Resolving the conflict over Fisher’s exact test. (English) Zbl 0766.62031

Routledge, R. D.

Can. J. Stat. 20, No. 2, 201-209 (1992).

Summary: Fisher’s exact test for two-by-two contingency tables

•  Wikipedia
•  Wolfram MathWorld

has repeatedly been criticized as being too conservative. These criticisms arise most frequently in the context of a planned experiment for which the numbers of successes in each of two experimental groups are assumed to be binomially distributed. It is argued here that the binomial model is often unrealistic, and that the departures from the binomial assumptions reduce the conservatism in Fisher’s exact test. Further discussion supports a recent claim that the residual conservatism is attributable, not to any additional information used by the competing method, but to the discrete nature of the test, and can be drastically reduced through the use of Lancaster’s mid-\(p\)-value. The binomial model is not recommended in that it depends on extra, questionable assumptions.

MSC:

62H17

Contingency tables

Keywords:

Lancaster’s mid-\(p\)-value; Fisher’s exact test; two-by-two contingency tables; binomial model; residual conservatism

References:

[1]	Anscombe, Computing in Statistical Science through APL. (1981) · doi:10.1007/978-1-4613-9450-1
[2]	Barnard, A new test for 2 {\(\times\)} 2 tables, Nature 156 pp 177– (1945) · Zbl 0060.30506
[3]	Barnard, Statistical inference, J. Roy. Statist. Soc. Ser. B 11 pp 115– (1949)
[4]	Barnard, On alleged gains in power from lower p-values, Statist. Med. 8 pp 1469– (1989)
[5]	Basler, Verbesserung des nicht-randomisierten exacten Testes von R.A. Fisher, Metrika 34 pp 287– (1987)
[6]	Berkson, In dispraise of the exact test. Do the marginal totals of a 2 {\(\times\)} 2 table contain relevant information respecting the table proportions?, J. Statist. Plann. Inference 2 pp 27– (1978) · Zbl 0374.62029
[7]	Cox, Theoretical Statistics (1974) · doi:10.1007/978-1-4899-2887-0
[8]	D’Agostino, The appropriateness of some common procedures for testing the equality of two independent binomial populations, Amer. Statist. 42 pp 198– (1988)
[9]	Fisher, A new test for 2 {\(\times\)} 2 tables, Nature 156 pp 388– (1945)
[10]	Franck, P-values for discrete test statistics, Biometrical J. 4 pp 403– (1986)
[11]	Gibbons, P-values: Interpretation and methodology, Amer. Statist. 29 pp 20– (1975) · Zbl 0361.62017
[12]	Haber, A comparison of some conditional and unconditional exact tests for 2{\(\times\)}2 contingency tables, Comm. Statist.-Simulation 16 pp 999– (1987) · Zbl 0695.62051
[13]	Hurlbert, Pseudoreplication and the design of ecological field experiments, Ecol. Monogr. 54 pp 187– (1984)
[14]	Lancaster, The combination of probabilities arising from data in discrete distributions, Biometrika 36 pp 370– (1949) · Zbl 0034.22901 · doi:10.1093/biomet/36.3-4.370
[15]	Lancaster, Statistical control of counting experiments, Biometrika 39 pp 419– (1952) · doi:10.1093/biomet/39.3-4.419
[16]	Lancaster, Significance tests in discrete distributions, J. Amer. Statist. Assoc. 56 pp 223– (1961) · Zbl 0104.13201
[17]	Lehmann, Testing Statistical Hypotheses (1959) · Zbl 0089.14102
[18]	Overall, Small-sample tests for homogeneity of response probabilities in 2 {\(\times\)} 2 contingency tables, Psychol. Bull. 102 pp 307– (1986)
[19]	Stone, The role of significance testing: Some data with a message, Biometrika 56 pp 485– (1969) · Zbl 0183.48105
[20]	Upton, A comparison of alternative tests for the 2 {\(\times\)} 2 comparative trail, J. Roy. Statist. Soc. Ser. A 145 pp 86– (1982)
[21]	Wilson, The controlled experiment and the four-fold table, Science 93 pp 557– (1941) · JFM 67.0485.04
[22]	Yates, Tests of significance for 2 {\(\times\)} 2 contingency tables, J. Roy. Statist. Soc. Ser. A 147 pp 426– (1984) · Zbl 0573.62050