Incremental criterion prediction of personality facets over factors: Obtaining unbiased estimates and confidence intervals
Introduction
Personality trait researchers have long been interested in how many personality traits are required to adequately capture individual differences. Hierarchical models of personality traits provide multiple levels of description, typified by the Five Factor Model in which the global factors of extraversion, neuroticism, conscientiousness, agreeableness and openness each consist of six facets representing a more detailed level of personality. Despite the popularity of the Big Five, there has been substantial debate about the relative merits of factor and facet assessments of personality (Ashton, 1998, Ashton et al., 1995, Ashton et al., 2014, Christiansen and Robie, 2011, O’Neill et al., 2013, Paunonen, 1998, Paunonen and Ashton, 2001, Paunonen et al., 2003, Paunonen et al., 1999, Salgado et al., 2013). Additionally, comparing the predictive value of a model with 30 facet predictors to one with only five factor predictors has presented a challenge for researchers concerned with issues of over fitting. Personality researchers seeking to predict outcomes such as well-being (Siegler & Brummett, 2000), job performance (Ashton, 1998, Christiansen and Robie, 2011, Ones and Viswesvaran, 1996, Salgado et al., 2013, Tett et al., 2003), and personality disorders (Bagby et al., 2005, Dyce and O’Connor, 1998) have then had to decide whether to include facets or only the Big Five factors as predictors.
Typically, incremental prediction of facets over factors has been estimated by subtracting the variance explained in a criterion by factors from that explained by facets. However, researchers have used many different estimators of variance explained, including unadjusted r-squared, adjusted r-squared, and cross-validated r-squared, combined with different regression procedures including direct entry (Mershon & Gorsuch, 1988) and stepwise regression (Baudin et al., 2011, Dyce and O’Connor, 1998, Ekehammar and Akrami, 2007, Quevedo and Abella, 2011, Schimmack et al., 2004); some studies have simply reported zero-order correlations (e.g., Rothmann and Coetzer, 2002, Siegler and Brummett, 2000). Thus, a principled selection of estimators is lacking (e.g., see critical review by O’Connor and Paunonen, 2007). Furthermore, the use of small sample sizes (e.g., Ashton et al., 1995, Mershon and Gorsuch, 1988, Schimmack et al., 2004) and incomplete facet–factor comparisons, based on the selection of subsets of either facets or factors (e.g., Ashton et al., 1995, Bagby et al., 2005, Dudley et al., 2006, Fruyt et al., 2006, Hastings and O’Neill, 2009, Paunonen and Ashton, 2001, Salgado et al., 2013, Stephan, 2009), has limited the available empirical evidence regarding the overall incremental value of facets over factors. Also, existing research has not explicitly specified a population parameter of interest. Furthermore, as will be shown many existing methods that have been used for estimating incremental prediction result in biased estimates. The lack of reporting of confidence intervals further compounds these issues. Further clarity is needed about these foundational issues in order to more clearly quantify the gains that can be achieved by the inclusion of facets in predictive models. Thus, existing approaches are insufficient for researchers seeking to make conclusions about the relative utility of facet- versus factor-level analysis in personality research.
The purpose of the current paper is (1) to identify the population parameter of interest for research on incremental prediction of facets over factors; (2) to compare methods for obtaining an estimate of this population parameter to demonstrate relative bias across methods through a series of simulations, and (3) to provide a method for reporting confidence intervals around this estimate. We also critically review the broader set of approaches that have been used to compare factor versus facet prediction of criterion variables. Based on our comparison of methods, we recommend the use of the Olkin–Pratt adjusted r-squared as an estimator, and the reporting of double-adjusted r-squared bootstrap (DAB) confidence intervals. We also review and make recommendations regarding methods for identifying which particular facets are of greatest incremental benefit. Finally, we present an R package that implements all the proposed methods.
Section snippets
Identifying the parameter of interest for incremental prediction research
The present paper focuses on the scenario where factors and facets come from a hierarchical measure, such as the 30 facets and five factors from the NEO-Personality Inventory (Costa & McCrae, 1995). Notably however, the method presented in this paper could readily be extended to scenarios where factors and facets are not derived from a hierarchical measure, such as when additional facet predictors are included. We also note that various other questions can meaningfully be asked, such as how
Description of estimators
We now review the different methods that have been used to estimate the incremental population variance explained by a regression with factors as predictors versus one with facets as predictors, denoted Δρ2. In general, estimates of Δρ2 are obtained by first obtaining estimates of ρ2 for facets and for factors, and then subtracting one from the other: , i.e., where the hats indicate estimates of corresponding population parameters. Three major classes of estimators of ρ2
Confidence intervals on incremental variance explained
There has been a movement in the reporting of psychological results that advocates the reporting of confidence intervals on effect sizes. Despite this recommendation, we were unable to find a single study in the personality factor versus facet comparison literature that has reported a confidence interval for Δρ2. This is presumably due to the lack of integration in standard statistical software for confidence intervals for ρ2, let alone Δρ2. Thus, a major contribution of this paper is the
The current study: a demonstration of relative bias across estimation methods
The preceding sections have critically evaluated estimators of Δρ2 that have been used in the literature. The analyses that follow apply the different estimators to a sample dataset and demonstrate the proposed procedure for obtaining confidence intervals. We then present a simulation study to assess bias and standards error of common estimators of Δρ2 for different sample sizes and datasets. Specifically, it was expected that Olkin–Pratt adjusted r-squared would provide the best estimator of Δρ
Illustration of recommended method
The current section compares different estimators of incremental variance explained and applies the Double Adjusted r-squared Bootstrap (DAB) method to calculate confidence intervals, to a sample dataset. The data comes from a study (Anglim & Grant, 2014) completed by 337 adults recruited from two Australian universities (76% female; mean age of 24.4, SD = 8.8). The 300 item version of the International Personality Item Pool modeled on the NEO-PI (Goldberg, 1999) was used to measure the five
Simulation comparing estimators
In order to evaluate the properties of various estimators of Δρ2 a simulation study was performed. The study assessed the bias and standard error of six estimators using two data generating mechanisms and four sample sizes. The six estimators were Olkin–Pratt adjusted r-squared, Ezekiel Adjusted r-squared, unadjusted r-squared, standard stepwise, penalized stepwise, and best facets as described in the previous data analysis section. The two data generating mechanisms labeled no effect and real
Identifying the importance of specific personality facets
We now make a few brief recommendations for researchers aiming to identify the role and importance of specific personality facets. If factors are taken as primary, a meaningful incremental prediction of facets overs factors is necessary to further examine the facet–criterion relationship. However, where such incremental prediction is present, this raises the issue of how best to describe the pattern of relationships. Several techniques have been adopted in the literature. Researchers commonly
Major recommendations
The major contribution of this paper was to provide recommendations for obtaining unbiased point estimates and confidence intervals for population incremental variance explained by facets over factors and therefore answer the question of how much more variance facets explain. In particular, principled examination of estimators and simulation both supported the use of adjusted r-squared formulas. While the simulation showed minimal differences between Olkin–Pratt and Ezekiel formulas, we still
Acknowledgment
We thank Sue Carmen for her assistance with data collection.
References (46)
- et al.
The criterion validity of broad factor scales versus specific facet scales
Journal of Research in Personality
(1995) - et al.
On the validity of narrow and broad personality traits: A response to
Personality and Individual Differences
(2014) - et al.
Predicting workplace deviance using broad versus narrow personality variables
Personality and Individual Differences
(2009) - et al.
Big Five personality predictors of post-secondary academic performance
Personality and Individual Differences
(2007) Openness to experience and active older adults’ life satisfaction: A trait and facet-level analysis
Personality and Individual Differences
(2009)- et al.
Asymptotic confidence limits for the difference between two squared multiple correlations: A simplified approach
Psychological Methods
(1999) - et al.
Confidence intervals for an effect size measure in multiple linear regression
Educational and Psychological Measurement
(2007) - et al.
Predicting psychological and subjective well-being from personality: Incremental prediction from 30 facets over the big 5
Journal of Happiness Studies
(2014) Personality and job performance: The importance of narrow traits
Journal of Organizational Behavior
(1998)- et al.
DSM-IV personality disorders and the five-factor model of personality: A multi-method examination of domain-and facet-level predictions
European Journal of Personality
(2005)
Note on unbiased estimation of the squared multiple correlation coefficient
Statistica Neerlandica
The role of personality in satisfaction with life and sport
Behavioral Psychology/Psicologia Conductual
Modeling general and specific variance in multifaceted constructs: A comparison of the bifactor model to other approaches
Journal of Personality
Further consideration of the use of narrow trait scales
Canadian Journal of Behavioural Science
Domains and facets: Hierarchical personality assessment using the Revised NEO Personality Inventory
Journal of Personality Assessment
Hierarchical regression analysis in structural equation modeling
Structural Equation Modeling: A Multidisciplinary Journal
Backward, forward and stepwise automated subset selection algorithms: Frequency of obtaining authentic and noise variables
British Journal of Mathematical and Statistical Psychology
The satisfaction with life scale
Journal of Personality Assessment
A meta-analytic investigation of conscientiousness in the prediction of job performance: Examining the intercorrelations and the incremental validity of narrow traits
Journal of Applied Psychology
Personality disorders and the five-factor model: A test of facet-level predictions
Journal of Personality Disorders
Personality and prejudice: From Big Five personality factors to facets
Journal of Personality
Methods of correlation and regression analysis: Linear and curvilinear
An R and S-Plus companion to applied regression
Cited by (34)
The influence of learning orientation on entrepreneurial performance: The role of business model innovation and risk-taking propensity
2023, Journal of Open Innovation: Technology, Market, and ComplexityBig Five factor and facet personality determinants of conflict management styles
2023, Personality and Individual DifferencesPredicting employee attitudes to workplace diversity from personality, values, and cognitive ability
2019, Journal of Research in PersonalityCitation Excerpt :Using adjusted r-squared is particularly important when comparing models with different numbers of predictors, such as when comparing models with 25 HEXACO facet predictors to one with 6 HEXACO domain predictors. Confidence intervals on adjusted r-squared change was assessed using bootstrap double adjusted-r-squared confidence intervals (Anglim & Grant, 2014). In addition to reporting regression coefficients, we also report a dominance analysis.
HEXACO personality predicts counterproductive work behavior and organizational citizenship behavior in low-stakes and job applicant contexts
2018, Journal of Research in Personality