Simpson's Paradox, Lord's Paradox, and Suppression Effects are the same phenomenon – the reversal paradox
Tóm tắt
This article discusses three statistical paradoxes that pervade epidemiological research: Simpson's paradox, Lord's paradox, and suppression. These paradoxes have important implications for the interpretation of evidence from observational studies. This article uses hypothetical scenarios to illustrate how the three paradoxes are different manifestations of one phenomenon – the reversal paradox – depending on whether the outcome and explanatory variables are categorical, continuous or a combination of both; this renders the issues and remedies for any one to be similar for all three. Although the three statistical paradoxes occur in different types of variables, they share the same characteristic: the association between two variables can be reversed, diminished, or enhanced when another variable is statistically controlled for. Understanding the concepts and theory behind these paradoxes provides insights into some controversial or contradictory research findings. These paradoxes show that prior knowledge and underlying causal theory play an important role in the statistical modelling of epidemiological data, where incorrect use of statistical models might produce consistent, replicable, yet erroneous results.
Tài liệu tham khảo
Barker DJ: Fetal origins of coronary heart disease. BMJ. 1995, 311:171-4.
Barker DJ, Eriksson JG, Forsen T, Osmond C: Fetal origins of adult disease: strength of effects and biological basis. Int J Epidemiol. 2002, 31:1235-9. 10.1093/ije/31.6.1235
Paneth N, Ahmed F, Stein AD: Early nutritional origins of hypertension: a hypothesis still lacking support. Journal of Hypertensio. 1996, 14 (5): S121-S129.
Lucas A, Fewtrell MS, Cole TJ: Fetal origins of adult disease-the hypothesis revisited. BMJ. 1999, 319: 245-9.
Huxley RR, Neil A, Collins R: Unravelling the fetal origins hypothesis: is there really an inverse association between birth weight and subsequent blood pressure?. Lancet. 2002, 360: 659-65. 10.1016/S0140-6736(02)09834-3
Tu YK, West R, Ellison GTH, Gilthorpe MS: Why evidence for the fetal origins of adult disease might be a statistical artifact: the "reversal paradox" for the relation between birth weight and blood pressure in later life. Am J Epidemiol. 2005, 161: 27-32. 10.1093/aje/kwi002
Weinberg CR: Invited commentary: Barker meets Simpson. Am J Epidemiol. 2005, 161: 33-5. 10.1093/aje/kwi003
De Stavola BL, Nitsch D, dos Santos Silva I, McCormack V, Hardy R, Mann V, Cole TJ, Morton S, Leon DA: Statistical issues in life course epidemiology. Am J Epidemiol. 2006, 163: 84-96. 10.1093/aje/kwj003
Pearl J: Causality: Models, Reasoning, and Inference. Cambridge: Cambridge University Press; 2000.
Greenland S, Robins JM, Pearl J: Confounding and collapsibility in causal inference. Stat Sci. 1999, 14: 29-46. 10.1214/ss/1009211805
Jewell NP: Statistics for Epidemiology. London: Chapman & Hall; 2004.
Cox DR, Wermuth N: A general condition for avoiding effect reversal after marginalisation. J R Statist Soc B. 2003, 65: 937-941. 10.1111/1467-9868.00424
Simpson EH: The interpretation of interaction in contingency tables. J R Stat Soc Ser B. 1951, 13: 238-41.
Yule GU: Notes on the theory of association of attributes in statistics. Biometrika. 1903, 2: 121-34. 10.1093/biomet/2.2.121
Pearson K, Lee A, Bramley-Moore L: Mathematical contributions to the theory of evolution: VI – Genetic (reproductive) selection: Inheritance of fertility in man, and of fecundity in thoroughbred racehorses. Philos Trans R Soc Lond A. 1899, 192: 257-330. 10.1098/rsta.1899.0006
Hennessy E, Alberman E: Intergenerational influences affecting birth outcome. II. Preterm delivery and gestational age in the children of the 1958 British birth cohort. Paediatr Perinat Epidemiol. 1998, 12 (1): 61-75. 10.1046/j.1365-3016.1998.0120s1061.x
Website of the Department of Health, United Kingdom. http://www.doh.gov.uk
Paik M: A graphical representation of a three-way contingency table: Simpson's paradox and correlation. Am Stat. 1985, 39: 53-54. 10.2307/2683907. 10.2307/2683907
Hernandez-Diaz S, Schisterman EF, Hernan M: The "birth weight" paradox uncovered?. Am J Epidemiol. 2006, 164: 1115-1120. 10.1093/aje/kwj275
Wilcox A: Invited Commentary: The perils of birth weight – a lesson from directed acyclic graphs. Am J Epidemiol. 2006, 164: 1121-1123. 10.1093/aje/kwj276
Lord FM: A paradox in the interpretation of group comparisons. Psychol Bull. 1967, 68: 304-5. 10.1037/h0025105
Lord FM: Statistical adjustments when comparing preexisting groups. Psychol Bull. 1969, 72: 337-8. 10.1037/h0028108. 10.1037/h0028108
Glymour MM, Weuve J, Berkman LF, Kawachi I, Robins JM: When is baseline adjustment useful in analysis of change? An example with education and cognitive change. Am J Epidemiol. 2005, 162: 267-278. 10.1093/aje/kwi187
Hand D: Deconstructuring statistical questions. J R Stat Soc Ser A Stat Soc. 1994, 157: 317-56. 10.2307/2983526. 10.2307/2983526
Campbell DT, Kenny DA: A primer on regression artefact. Guildford: The Guilford Press; 1999.
Mohr LB: Regression artifacts and other customs of dubious desert. Eval Program Plann. 2000, 23: 397-409. 10.1016/S0149-7189(00)00029-X. 10.1016/S0149-7189(00)00029-X
Reichardt CS: Regression facts and artifacts. Eval Program Plann. 2000, 23: 411-4. 10.1016/S0149-7189(00)00030-6. 10.1016/S0149-7189(00)00030-6
Wainer H: Adjusting for differential base rates: Lord's paradox again. Psychol Bull. 1991, 109: 147-51. 10.1037/0033-2909.109.1.147
Stigler SM: Statistics on the Table. Cambridge, Massachusetts: Harvard University Press; 1999.
Vickers AJ, Altman DG: Analysing controlled trials with baseline and follow up measurements. BMJ. 2001, 323: 1123-4. 10.1136/bmj.323.7321.1123
Tu YK, Blance A, Clerehugh V, Gilthorpe MS: Statistical power for analyses of changes in randomized controlled trials. J Dent Res. 2005, 84: 283-287.
Lewis JW, Escobar LA: Suppression and enhancement in bivariate regression. Statistician. 1986, 35: 17-26. 10.2307/2988294. 10.2307/2988294
Bertrand PV, Holder RL: A quirk in multiple regression: the whole regression can be greater than the sum of its parts. Statistician. 1988, 37: 371-4. 10.2307/2348761. 10.2307/2348761
Sharpe NR, Roberts RA: The relationship among sums of squares, correlation coefficients and suppression. Am Stat. 1997, 51: 46-48. 10.2307/2684693. 10.2307/2684693
Friedman L, Wall M: Graphical views of suppression and multicollinearity in multiple linear regression. Am Stat. 2005, 127-136.
Cohen J, Cohen P: Applied multiple regression/correlation analysis for the behavioural sciences. London: LEA; 1983.
Pedhazur EJ: Multiple regression in behavioral research: Explanation and prediction. Fort Worth: Harcourt; 1997.
Stocks NP, Davey Smith G: Blood pressure and birth weight in the first year university student aged 18–25. Public Health. 1999, 113: 273-7. 10.1016/S0033-3506(99)00179-1
Williams S, Poulton R: Birth size, growth, and blood pressure between the ages of 7 and 26 years: failure to support the fetal origins hypothesis. Am J Epidemiol. 2002, 155: 849-52. 10.1093/aje/155.9.849
McNeill G, Tuya C, Campbell DM, Haggarty P, Smith WCS, Masson LF, Cumming A, Broom I, Haites N: Blood pressure in relation to birth weight in twins and singleton controls matched for gestational age. Am J Epidemiol. 2003, 158: 150-5. 10.1093/aje/kwg130
Tu YK, Gilthorpe MS, TH Ellison GTH: What is the effect of adjusting for more than one measure of current body size on the relation between birth weight and blood pressure?. J Hum Hypertens. 2006, 20: 646-657. 10.1038/sj.jhh.1002044
Horst P: The role of prediction variables which are independent of the criterion. The Prediction of Personal Adjustment. Edited by: Horst P. New York: Social Science Research Council; 1941, 431-6.
MacKinnon DP, Krull JL, Lockwood CM: Equivalence of the mediation, confounding and suppression effect. Prev Sci. 2000, 1: 173-81. 10.1023/A:1026595011371
Tu YK, Ellison GTH, Gilthorpe MS: Growth, current size and the role of the 'reversal paradox' in the foetal origins of adult disease: an illustration using vector geometry. Epidemiol Perspect Innov. 2006, 3: 9. 10.1186/1742-5573-3-9
Von Elm E, Egger M: The scandal of poor epidemiological research. BMJ. 2004, 329: 868-9. 10.1136/bmj.329.7471.868