. E.; Larntz, K.: Loglinear representation for paired and multiple comparison models. Biometrika 63, 245-254 (1976) · Zbl 0339.62051
[20] Forster, J. J.; Mcdonald, J. W.; Smith, P. W. F.: Monte Carlo exact conditional tests for log-linear and logistic models. J. roy. Statist. soc. Sec. B 58, 445-453 (1996) · Zbl 0853.62050
[21] Goodman, L. A.: Simple models for the analysis of association in cross-classifications having ordered categories. J. amer. Statist. assoc. 74, 537-552 (1979)
[22] Kim, D.; Agresti, A.: Improved exact inference about conditional association in three-way contingency tables. J. amer. Statist. assoc. 90, 632-639 (1995) · Zbl 0925.62207
[23] Kolassa, J. E.; Tanner, M. A.: Approximate conditional inference in exponential families via the Gibbs sampler. J. amer. Statist. assoc. 89, 697-702 (1994) · Zbl 0803.62013
[24] Kreiner, S.: Analysis of multidimensional contingency tables by exact conditional tests: techniques and strategies. Scand. J. Statist. 14, 97-112 (1987) · Zbl 0636.62048
[25] Kreiner, S.: User guide to DIGRAM, a program for discrete graphical modeling. Research report 89/10 (1989)
[26] Landis, J. R.; Heyman, E. R.; Koch, G. G.: Average partial association in three-way contingency tables: a review and discussion of alternative tests. Internat. statist. Rev. 46, 237-254 (1978) · Zbl 0397.62038
[27] Mantel, N.: Chi-square tests with one degree of freedom: extensions of the mantel-haenszel procedure. J. amer. Statist. assoc. 58, 690-700 (1963) · Zbl 0114.11601
[28] Mehta, C. R.; Patel, N. R.; Senchaudhuri, P.: Importance sampling for estimating exact probabilities in permutational inference. J. amer. Statist. assoc. 83, 999-1005 (1988)
[29] Morgan, W. M.; Blumenstein, B. A.: Exact conditional tests for hierarchical models in multidimensional contingency tables. Appl. statist. 40, 435-442 (1991) · Zbl 0825.62500
[30] Patefield, W. M.: An efficient method of generating random R \({\times}\) C tables with given row and column totals. J. roy. Statist. soc. Ser. C 30, 91-97 (1981) · Zbl 0467.62050
[31] Pierce, D. A.; Peters, D.: Practical use of higher order asymptotics for multiparameter exponential families. J. roy. Statist. soc. Ser. B 54, 701-737 (1992)
[32] Rasch, G.: On general laws and the meaning of measurement in psychology, 321–333. Proc. 4th Berkeley symp. Math. statist. Probab. 4 (1961) · Zbl 0107.36805
[33] Somes, G. W.: The generalized mantel-haenszel statistic. Amer. statist. 40, 106-108 (1986) · Zbl 0594.62049
[34] Statxact: Statxact: statistical software for exact nonparametric inference. (1991)
[35] Yao, Q.; Tritchler, D.: An exact analysis of conditional independence in several \(2 {\times} 2\) contingency tables. Biometrics 49, 233-236 (1993)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.