Some benefits of dichotomization in psychiatric and criminological research

Background The product-moment correlation r is widely used in criminology and psychiatry to measure strength of association. However, most criminological and psychiatric variables contravene its underlying assumptions. Aim To compare statistical measures of association based on dichotomous variables with the use of r. Method Explanatory variables for delinquency are investigated in the Pittsburgh Youth Study using a sample of 506 boys aged 13–14. Results Dichotomization does not necessarily cause a decrease in measured strength of associations. Conclusions about the most important explanatory variables for delinquency were not greatly affected by using dichotomous as opposed to continuous variables, by different dichotomization splits, or by using logistic versus OLS multiple regression. Non-linear relationships, interaction effects and multiple risk factor individuals were easily studied using dichotomous data. Conclusions Dichotomization produces meaningful findings that are easily understandable to a wide audience. Measures of association for dichotomous variables, such as the odds ratio, have many advantages and are often more realistic and meaningful measures of strength of relationship than the product-moment correlation r. Copyright © 2000 Whurr Publishers Ltd.

[1]  E. Sonuga-Barke.  Categorical models of childhood disorder: a conceptual and empirical analysis. , 1998, Journal of child psychology and psychiatry, and allied disciplines.

[2]  M. Rutter.  Comorbidity: concepts, claims and choices , 1997 .

[3]  D. Fergusson,et al.  Attentional difficulties in middle childhood and psychosocial outcomes in young adulthood. , 1997, Journal of child psychology and psychiatry, and allied disciplines.

[4]  J E Richters,et al.  The Hubble hypothesis and the developmentalist's dilemma , 1997, Development and Psychopathology.

[5]  H. Kraemer,et al.  Coming to terms with the terms of risk. , 1997, Archives of general psychiatry.

[6]  David P. Farrington,et al.  THE FAMILIAL TRANSMISSION OF CRIMINAL CONVICTIONS , 1997 .

[7]  D. Farrington.  Key Issues in Studying the Biosocial Bases of Violence , 1997 .

[8]  A. Raine,et al.  Biosocial Bases of Violence , 1997, Nato ASI Series.

[9]  W. R. Smith.  The effects of base rate and cutoff point choice on commonly used measures of association and accuracy in recidivism research , 1996 .

[10]  P. Cohen.  Childhood Risks for Young Adult Symptoms of Personality Disorder: Method and Substance. , 1996, Multivariate behavioral research.

[11]  D. Fergusson,et al.  Antisocial behaviour, unintentional and intentional injuries during adolescence , 1995 .

[12]  G. Harris,et al.  Violent recidivism: assessing predictive validity. , 1995, Journal of consulting and clinical psychology.

[13]  S. Greenland.  Dose‐Response and Trend Analysis in Epidemiology: Alternatives to Categorical Analysis , 1995, Epidemiology.

[14]  R. Loeber,et al.  Which boys will fare worse? Early predictors of the onset of conduct disorder in a six-year longitudinal study. , 1995, Journal of the American Academy of Child and Adolescent Psychiatry.

[15]  D. Fergusson,et al.  Predictive validity of categorically and dimensionally scored measures of disruptive childhood behaviors. , 1995, Journal of the American Academy of Child and Adolescent Psychiatry.

[16]  D. Mossman.  Assessing predictions of violence: being accurate about accuracy. , 1994, Journal of consulting and clinical psychology.

[17]  D. Fergusson,et al.  The effects of conduct disorder and attention deficit in middle childhood on offending and scholastic ability at age 13. , 1993, Journal of child psychology and psychiatry, and allied disciplines.

[18]  C. Judd,et al.  Statistical difficulties of detecting interactions and moderator effects. , 1993, Psychological bulletin.

[19]  S. Maxwell,et al.  Bivariate median splits and spurious statistical significance. , 1993 .

[20]  L. P. Zhao,et al.  Efficiency loss from categorizing quantitative exposures into qualitative exposures in case-control studies. , 1992, American journal of epidemiology.

[21]  M. Rutter,et al.  Comorbidity in child psychopathology: concepts, issues and research strategies. , 1991, Journal of child psychology and psychiatry, and allied disciplines.

[22]  David M. Fergusson,et al.  A latent class model of child offending , 1991 .

[23]  D. Lykken,et al.  Heterogeneity in the inheritance of alcoholism. A study of male and female twins. , 1991, Archives of general psychiatry.

[24]  Robert Rosenthal,et al.  How are we doing in soft psychology , 1990 .

[25]  John E. Hunter,et al.  Dichotomization of continuous variables: the implications for meta-analysis , 1990 .

[26]  D. Farrington,et al.  Relative improvement over chance (RIOC) and phi as measures of predictive efficiency and strength of association in 2×2 tables , 1989 .

[27]  H. Veiel.  Base-rates, cut-points and interaction effects: the problem with dichotomized continuous variables , 1988, Psychological Medicine.

[28]  D. Magnusson,et al.  Individual and variable-based approaches to longitudinal research on early risk factors. , 1988 .

[29]  R. Fowler,et al.  Power and Robustness in Product-Moment Correlation , 1987 .

[30]  J. Fleiss,et al.  The logistic regression analysis of psychiatric data. , 1986, Journal of psychiatric research.

[31]  J. Swets.  Indices of discrimination or diagnostic accuracy: their ROCs and implied models. , 1986, Psychological bulletin.

[32]  Jacob Cohen.  The Cost of Dichotomization , 1983 .

[33]  David M. Fergusson,et al.  Signal Detectability Theory and the Evaluation of Prediction Tables , 1977 .

[34]  J. Fleiss,et al.  Statistical methods for rates and proportions , 1973 .

[35]  M. Wolfgang,et al.  Delinquency in a birth cohort , 1972 .

[36]  A. W. Edwards.  The Measure of Association in a 2 × 2 Table , 1963 .