The Cost of Dichotomization

“The use of dichotomous measures, necessitated by the limitations of the data and the technical constraints of mediation analysis, has the undesirable effect of reducing variance and diminishing statistical power (Aiken & West, 1991; Cohen, 1983).”

“…21 February 2012; Accepted 29 February 2012 Academic Editor: Alberto Guillén Copyright © 2012 Enayatollah Bakhshi et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium,…”

“While our memory measure was a discrete variable, brain activity signals were continuous variables and should have greater statistical power (e.g., Cohen, 1983), which could explain significant effects in the neural measures but not in the behavioral measures.”

“Dichotomizing continuous variables also has many statistical costs such as reduced power and effect size (see MacCallum et al. 2002; also see Aiken and West 1991; Cohen 1983; Frazier et al. 2004).”

“To begin with, conducting group analyses when the variable of interest may vary on a continuum diminishes statistical power and increases the risk of rejecting the null hypothesis (Cohen, 1983; Altman and Royston, 2006; DeCoster et al., 2011).”

“The authors recognize that the sample size is marginal for this analysis and using the median to dichotomize participants may reduce effect sizes and statistical power, and increase the risk for Type II error (Cohen, 1983, Iacobucci et al., 2015), hence we consider this analysis exploratory.”

“We used the instructions given in the appendix of [40], but we did not dichotomize the prosocial (cooperative and equality-seeking), individualistic, and competitive subscales of the Triple-Dominance Measure in the manner recommended by the scale developers, using six or more consistent responses out of nine as the cr…”

“Although dichotomizing the data can result in loss of power(15) more than 25% of the sample measurements were censored, and by dichotomizing the sample measurements the censored values, otherwise known as non-detects (NDs), were classified as below the PEL.”

“– 23 Categorisation leads to loss of power, 36 37 which might explain the poorer fi t to the data in terms of AIC for this model.”

“Symptom scores thus ranged from 10 to 60, providing a dimensional understanding of BPD symptoms [70, 71].”

“When the construct of interest is an outcome variable, such as in a randomized trial or an observational study, having a continuous measurement will increase statistical power compared to a dichotomous or ordinal variable (Cohen 1983; Senn 2003).”

“This is usually a poor statistical method of analysis because it adds error, reducing statistical power and increasing the likelihood of Type II errors in many cases (Humphreys and Fleishman, 1974; Cohen, 1983).”

“Using bivariate correlations, which is a more statistically powerful approach, we found a significant decrease in scores with age (Cohen, 1983).”

“learning characteristic reduces variance [43].”

“, ASD or ADHD) or examination grades, require a larger sample (Cohen, 1983), thus lending themselves towards secondary use of population level administrative records (Hjorth et al.”

“Second, we lose statistical power by the dichotomisation (Cohen, 1983; MacCallum et al., 2002).”

“This presents an important counterexample to previous authors, such as Cohen [5], Irwin & McClelland [10], and MacCallum et al. [9] who argued that “tail splits” are still worse than median splits.”

“Statistically, dichotomisation results in loss of individual-level information, resulting in a reduction in statistical power and increasing the risk of type-II error.([29,30]) Similarly, the categorisation of patients into groups using GBTM results in a loss of information and a reduction in statistical power.”

“Effect sizes were assessed according to Cohen (0.1–0.3 weak; 0.3–0.5 medium, 0.5–1 strong) (Cohen, 1983).”

“Although dichotomizing continuous data may assist with scale correspondence, it may be costly to data interpretation in other ways (cf Cohen et al.(40)).”

“1 Creating dichotomous variables from continuous measures can create bias in estimates and it is generally recommended that researchers avoid this procedure and instead rely on regression analyses (eg, Cohen, 1983 ; Maxwell & Delaney, 1993 ; Tellegen & Lubinski, 1983 ).”

“30,31 To overcome these limitations, we developed a novel scoring method that quantifies GDMT use by integrating the number of eligible GDMT and treating dose intensity as a continuous variable, rather than using thresholds to assess dose intensity.”

“qualitative), effect sizes are not comparable in a multivariable logistic regression with periostin (39).”

“If null hypothesis statistical significance testing is going to be used, sufficient power is necessary to provide context for interpreting obtained p values (Cohen, 1983).”

“Data was evaluated both dimensionally and categorically (Altman, 2006; Cohen, 1983; Naggara et al., 2011; Royston, Altman, & Sauerbrei, 2006).”

“Effect sizes are reported using partial eta-squared ( p(2)) classified by Cohen (1988) as small (.”

“Continuous predictors are used rather than the traditional cutpoint approach to avoid reduced power (Cohen, 1983) and the possibility of significant but spurious effects (West, Aiken, & Krull, 1996).”

“Although heavily criticised by a many scholars (Cohen 1983; MacCallum et al. 2002), the dichotomisation of a variable offers a number of advantages, mainly in the simplification of analysis and evaluation.”

“A continuous scale may have revealed greater variability in orientation (Drucker, 2010), and dichotomizations and other truncations of continuous variables are known to result in attenuated effect sizes and reduced statistical power (Cohen, 1983; MacCallum, Zhang, Preacher, & Rucker, 2002).”

“…into high and low performing groups may have led to a loss of statistical power due to dichotomization of the continuous memory score (see e.g., Cohen, 1983), we additionally calculated a correlation between overall d’ and the ORB for young faces, which revealed no significant result, r =…”

“creating a dichotomous measure of depressive symptoms may have resulted in both a loss of variance accounted for as well as a reduction in power (Cohen, 1983).”

“Although widely used, the practice of creating groups from data measured as continuous variables has been a point of concern for some time (Cohen, 1983).”

“Although it is not known how dichotomi-zation affects the predictive strength of expectation [15], dichotomization always results in a loss of information [52,53].”

“A possible explanation is that the effect of LSU was found for accuracy when groups were used only due to a Type I error, since dividing a continuous variable into categories increases the risk of a Type I error (Cohen, 1983).”

“To the contrary, the model has both greater power and interpretability when X is included as a continuous covariate (Cohen, 1983; Preacher et al., 2005).”

“However, we acknowledge that categorizing PRD has the potential to lose important information (Cohen, 1983).”

“…such as the loss of information and statistical power and the population-dependency of a participant’s group membership (e.g., Cohen, 1983; Cohen & Cohen, 1983; Irwin & McClelland, 2003; Maxwell & Delaney, 1993), conducting moderation analyses and by this means keeping the continuous nature of…”

“A general rule of thumb is that dichotomizing normally distributed predictor variables reduces the squared multiple correlation to about 64% of what it otherwise would have been (Cohen, 1983).”

“Indeed, the argument against artificially bifurcating continuous data is not new (e.g., Cohen, 1983), however past attempts to make inferences about generational effects in cross-sectional designs present an opportunity to investigate the particularly insidious nature of this practice and its Cort…”

“The lesion delineation technique we used compares CT image intensity from a single patient with a group of images from control participants to identify outlier voxels (Crawford et al., 2009; Gillebert et al., 2014).”

“Validities for Valentine (1977) were corrected for dichotomization of criterion (Cohen, 1983).”

“Rather than dichotomize the outcome variable, which would reduce statistical power (Cohen, 1983), or use transformations to reduce skew, we followed recommendations outlined by Atkins and Gallop (2007) and used negative binomial distributions and log link functions to account for the nonnormality…”

“if they do not contribute usefully to interpretation of the trial nor it is statistically sensible to disregard the richness of the information that they provide; indeed, an ordinal approach to analysis typically contributes 36% more information and thus statistical power than a dichotomized approach.(3) Third, dichot…”

“Correlations involving the first OPR also were corrected for dichotomization (Cohen, 1983) of the OPR stratification score.”

“For example, dichotomizing variables at the median point of the distribution reduces statistical power in the same way that eliminating nearly 40% of the data would, and is even more costly when the cutting line deviates from the midpoint,[30] which is often the case in a responder analysis.”

“Although the use of median splits has been criticized on the basis that it reduces statistical power (Cohen, 1983), median splits are common in personality research of extraversion and sensation seeking, and several hypotheses about these traits have been confirmed using this methodology (Bluementhal, 2001; Helgason,…”

“We chose to use the continuous CES-D score; a dichotomous variable created by a cut point would result in loss of power (28).”

“and reduces statistical power to detect effects [65].”

“…removing demographic information directly results in the inability to study these variables; binning the data as a solution to keeping individual demographics anonymous would limit the variability of the data (Cohen, 1983) and limit the appropriate analyses and inferences that can be made as well.”

“The MEG data were analysed with a trial-level regression approach because regression-based analyses are more suited for continuous variables such as phonological neighbourhood and word frequency than factorial designs (Cohen, 1983; Hauk et al., 2006).”

“For example, there were a number of articles that dichotomised the interval, limiting their ability to observe any non-linear effects, and substantially decreasing the statistical power of the analyses [165, 166].”

“Such convertion of continuous measurements to categorial variables will amount to what Cohen and Cohen (1983) describe as dichotomisation.”

“Based on the body of research, one could select the average (Balmford et al., 2019; Becker et al., 1998; Cohen, 1983; Creel and Loomis, 1997; Dougados et al., 2012) and/or the median (Kim, 2018,; Teshima et al.”

“This could be a limitation as dichotomizations generally lead to reductions in statistical power, which could potentially increase the risk of Type II errors (Cohen, 1983) and, in extreme cases, spurious patterns of significance (Maxwell & Delaney, 1993).”

“Rather, researchers interested in experimentally testing interventions often rely on continuous (scaled) outcome measures – and there are, of course, good statistical arguments to do so (Cohen, 1983; DeCoster et al., 2009; MacCallum et al., 2002).”

“However, methodologists have argued for 171 more than two decades that the dichotomisation of c ontinuous variables is not valid for hypothesis 172 testing [1, 9-14, 16-18].”

“While dichotomization of data retains ordinal information, this technique can reduce statistical power by the same amount as deliberately discarding roughly a third of the data (Cohen, 1983; MacCallum et al., 2002).”

“Consistent with data from the general population (42, 43), we found that older age and the presence of physical health problems such as pain and mobility difficulties were associated with less physical activity.”

“We treated proficiency as a continuous variable (i.e., using the exact cloze test scores) because this analysis incorporates the full range of variation in proficiency and is more powerful (Cohen, 1983).”