After online publication, subscribers (personal/institutional) to this journal will haveaccess to the complete article via the DOI using the URL:

If you would like to know when your article has been published online, take advantageof our free alert service. For registration and further information, go to:http://www.springerlink.com.

Due to the electronic nature of the procedure, the manuscript and the original figureswill only be returned to you on special request. When you return your corrections,please inform us, if you would like to have these documents returned.

Dear Author

Here are the proofs of your article.

• You can submit your corrections online, via e-mail or by fax.

• For online submission please insert your corrections in the online correction form.

Always indicate the line number to which the correction refers.

• You can also insert your corrections in the proof PDF and email the annotated PDF.

• For fax submission, please ensure that your corrections are clearly legible. Use a fine

black pen and write the correction in the margin, not too close to the edge of the page.

• Remember to note the journal title, article number, and your name when sending your

response via e-mail or fax.

• Check the metadata sheet to make sure that the header information, especially author

names and the corresponding affiliations are correctly shown.

• Check the questions that may have arisen during copy editing and insert your

answers/corrections.

• Check that the text is complete and that all figures, tables and their legends are included.

Also check the accuracy of special characters, equations, and electronic supplementary

material if applicable. If necessary refer to the Edited manuscript.

• The publication of inaccurate data such as dosages and units can have serious

consequences. Please take particular care that all such details are correct.

• Please do not make changes that involve only matters of style. We have generally

introduced forms that follow the journal’s style.

• Substantial changes in content, e.g., new results, corrected values, title and authorship are

not allowed without the approval of the responsible editor. In such a case, please contact

the Editorial Office and return his/her consent together with the proof.

• If we do not receive your corrections within 48 hours, we will send you a reminder.

• Your article will be published Online First approximately one week after receipt of your

corrected proofs. This is the official first publication citable with the DOI. Further

changes are, therefore, not possible.

• The printed version will follow in a forthcoming issue.

Please note

http://dx.doi.org/10.1007/s10862-013-9364-3

AUTHOR'S PROOF

Metadata of the article that will be visualized in OnlineFirst

1 Article Title Age-Neutrality of the Trait Facets Proposed for Personality

Disorders in DSM-5: A DIFAS Analysis of the PID-5

2 Article Sub- Title

3 Article Copyright -Year

Springer Science+Business Media New York 2013(This will be the copyright line in the final PDF)

4 Journal Name Journal of Psy chopathology and Behav ioral Assessment

5

Corresponding

Author

Family Name Broeck

6 Particle Van den

7 Given Name Joke

8 Suffix

9 Organization Vrije Universiteit Brussel

10 Division Faculty of Psychology and Educational Sciences,Department of Clinical and Lifespan Psychology

11 Address Pleinlaan 2, Brussels 1050, Belgium

12 Organization Vrije Universiteit Brussel

13 Division Department of Clinical and Lifespan Psychology

14 Address Brussels, Belgium

15 e-mail Joke.Van.Den.Broeck@vub.ac.be

16

Author

Family Name Bastiaansen

17 Particle

18 Given Name Leen

19 Suffix

20 Organization Vrije Universiteit Brussel

21 Division Department of Clinical and Lifespan Psychology

22 Address Brussels, Belgium

23 e-mail

24

Author

Family Name Rossi

25 Particle

26 Given Name Gina

27 Suffix

28 Organization Vrije Universiteit Brussel

29 Division Department of Clinical and Lifespan Psychology

30 Address Brussels, Belgium

AUTHOR'S PROOF

31 e-mail

32

Author

Family Name Dierckx

33 Particle

34 Given Name Ev a

35 Suffix

36 Organization Vrije Universiteit Brussel

37 Division Department of Clinical and Lifespan Psychology

38 Address Brussels, Belgium

39 e-mail

40

Author

Family Name Clercq

41 Particle De

42 Given Name Barbara

43 Suffix

44 Organization Personality and Social Psychology, GhentUniversity

45 Division Department of Developmental

46 Address Ghent, Belgium

47 e-mail

48

Schedule

Received

49 Revised

50 Accepted

51 Abstract An age-neutral measurement sy stem is one of the basic conditions f or anaccurate personality assessment across the lif espan, both longitudinallyand cross-sectionally . In this study the age-neutrality of the PersonalityInventory for DSM-5 PID-5; Krueger et al. (Psy chological Medicine 42,1879–1890, 2012) was inv estigated. Potential Dif f erential Item Functioning(DIF) was examined f or the 25 trait f acets in older v ersus y ounger adults.Ov erall, 33 items display ed large DIF, according to the adjusted Bonf erronicorrected cutof f s (Mantel Chi-square, Liu-Agresti Cumulativ e CommonLog-Odds Ratio [L-A LOR], and Cox’s Noncentrality Parameter Estimator[Cox’s B]). In a next step, the implications of the item lev el DIF across agegroups was inv estigated on scale (i.e., f acet) lev el. These Dif f erential TestFunctioning (DTF) analy ses rev ealed large DTF f or f our of the 25 PID-5f acets (i.e., Withdrawal, Attention Seeking, Rigid Perf ectionism and UnusualBelief s). Current initial results show that most PID-5 traits are measuredequally well across age, howev er, f urther research is needed to f urtherref ine this instrument and make it entirely age-neutral.

52 Keywordsseparated by ' - '

Assessment - Personality disorders - Age-neutrality - Dif f erential itemf unctioning - DSM-5 - PID-5

53 Foot noteinformation

AUTHOR'S PROOF

UNCORRECTEDPROOF

1

23

4 Age-Neutrality of the Trait Facets Proposed for Personality5 Disorders in DSM-5: A DIFAS Analysis of the PID-5

7 Joke Van den Broeck & Leen Bastiaansen & Gina Rossi &8 Eva Dierckx & Barbara De Clercq9

10

1112 # Springer Science+Business Media New York 2013

13 Abstract An age-neutral measurement system is one of the14 basic conditions for an accurate personality assessment across15 the lifespan, both longitudinally and cross-sectionally. In this16 study the age-neutrality of the Personality Inventory for DSM-17 5 PID-5; Krueger et al. (Psychological Medicine 42, 1879–18 1890, 2012) was investigated. Potential Differential Item19 Functioning (DIF) was examined for the 25 trait facets in20 older versus younger adults. Overall, 33 items displayed large21 DIF, according to the adjusted Bonferroni corrected cutoffs22 (Mantel Chi-square, Liu-Agresti Cumulative Common Log-23 Odds Ratio [L-A LOR], and Cox’s Noncentrality Parameter24 Estimator [Cox’s B]). In a next step, the implications of the25 item level DIF across age groups was investigated on scale26 (i.e., facet) level. These Differential Test Functioning (DTF)27 analyses revealed large DTF for four of the 25 PID-5 facets28 (i.e., Withdrawal, Attention Seeking, Rigid Perfectionism and29 Unusual Beliefs). Current initial results show that most PID-530 traits are measured equally well across age, however, further31 research is needed to further refine this instrument and make it32 entirely age-neutral.

33 Keywords Assessment .Personalitydisorders .Age-neutrality .

34 Differential item functioning . DSM-5 . PID-5

35Introduction

36The Diagnostic and Statistical Manual of Mental Disorders,37Fifth Edition (DSM-5; American Psychiatric Association38[APA] 2013) retained the categorical DSM-IV (APA 1994)39approach to personality disorders, describing ten different40personality disorder categories. Many prominent researchers41in the field advocated a dimensional approach (e. g., Widiger42and Trull 2007) and during the development process of43DSM-5, the DSM-5 Personality and Personality Disorders44Work Group (http://www.dsm5.org/MeetUs/Pages/Personality45Disorders.aspx) proposed a hybrid dimensional-categorical46model for personality and personality disorder assessment47and diagnosis in which six specific personality disorder types48are defined by two fundamental criteria, being impairments in49personality functioning and the presence of pathological per-50sonality traits. Regarding the latter, a multidimensional mal-51adaptive personality trait system has been developed, in order52to represent individual differences in personality disorder53expression (Krueger et al. 2012; Wright et al. 2012). In this54model, 25 primary traits are organized by five higher-order55dimensions: Negative Affect, Detachment, Antagonism, Dis-56inhibition, and Psychoticism. The Personality Inventory for57DSM-5 (PID-5; Krueger et al. 2012) was developed to58operationalize these DSM-5 traits. The above mentioned al-59ternative dimensional proposal is included in a separate chap-60ter of DSM-5 (APA 2013), “Results”, and aims to encourage61further research on how this new methodology can be used to62assess and diagnose personality disorders. Unfortunately, the63Work Group’s dimensional focus detracted attention from64another important issue, namely the suitability of the criteria65for measuring personality in later life (Tackett et al. 2009;66Oltmanns and Balsis 2011). This is regrettable, since “even if67a dimensional shift is made, there will be continued psycho-68metric and conceptual problems if the criteria do not closely69consider the presentation of personality in later life” (Tackett70et al. 2009, p. 14). It is commonly assumed that the current

J. Van den Broeck : L. Bastiaansen :G. Rossi : E. DierckxDepartment of Clinical and Lifespan Psychology,Vrije Universiteit Brussel, Brussels, Belgium

B. De ClercqDepartment of Developmental, Personality and Social Psychology,Ghent University, Ghent, Belgium

J. Van den Broeck (*)Faculty of Psychology and Educational Sciences,Department of Clinical and Lifespan Psychology,Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgiume-mail: Joke.Van.Den.Broeck@vub.ac.be

J Psychopathol Behav AssessDOI 10.1007/s10862-013-9364-3

JrnlID 10862_ArtID 9364_Proof# 1 - 22/06/2013

AUTHOR'S PROOF

UNCORRECTEDPROOF

71 DSM-IV criteria for personality disorders are not adequately72 attuned to the living situations and experiences of older people73 (Agronin and Maletta 2000; Segal et al. 2000). Based on item74 response theory analyses in a large, cross-sectional study of75 37.000 participants, Balsis et al. (2007) concluded that 29 %76 of the DSM-IVAxis II criteria lack face validity in older age77 groups, hence possibly leading to over- or underdiagnosis of78 personality pathology in old age populations. Unfortunately,79 the presentation of later life was not explicitly considered in80 the construction of the PID-5 either. Yet an age-neutral mea-81 surement system is one of the basic conditions for an accurate82 personality assessment across the lifespan, both longitudinally83 and cross-sectionally.84 Since the PID-5’s age-neutrality has, to our knowledge,85 not been empirically investigated thus far, we set out to86 detect possible Differential Item Functioning (DIF) in older87 versus younger adults. An item is said to exhibit DIF when88 younger and older adults with a similar position on the89 underlying trait of interest do not have the same probability90 of endorsing that item (Edwards and Edelen 2009). If DIF91 occurs, the assumption of measurement invariance is violat-92 ed, leading to possibly flawed interpretations of observed93 between-group differences (Millsap 2011). DIF analyses can94 be done in both a Classical Test Theory (CTT) and an Item95 Response Theory (IRT) framework. The current analyses96 were conducted using an odds ratio approach (CTT), which97 is, in contrast to the IRT approach, not hampered by require-98 ments of model fit and large sample sizes, and can be99 conducted using the easily accessible DIFAS program100 (Penfield 2005). In addition, it has been shown that the101 cumulative common log-odds ratio is very similar to the102 IRT based approaches (Penfield 2010).

103 Method

104 Participants and Procedure

105 A total of 464 participants were included in the current study,106 subdivided into a younger and an older sample. The younger107 sample consisted of 288 undergraduate psychology students108 with ages ranging from 17 to 40 (M=21.05, SD=3.70, 27 %109 male). Participants in the older sample were 176 Dutch-110 speaking community-dwelling adults recruited by under-111 graduate psychology students. Students were requested to112 recruit at least one person aged 60 or older. No other spec-113 ifications or conditions were provided. As a return for par-114 ticipation, the students received course credits. Participants’115 age in the older sample ranged between 61 and 99 years116 (M=72.73; SD=6.09), with 40 % male participants. All117 participants volunteered their participation and provided a118 written informed consent.

119Measure

120The Dutch authorized translation of the Personality Invento-121ry for DSM-5 (PID-5; Krueger et al. 2012; De Clercq et al.1222011), a 220-item self-report questionnaire, was used to123measure the proposed DSM-5 traits. Items are rated on a 4-124point Likert format scale, ranging from very false or often125false to often true or very true. The PID-5 has 25 primary126lower-order scales or facets that load onto five higher-order127personality pathology dimensions (Negative Affectivity, De-128tachment, Antagonism, Disinhibition, and Psychoticism).129Lower-order scale internal consistencies (Cronbach’s alpha)130ranged from 0.68 (Suspiciousness) to 0.95 (Eccentricity) in131the current younger sample, and from 0.25 (Suspiciousness)132to 0.91 (Eccentricity) in the older sample (Mdn=.82 in both133samples). These values are largely in line with data reported134in previous research (e.g., Wright et al. 2012; Hopwood et al.1352012), however the internal consistency of Suspiciousness is136remarkably low in the current older sample (see Table 1).137Closer examination of the low alpha value for Suspicious-138ness (i.e., 0.25) revealed that this is mainly due to the two139reversed scored items in this scale (i.e., i131 ‘people are140basically trustworthy’, and i177 ‘I rarely feel that people I141know are trying to take advantage of me’). After deleting142these two items from the scale, the alpha value increased143to 0.54.

144Statistical Analyses

145Between-group differences between the younger and older146age group on the PID-5 facets were computed with t-tests for147independent samples. Cohen’s d was used as a measure of148effect size (Cohen 1988), with d≥ .20 indicating a small149effect, d≥ .50 a medium effect, and d≥ .80 a large effect).150Next, it was investigated whether there are PID-5 items151that measure the personality trait of interest differently in152younger versus older adults, after controlling for the overall153level of underlying trait between both age groups. To detect154possible DIF as a function of age, the Mantel Chi-square155(Mazor et al. 1992), the Liu-Agresti Cumulative Common156Log-Odds Ratio (L-A LOR; Liu and Agresti 1996), and the157Cox’s Noncentrality Parameter Estimator (Cox’s B; Camilli158and Congdon 1999) were used for polytomous items, using159the DIFAS 5.0 software program (Penfield 2007). According160to these three approaches, individuals are first divided across161different “bands” or “strata”, based on their total scale scores162(measuring the personality trait of interest). All individuals163within the same stratum are considered equated, so that any164difference between individuals in the younger and older age165group are considered indicative of DIF (Baer et al. 2011).166The Mantel chi-square statistic is based on a group (2) x167response option (4) contingency table, distributed as a chi-168square with 1° of freedom. The higher the chi-square value,

J Psychopathol Behav Assess

JrnlID 10862_ArtID 9364_Proof# 1 - 22/06/2013

AUTHOR'S PROOF

UNCORRECTEDPROOF

169 the higher the probability the item displays DIF. In line, the170 L-A LOR considers the log odds ratio of one group endors-171 ing a response option relative to another. The Cox’s B sta-172 tistic is similar to the Mantel-Haenszel statistic but uses the173 hypergeometric mean. For both the L-A LOR and Cox’s B174 statistic negative values indicate DIF against the reference175 group (younger adults), and positive values indicate DIF176 against the focal group (older adults). For example, an item177 revealing a negative L-A LOR and Cox’s B value is more178 readily endorsed by older adults, despite the fact that both179 groups were matched on the underlying personality trait of180 interest (e.g., Anxiousness). The following cut-off criteria181 are available to flag items with large DIF: |L-A LOR| > .64182 (Penfield 2007), and |Cox’s B| > .40 (Camilli and Congdon183 1999). Critical values for the Mantel Chi-square statistic are184 3.84 for a Type I error rate of 0.05, and 6.63 for a Type I error185 rate of 0.01 (Penfield 2007). However, to reduce the Type I186 error of incorrectly flag items as problematic, we used a187 stringent Bonferroni corrected critical chi-square value,188 ranging from 8.28 to 10.83 (depending on the number of

189items per scale; for example for Anhedonia the Bonferroni190correction was applied across all 8 items by three test statis-191tics [0.05/8*3], leading to a critical chi-square value of 9.55).192Similar stringent Bonferroni cut-offs were applied to the193L-A-LOR (ranging from 0.85 to 0.99), and Cox’s B (ranging194from 0.53 to 0.62) statistics to flag items with large DIF (also195depending on the number of items per scale)1.196The impact of DIF at scale level was examined by Differ-197ential Test Functioning (DTF) analyses. DTF concerns the198additive effect of item DIF across all items of a scale.199Penfield and Algina (2006) argued that a collective large200level of DIF in a group of items exist if 25 % or more of201the items show moderate to large DIF. They propose to202define DIF effect variance (indicated by the weighted v2

203statistic) as small for v2<.07, medium for .07 ≤ v2 ≤ .14,204and large for v2>.14. In line with the Bonferroni corrected205DIF thresholds, similar adjustments were made for the DTF

t1:1 Table 1 Descriptive statistics for the PID-5 primary traits for the younger (n=288) and older (n=176) sample

t1:2 Facet Cronbach alpha Raw means (SD) Effect size

t1:3 Young Old Young Old Cohen’s d

t1:4 Anhedonia .83 .74 4.97 (3.60) 5.73 (3.88) −.20*

t1:5 Anxiousness .88 .85 11.05 (5.59) 8.44 (5.61) .47***

t1:6 Depressivity .89 .88 7.25 (6.01) 6.18 (6.45) ns

t1:7 Emotional Lability .88 .85 9.24 (4.94) 7.17 (4.86) .42***

t1:8 Hostility .81 .78 10.86 (4.88) 7.76 (5.13) .62***

t1:9 Perseveration .78 .74 8.56 (4.22) 7.32 (4.49) .28**

t1:10 Rigid Perfectionism .87 .85 9.93 (5.77) 11.02 (6.10) ns

t1:11 Separation Insecurity .74 .72 9.02 (3.87) 8.23 (4.19) .20*

t1:12 Submissiveness .76 .73 4.18 (2.24) 3.94 (2.76) ns

t1:13 Suspiciousness .68 .24 7.02 (3.15) 7.39 (2.64) ns

t1:14 Withdrawal .87 .87 4.66 (4.52) 6.34 (5.67) −.33***

t1:15 Attention Seeking .86 .86 8.13 (4.39) 5.34 (4.83) .60***

t1:16 Callousness .81 .77 5.58 (4.34) 5.60 (5.24) ns

t1:17 Deceitfulness .84 .84 7.20 (4.61) 4.74 (4.97) .51***

t1:18 Grandiosity .72 .83 2.83 (2.43) 2.57 (3.26) ns

t1:19 Manipulativeness .83 .82 5.07 (3.13) 2.91 (3.12) .69***

t1:20 Intimacy Avoidance .81 .68 2.15 (2.74) 5.79 (3.67) −1.12***

t1:21 Restricted Affectivity .82 .70 5.69 (3.97) 6.01 (3.62) ns

t1:22 Distractibility .89 .84 10.33 (5.42) 7.10 (5.29) .60***

t1:23 Eccentricity .95 .91 9.05 (8.01) 6.33 (7.00) .36***

t1:24 Perceptual Disregulation .79 .86 5.13 (4.51) 4.97 (5.53) ns

t1:25 Risk Taking .89 .74 18.64 (6.87) 13.27 (5.83) .84***

t1:26 Unusual Beliefs .80 .81 2.88 (3.42) 3.57 (4.03) ns

t1:27 Impulsivity .80 .72 6.57 (3.26) 5.06 (3.44) .45***

t1:28 Irresponsibility .71 .71 4.91 (3.05) 3.08 (3.13) .59***

* p≤ .05, ** p≤ .01, *** p≤ .001

1 The exact cutoff values for each of the three DIF indicators for eachanalysis can be obtained on requested from the first author.

J Psychopathol Behav Assess

JrnlID 10862_ArtID 9364_Proof# 1 - 22/06/2013

AUTHOR'S PROOF

UNCORRECTEDPROOF

206 thresholds2. Both DIF and DTF analyses were conducted at207 facet level, so the total facet scores were used as stratifying208 variables. The stratum size was set at 1, which is the default209 option in DIFAS 5.0.

210 Results

211 Descriptive Statistics

212 Overall, for 16 of the 25 facets, the mean scale scores of the213 older sample differed significantly (p≤ .05) from the mean214 scale scores of the younger sample (see Table 1). Large215 significant differences (d>.80) were found for Intimacy216 Avoidance and Risk Taking, and moderate significant differ-217 ences (d>.50) for Hostility, Attention Seeking, Deceitful-218 ness, Manipulativeness, Distractibility, and Irresponsibility.219 For all but the Intimacy Avoidance scale, mean scale scores220 for these facets were significantly higher for younger com-221 pared to older adults (p<.001).

222 Differential Item and Test Functioning

223 Twenty-five separate DIF and DTF analyses were conducted224 for each of the primary traits of the PID-5, whereby the225 respective total facet-score served as stratifying variable.226 Overall, DIF analyses revealed 30 items showing significant227 DIF, divided across 15 facets. Table 2 shows the paraphrased228 item content and statistical details of these items, and the229 facets they belong to. In order to evaluate the impact of these230 DIF items at scale (i.e., facet) level, additional DTF analyses231 were conducted. According to the Bonferroni corrected cut-232 off (> 0.35)2, DTF analyses revealed large DTF forWithdraw-233 al, Attention Seeking, Rigid Perfectionism, and Unusual Be-234 liefs (with weighted v2 values of 0.64, 0.65, 0.37, and 0.44,235 respectively). Withdrawal contained five items displaying236 significant DIF. Two of them displayed DIF against the older237 age group (“I keep to myself” and “I keep my distance from238 people”), indicating they were more readily endorsed by239 younger adults with the same level of underlying personality

240trait (i.e., Withdrawal). Three items showedDIF against youn-241ger adults (“I don’t like spending time with others”, “I’m not242interested in making friends”, and “I say as little as possible243when dealing with people”). Attention Seeking contained four244items displaying significant DIF, of which two displayed DIF245against younger (“I do things so that people just have to246admire me”, and “I crave attention”), and two against older247adults (“I love getting attention”, and “I like standing out in a248crowd”). Rigid Perfectionism contained one item displaying249DIF against older (“I simply won’t put up with things being250out of their proper places”), and one item displaying DIF251against younger adults (“I focus too much on minor details”).252Similarly, Unusual Beliefs contained one item displaying DIF253against older (“I believe that some people can move things254with their minds”), and two displaying DIF against younger255adults (“Other people seem to think my behavior is weird”,256and “I see unusual connections between things”), indicating257that the latter two are more readily endorsed by younger258adults, matched on underlying personality trait.

259Discussion

260The primary aim of this study was to investigate the age-261neutrality of the PID-5 facets by examining potential DIF for262older versus younger adults. According to the stringent263Bonferroni corrected cutoffs, analyses revealed a total of 33264items displaying significant DIF, divided across 15 facets.265The impact hereof at scale level was relatively small. Large266DTF was confirmed for four facets, namely Withdrawal,267Attention Seeking, Rigid Perfectionism and Unusual Beliefs.

268Withdrawal Three items showed negative DIF, indicating269they were more readily endorsed by older adults with similar270levels of the latent personality trait. These items focus on (the271absence of) close relationships (e.g., ‘I’m not interested in272making friends’). In this respect, it is possible that endorsing273these items does not reflect personality pathology, but rather274dealing with the death of loved ones, or, for example, isola-275tion caused by physical illness (Van Alphen et al. 2006).276Also, as people age, they tend to engage in selective social277interaction, maintaining only the most rewarding contacts to278satisfy their emotional needs (Carstensen 1991). Notably,279two other items from theWithdrawal facet were more readily280endorsed by younger adults (“I keep to myself” and “I keep281my distance from people”). Although at first sight substantial-282ly very similar to the items displaying negative DIF, these283findings might reveal a difference in the interpretation of these284items that varies with age. It is not unlikely that older adults285are less inclined to endorse these items that probe social286isolation and withdrawal, because the diminishing of a social287network is more common in later life, due to loss experiences288or physical deterioration. Both younger and older participants

2 In deriving their thresholds Penfield and Algina (2006) argued that acollective large level of DIF in a group of items exist if 25 % or more ofthe items are categorized as having moderate or large magnitudes ofDIF based on the ETS classification scheme (i.e. if 25 % or more of theitems have an absolute value of log(αMH) greater than or equal to .43.They also suggest that MH and L-A LOR have similar meanings interms of DIF magnitude. Because we wanted to reduce the Type I error,a Bonferroni correction was applied and adjusted L-A LOR cut-offvalues were used to flag items with large DIF (depending on the numberof items per scale). In line, we made a similar adjustment for the DTFthresholds. For example: Penfield and Algina consider the variance ofDIF effect large when weighted v2>.14, using an L-A LOR value of .43as critical value. Since we adhere to stringent L-A LOR critical values(for example>.92 for Anhedonia) we adjusted this to v2>.35 (i.e.,.14/.43*.92).

J Psychopathol Behav Assess

JrnlID 10862_ArtID 9364_Proof# 1 - 22/06/2013

AUTHOR'S PROOF

UNCORRECTEDPROOF

289 probably compared themselves to peers when filling out the290 PID-5 questionnaire, thus an item as “I keep to myself” may291 lead to less extremely high scores in an older population,292 because it is not so much seen as “deviant behavior”, but293 rather as a normative phenomenon related to aging.

294 Attention Seeking The item “I love getting attention” was295 more readily endorsed by younger adults, whereas the item296 “I crave attention” was more readily endorsed by older297 adults. Although the contents of these items show consider-298 able overlap, it seems as though there really is a difference299 between “love attention” and “crave attention”. Craving300 attention might be a more desirable statement for those301 who are lonely, and, supposing elderly people might in

302general be more lonely (e.g., Holmén and Furukawa 2002),303this might explain the differences found for this item.

304Rigid Perfectionism Younger adults scored generally higher305on the item “I focus too much on minor details”. This item is306possibly more related to an occupational context, and there-307fore less relevant (and less readily endorsed) by retirees. The308other DIF item in this facet (“I simply won’t put up with309things being out of their proper places”) was more readily310endorsed by older adults, suggesting that older adults might311be in general more orderly and/or rigid, and less resistant312against changes in their personal habitat than younger adults.313Another explanation might lie in the cognitive decline and314memory problems associated with advancing age. Older

t2:1 Table 2 Items meeting Bonferroni adjusted criteria for large DIF

t2:2 Facet Item Paraphrased context Mantel χ2 L-A LOR Cox’s B

t2:3 Items displaying DIF against younger adults (older > younger)

t2:4 Anxiousness 110. I worry about almost everything. 9.72 −.69 −.61

t2:5 Anxiousness 174. I’m fearful about bad things that might happen. 10.98 −.71 −.58

t2:6 Emotional lability 165. I get emotional over every little thing. 18.54 −1.01 −.79

t2:7 Restricted affectivity 167. I never show emotions to others. 11.9 −.72 −.60

t2:8 Withdrawal 136. I don’t like spending time with others. 12.58 −.98 −.78

t2:9 Withdrawal 146. I’m not interested in making friends. 17.42 −1.16 −.66

t2:10 Withdrawal 147. I say as little as possible when dealing with people. 27.64 −1.20 −.95

t2:11 Attention seeking 43. I do things so that people just have to admire me. 24.61 −1.24 −.84

t2:12 Attention seeking 191. I crave attention. 35.56 −1.36 −1.03

t2:13 Callousness 207. I don’t see the point in feeling guilty about things. 17.29 −1.23 −.94

t2:14 Irresponsibility 201. I skip appointments if I’m not in the mood. 10.63 −.84 −.62

t2:15 Rigid perfectionism 196. I simply won’t put up with things being out of their proper places. 37.40 −1.35 −.85

t2:16 Risk taking 195. I don’t think about getting hurt when I’m doing things that might be dangerous. 16.36 −.98 −.66

t2:17 Unusual beliefs 143. People can move things with their minds. 12.60 −.99 −.68

t2:18 Items displaying DIF against older adults (younger > older)

t2:19 Hostility 28. I snap at people when they irritate me. 12.58 .78 .67

t2:20 Hostility 32. I can be mean when I need to be. 32.11 1.20 .77

t2:21 Intimacy avoidance 89. I keep romance out of my life. 11.46 1.02 .71

t2:22 Withdrawal 20. I keep to myself. 14.40 .87 .63

t2:23 Withdrawal 82. I keep my distance from people. 39.40 1.61 1.13

t2:24 Attention seeking 74. I love getting attention. 12.74 .79 .60

t2:25 Attention seeking 111. I like standing out in a crowd. 18.93 .97 .76

t2:26 Emotional lability 18. My emotions change for no good reason. 24.86 1.04 .74

t2:27 Callousness 200. I enjoy making people in control look stupid. 12.12 .86 .64

t2:28 Deceitfulness 134. I don’t hesitate to cheat if it gets me ahead. 10.37 .79 .64

t2:29 Deceitfulness 214. Lying comes easily to me. 10.60 .77 .61

t2:30 Manipulativeness 125. Sweet-talking others helps me get what I want. 8.91 .68 .62

t2:31 Distractibility 132. I am easily distracted. 15.18 .92 .78

t2:32 Rigid perfectionism 49. I focus too much on minor details. 61.62 1.74 1.03

t2:33 Eccentricity 24. Others think my behavior is weird. 15.06 .93 .63

t2:34 Unusual beliefs 194. I see unusual connections between things. 14.31 .96 .67

L-A LOR Liu-Agresti common log odds ratio, Cox’s B Cox’s Noncentrality parameter estimator. R Reversed scored items. Facets with large DTF aregiven in bold

J Psychopathol Behav Assess

JrnlID 10862_ArtID 9364_Proof# 1 - 22/06/2013

AUTHOR'S PROOF

UNCORRECTEDPROOF

315 adults are possibly more prone to compensate with order and316 regularity, as a way to cope with their forgetfulness.

317 Unusual Beliefs Two items displayed DIF, of which one was318 more readily endorsed by younger adults (“I see unusual con-319 nections between things”), and one by older adults (“People can320 move things with their minds”). The item “I see unusual321 connections between things” possibly reflects the tendency of322 young people to challenge traditional norms and their striving323 to be seen as unique individuals whowant to gather information324 and expand their horizons, whereas older adults are less325 knowledge-oriented (Reed and Carstensen 2012).326 Overall, the current initial results validate the comparison of327 mean facet scores across younger and older age groups for 21328 of the 25 PID-5 traits. Given the lack of measurement invari-329 ance for Withdrawal, Attention Seeking, Rigid Perfectionism,330 and Unusual Beliefs, caution is warranted in the interpretation331 of age differences based on these particular traits. Concerning332 the current between-group differences on mean facet scores,333 largest mean differences (as evidenced by moderate and large334 effect sizes) were found for Hostility, Attention Seeking, De-335 ceitfulness, and Manipulativeness, all four belonging to the336 higher-order trait domain of Antagonism. Younger adults had337 significantly higher mean level scores for these traits. Accord-338 ingly, younger adults scored also significantly higher on Risk339 Taking, Distractibility, and Irresponsibility, three facets of the340 Disinhibition domain. When subjected to a hierarchical struc-341 ture analysis, it is demonstrated that Antagonism and Disinhi-342 bition are both split-offs of a higher-level Externalizing factor343 (Wright et al. 2012). These findings support the notion that344 externalizing personality traits tend to remit with age. Due to345 physical changes associated with aging and consequently a346 reducedmobility and slower pace, elderly people are less likely347 to act impulsively or manifest risky, irresponsible behavior348 (Roberts et al. 2006). Furthermore, the PID-5 five-factor struc-349 ture shows clear resemblance to other five factor models like350 the Personality Psychopathology Five (PSY-5) domains, in-351 cluding Negative Emotionality/Neuroticism, Introversion/Low352 Positive Emotionality, Aggressiveness, Disconstraint, and353 Psychoticism (Anderson et al. 2013; Harkness and McNulty354 1994), or the structure of normal personality as represented by355 the FFM, whereby Antagonism is the pathological variant of356 (low) Agreeableness, and Disinhibition the pathological vari-357 ant of (low) Conscientiousness (Thomas et al. 2012). Parallels358 can thus be drawn between current findings and established359 research on age-related mean-level changes in general person-360 ality trait scores, describing an increase of Agreeableness and361 Conscientiousness traits throughout the lifespan (e.g.,362 Terracciano et al. 2005; Roberts et al. 2006). Older adults, in363 turn, scored significantly higher on Intimacy Avoidance, a trait364 facet belonging to the higher-order domain Detachment, the365 pathological variant of (low) Extraversion. Again, these results366 corroborate the decline in Extraversion with advancing age.

367Limitations

368A few limitations should be considered. First of all, the cross-369sectional design of the current study makes it difficult to370distinguish real age effects from cohort effects, hampering371the interpretation of the DIF results. It was not our intention372however, to provide clear explanations for the current DIF373results, but only to detect for possible DIF as a function of age374in the recently proposed maladaptive personality traits for375DSM-5. In line, the possibility cannot be ruled out that some376of these initial findings are due to other factors beyond age that377define differences between the current subsamples (e.g., gen-378der, or socio-economic status). Future studies should therefore379strive for larger, more representative samples and explore the380possible underlying causes of DIF, and their consequences for381the assessment of personality pathology across the lifespan in382demographically matched samples of younger and older sam-383ples. The current sample sizes were too small to obtain stable384parameter estimates using the graded response model or other385polytomous IRT models, yet these more sophisticated re-386search methods may play an important role in further explor-387ing the possible influence of the above mentioned confound-388ing variables [see e.g., Penfield (2010) for a detailed compar-389ison between CTT and IRT approaches for identifying DIF].390Replicating the current findings within clinical samples is391another important avenue for further research, since the PID-3925 was primarily designed to identify personality pathology, a393clinically-relevant phenomenon. A second limitation pertains394to the Suspiciousness scale. Although no significant differ-395ences in mean scale scores were found between the younger396and older age group for this trait, the internal consistency of397this scale was clearly low in the current older sample and398warrants further investigation. Possibly the two reverse scored399items in this scale share variance above and beyond the400general factor of suspiciousness, making them qualitatively401distinct from the other items in the scale, and violating the402assumption of local independence (e.g., Hopwood and403Donnellan 2010). In addition, further research should explore404whether a similar low internal consistency would appear in a405clinical sample. Third, the current analyses and results apply406to the Dutch version of the PID-5, so that further research is407needed to investigate whether the same results would apply to,408for instance, a North American population. Such cross-409cultural research could possibly reveal interesting findings410towards the universality of age-related changes in personality411pathology and the course of personality disorders throughout412the lifespan, aspects that are only just beginning to be under-413stood. A last limitation considered here is the lower bound of414age 60 as inclusion criteria for the older age group. The415heterogeneity of this older age group should not be416underestimated, since considerable differences might exist417between, for example, 60–65 and 80–85 year-olds. With the418current software used to investigate DIF we were limited to

J Psychopathol Behav Assess

JrnlID 10862_ArtID 9364_Proof# 1 - 22/06/2013

AUTHOR'S PROOF

UNCORRECTEDPROOF

419 the comparison of two age groups, but investigating measure-420 ment invariance across different age groups covering the421 whole lifespan and using smaller age-ranges might reveal422 interesting findings about more nuanced age-related changes423 in personality traits.

424 Conclusions

425 Despite not having explicitly considered the later life context426 during its development, current initial results show that most427 PID-5 traits are measured equally well across both a younger428 and an older age group. These results are promising in light429 of the growing awareness that an age-neutral measurement is430 crucial for a valid assessment of personality pathology431 throughout the lifespan. Additional research is certainly432 needed however to further refine this instrument and make433 it entirely age-neutral, since 33 items appeared to display434 large DIF, resulting in four scales exhibiting significant DTF.435 To this end, a set of alternative items could be written that436 works equally well for younger and older adults, regardless437 of their somewhat different living conditions. This set of438 items can then be tested for DIF across important demo-439 graphic groups (e.g., gender, age, ethnic status, etc.), to440 finally reach a scale that contain no measurement artifacts441 (Oltmanns and Balsis 2011). We do realize that creating such442 items is a challenging task, however we hope that the current443 exploratory analyses point out the need and inspire further444 researchers towards developing an age-neutral measurement445 system.

446

447 References448

449 Agronin, M. E., &Maletta, G. (2000). Personality disorders in later life.450 Understanding and overcoming the gap in research. The American451 Journal of Geriatric Psychiatry, 8, 4–18.452 American Psychiatric Association. (1994). Diagnostic and statistical453 manual of mental disorders (4th ed.). Washington, DC: Author.454 American Psychiatric Association. (2013). Diagnostic and statistical455 manual of mental disorders (5th ed.). Washington, DC: Author.456 Anderson, J. L., Sellbom, M., Bagby, M. R., Quilty, L. C., Veltri, C. O.457 C., Markon, K. E., & Krueger, R. F. (2013). On the convergence458 between PSY-5 domains and PID-5 domains and facets: implica-459 tions for assessment of DSM-5 personality traits. Assessment.460 doi:10.1177/1073191112471141.461 Baer, R. A., Samuel, D. B., & Lykins, E. L. B. (2011). Differential item462 functioning on the five facet mindfulness questionnaire is minimal463 in demographically matched mediators and nonmediators. Assess-464 ment, 18(1), 3–10.465 Balsis, S., Gleason, M. E. J., Woods, C. M., & Oltmanns, T. F. (2007).466 An item response theory analysis of DSM-IV personality disorder467 criteria across younger and older age groups. Psychology and468 Aging, 22(1), 171–185.

469Camilli, G., & Congdon, P. (1999). Application of a method of estimat-470ing DIF for polytomous test items. Journal of Educational and471Behavioral Statistics, 24, 323–341.472Carstensen, L. L. (1991). Selectivity theory: social activity in life-span473context. Annual Review of Gerontology and Geriatrics, 11, 195–217.474Cohen, J. (1988). Statistical power analysis for the behavioral sciences475(2nd ed.). Hillsdale: Erlbaum.476De Clercq, B., De Fruyt, F., Mervielde, I., Krueger, R. F., &Markon, K.477(2011). Dutch translation of the Personality Inventory for DSM-5478(PID5).479Edwards, M. C., & Edelen, M. O. (2009). Special topics in item480response theory. In R. E. Millsap & A. Maydeu-Olivares (Eds.),481The SAGE handbook of quantitative methods in psychology (pp.482178–185). London: Sage.483Harkness, A. R., & McNulty, J. L. (1994). The personality psychopathol-484ogy five (PSY-5): Issue from the pages of a diagnostic manual instead485of a dictionary. In S. Strack &M. Lorr (Eds.),Differentiating normal486and abnormal personality (pp. 291–315). New York: Springer.487Holmén, K., & Furukawa, H. (2002). Loneliness, health and social488network among elderly people—a follow-up study. Archives of489Gerontology and Geriatrics, 35, 261–274.490Hopwood, C. J., & Donnellan, M. B. (2010). How should the internal491structure of personality inventories be evaluated? Personality and492Social Psychology Review, 14(3), 332–346.493Hopwood, C. J., Thomas, K. M., Markon, K. E., Wright, A. G. C., &494Krueger, R. F. (2012). DSM-5 personality traits and DSM-IV495personality disorders. Journal of Abnormal Psychology, 121(2),496424–432.497Krueger, R. F., Derringer, J., Markon, K. E., Watson, D., & Skodol, A.498E. (2012). Initial construction of a maladaptive personality trait499model and inventory for DSM-5. Psychological Medicine, 42,5001879–1890.501Liu, I. M., & Agresti, A. (1996). Mantel-Haenszel-type inference for502cumulative odds ratio with a stratified ordinal response. Biomet-503rics, 52, 1223–1234.504M Q1arkon, K. E., Krueger, R. F., & Watson, D. (2005). Delineating the505structure of normal and abnormal personality: an integrative hier-506archical approach. Journal of Personality and Social Psychology,50788(1), 139–157.508Mazor, K. M., Clauser, B. E., & Hambleton, R. K. (1992). The effect of509sample size on the functioning of the Mantel-Haenszel statistic.510Educational and Psychological Measurement, 52, 443–451.511Millsap, R. E. (2011). Statistical approaches to measurement invari-512ance. New York: Routledge.513M Q2roczek, D. K., Hurt, S. W., & Berman, W. H. (1999). Conceptual and514methodological issues in the assessment of personality disorders in515older adults. In E. Rosowsky, R. C. Abrams, & R. A. Zweig (Eds.),516Personality disorders in older adults (1st ed., pp. 135–150).517Mahwah: Lawrence Erlbaum Associates.518Oltmanns, T. F., & Balsis, S. (2011). Personality disorders in later life:519questions about the measurement, course, and impact of disorders.520Annual Review of Clinical Psychology, 7, 321–349.521Penfield, R. D. (2005). DIFAS: differential item functioning analysis522system. Applied Psychological Measurement, 29, 150–151.523Penfield, R. D. (2007). DIFAS 4.0. Differential item functioning anal-524ysis system: user’s manual. Unpublished manuscript.525Penfield, R. D. (2010). Distinguishing between net and global DIF in526polytomous items. Journal of EducationalMeasurement, 47, 129–149.527Penfield, R. D., & Algina, J. (2006). A generalized DIF effect variance528estimator formeasuring unsigned differential test functioning inmixed529format tests. Journal of Educational Measurement, 43(4), 295–312.530Reed, A. E., & Carstensen, L. L. (2012). The theory behind the age-531related positivity effect. Frontiers in Psychology, 3, doi:10.3389/532fpsyg.2012.00339.533Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of534mean-level change in personality traits across the life course: a

J Psychopathol Behav Assess

JrnlID 10862_ArtID 9364_Proof# 1 - 22/06/2013

AUTHOR'S PROOF

UNCORRECTEDPROOF

535 meta-analysis of longitudinal studies. Psychological Bulletin, 132,536 3–27.537 Segal, D. L., Coolidge, F. L., & Rosowsky, E. (2000). Personality538 disorders. In S. Krauss (Ed.), Psychopathology in later adulthood539 (pp. 89–115). Wiley: New York.540 Tackett, J. L., Balsis, S., Oltmanns, T. F., & Krueger, R. F. (2009). A541 unifying perspective on personality pathology across the life span:542 developmental considerations for the fifth edition of the diagnostic543 and statistical manual of mental disorders. Development and Psy-544 chopathology, 21, 687–713.545 Terracciano, A., McCrae, R. R., Brant, L. J., & Costa, P. T., Jr. (2005).546 Hierarchical linear modeling analyses of NEO PI-R scales in the547 Baltimore longitudinal study of aging. Psychology and Aging, 20548 (3), 493–506.549 Thomas, K. M., Yalch, M. M., Krueger, R. F., Wright, A. G. C.,550 Markon, K. E., & Hopwood, C. J. (2012). The convergent struc-551 ture of DSM-5 personality trait facets and Five Factor Model trait552 domains. Assessment.553 Van Alphen, S. P. J., Engelen, G. J. J. A., Kuin, Y., & Derksen, J. J. L.554 (2006). The relevance of a geriatric sub-classification of personality

555disorders in the DSM-V. International Journal of Geriatric Psychi-556atry, 21, 205–209.557V Q3an den Broeck, J., Rossi, G., & Dierckx, E. (2010). Diagnostiek558van persoonlijkheid en persoonlijkheidspathologie bij ouderen.559[Assessment of personality and personality pathology in older560adults.]. Tijdschrift voor Gerontologie en Geriatrie, 41(2), 68–56178.562V Q4enables, P., & Bailes, K. (1994). The structure of schizotypy, its563relation to subdiagnoses of schizophrenia and to sex and age.564British Journal of Clinical Psychology, 33, 277–294.565Widiger, T. A., & Trull, T. J. (2007). Plate tectonics in the classification566of personality disorder: shifting to a dimensional model. American567Psychologist, 62, 71–83.568Wright, A. G. C., Thomas, K. M., Hopwood, C. J., Markon, K. E.,569Pincus, A. L., & Krueger, R. F. (2012). The hierarchical structure570of DSM-5 pathological personality traits. Journal of Abnormal571Psychology. doi:10.1037/a0027669.572Z Q5weig, R. A. (2008). Personality disorder in older adults: assessment573challenges and strategies. Professional Psychology: Research and574Practice, 39(3), 298–305.

575

J Psychopathol Behav Assess

JrnlID 10862_ArtID 9364_Proof# 1 - 22/06/2013

AUTHOR'S PROOF

UNCORRECTEDPROOF

AUTHOR QUERIES

AUTHOR PLEASE ANSWER ALL QUERIES.

Q1. Markon et al. (2005) was not cited anywhere in the text. Please provide a citation. Alternatively,delete the item from the list.

Q2. Mroczek et al. (1999) was not cited anywhere in the text. Please provide a citation. Alternatively,delete the item from the list.

Q3. Van den Broeck et al. (2010) was not cited anywhere in the text. Please provide a citation.Alternatively, delete the item from the list.

Q4. Venables & Bailes (1994) was not cited anywhere in the text. Please provide a citation.Alternatively, delete the item from the list.

Q5. Zweig (2008) was not cited anywhere in the text. Please provide a citation. Alternatively, deletethe item from the list.