A test of the four-factor model of impulsivity-related traits

www.elsevier.com/locate/paid

A test of the four-factor model of impulsivity-related traits

Joshua Miller*, Kate Flory, Donald Lynam, Carl Leukefeld

Department of Psychology, University of Kentucky, 115 Kastle Hall, Lexington, KY 40506-0044, USA

Received 14 September 2001; received in revised form 12 March 2002; accepted 21 April 2002

Abstract

In a recent article, Whiteside and Lynam [Person. Indiv. Diff. 30 (2001) 669] proposed a new model forunderstanding personality pathways to impulsive behavior. Their UPPS model maintains that there arefour personality dimensions that are related differentially to impulsive behaviors: urgency, sensation seek-ing, (lack of) premeditation, and (lack of) perseverance. The purpose of the current study was to providevalidation for this model and to test some of the specific hypotheses offered by Whiteside and Lynam. Thesample used in this study consisted of 481 individuals who completed mailed surveys at age 20 and an in-depth laboratory protocol at age 21. Zero-order correlations and multiple regression analyses were used toexamine the relations between the four personality traits and various forms of externalizing behaviors,several types of internalizing psychopathology, and intelligence. Results revealed that the four traits weredifferentially related to various behaviors and forms of psychopathology consistent with many of White-side and Lynam’s predictions. Therefore, the UPPS model does appear to offer a novel and useful way ofunderstanding behaviors and forms of psychopathology considered to be characterized by some form ofimpulsivity.# 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Impulsivity; Personality; UPPS; Externalizing; Internalizing

Impulsivity is one of the most widely researched, but poorly agreed upon, personality con-structs. Despite many different nominal representations, some variant of impulsivity can be foundin every major model of personality. Various researchers have described impulsivity-related con-structs using terms such as control, deliberation, excitement seeking, impulsivity, novelty seeking,psychoticism, self-discipline, and venturesomeness. Although there is clearly some conceptual and

0191-8869/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.

PI I : S0191-8869(02 )00122-8

Personality and Individual Differences 34 (2003) 1403–1418

* Corresponding author. Tel.: +1-859-257-6683; fax: +1-859-323-1979.

E-mail address: [email protected] (J. Miller).

http://www.elsevier.com/locate/paid/a4.3d

mailto:[email protected]

content overlap among these impulsivity constructs, there may also be real and important dis-tinctions among them.Over the past 30 years, there have been several attempts to consolidate the many impulsivity-

related constructs by locating impulsivity within a broader framework of personality. Forinstance, Eysenck and Eysenck (1985) argued that impulsivity had multiple forms, one of which(venturesomeness) existed on the superfactor of Extraversion, while a more ‘‘pure’’ form ofimpulsivity existed on the superfactor of Psychoticism. Cloninger’s personality model (Cloninger,Przybeck, & Svrakic, 1991; Cloninger, Svrakic, & Przybeck, 1993) gave impulsivity its own factorcalled Novelty Seeking. It is clear from this name that Cloninger’s view of impulsivity is onecharacterized primarily by a need for engagement in novel and risky behaviors. Similarly, Zuck-erman and colleagues (Zuckerman, Kuhlman, Joireman, Teta, & Kraft, 1993) placed the impul-sivity construct on a superfactor entitled Impulsive Sensation Seeking, which combines a need forexcitement and danger with a difficulty in delaying one’s behavioral responses in order to firstconsider the possible consequences. Barratt (1959, 1965, 1972) developed a personality-basedmodel of impulsivity with an explicit goal of distinguishing impulsivity-related traits from anxi-ety-related traits, in order to capture a ‘‘purer’’ form of impulsivity. Although these variousconceptualizations of impulsivity may have merit, they do not seem to capture fully the variety ofimpulsivity-related constructs found in the literature.Recently, Whiteside and Lynam (2001) have suggested that the Five Factor Model (FFM) of

personality, as assessed by the NEO-PI-R (Costa & McCrae, 1992), might provide a way of par-sing impulsivity into its different forms. They argued that the FFM model includes four person-ality traits that represent distinct pathways to impulsive behavior. These four traits exist on threeseparate higher-order factors of personality. The first trait they identified was the NEO-PI-R traitof impulsiveness, which is situated on the Neuroticism factor and assesses an individual’s ten-dency to give in to strong impulses, specifically when accompanied by negative emotions such asdepression, anxiety, or anger. Next, they suggested that excitement seeking, a specific trait on thehigher-order factor of Extraversion, which measures an individual’s preference for excitementand stimulation, is another route to impulsive behavior. The third and fourth traits, self-disciplineand deliberation, both exist on the broader factor of Conscientiousness, and a lack of these traitsis hypothesized to lead to impulsive behavior. Self-discipline measures an individual’s ability topersist in completing jobs or obligations despite boredom and/or fatigue, while deliberationassesses an individual’s ability to think through the potential consequences of his or her behaviorbefore acting. Whiteside and Lynam (2001) went on to argue that these four personality dimen-sions actually subsume the major conceptualizations of impulsivity and that the four dimensionsare best understood as four personality pathways to impulsive behavior. That is, rather than oneoverarching impulsivity factor, there may be multiple impulsivity-related personality traits thatlead to various impulsive behaviors.In order to test these hypotheses, Whiteside and Lynam (2001) conducted factor analyses with

the most widely-used impulsivity scales and the four selected scales from the FFM. The resultssupported the existence of a four-factor model, which accounted for 66% of the variance in theimpulsivity instruments. Importantly, each of the four traits from the FFM loaded on a separatefactor. Following this, Whiteside and Lynam (2001) created four new scales to assess each of theunderlying dimensions. Although the new factors were defined by the FFM scales, in an effort tobring clarity the authors titled the new dimensions urgency (FFM-impulsiveness), perseverance

1404 J. Miller et al. / Personality and Individual Differences 34 (2003) 1403–1418

(FFM-self-discipline), premeditation (FFM-deliberation), and sensation seeking (FFM-excite-ment seeking), and named the new scale the UPPS Impulsive Behavior scale.Before this new conceptual approach to parsing impulsivity can be considered useful, however,

convergent and discriminant validity must be demonstrated. Whiteside and Lynam (2001) suggestvery specific ways in which the dimensions might be related to psychopathology and other prob-lematic behaviors. For instance, they suggest that urgency, the difficulty resisting strong impulsesdriven by negative affect, might be related to borderline personality disorder and bulimia. Con-sistent with this hypothesis, Linehan (1993) has argued that parasuicidal and suicidal behaviors,common to individuals with borderline personality disorder, ‘‘are usually maladaptive solutionbehaviors to the problem of overwhelming, uncontrollable, intensely painful negative affect’’ (p.60). Similarly, several studies have documented the role of impulsivity in symptoms of bulimianervosa (e.g., Wiederman & Pryor, 1996). With regards to perseverance, Whiteside and Lynam(2001) argued that the lack of ability to stay on task despite boredom might be related to theinattention problems that are at the core of attention deficit hyperactivity disorder [AmericanPsychological Association (APA), 1994]. Next, they hypothesized that sensation seeking, anindividual’s need for excitement and stimulation, might be related to involvement in exciting butdangerous activities such as substance use. Consistent with this idea, numerous studies havedocumented relations between sensation seeking and substance use (e.g., Andrew & Cronin, 1997;Donohew et al., 1999; Earleywine & Finn, 1991). Finally, Whiteside and Lynam (2001) statedthat a lack of premeditation might lead to psychiatric problems such as antisocial personalitydisorder and psychopathy, each of which are characterized by an inability to consider thepotential consequences of one’s behavior. Consistent with this hypothesis, a recent meta-analysis(Miller & Lynam, 2001) found that several personality predictors related to premeditation weresignificantly associated with antisocial or delinquent behavior.In the current study, we examine the utility of Whiteside and Lynam’s (2001) model by

exploring the relations between the four personality factors and a variety of externalizing (i.e.,crime, aggression, psychopathy, risky sex, ADHD, Borderline Personality, and eating problems)and internalizing problems (i.e. Generalized Anxiety Disorder and Depression). We used the fouroriginal FFM scales to represent each of the UPPS traits: urgency was assessed with FFM-impulsiveness, perseverance was assessed with FFM-self-discipline, premeditation was assessedwith FFM-deliberation, and sensation seeking was assessed with FFM-excitement seeking.Although our primary goal was to look for evidence of convergent and discriminant validityamong these personality dimensions, we were also interested in possible gender differences and intesting the specific hypotheses offered by Whiteside and Lynam (2001).

1. Method

1.1. Participants

The sample used in this study was composed of 481 individuals who were part of the 10-yearLexington Longitudinal Study. These individuals were part of a larger sample (N=1431) that hadbeen followed since the 1987–1988 school year, at which time participants were in the 6th grade.Data were collected from the participants through school-based questionnaires over a five-year

J. Miller et al. / Personality and Individual Differences 34 (2003) 1403–1418 1405

period from 6th through to 10th grades and again via mailed survey at age 20. Following themailed survey, 481 (approximately 47% of the mailed-survey sample) individuals participated inan extensive, 3–4 h protocol that included personality and intelligence assessments, a diagnosticinterview, and detailed histories of past and current substance use and delinquency. The averageage of the participants during this protocol was 21.The 481 individuals who participated in the laboratory protocol were selected from the larger

sample of 1431 to over-represent individuals with histories of substance use in order to compen-sate for previous sample attrition. Attrition analyses were conducted to examine the similarity ofour laboratory sample to the larger longitudinal sample. The laboratory sample of 481 wascompared to the 950 individuals who were also in the longitudinal study, but were not selected forthe laboratory protocol. We examined whether the 481 differed on a number of variables assessedat 6th grade, 7th and 8th grades, and 9th and 10th grades. These variables included the following:gender; age; race; lifetime, yearly, and monthly use of cigarettes, chewing tobacco, alcohol, andmarijuana at each of the three assessment periods; and sensation seeking, peer relations, familyrelations, self-esteem, and peer pressure resistance at each assessment period.The laboratory sample differed in only a few ways from the larger longitudinal sample. Indivi-

duals in the laboratory sample tended to be younger, t (981)=�3.15, P<0.01, and reported moremarijuana use during 7th/8th grade, t (699)=�2.11, P<0.05. In addition, the laboratory samplewas more likely to be white, w2=11.98, d.f.=1, P<0.001. No other differences were significant.Finally, a pairwise regression analysis was conducted in which all of the variables were entered(except age, race, and gender which were only entered once) for all three assessment periods inorder to predict an individual’s present versus absent status in the laboratory sample. The 54variables (17 variables measured at three times, and gender, race, and age) accounted for a smallbut significant amount of the variance, R2=0.06, F (54, 1214)=1.45, P<0.05. Overall, theseresults suggest that the laboratory sample is sufficiently comparable to the remainder of thesample that was not chosen for the more intensive protocol.

1.2. Procedure

Before the participants were brought into the laboratory, they signed a consent form andcompleted a packet of questionnaires mailed to them a week before the laboratory protocol. Themailed packet consisted of several self-report measures including the NEO-PI-R (Costa &McCrae, 1992) and the Levenson Self-Report Psychopathy Scale (Levenson, Kiehl, & Fitzpatrick,1995).The laboratory protocol itself lasted approximately three to four hours and was administered

by trained research assistants. During the protocol, participants completed two Life HistoryCalendars (Caspi et al., 1996), one for delinquency and one for substance use, segments of theDiagnostic Interview Schedule DIS; (Robins, Cottler, Bucholz, & Compton, 1997), several subt-ests from the Wechsler Adult Intelligence Scale-Revised WAIS-R (Wechsler, 1981); and a ques-tionnaire designed to assess attention deficit hyperactivity disorder symptoms.Although the present study primarily utilized measures collected during the laboratory protocol

described above, several variables (e.g., number of sexual partners, age of sexual debut, border-line personality disorder symptoms, eating problems, aggression) were taken from the mailedsurvey administered when participants were 20 years old.


1.3. Measures

1.3.1. Impulsivity-related traitsThe four impulsivity-related personality traits were measured via four scales from the Revised

NEO-Personality Inventory (NEO-PI-R), a self-report questionnaire developed by Costa andMcCrae (1992) to assess normal personality dimensions based on the Five Factor Model of per-sonality. Specifically, the NEO-PI-R traits represent the four components of the UPPS model ofimpulsivity as follows: urgency=NEO-PI-R impulsivity; perseverance=NEO-PI-R self-dis-cipline; premeditation=NEO-PI-R deliberation, sensation seeking=NEO-PI-R excitement seek-ing. Each of these scales marked one of the four impulsivity factors identified by the Whitesideand Lynam (2001) study. The NEO-PI-R consists of 240 items that are rated on a five-point scale,anchored by ‘‘strongly disagree’’ and ‘‘strongly agree’’. The NEO-PI-R yields a score for all fivehigher-order factors (neuroticism, extraversion, agreeableness, conscientiousness, and openness)based on 48 questions per factor, and assesses six traits within each factor using 8 items per trait.Many studies have been conducted using the NEO-PI-R and it has consistently shown goodreliability and validity. In the present study, we were not able to examine the internal con-sistencies, however, reliability estimates have been reported previously in the normative sample.Internal consistency for the four impulsivity related traits were as follows: impulsiveness(urgency) =0.70, excitement seeking (sensation seeking)=0.65, self-discipline (lack of persever-ance) =0.75, and deliberation (lack of premeditation) =0.71 (Costa, McCrae, & Dye, 1991). Inthe current study, items on the perseverance and premeditation scales were reverse-scored so thathigh scores on each of the four factors represented greater impulsivity.

1.3.2. Diagnostic interview scheduleThe diagnostic interview schedule (DIS; Robins et al., 1997) is a structured interview used by

non-clinicians to assess the presence or absence of psychiatric disorders. The version of the DISused in the current study corresponds with diagnoses included in the DSM-IV (1994). Althoughthe DIS has been changed periodically to correspond to the criteria of the most current version ofthe DSM, these changes do not alter the DIS as an assessment tool. Thus, reliability and validityevidence is believed to be the same across versions. The DIS-II, which corresponded to the DSM-III, proved to have good sensitivity (0.75), excellent specificity (0.94), and moderate positive pre-dictive power (0.76) (Robins, Helzer, Ratcliff, & Seyfried, 1982). Kappa, a measure of agreement,was conducted to see how reliable the DIS was for administration by non-clinicians versusadministration by psychiatrists. Kappa ranged from a low of 0.4 (only for a diagnosis of panicdisorder) to a high of 1 (for a diagnosis of anorexia nervosa) with all other diagnoses at least ashigh as 0.51 (Robins et al., 1982). For the current study, only selected sections of the DIS wereused. Participants were assessed for antisocial personality symptoms since age 15, generalizedanxiety, and depression.

1.3.3. Life history calendars for substance use and delinquencyThe life history calendar (LHC) is a retrospective method for collecting data on a wide range of

life events and behaviors (Caspi et al., 1996). The LHC is a large grid that is used to document theoccurrence of certain events in the subject’s life. The rows represent the different activities andevents of interest, while the columns partition the grid into different blocks of time. In our LHC,


we asked about the occurrence of delinquent acts (i.e., truancy, stealing, fire-setting, fighting) andsubstance use beginning in 1986, when participants were in the fifth grade, up until the time of thelaboratory protocol. Each year was broken into three four-month segments and participants wereasked about the occurrence and frequency of acts and use during these time intervals. Previousstudies (e.g., Caspi et al., 1996) and data from our project document the reliability and validity of theLHC. For example, in our study, agreement between survey reports of ever having used a substance(measured in 7th, 8th, 9th, and 10th grades and at age 20–21) and the retrospective LHC was quitegood with average kappas of 0.47, 0.46, and 0.57 for cigarettes, alcohol, and marijuana respectively.For the present study, we created a variable that assessed the total number of different types of

delinquent acts that a participant reported across the course of the calendar. For substance use,we created variables that represented the number of periods during which a participant reportedregular use of cigarettes, alcohol, and drugs (i.e., marijuana, cocaine, PCP).

1.3.4. Levenson Self-Report Psychopathy ScaleUnlike previous self-report measures of psychopathy that measure only the social deviance

component of the construct, the Levenson Self-Report Psychopathy Scale (LSRP) (Levenson etal., 1995; Lynam, Whiteside, & Jones, 1999) was created specifically to assess both social devianceand the callous, remorseless view of others. The 26 items of the LSRP are scored on a four-point scaleranging from ‘‘disagree strongly’’ to ‘‘agree strongly’’. Factor analyses of the LSRP have shown thatthe items do indeed load on two factors consistent with its conceptualization (Levenson et al., 1995;Lynam et al., 1999). Scores on the LSRP have been found to relate in predicted directions to seriousantisocial behavior, personality dimensions of disinhibition, neuroticism, agreeableness, con-scientiousness, and boredom susceptibility, and performance tasks on response modulation (Lynamet al., 1999). Internal consistency for the overall scale was excellent, �=0.98.

1.3.5. Performance and verbal IQPerformance IQ was assessed during the laboratory protocol using the block design subtest

from the Wechsler Adult Intelligence Scale-Revised WAIS-R; (Wechsler, 1981). Verbal IQ wasassessed using the Vocabulary subtest of the WAIS-R. Both scales are reliable (block design �=0.88;vocabulary �=0.94) and have amongst the highest overall g loading for their respective IQ category.

1.3.6. Attention deficit hyperactivity disorderChildhood attention deficit hyperactivity disorder (ADHD) symptoms were assessed retro-

spectively during the laboratory protocol using an 18-item questionnaire with items constructedto correspond directly to the DSM-IV (1994) criteria for ADHD (e.g., ‘‘When you were younger,did you fail to give close attention to details or make careless mistakes in schoolwork, work, orother activities?’’). For each symptom, individuals indicated whether they had experienced theproblem in childhood and whether it had resulted in a significant degree of impairment. For thepresent study, total inattentive symptoms (�=0.74) and total hyperactive/impulsive symptoms(�=0.75) were examined.

1.3.7. Number of sexual partnersThis variable was assessed by an item that asked, ‘‘During your life, howmany sexual partners have

you had sexual intercourse with?’’ The answer choices for this item ranged from ‘‘0’’ to ‘‘6 or more’’.


1.3.8. Age at first sexual intercourseThis variable was assessed by an item that asked ‘‘How old were you when you had sexual

intercourse for the first time?’’ The possible answers for this question ranged from ‘‘I have neverhad intercourse’’ to ‘‘18 years old or more’’. Individuals that had not engaged in sexual inter-course at the time of the mailed survey were not included in analyses that used this variable.

1.3.9. Borderline personality disorder symptomsThis scale was created by taking the sum of 10 items from the Brief Symptom Inventory (BSI;

Derogatis, 1993) that assessed relevant DSM-IV (1994) borderline personality disorder diagnosticinformation as agreed upon by all three authors. For example, one of the items used assessedemotional instability, while others assessed loneliness and difficulty being alone. Item choices rangedfrom 1 (‘‘not at all’’) to 5 (‘‘extremely’’). The internal consistency for this scale was high (�=0.89).

1.3.10. Eating problemsThis scale was created by calculating the sum of five items that asked whether the participant

was satisfied with his or her weight, considered himself or herself over, under, or normal weight,was currently dieting, or was restricting caloric intake, as well as the number of months duringwhich the participant had dieted over the past year. The internal consistency for this scale wasadequate (�=0.69).

1.3.11. AggressionThe aggression scale, derived from the Conflict Tactics Scale (Straus & Gelles, 1990), consisted

of the number of different aggressive acts the participant reported engaging in over the past 90days. The aggressive acts were as follows: (1) insulted or swore at someone, (2) threatened to hitor throw something at another person, (3) threw something at someone, (4) pushed, grabbed, orshoved someone, (5) slapped another person, (6) kicked, bit, or hit someone, (7) hit or tried to hitanyone with an object, (8) beat up someone, (9) threatened anyone with a knife or gun, and/or(10) used a knife or gun on another person. We added all 10 items to make up the total score. Theinternal consistency for this scale was adequate (�=0.84).

3. Results

The purpose of this study was to examine the correlates of the four personality traits hypothe-sized by Whiteside and Lynam (2001) to be related to impulsive behavior. Specifically, we wereinterested in examining whether different types of externalizing and internalizing behaviors andsymptoms would be related differentially to urgency, sensation seeking, lack of perseverance, andlack of premeditation. Before examining these external correlates however, the correlations ofthese four traits with one another were examined. As can be seen in Table 1, five of the sixintercorrelations were significant with a mean correlation of 0.36. This suggests that these fourtraits are related, yet distinct.


3.1. Zero-order relations

As can be seen in Table 2, we next examined the zero-order correlations between the four per-sonality traits and measures of substance use, delinquency, aggression, sexual behavior, psycho-pathology, and IQ. We have separately listed correlations that significantly differed by gender.Out of 68 correlations, only eight demonstrated significant gender differences and most were dueto only minor differences in the size of the relations for males and females.1

With regards to externalizing behaviors such as antisocial personality disorder, psychopathy,and variety of delinquent acts, the strongest correlation with these behaviors was lack of pre-meditation with an average correlation of 0.36, followed by urgency (0.25), lack of perseverance(0.23), and sensation seeking (0.21). In several cases (e.g., ASPD), the correlation for lack ofpremeditation was significantly stronger than the other correlations.With regards to aggression, although not significantly different from all other correlations, the

strongest correlate was urgency with a correlation of 0.20, followed by lack of premeditation(0.18), sensation seeking (0.12), and lack of perseverance (0.10).We also examined the relations between the four personality traits and several different aspects

of substance use. As can be seen in Table 2, substance use, including cigarette, alcohol, and druguse, was consistently related to all four traits, although the size of these relations differed sig-nificantly across the traits. Lack of premeditation was again the strongest correlate of thesebehaviors with a mean correlation of 0.35, followed by sensation seeking (0.22), urgency (0.22),and lack of perseverance (0.12).All four personality traits were also significantly related to both measures of sexual behavior

included in this study: number of sexual partners and age of sexual debut. As can be seen inTable 2, for both males and females, lack of premeditation was the strongest correlate, althoughit differed significantly only from the correlation for lack of perseverance. Interestingly, the rela-tion of lack of perseverance with number of sexual partners and age of sexual debut differed indirection across gender. For women, lack of perseverance was related to having more sexualpartners. However, for men, lack of perseverance was not related to number of sexual partners,

Table 1Intercorrelations between four personality dimensions

Urgency
Sensationseeking
Perseverance(lack of)

Premeditation(lack of)

Urgency
– – Sensation seeking 0.29�� – Perseverance (lack of) 0.41�� 0.06 – Premeditation (lack of) 0.56�� 0.31�� 0.50�� –
**P<0.01.

1 Gender differences in the zero-order correlations between the four personality dimensions and each outcome wereexamined using a series of hierarchical regression equations with gender and the personality dimension entered in the

first step and the interaction term between gender and personality entered in the second step. In instances where theinteraction term was significant at P4 0.05, zero-order correlations were reported separately for males and females(see Table 2).


but was positively related to age of sexual debut, such that men who scored higher in lack ofperseverance had a later age of sexual debut.Additionally, we examined the relations between other forms of externalizing psychopathol-

ogy and the four UPPS traits. Again, almost all of the variables we examined in this categorywere significantly correlated with the four personality traits, although the size of the correlations

Table 2

Zero-order correlations between personality dimensions and externalizing and internalizing behaviors

Urgency
Sensation seeking Perseverance(lack of)

Externalizing problems

Crime and delinquency
ASPD 0.19**a 0.21**a 0.17**a 0.32**b
LSRP total
0.29**ab/a 0.17**a/a 0.26**ab/.46**b 0.32**b/0.52**b Variety conduct problems 0.31**a/0.25**ab 0.24**ab/ab 0.15**b/a 0.34**a/b
Aggression
CTS total aggression 0.20**a 0.12*ab 0.10*b 0.18**ab
Alcohol and drug use
Cigarette use 0.18**a 0.10*a 0.13**a 0.32**b Alcohol use 0.22**a/a 0.31**ab/ab �0.06c/.26**ab 0.34**b/b
Drug use
0.25**a 0.25**a 0.13**b 0.38**c
Risky sexual activity
# Sex partners 0.22**a/ab 0.23**a/ab �0.07b/0.15*a 0.29**a/b Age sexual debut �0.13*a/a �0.06a/a 0.17*b/-0.12a �0.15**a/a
ADHD
Inattention symptoms 0.09 0.10* 0.17** 0.17** Hyper/impulsivity symptoms 0.12**ab 0.12**ab 0.11*a 0.20**b
Other psychopathology

Borderline symptoms
0.08a/0.34**a �0.11b/0.08b 0.23**c/ab 0.17**ac/b Eating problems 0.25**a 0.01b 0.10*b 0.11*b
Discriminant relations

Internalizing problems

GAD symptoms
0.13**a �0.04b 0.16**a 0.09*a Depression symptoms 0.23**a 0.07b 0.21**a 0.23**a
Intelligence
Performance IQ �0.03a 0.07ab 0.07b 0.00ab
Verbal IQ
�0.02ab 0.01ab 0.07a �0.05b
*P<0.05, **P<0.01.ASPD=antisocial personality disorder, LSRP=Levenson Self-Report Psychopathy Scale, ADHD=attention deficit

hyperactivity disorder, GAD=generalized anxiety disorder.

For outcomes where any correlations were significantly different by gender, the correlation for males is presentedfirst followed by the correlation for females.Correlations with different letters differ significantly at P40.05; in the case of variables for which there was a gender

difference, subscripts appearing to the left represent males, while subscripts to the right represent relations corre-sponding to females.


differed as a function of which trait was examined. As Table 2 demonstrates, symptoms of inat-tention and hyperactivity/impulsivity, were most strongly related to lack of premeditation, fol-lowed by lack of perseverance, sensation seeking, and urgency. With regards to borderlinepersonality disorder symptoms, for women, urgency was the strongest correlate, whereas thisdimension was not a significant correlate in men. However, both lack of perseverance and lack ofpremeditation were significantly correlated with borderline personality disorder symptoms inboth genders. Similarly, eating problems were significantly related to urgency, lack of persever-ance and lack of premeditation, although these latter relations were significantly smaller than therelations for urgency.

Table 3Standardized regression coefficients for personality traits for simultaneous analyses

R2
Urgency Sensationseeking
Perseverance(lack of)


Externalizing problems

Crime and delinquency
ASPD 0.12** �0.02 0.13** 0.03 0.28** LSRP total 0.22** 0.02 0.06 0.18** 0.32**
Variety conduct problems
0.14** 0.04 0.15** �0.01 0.28**
Aggression
CTS—total aggression 0.05** 0.14* 0.05 �0.01 0.10
Alcohol and drug use

Cigarette use
0.10** 0.01 �0.00 �0.03 0.33** Alcohol use 0.17** 0.02 0.21** �0.07 0.30** Drug use 0.17** 0.05 0.13** 0.07 0.16**
Risky sexual activity

# Sex partners
0.12** 0.10 0.12** �0.13* 0.26** Age sexual debut 0.04** �0.10 �0.01 0.13* �0.15*
ADHD
Inattention symptoms 0.04** �0.05 0.07 0.13* 0.12
Hyper/imp symptoms
0.04** �0.00 0.07 0.03 0.16**
Other psychopathology
Borderline symptoms 0.08** 0.18** �0.08 0.15** 0.02
Eating problems
0.07** 0.29** �0.06 0.00 �0.03
Discriminant relations

Internalizing problems
GAD symptoms 0.04** 0.11 �0.10* 0.05 0.08
Depression symptoms
0.08** 0.12* �0.01 0.10 0.12*
Intelligence
Performance IQ 0.02 �0.08 0.09 0.11* �0.03 Verbal IQ 0.01 �0.02 0.04 0.13* �0.11
*P<0.05, **P<0.01.

ASPD=antisocial personality disorder, LSRP=Levenson Self-Report Psychopathy Scale, ADHD=attention deficithyperactivity disorder, Hyper/Imp=hyperactivity/impulsivity, GAD=generalized anxiety disorder.


Finally, we examined the relations of the four personality traits to internalizing problems andintelligence, primarily for the purpose of demonstrating discriminant validity. However, we foundthat both generalized anxiety disorder symptoms and depression symptoms were significantlyrelated to urgency, lack of perseverance, and lack of premeditation, although these relationstended to be somewhat smaller than the relations between the traits and externalizing problems.With regards to the relations between personality and intelligence, neither verbal nor perfor-mance IQ was significantly related to any of the four traits.

3.2. Simultaneous analyses

Following these zero-order analyses, we were interested in two further questions. First, howmuch variance did the personality traits account for in the behaviors and symptoms examined,and second, when controlling for statistical overlap between the traits, which traits would proveto be the most useful in understanding the various behaviors. Because so few gender differenceswere found, gender was not included in the simultaneous analyses. As can be seen in Table 3, weconducted regression analyses for each outcome variable in which the outcome variable waspredicted simultaneously by all four personality traits. Because of our large sample size, the R2’swere significant for all of our variables, except for verbal and performance IQ.With regard to crime and delinquent activities, the R2’s ranged from 0.12 to 0.22. With all four

personality traits entered at the same time, lack of premeditation made the largest and mostconsistent independent contribution. Sensation seeking was also significantly related to two of thethree variables. A lack of perseverance was a significant predictor of psychopathy only, whileurgency was not significantly related to any of the crime/delinquency outcomes.Different relations to aggression were also observed. The personality traits accounted for 5% of

the variance in the outcome, but only urgency made a significant independent contribution whenall four traits were in the regression model.The impulsivity-related traits accounted for significant portions of variance in all of the sub-

stance use variables, ranging from 10 to 17%. The results for these three variables were strikinglysimilar across types of substances, as well as with results for crime and delinquency, in that a lackof premeditation was a significant independent predictor for all three variables, while sensationseeking significantly predicted two of the three. However, neither urgency nor lack of persever-ance was significantly, independently related to substance use.As can be seen in Table 3, the personality traits accounted for 12% of the variance in the

number of sexual partners. Again, lack of premeditation made the largest independent contribu-tion, followed by smaller, but significant, contributions by lack of perseverance and sensationseeking. In addition, the traits were able to explain 4% of the variance in age of sexual debut,with lack of premeditation and lack of perseverance significantly predicting this variable when allfour traits were included in the regression model. Unexpectedly, the relations between lack ofperseverance and the sexual activity variables were in the direction opposite than expected. Lessperseverance was associated with fewer partners and a later sexual debut.For the ADHD variables, the personality traits accounted for 4% of the variance. Interestingly,

lack of perseverance was the only significant independent predictor of inattention problems, whilea lack of premeditation was the sole independent predictor of the hyperactivity/impulsivitysymptom count.


The personality factors explained 8 and 7% of the variance in borderline personality disordersymptoms and eating problems, respectively. For both of these variables, urgency was the stron-gest predictor. In fact, urgency was the only independent contributor to the eating problemsvariable. In addition, lack of perseverance was significantly related to borderline symptomatologywhen all four personality traits were included in the model.Finally, the personality traits accounted for 4% of the variance in anxiety symptoms and 8% of

the variance in depressive symptoms. For anxiety symptoms, only sensation seeking made a sig-nificant independent contribution such that low sensation seeking predicted high anxiety. Fordepression, both urgency and lack of premeditation were significant predictors such that highurgency and a lack of premeditation predicted depression.

4. Discussion

Recently, Whiteside and Lynam (2001) argued that four specific personality traits might bestaccount for a variety of behaviors and actions that are considered impulsive. They argued that,‘‘impulsivity is an artificial umbrella term that actually encompasses four distinct facets of per-sonality associated with impulsive behavior’’ (Whiteside & Lynam, 2001, p. 687). That is, ratherthan being a unitary trait, impulsivity is best conceived as being made up of multiple traits orpathways that can lead to a variety of impulsivity related behaviors. Using the FFM as a startingpoint, Whiteside and Lynam developed the UPPS Impulsive Behavior scale, which assesses thefour personality traits of urgency, sensation seeking, perseverance, and premeditation. In thecurrent study, we have used the corresponding scales from the FFM (impulsivity, excitementseeking, self-discipline, and deliberation) to assess the four impulsivity-related traits discussed byWhiteside and Lynam. The goal of the current paper was to examine the convergent and diver-gent relations between these four traits and externalizing and internalizing behaviors in whichimpulsive behavior is considered a defining feature (e.g., delinquency, psychopathy, substanceuse, sexual behavior, and several forms of psychopathology). Another aim of our study was totest several of the specific hypotheses about the UPPS model offered by Whiteside and Lynam(2001). For example, they argued that urgency should be related to borderline personality dis-order and bulimia, while sensation seeking would be related to exciting but dangerous behavior,such as substance use. In addition, they postulated that a lack of perseverance might be relatedspecifically to the inattention difficulties found in children with ADHD and that a lack of pre-meditation would be related to antisocial personality disorder and psychopathy. Finally, wesought to examine gender differences in the relations between the impulsivity-related personalitytraits and our outcome variables.Although zero-order correlations revealed that all four traits were significantly related to

almost all of the externalizing disorders and even several of the internalizing ones, several inter-esting trends did emerge. Specifically, lack of premeditation was consistently the strongest corre-late of the externalizing problems, whereas urgency appeared to be more strongly related toseveral other forms of psychopathology, such as borderline personality disorder symptoms, eat-ing problems, and depression.A series of regression analyses enabled us to more closely examine both the joint contributions

that the four traits made in predicting the outcomes, as well as their independent contributions.


These analyses suggested that the four personality traits made important and significant con-tributions to the prediction of a variety of behaviors and problems. The regression resultsrevealed that lack of premeditation, the inability to think through possible consequences of one’sbehavior before acting, was the most consistent of the four personality traits in explaining mostexternalizing behaviors. Unquestionably, it appears that this inability to consider potential con-sequences, often considered the prototypical concept of ‘‘impulsivity’’, is linked to a variety ofproblematic behaviors. Following lack of premeditation, the dimension of sensation seeking wasthe next most consistent and important personality dimension in accounting for a range ofexternalizing behaviors. Consistent with previous research (Zuckerman, 1994), this preference fordangerous and exciting activities was a significant predictor of involvement in delinquent acts,drug and alcohol use, and risky sexual behavior.In contrast, lack of perseverance, the inability to force oneself to persist in fulfilling jobs or

obligations, despite boredom or fatigue, explained very few of the externalizing behaviors. One ofthe few exceptions occurred for the total score of the Levenson Self-Report psychopathy scale(LSRP), which was predicted by a lack of perseverance and a lack of premeditation. These find-ings support the notion that psychopathy is a disorder characterized by extreme impulsivity(Lynam, 2002; Miller, Lynam, Widiger, Leukefeld, & Clayton, 2001). In addition, a lack of per-severance was also able to account for a significant amount of the variance in borderline per-sonality symptomatology and risky sexual behaviors (early debut, number of partners), althoughthis latter relation was in an unexpected direction. Finally, as predicted by Whiteside and Lynam(2001) a lack of perseverance was a significant predictor, in fact, the only significant predictor, ofinattention problems related to ADHD.Like lack of perseverance, the personality factor of urgency, the ‘‘tendency to experience strong

impulses, frequently under conditions of negative affect’’ (Whiteside & Lynam, 2001, p. 685), waslargely unrelated to externalizing problems, with a few exceptions. Interestingly, we did find thaturgency was the strongest predictor of aggression. One possible explanation for the role of thisneurotic form of impulsivity in accounting for aggressive behavior lies in the distinction betweenreactive and proactive aggression (Dodge & Coie, 1987). Although highly correlated with proac-tive aggression, reactive aggression is typically considered a hostile or angry counterstrike by aperson as a result of a belief that he or she has been attacked or is in danger. In contrast, pro-active aggression, which occurs less frequently, is considered an instrumental form of aggressionin which a person acts aggressively in order to achieve a goal such as increasing his or her statusor obtaining desired material goods. To the extent that the Conflict Tactics Scale (Straus &Gelles, 1990) captures reactive aggression, the relation to urgency may be due to the presence ofnegative affect across both constructs.Urgency was also found to be the strongest predictor of borderline personality disorder symp-

toms and eating problems, which is consistent with the usual characterization of these types ofpsychopathology as including intense negative affect, as well as impulsive behaviors (e.g., self-mutilation, binging, purging) that attempt to alleviate some of the individual’s distress.With regards to gender, the findings from this study are similar to those offered by Whiteside

and Lynam (2001). In their study, it was demonstrated that the four-factor structure of impul-sivity-related traits was consistent across men and women. Similarly, in the current study, wefound that there were only a small number of cases in which the correlations between the fourpersonality traits and the outcome variables were significantly different across gender. In fact, of


68 correlations examined, only 8 demonstrated significant differences in the examined relationsfor men and women. This suggests that urgency, sensation seeking, lack of perseverance, and lackof premeditation behave similarly across gender.In order for the four-factor conceptualization of impulsivity-related personality traits to be

valid and useful, the traits must show divergent as well as convergent validity. As expected, thefour traits were unrelated to both performance and verbal intelligence. However, unexpectedly,the traits did relate to both anxiety and depressive symptoms. This finding is somewhat surprisingin that neither disorder is thought to be characterized by impulsive thinking or acting. The sig-nificant associations among urgency, depression, and anxiety might be due to the fact that allthree constructs deal with item content that reflects some form of negative affectivity. However,the relations of lack of premeditation and lack of perseverance to depression are more difficult toexplain. It is possible that, given the cross-sectional nature of our data, being depressed leads tosome temporary depression-related changes in the other personality traits. That is, individualswho are depressed may have a difficult time persisting in their duties because depression tends tosap one’s motivation (i.e., lack of perseverance). Similarly, depression may make thinking clearlyand concentrating more difficult and therefore, depressed individuals may rate themselves asexperiencing difficulty thinking through the consequences of their behavior (i.e., lack of pre-meditation). Alternatively, it is possible that impulsivity is truly related to depression; this idea isconsistent with studies documenting the role of low serotonin in both impulsivity (e.g., Brown etal., 1989; Davidson, Putnam, & Larson, 2000) and depression (Deakin, 1991).Overall, the specific hypotheses offered by Whiteside and Lynam (2001) regarding the relations

of the personality traits of urgency, sensation seeking, perseverance, and premeditation with anumber of psychological disorders and problematic behaviors were supported. However, therelations between these traits and the outcome variables were not quite as clean as one mighthope. In fact, of the 15 outcome variables that were significantly predicted by the personalitytraits, only 8 were predicted by one specific dimension. This finding is not surprising given theinitial correlations showing significant interrelations among the traits. Thus, although not ideal, itis to be expected that often more than one of these personality traits will make a contribution inexplaining various impulsive behaviors. It must be noted, however, that the regression analysesdid provide some important clarity beyond that offered by the zero-order correlations in exam-ining the personality traits related to specific behaviors.The major limitation of this study is its cross-sectional nature. As with any cross-sectional

design, the current study cannot offer evidence directly supporting a causal link between person-ality and various outcomes. However, given the early rise of temperament (e.g., Caspi & Silva,1995) and the stability of personality (e.g., McCrae & Costa, 1990; Roberts & DelVecchio, 2000),one may be somewhat confident that it is unlikely that the impulsive behaviors created the per-sonality configurations. It is more likely that individuals’ personality traits led to, either directlyor indirectly, their impulsive behaviors. Future studies may help tease out these distinctions byexamining longitudinally the relation between the impulsivity-related traits and later impulsivebehavior and symptomatology. A second limitation of this study is the reliance on self-reportmeasures of both personality and the behavioral outcomes such as aggression, substance use, andpsychopathology. As a result, common method variance could have artificially inflated the find-ings. In addition, the lack of validity scales on the self-report measures may be problematicbecause the results could have been influenced by individuals responding in a socially desirable


manner. However, the differential pattern of results observed argues somewhat against thesepossibilities.The goal of the current study was to examine criterion validity of the UPPS model by examin-

ing relations between the four traits and a number of outcomes. We feel that this is an importantfirst step in establishing the validity and utility of the UPPS model. However, we acknowledgethat there are other steps necessary to fully validate this model. For instance, an important test ofthis model will be to compare the model against other, empirically supported models of impul-sivity such as Barratt’s (1965), Dickman’s (1990), and Eysenck’s (Eysenck, Pearson, Easting, &Allsopp, 1985). Our goal in introducing this model is not to supplant these pre-existing modelsbut to argue for a model of impulsivity that explicitly addresses the multidimensional nature ofthis construct. The current study offers some convincing evidence that the UPPS model ofimpulsivity-related traits developed by Whiteside and Lynam (2001) has the potential for intro-ducing order among the plethora of impulsivity constructs currently existing in the literature. TheUPPS model also appears to offer a novel and useful way of understanding behaviors and formsof psychopathology considered to be characterized by some form of impulsivity. Given theimportance of impulsivity in a wide variety of behaviors and disorders, this model may help bringadditional clarity to the field.

Acknowledgements

This research was supported by the University of Kentucky Research Challenge Trust FundFellowship to the first author.

References

American Psychiatric Association, 1994. Diagnostic and statistical manual of mental disorders (4th ed.). Washington,DC: Author.

Andrew, M., & Cronin, C. (1997). Two measures of sensation seeking as predictors of alcohol use among high schoolmales. Personality and Individual Differences, 22, 393–401.

Barratt, E. S. (1959). Anxiety and impulsiveness related to psychomotor efficiency. Perceptual and Motor Skills, 9, 191–198.

Barratt, E. S. (1965). Factor analysis of some psychometric measures of impulsiveness and anxiety. PsychologicalReports, 16, 547–554.

Barratt, E. S. (1972). Anxiety and impulsiveness: toward a neuropsychological model. In Spielberger, C. (Ed.), Currenttrends in theory and research (Vol. 1, pp. 195–222). New York: Academic Press.

Brown, C. S., Kent, T. A., Bryant, S. G., Gevedon, R. M., Campbell, J. L., Felthaus, A. R., Barrett, E. S., & Rose, R.M. (1989). Blood platelet uptake of serotonin in episodic aggression. Psychiatry Research, 27, 5–12.

Caspi, A., Moffitt, T., Thornton, A., Freedman, D., Amell, J., Harrington, H., Smeijers, J., & Silva, P. (1996). The life

history calendar: a research and clinical assessment method for collecting retrospective event-history data. Inter-national Journal of Methods in Psychiatric Research, 6, 101–114.

Caspi, A., & Silva, P. A. (1995). Temperamental qualities at age 3 predict personality traits in young adulthood:

longitudinal evidence from a birth cohort. Child Development, 66, 486–498.Cloninger, C. R., Przybeck, T. R., & Svrakic, D. M. (1991). The Tridimensional Personality Questionnaire: US nor-mative data. Psychological Reports, 69, 1047–1057.

Cloninger, C. R., Svrakic, D. M., & Przybeck, T. R. (1993). A psychobiological model of temperament and character.Archives of General Psychiatry, 10, 975–990.


Costa, P. T., & McCrae, R. R. (1992). Revised NEO personality inventory manual. Odessa, FL: Psychological Assess-

ment Resources.Costa, P. T., McCrae, R. R., & Dye, D. A. (1991). Facet scales for agreeableness and conscientiousness: a revision ofthe NEO personality inventory. Personality and Individual Differences, 12, 887–898.

Davidson, R. J., Putnam, K. M., & Larson, C. L. (2000). Dysfunction in the neural circuitry of emotion regulation—apossible prelude to violence. Science, 289, 591–594.

Deakin, J. F. (1991). Depression and 5HT. International Clinical Psychopharmacology, 6, 23–31.

Derogatis, L. R. (1993). Brief symptom inventory: administration, scoring, and procedures manual. Minneapolis:National Computer Systems.

Dickman, S. J. (1990). Functional and dysfunctional impulsivity: personality and cognitive correlates. Journal of Per-sonality and Social Psychology, 58, 95–102.

Dodge, K. A., & Coie, J. D. (1987). Social information-processing factors in reactive and proactive aggression inchildren’s playgroups. Journal of Personality and Social Psychology, 53, 1146–1158.

Donohew, R. L., Hoyle, R. H., Clayton, R. R., Skinner, W. F., Colon, S. E., & Rice, R. E. (1999). Sensation seeking and

drug use by adolescents and their friends: models for marijuana a and alcohol. Journal of Studies on Alcohol, 60, 622–631.Earleywine, M., & Finn, P. R. (1991). Sensation seeking explains the relation between behavioral disinhibition andalcohol consumption. Addictive Behaviors, 16, 123–128.

Eysenck, H. J. (1985). Personality and individual differences: a natural science approach. New York: Plenum Press.Eysenck, S. B. G., Pearson, P. R., Easting, G., & Allsopp, J. F. (1985). Age norms for Impulsiveness, venturesomeness,and empathy in adults. Personality and Individual Differences, 6, 613–619.

Levenson, M., Kiehl, K., & Fitzpatrick, C. (1995). Assessing psychopathic attributes in a noninstitutionalized popu-lation. Journal of Personality and Social Psychology, 68, 151–158.

Linehan, M. (1993). Cognitive-behavioral treatment of borderline personality disorder. New York: Guilford Press.Lynam, D. R. (2002). Psychopathy from the perspective of the Five Factor Model. In P. T. Costa & T. A. Widiger

(Eds.), Personality disorders and the five-factor model of personality (2nd ed., pp. 325–350). Washington, DC: Amer-ican Psychological Association.

Lynam, D. R., Whiteside, S., & Jones, S. (1999). Self-reported psychopathy: a validation study. Journal of Personality

Assessment, 73, 110–132.McCrae, R. R. (1990). Personality in adulthood. New York: Guilford.Miller, J. D., & Lynam, D. (2001). Structural models of personality and their relation to antisocial behavior: A meta-

analytic review. Criminology, 39, 765–792.Miller, J., Lynam, D., Widiger, T., Leukefeld, C., & Clayton, R. (2001). Personality disorders as an extreme variant ofcommon personality dimensions: can the Five Factor Model represent psychopathy. Journal of Personality, 69, 253–276.

Roberts, B., & DelVecchio, W. (2000). The rank-order consistency of personality traits from childhood to old age: aquantitative review of longitudinal studies. Psychological Bulletin, 126, 3–25.

Robins, L., Cottler, L., Bucholz, K., & Compton, W. (1997). Diagnostic Interview Schedule for the DSM-IV (DIS-IV).St. Louis: Washington University School of Medicine.

Robins, L., Helzer, J., Ratcliff, K., & Seyfried, W. (1982). Validity of the Diagnostic Interview Schedule, Version II:DSM-III diagnoses. Psychological Medicine, 12, 855–870.

Straus, M., & Gelles, R. (1990). Physical violence in American families: risk factors and adaptation to violence in 8145

families. New Brunswick, NJ: Transaction Publications.Wechsler, D. (1981).Wechsler Adult Intelligence Scale-Revised (WAIS-R). San Antonio: The Psychological Cooperation.Whiteside, S. P., & Lynam, D. R. (2001). The Five Factor Model and impulsivity: using a structural model of per-

sonality to understand impulsivity. Personality and Individual Differences, 30, 669–689.Wiederman, M. W., & Pryor, T. (1996). Multi-impulsivity among women with bulimia nervosa. International Journalof Eating Disorders, 20, 359–365.

Zuckerman, M. (1994). Behavioral expressions and biosocial bases of sensation seeking. New York: Cambridge Uni-

versity Press.Zuckerman, M., Kuhlman, D. M., Joireman, J., Teta, P., & Kraft, M. (1993). A comparison of three structural modelsof personality: the big three, the big five, and the alternative five. Journal of Personality and Social Psychology, 65,

757–768.


A test of the four-factor model of impulsivity-related traits

Documents

Transcript of A test of the four-factor model of impulsivity-related traits