Some Practical Considerations and a Crude Formula for Estimating Sample Size for McNemar’s Test

Autor:	Robert Lehr
Rok vydání:	2001
Předmět:	Correlation McNemar's test Sample size determination Drug Guides Statistics Public Health Environmental and Occupational Health Pharmacology (medical) Pharmacology (nursing) Mathematics Test (assessment) Statistical hypothesis testing
Zdroj:	Drug Information Journal. 35:1227-1233
ISSN:	2164-9200 0092-8615
DOI:	10.1177/009286150103500419
Popis:	Although McNemar’s test for correlated proportions is easy to apply and interpret, planning studies based on this test with clinicians is not trivial. Trials that involve two measurements made on the same or matched subjects under different treatments or experimental conditions may be analyzed using this test. I present an intuitive and easily-remembered equation for calculating approximate sample size requirements for experiments whose primary variables are to be analyzed using McNemar’s test. This formula, n = 16pq(1 - r/d2, makes it easier to see and explain the ramifications of different choices of parameter estimates. After presenting the formula, I provide some guidance for the choice of r. I also give suggestions for situations where some parameter estimates are not available. A useful relationship between the correlation r and ψ, the proportion of discordant cells, is presented and discussed. Then results of the approximation are compared to those from other formulae. Finally, some examples are provided with suggested solutions.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::2d2505e94e33303748f2e9cbc47f8b8a https://doi.org/10.1177/009286150103500419 Zobrazit plný text záznamu