Stewart (2004) PAID Tpq Structure

Embed Size (px)

Citation preview

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    1/16

    The structure of Cloningers Tridimensional Personality

    Questionnaire in a British sample

    Mary E. Stewarta,*, Klaus P. Ebmeierb, Ian J. Dearyc

    aDepartment of Psychiatry, McKinnon House, Royal Edinburgh Hospital, Morningside Park,

    Edinburgh EH10 5HF, UKbDepartment of Psychiatry, University of Edinburgh, Royal Edinburgh Hospital, Morningside Park,

    Edinburgh EH10 5HF, UKcDepartment of Psychology, University of Edinburgh, 7 George Square, Edinburgh EH8 9JZ, UK

    Received 31 October 2002; received in revised form 6 May 2003; accepted 9 June 2003

    Abstract

    This study adds to the very few published reports of the structure of the Tridimensional Personality

    Questionnaire (TPQ) at both the item and subscale levels. The TPQ was completed by 897 students from

    Universities within Edinburgh. Exploratory factor analysis was run on the items and the 12 subscales as

    described by Cloninger, Przybeck, and Svrakic (1991). Harm Avoidance showed high internal consistencyboth for the whole scale and the subscales; however, this was not the case for Reward Dependence and

    Novelty Seeking. A three factor solution was extracted from analysis at the scale level which gives support

    to Cloningers model. However, when analysis was carried out at the item level, three and four factor

    solutions were extracted with only one factor, that of Harm Avoidance, resembling Cloningers model. The

    four factors extracted were provisionally named Harm Avoidance, Conscientiousness, Sociability and

    Impulsiveness. These more closely resemble factors from the Five Factor Model (Costa & McCrae, 1992)

    and Eysencks three factor model (Eysenck, Eysenck, & Barrett, 1985) than Cloningers theory. It may be

    necessary to adapt Cloningers model for a British sample, and more generally to question the psychometric

    structure of the TPQ.

    # 2003 Elsevier Ltd. All rights reserved.

    Keywords: Tridimensional Personality Questionnaire; TPQ; Exploratory factor analysis

    The majority of researchers working on trait models of personality agree that there are between

    three and five independent, reliable and stable broad personality factors (Barrett, Petrides,

    Eysenck, & Eysenck, 1998; DiLalla, Gottesman, Carey, & Bouchard, 1999; Jang, McCrae,

    0191-8869/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.doi:10.1016/S0191-8869(03)00237-X

    Personality and Individual Differences 36 (2004) 14031418www.elsevier.com/locate/paid

    * Corresponding author. Tel.: +44-131-537-6591.

    E-mail address: [email protected] (M.E. Stewart).

    http://www.elsevier.com/locate/paid/a4.3dmailto:[email protected]:[email protected]://www.elsevier.com/locate/paid/a4.3d
  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    2/16

    Angleitner, Riemann, & Livesley, 1998; McCrae et al., 2000). Recently attention has turned to the

    determinants of personality and its biological basis.

    Cloninger (1987) proposed a biological model derived from animal research. He suggests that

    three personality traits: Harm Avoidance (HA); Reward Dependence (RD); and Novelty Seeking(NS) are heritable and that they relate to the monoamine systems: serotonin; noradrenaline; and

    dopamine, respectively. They are genetically independent but the systems are interconnected.

    Cloningers theory has received much attention both within psychology and also within

    psychiatry, with relationships being found between these personality traits and mental illness.

    For instance, Harm Avoidance scores are increased in patients with depression, panic disorder,

    generalised anxiety disorder, obsessive-compulsive disorder and eating disorders (Kleifield,

    Sunday, Hurt, & Halmi, 1994; Reich, Noyes, Hirschfeld, Coryell, & OGorman, 1987; Richter,

    Summerfeldt, Joffe, & Swinson, 1996; Starcevic, Uhlenhuth, Fallon, & Pathak, 1996).

    However, the measures derived from Cloningers (1987) theory of personality must yet be

    shown to be both reliable and valid. To date only one study has tested Cloningers model in aBritish sample (Otter, Huber, & Bonner, 1995). In this study exploratory analysis at the scale level

    and confirmatory factor analysis were used in only 413 subjects. Although the Harm Avoidance

    Scale and its subscales were found to be reliable, the a coefficients were less than 0.6 for the

    Reward Dependence subscales, RD1, RD2 and RD4 and the Novelty Seeking subscales NS1 and

    NS4, with only NS3 and RD3 achieving coefficients above 0.7. Kline (1993) recommends that the

    coefficents should be nearing 0.9 and not fall below 0.7. Therefore in Otter et al.s sample it would

    appear that only the Harm Avoidance scale and its subscales are reliable. When an exploratory

    factor analysis was carried out at the scale level a three factor solution in line with Cloningers

    was found, however, confirmatory factor analysis, also at the scale level, suggested that Clo-

    ningers model provided a poor fit for the data. The w2 goodness-of-fit did not support the model.

    Cloningers model has been tested in other groups and cultures (Aschauer et al., 1994; Le Bon,Staner, Tecco, Pull, & Pelc, 1998; Lepine, Pelissolo, Teodorescu, & Teherani, 1994; Weyers,

    Krebs, & Janke, 1995) as well as in North America (Bagby, Parker, & Joffe, 1992; Cannon, Clark,

    Leeka, & Keefe, 1993; Giancola, Zeichner, Newbolt, & Stennett, 1994; Parker, Bagby, & Joffe,

    1996; Sher, Wood, Crews, & Vandiver, 1995; Waller, Lilienfeld, Tellegen, & Lykken, 1991) with

    mixed results. Using confirmatory factor analysis on both a community based adult population

    (n=360), and a group of undergraduate students (n=233) Parker et al. (1996) suggest that their

    data are consistent with Cloningers model. Similarly Bagby and colleagues (1992) suggested from

    a group of undergraduate students (n=216) that a good fit of Cloningers model could be found.

    However, the w2 goodness-of-fit did not support the model in either study although three other

    indices did the goodness-of-fit, the adjusted-goodness-of-fit, and the root mean-square residual.Earleywine (1993) disagreed with Bagby et al.s interpretation, suggesting both from further data

    (Earleywine, Finn, Peterson, & Pihl, 1992) and from Bagby et al.s (1992) study that Cloningers

    model was not a good fit. Earleywine (1993) suggests that these three indices (the goodness-of-fit,

    the adjusted-goodness-of-fit, and the root mean-square residual) should not be taken in isolation.

    Support for Earleywines position comes from Raykov (1993) and Cannon and colleagues (1993),

    who suggest that these three indices are descriptive and that the w2 shows that the model is a poor fit.

    French and German samples, where the questionnaire has been translated, find a somewhat

    similar model to Cloningers (Le Bon et al., 1998; Weyers et al., 1995), however, both of these

    studies analyse the TPQ at the scale level and the samples are small (n=104 and n=160, respectively).

    1404 M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    3/16

    Similarly at the scale level modest support is given from a Taiwanese sample, however, the authors

    suggest refinement of the Reward Dependence Scale (Chen, Chen, Chen, Yu, & Cheng, 2002).

    Five studies factor analyse the TPQ at the item level ( Cannon et al., 1993; Chen et al., 2002;

    Sher et al., 1995; Stallings, Hewitt, Cloninger, Heath, & Eaves, 1996; Zohar, Lev-Ari, Benja-min, Ebstein, Lichtenberg, & Osher, 2001). Three of these were in samples from the US

    (Cannon et al., 1993; Sher et al., 1995), one in a Hebrew speaking population from Israel (Zohar

    et al., 2001), and one in a Chinese speaking population in Taiwan. Sher et al. recruited a mixed

    group of volunteers, some who had a family history of alcoholism ( n=490) (Sher et al., 1995),

    while the group that Cannon and colleagues recruited were all alcoholics (Cannon et al., 1993).

    Interestingly, Stallings et als group was a twin sample (n=3290). They assessed Cloningers

    model by extracting 12 factors using iterative principal factor analysis. Nine of the subscales

    corresponded very closely to Cloningers model and only one (RD4) could not be recovered.

    However, a three or four factor model was not attempted at the item level in this study. Chen et al

    (2002) found that factor loading at the item level was not coherent with Cloningers scales,however the sample was small with only 201 subjects. The other three studies find moderate

    support for Cloningers model with the first factor clearly representing Harm Avoidance, but there

    was considerable overlap between Novelty Seeking and Reward Dependence. Zohar et al. (2001)

    found that over 50% of the items did not load on the designated factor in 1139 subjects.

    Cloningers model of personality has been widely accepted and utilised throughout psychiatry

    and psychology. It is perhaps surprising that Cloningers TPQ has been so widely accepted within

    research on personality although it has not been subject to equivalent reliability and validity

    analysis as questionnaires such as the NEO-PI-R or the EPQ-R. Both of these personality the-

    ories have received support from a wide range of sources and across cultures ( Barrett et al., 1998;

    Costa & McCrae, 1992; Digman, 1990; Heath, Eaves, & Martin, 1989; Katigbak, Church, &

    Akamine, 1996; McCrae, Jang, Livesley, Riemann, & Angleitner, 2001; Ostendorf & Angleitner,1994; Yang & Bond, 1990). The present study is the largest analysis of the TPQ in a British

    sample and the only exploratory item level analysis in a British sample. This study aims to assess

    Cloningers three factor personality model in a large British sample by using exploratory factor

    analysis at both the scale and item levels.

    1. Methods

    1.1. Recruitment of subjects

    Subjects were recruited from Edinburgh Universities. They were included if English was theirfirst language and if they were over 18 years of age. Participants were given a stamped, addressed

    envelope to return a questionnaire, and were asked to take a questionnaire only if they were

    willing to be considered for the experimental [acute tryptophan depletion stage of the study (see

    Stewart, Deary, & Ebmeier, 2002). Approximately 2000 questionnaires assessing mood and per-

    sonality were distributed; 1032 were completed and returned. The questionnaire pack included

    the Tri-dimensional Personality Questionnaire (TPQ; Cloninger, 1987), the Eysenck Personality

    Questionnaire-Revised (EPQ-R; Eysenck, Eysenck, & Barrett, 1985) as well as other measures

    assessing personality and mood. The questionnaire pack took approximately 1 h to complete.

    Here we report results from the TPQ responses.

    M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418 1405

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    4/16

    1.2. Factor analysis

    The psychometric properties of the TPQ were explored using exploratory factor analysis. At

    both the scale and item level factors were extracted using prinicipal axis factoring (communalitieswere iterated) and rotated orthogonally using Varimax.

    2. Results

    2.1. Subjects

    Those who did not complete both the TPQ and EPQ-R were excluded. A total of 347 males and

    550 females remained in the data set. Six males and five females omitted their dates of birth from

    the questionnaire pack. Male ages ranged from 17.4 to 50.4 (mean=21.1, standarddeviation=4.7). Female ages ranged from 17.1 to 48.9 (mean=20.7, standard deviation=4.0).

    2.2. Psychometric properties of the TPQ

    In order to attempt confirmation of Cloningers model analysis was conducted on the three

    higher factors and 12 subscales as described by Cloninger, Przybeck, and Svrakic (1991). Table 1

    shows the number of items, the mean scores, and internal consistency (Cronbach a reliability) for

    the three dimensions of Harm Avoidance, Novely Seeking and Reward Dependence and their

    subscales for males and females separately. The alpha reliabilities for the total scales range from

    0.72 to 0.90 (mean=0.81) and are all within the satisfactory range. The TPQ Harm Avoidance

    Scale has the highest reliabilities both for the total scale (0.90 and 0.89 for males and femalesrespectively) and the subscales (range between 0.72 and 0.90). Reward Dependence has the lowest

    reliabilities both for the total scale (0.72 in both males and females) and for the subscales (range

    between 0.39 and 0.76) and both RD1 (Sentimentality) and RD4 (Dependence) have a less than

    0.5 in both men and women. The Novelty Seeking subscales also have low reliabilities and only

    NS3 (Extravagance) has an a above 0.7 with both NS1 (Exploratory Excitability) and NS4

    (Disorderliness) having a less than 0.6. In sum, the reliabilities for both the Novelty Seeking and

    Reward Dependence subscales are poor.

    All the TPQ Scales are normally distributed. Gender differences for the total scale scores and

    the subscales are shown in Table 1. Females scored higher on the TPQ HA total scale ( t=5.35,

    df=895, P0.3 are in bold (the oblique solution is very similar and is available from the

    authors).

    Factor 1 is evidently Harm Avoidance, with high loadings from all four of the subscales and

    these subscales do not load above 0.2 on any of the other factors. However, Novelty Seekings

    exploratory excitability subscale does load negatively to a high degree on this factor. Cloninger et

    al. (1991) suggests that this subscale represents exploratory excitability versus stoic rigidity.

    Factor 2 is made up mainly of Novelty Seeking with all four of the subscales having loadingsgreater than 0.3 on this factor. However, Novelty Seekings exploratory excitability subscale,

    loads more highly on factor 1 than on factor 2. Also the Persistence subscale (RD2) loads nega-

    tively to a moderate degree on factor 2. Factor 3 is Reward Dependence being made up of 3 of

    the Reward Dependence subscales and none of the other subscales. The Persistence subscale

    loads below 0.1 on this, its home, factor. In sum, Cloningers structure is fairly well represented

    when exploratory factor analysis is carried out at the scale level. Three factors emerge with the

    highest loadings coming from the hypothesised subscales. However, there are some significant

    apparent mis-placements of the exploratory excitability and persistence subscales.

    2.4. Exploratory item analysis of the TPQ

    Principal axis factoring of the items from the TPQ was carried out and an orthogonal rotation

    was used (SPSS version 10.0 for Windows). As a guide, items with loadings lower than 0.3 were

    considered not to load on given factors. Loadings are reported to two decimal places. A Scree

    plot of eigenvalues derived from the correlation matrix of the items from the TPQ is shown in

    Fig. 2. The graph starts to flatten off after four factors. However, Cloninger predicts three fac-

    tors. Therefore both the three and four factor solution were extracted. Tables 3 and 4 show the

    rotated factor matrix for the three and four factor solutions respectively (the oblique solution is

    very similar, available via the authors).

    Table 2

    Structure matrix for three factor solution of the TPQ at the scale level

    Scale Description Factor 1 Factor 2 Factor 3 Communality

    HA1 Anticipatory worry 0.72 0.03 0.14 0.45

    HA2 Fear of uncertainty 0.75 0.19 0.15 0.46

    HA3 Shyness with strangers 0.68 0.11 0.20 0.40

    HA4 Fatigability 0.60 0.14 0.07 0.33

    NS1 Exploratory excitability 0.50 0.38 0.24 0.34

    NS2 Impulsiveness 0.14 0.65 0.09 0.30

    NS3 Extravagance 0.05 0.62 0.29 0.29

    NS4 Disorderliness 0.19 0.66 0.08 0.32

    RD1 Sentimentality 0.01 0.02 0.65 0.26

    RD2 Persistence 0.01 0.37 0.06 0.18

    RD3 Attachment 0.12 0.16 0.66 0.28

    RD4 Dependence 0.18 0.10 0.52 0.25

    1408 M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    7/16

    2.5. Three factor solution of the TPQHarm Avoidance, Conscientiousness and Sociability

    (HCS) (Table 3)

    The first factor in the three factor solution is clearly Harm Avoidance. However, Novelty

    Seeking and Reward Dependence are replaced by two factors which appear to represent Con-

    scientiousness and Sociability. Thirty of the 34 HA items load predominantly on the first factor

    with loadings ranging between 0.31 and 0.67. Only three items from other scales load on this

    Harm Avoidance factor, two Novelty Seeking items load negatively and one Reward Dependence

    item loads positively. Three of the HA items do not load on any factor. Two items load on both

    Harm Avoidance and factor 3 (HA 33, HA 37). However, their loadings are higher on Harm

    Avoidance than on the other factors. There are a number of themes within this factor. The first

    relates to feelings of worry and anxiety for unfamiliar and future experiences, for instance: HA2

    19 I often feel tense and worried in unfamiliar situations, even when others feel there is little to worryabout and HA1 5Usually I am more worried than most people that something might go wrong in

    the future. The second theme seems to relate to fatigue for instance: HA4 73It is extremely

    difficult for me to adjust to changes in my usual way of doing things because I get so tense, tired or

    worried and HA4 84I recover more quickly than most people from minor illnesses or stress. The

    third theme could be said to relate to confidence and anxiousness in social situations, for exam-

    ple: HA3 37I often avoid meeting strangers because I lack confidence with people I do not know

    and HA3 33When I have to meet a group of strangers, I am more shy than most people. Lastly

    there are a few questions relating to a search for excitement and inability to settle which load

    negatively on this factor: NS1 4When nothing new is happening, I usually start looking for

    Fig. 2. Scree plot of a factor analysis for the TPQ.

    M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418 1409

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    8/16

    something that is thrilling or exciting and HA3 100- It is easy for me to organise my thoughts while

    talking to someone.

    Factors 2 and 3 are not quite so clear. Factor 2 (Conscientiousness-HCS) consists of 15 load-

    ings from Novelty Seeking items, 5 from Reward Dependence and one from Harm Avoidance.Loadings ranged between 0.31 and 0.56. It would seem that factor 2 is Novelty Seeking, accord-

    ing to Cloningers scheme. However, only five items load above 0.5 and as many as 15 NS items

    do not load on any factor while three load on Harm Avoidance and two load on Sociability. Less

    than 50% of the NS items load on this factor within this the subscales are not equally well

    represented. There is a range from NS2, 3 and 4 subscales but there is only one question from

    NS1. There are a number of themes within this factor. The first relates to an unwillingness to

    Table 3

    Three factor varimax rotated solution of the TPQ

    Factor one Factor two Factor three

    No. Item Loading No. Item Loading No. Item Loading

    19 HA2 0.67 55 NS2 0.54 88 RD3 0.70

    23 HA2 0.66 66 NS3 0.52 90 RD3 0.66

    18 HA2 0.62 72 NS3 0.52 86 RD3 0.64

    1 HA1 0.60 70 NS3 0.50 12 RD3 0.63

    89 HA3 0.59 49 HA4 0.45 3 RD3 0.60

    5 HA1 0.57 48 NS2 0.45 31 RD1 0.44

    8 HA1 0.54 99 NS2 0.44 32 NS3 0.42

    75 HA4 0.54 22 NS4 0.43 36 RD4 0.36

    73 HA4 0.53 24 NS4 0.40 40 NS1 0.35

    95 HA1 0.53 56 NS2 0.38 74 RD3 0.3584 HA1 0.52 87 NS3 0.38 15 RD3 0.34

    37 HA3 0.48 81 NS2 0.37 58 RD4 0.33

    80 HA4 0.48 46 NS2 0.36 7 RD3 0.31

    10 HA1 0.48 36 RD4 0.35 34 RD1 0.30

    79 RD2 0.48 43 NS1 0.35 37 HA3 0.31

    59 HA4 0.47 13 NS4 0.31 33 HA3 0.33

    26 HA2 0.45 78 NS3 0.31

    42 HA3 0.45 97 RD2 0.41

    33 HA3 0.44 45 RD2 0.45

    14 HA1 0.44 39 RD2 0.49

    54 HA4 0.43 92 RD2 0.56

    63 HA4 0.42

    82 HA1 0.4291 HA1 0.41

    98 HA1 0.40

    44 HA3 0.38

    38 HA3 0.38

    69 HA4 0.36

    51 HA2 0.34

    57 HA4 0.33

    100 HA3 0.31

    4 NS1 0.31

    2 NS1 0.40

    1410 M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    9/16

    push hard at work, for instance, RD2 92 loads negatively I am more hard-working than most

    people and RD2 39 loads negatively I usually push myself harder than most people do because I

    want to do as well as I possibly can . The second theme relates to a preference in spending rather

    than saving money for instance NS3 66 loads negatively I am better at saving money than mostpeople and NS3 70 I often spend money until I run out of cash or get into debt from using too much

    credit. The third to a willingness to make rash decisions rather than think over consequences, for

    instance, NS2 55 is answered negatively I usually think about all the facts in detail before I make a

    decision and NS2 48 I often follow my instincts, hunches, or intuition without thinking through all

    the details. This factor could represent a negative Conscientiousness or Irresponsibility and

    appears to also relate to Impulsiveness.

    Factor 3 (Sociability-HCS) is predominantly Reward Dependence, with 12 RD, two NS items

    and two HA items loading negatively. The loadings on this factor ranged between 0.30 and 0.70.

    Table 4

    Four factor solution of the TPQ orthogonally rotated

    Factor one Factor two Factor three Factor four

    No. Item Loading No. Item Loading No. Item Loading No. Item Loading

    23 HA2 0.68 66 NS3 0.53 88 RD3 0.69 75 HA4 0.46

    19 HA2 0.67 55 NS2 0.50 90 RD3 0.65 47 HA2 0.41

    73 HA4 0.62 72 NS3 0.48 86 RD3 0.64 26 HA2 0.39

    5 HA1 0.60 70 NS3 0.48 12 RD3 0.62 29 HA2 0.38

    18 HA2 0.59 99 NS2 0.46 3 RD3 0.61 1 HA1 0.38

    10 HA1 0.55 49 HA4 0.43 31 RD1 0.45 51 HA2 0.38

    14 HA1 0.50 87 NS3 0.39 32 NS3 0.42 89 HA3 0.37

    1 HA1 0.49 56 NS2 0.38 40 NS1 0.37 59 HA4 0.3789 HA3 0.46 48 NS2 0.38 36 RD4 0.36 82 HA1 0.35

    84 HA1 0.46 22 NS4 0.37 74 RD3 0.35 20 RD4 0.34

    79 RD2 0.46 24 NS4 0.35 58 RD4 0.32 17 RD4 0.33

    8 HA1 0.46 81 NS2 0.34 15 RD3 0.32 8 HA1 0.32

    37 HA3 0.45 36 RD4 0.34 7 RD3 0.31 95 HA1 0.31

    95 HA1 0.44 78 NS3 0.34 34 RD1 0.31 80 HA4 0.30

    54 HA4 0.44 16 NS4 0.33 83 RD1 0.31 41 RD2 0.30

    69 HA4 0.43 46 NS2 0.33 35 NS4 0.30 22 NS4 0.31

    91 HA1 0.40 21 NS4 0.32 38 HA3 0.30 97 RD2 0.34

    30 NS2 0.40 43 NS1 0.31 37 HA3 0.34 62 NS4 0.36

    33 HA3 0.38 45 RD2 0.47 33 HA3 0.36 48 NS2 0.40

    80 HA4 0.38 97 RD2

    0.50 4 NS1

    0.4138 HA3 0.37 39 RD2 0.56 60 NS4 0.41

    42 HA3 0.37 92 RD2 0.62 2 NS1 0.47

    75 HA4 0.36

    64 RD3 0.36

    57 HA4 0.35

    63 HA4 0.34

    59 HA4 0.33

    26 HA2 0.33

    68 HA4 0.32

    11 NS1 0.36

    M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418 1411

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    10/16

    The highest loadings are all from Reward Dependence 3 subscale. These questions relate to

    openness with other people and discussing problems with others, Reward Dependence items are

    not representative of the factor as 14 RD items do not load on any factor, five load on Factor 2,

    one loads on factor 1. The first theme relates to a willingness to open up and listen to others RD388 Even when I am with friends, I prefer not to open up very much, and RD3 90 answers nega-

    tively I usually like to stay cool and detached from other people. The second main theme is a

    preference to be around others than being alone, for instance, RD3 7 I would like to have warm

    and close friends with me most of the time and RD3 15 It wouldnt bother me to be alone all the

    time. This factor appears to relate to sociability or openness to others.

    2.6. Four factor solution of the TPQHarm Avoidance, Conscientiousness, Sociability and

    Impulsiveness (HCSI) (Table 4)

    Factor 1 (Harm Avoidance-HCSI) is clearly Harm Avoidance, as with the three-factor HCSsolution. In this case 26 of the HA items load on factor 1, with 24 of these having their highest

    loadings on factor 1. The loadings range between 0.32 and 0.68. It also has moderate loadings

    from 2 Reward Dependence items and 2 Novelty Seeking items, one loading positively and one

    loading negatively. The factor seems to be describing emotionality. The themes are similar to that

    of the three factor solution, with a theme relating to worry, a theme relating to confidence in

    social situations and a theme relating to fatigue. However, this factor is slightly less broad than

    the one described in the three factor HCS solution and does not contain a theme relating to a

    search for excitement and inability to settle.

    Factor 2 (Conscientiousness-HCSI) is a mix of Reward Dependence and Novelty Seeking items.

    The Reward Dependence items are all but one from the subscale Persistence and these all load

    negatively on the factor with items also from the Novelty Seeking sub-scales of Impulsiveness,Extravagance and Disorderliness. The loadings range between 0.31 and 0.62. It has an almost

    identical item content to the Conscientiousness factor found in the three factor HCS solution

    with one item not loading on this factor and two further items (all three have loadings of less than

    0.35). The items with the highest loadings relate to work attitudes and pushing hard at work. The

    second theme relates to spending money with a preference to spending rather than saving. The

    third theme relates to rashness and not willing to follow routines or rules. The themes and item

    content are the same as in factor 2 of the three factor solution.

    Factor 3 (Sociability-HCSI) is mainly made up of Reward Dependence items with the highest

    loadings being five items from the Reward Dependence subscale Attachment (RD3) with the loadings

    ranging between 0.30 and 0.69. The item content is almost identical to factor 3 in the three factorsolution with an additional three items loading on this factor with loadings less than 0.31. The first

    theme relates to openness with friends and a willingness to discuss problems and private thoughts.

    The second corresponds to the individuals response to others, and wish to be with others. This factor

    appears to relate to Sociability and is almost identical to the factor in the three factor solution.

    Factor 4 (Impulsiveness-HCSI) has a mix of predominantly Novelty Seeking and Harm

    Avoidance with three Reward Dependence items with loadings ranging between 0.30 and 0.47.

    The items appear to relate to searching for new ideas and thrills, risk taking, impulsiveness and

    not thinking about the consequences. For instance the highest loading items are: NS1 2 I often try

    new things just for fun or thrills, even if most people think it is a waste of time, HA4 75 I usually feel

    1412 M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    11/16

    much more confident and energetic than most people, even after minor illnesses or stress, NS4 60 I

    am usually able to get other people to believe me, even when I know that what I am saying is exag-

    gerated or untrue, HA2 47 Most of the time I would prefer to do something a little risky (like riding

    in a fast automobile over steep hills and sharp turns)rather than having to stay quiet and inactivefor a few hours, NS1 4 When nothing new is happening, I usually start looking for something that is

    thrilling or exciting and NS2 48 I often follow my instincts, hunches, or intuition without thinking

    through all the details. It seems to represent a factor relating to Impulsiveness (I).

    3. Discussion

    Cloningers model appears to be relatively well replicated when exploratory analysis is carried

    out at the scale level. The three factors did, in the main, contain loadings from the hypothesised

    subscales with only a few exceptions. However the subscale of Persistence (RD2) did not loadhighly on any factor but had a small negative loading on the Novelty Seeking factor (0.37), and

    exploratory excitability loaded higher on HA than on its home factor of Novelty Seeking. Some

    of the scales and subscales showed unacceptably low internal consistency suggesting that the

    items are not measuring the same construct. Only the subscales from Harm Avoidance show

    satisfactory reliabilities with the total scale showing high alphas for both males and females. At

    the item level the three factor model as proposed by Cloninger was not found. Three and four

    factor solutions were extracted with only Harm Avoidance from Cloningers model being

    extracted as a single factor. The three factor solution, imposed on the data because of Cloningers

    model (HCS) suggested factors resembling Harm Avoidance, Conscientiousness and Sociability.

    The four factor solution, indicated by the data, (HCSI) suggested factors representing Harm

    Avoidance, Conscientiousness, Sociability and Impulsiveness.Surprisingly, Cloningers model has not been subject to the level of psychometric debate or testing

    of other personality models such as Eysencks three factor model or Costa and McCrae Five

    Factor Model. This study does not add support to the psychometric properties of the Tridimen-

    sional Personality Model as proposed by Cloninger. The personality factor, Harm Avoidance, is

    apparent whether Cloningers questionnaire is analysed at the item or the scale level. However the

    other two factors of Reward Dependence and Novelty Seeking are weakly supported by these

    analyses. The factors found in this study are similar to Eysencks three factor model of person-

    ality and to Costa and McCrae Five Factor Model. The four factor solution contains dimensions

    very similar to those in Eysencks model of personality. The Harm Avoidance factor from the four

    factor solution is similar to Eysencks Neuroticism where the items relate to anxiety and depres-sion with no items relating to impulsivity. Within both the three and four factor solutions is a

    Sociability factor that is similar to Eysencks Extraversion from the EPQ-R and EPQ. The Extraver-

    sion factor from the Eysenck Personality Inventory (EPI, Eysenck & Eysenck, 1964) measures both

    Sociability and Impulsivity whereas Extraversion from the EPQ scales measures Sociability alone

    (Roger & Morris, 1991). Interestingly there is a factor in the four factor solution relating to

    Impulsivity. There is a further factor of Conscientiousness which could be said to relate to Tough

    Mindedness. Therefore in the four factor solution there is a factor similar to Neuroticism (Harm

    Avoidance), Extraversion appears to be split into Sociability and Impulsiveness and Con-

    scientiousness which relates to Tough Mindedness.

    M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418 1413

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    12/16

    The Harm Avoidance factor in the three factor solution appears more similar to the Five

    Factor Models Neuroticism with themes relating to worry and anxiety, to fatigue, to confidence

    and anxiousness in social situations and to excitement and an inability to settle. Costa and

    McCraes Neuroticism factor contains six subscales which are Anxiety (N1), Angry Hostility(N2), Depression (N3), Self-Consciousness (N4), Impulsiveness (N5) and Vulnerability (N6). The

    Conscientiousness factor in both the solutions is clearly very similar to Costa and McCraes

    Conscientiousness. Costa and McCrae include six subscales of Competence (C1), Order (C2),

    Dutifulness (C3), Achievement Striving (C4), Self-Discipline (C5), Deliberation (C6). The themes

    in the solutions from the TPQ are working hard which may relate to Dutifulness and Achieve-

    ment Striving, a preference to saving money rather than spending which may relate to Self-Dis-

    cipline and Order, and a wish to think things through and in detail which would seem to relate to

    Deliberation. Without further analysis it is difficult to ascertain the extent to which these scales

    overlap. It is easy to see how what one person may call Conscientiousness another may call

    Tough Mindedness and so on. It is only when these factors are tested against other constructsthat they can truly be identified.

    Other studies which analyse the TPQ at the item level similarly find an Emotionality factor

    whether a three, four or five factor solution is extracted (Cannon et al., 1993; Sher et al., 1995;

    Zohar et al., 2001). Interestingly a Conscientiousness factor is also extracted in all but Sher et al.s

    three factor solution. A factor that relates to conformity, sociability and agreeableness can be

    seen in all the solutions. Table 5 summarises these solutions and the solutions from the current

    study. Notably all of the solutions contain a factor similar to Harm Avoidance. The second most

    consistent factor is one relating to Conscientiousness (Conscientiousness in the three and four

    factor solutions from this study, Persistence from Sher et al. and Orderliness from Cannon et al).

    A third factor that is represented is Sociability or Reward Dependence. This factor appears to

    encompass a wish to be with others and an ability to be open with others. In Cannon et al.s fivefactor solution this is represented negatively by their second factor Detachment. In sum, none of

    the published exploratory item level analyses of the TPQ or the current study could reproduce

    Cloningers three factors to a satisfactory level.

    Females scored higher on the Harm Avoidance Scale and on the Reward Dependence Scale.

    Similar differences in these scales have been found on these scales in other languages and cultures

    (e.g. Cloninger et al., 1991; Giancola et al., 1994; Otter et al., 1995; Zohar et al., 2001). There are

    a number of possibilities why women score higher on these scales. There may be a genuine gender

    bias, a sampling error, it may be due to the scale itself, a reporting bias or it may be due to

    Table 5

    Factors found from exploratory item analysis of the TPQ

    Author Factor 1 Factor 2 Factor 3 Factor 4 Factor 5

    (Cannon et al., 1993) Subjective Distress Detachment Disinhibition Relaxed Confidence Orderliness

    (Sher et al., 1995) Harm Avoidance Reward Dependence Novelty Seeking

    (Sher et al., 1995) Harm Avoidance Reward Dependence Novelty Seeking Persistence

    HCS this study Harm Avoidance Sociability Conscientiousness

    HCSI Harm Avoidance Sociability Impulsiveness Conscientiousness

    HCS H, Harm Avoidance; C, Conscientiousness; S, Sociability; HCSI H, Harm Avoidance; I, Impulsivity.

    1414 M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    13/16

    chance. There may be a gender bias in the questions, for instance Jorm (1987) (when discussing

    Neuroticism, a scale which is highly correlated to Harm Avoidance) suggests that some of the

    questions might be weighted to contain more items which are more appropriate for females rather

    than males. A male might be more likely to show anger than cry. Zohar et al. (2001) suggest thatthe gender differences in Harm Avoidance represent a genuine difference as a similar pattern is

    shown across other languages and cultures and that this difference is shown in scales beyond the

    personality questionnaires for instance in anxiety and mood measures. However, they suggest

    that the differences in Reward Dependence are more related to social desirability in responding

    than in genuine differences in personality. Two components have been found in the Neuroticism

    Scale, one that was influenced by gender, and one that was not (Francis, 1993). It would be

    interesting to assess whether this is also the case with the TPQ.

    The mean scores, for this sample if compared to Cloningers normative sample ( Cloninger et

    al., 1991) and the previous English sample (Otter et al., 1995), differ. The present sample showed

    slightly higher means for Harm Avoidance compared with the other samples, higher NoveltySeeking values than Cloninger et al.s (1991) samples though similar to Otter et als (1995) sam-

    ple, but similar Reward Dependence values. There could be a number of reasons for the differ-

    ence in scores. When the subjects were recruited they were asked to volunteer only if they would

    be willing to take part in the acute tryptophan depletion study. Perhaps those higher in Novelty

    Seeking would be more likely to apply. A further reason for a difference in scores may be age. In

    comparison to the other two studies, the present group was younger and all were recruited from

    the student population. Novelty Seeking is likely to decrease with age therefore it is likely that the

    younger group would have higher scores. Cloninger et al. (1991) found a negative correlation

    (r=0.36 to 0.42) between age and Novelty Seeking.

    Cloningers model is attractive in that a trait is directly related to a monamine system. One of

    the first quantitative trait loci identified for personality was connected with Novelty Seeking(Benjamin, Li, L., Patterson, Greenberg, Murphy, & Hamer, 1996; Ebstein et al., 1996), an

    association between the D4 dopamine receptor gene and Novelty Seeking. However only one of

    these groups used the TPQ and Cloningers personality trait of Novelty Seeking (Ebstein et al.,

    1996), the other used the NEO personality inventory (Benjamin et al., 1996). This association has

    proved inconsistent (Malhotra, Virkkunen, Rooney, Eggert, Linnoila, & Goldman, 1996). Jang,

    Vernon, and Livesley (2001) suggest that the ambiguity in this area may be due to the measure of

    personality employed, and most of the studies which have found negative results with dopamine

    and Novelty Seeking have used the TPQ.

    Extraversion and Neuroticism are found consistently within research on personality but there is

    still a debate as to whether there is one further factor or three. Extraversion and Neuroticismwere found from Cloningers TPQ with two further personality factors Conscientiousness and

    Impulsivenss. Zuckerman, Kuhlman and Camac (1988) asked what lies beyond E and N. This

    question is still valid.

    Acknowledgements

    We thank the volunteers who took part in this study. Norma Brearley, George Tait and

    Margaret van Beck for their help and support.

    M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418 1415

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    14/16

    References

    Aschauer, H. N., Meszaros, K., Willinger, U., Fischer, G., Berger, P., Reiter, E., Lenzinger, E., & Berger, K. (1994).

    Tridimensional Personality Questionnairesome results on the validity of a German version of the questionnaire.Neuropsychiatrie, 8(3), 177181.

    Bagby, R. M., Parker, J. D. A., & Joffe, R. T. (1992). Confirmatory factor analysis of the Tridimensional Personality

    Questionnaire. Personality and Individual Differences, 13, 12451246.

    Barrett, P. T., Petrides, K. V., Eysenck, S. B. G., & Eysenck, H. J. (1998). The Eysenck Personality Questionnaire: an

    examination of the factorial similarity of P, E, N, and L across 34 countries. Personality and Individual Differences,

    25(5), 805819.

    Benjamin, J., Li, L., Patterson, C., Greenberg, B. D., Murphy, D. L., & Hamer, D. H. (1996). Population and familial

    association between the D4 dopamine receptor gene and measures of novelty seeking. Nature Genetics, 12, 8184.

    Cannon, D. S., Clark, L. A., Leeka, J. K., & Keefe, C. K. (1993). A reanalysis of the tridimensional personality ques-

    tionnaire (TPQ) and its relation to Cloningers Type 2 Alcoholism. Psychological Assessment, 5(1), 6266.

    Chen, W. J., Chen, H. M., Chen, C. C., Yu, W. Y., & Cheng, A. T. A. (2002). Cloningers tridimensional personality

    questionnaire: psychometric properties and construct validity in Taiwanese adults. Comprehensive Psychiatry, 43(2),

    158166.

    Cloninger, C. R. (1987). A systematic method for clinical description and classification of personality variants. A

    proposal. Archives of General Psychiatry, 44(6), 573588.

    Cloninger, C. R., Przybeck, T. R., & Svrakic, D. M. (1991). The Tridimensional Personality Questionnaire: US

    normative data. Psychological Reports, 69(3), 10471057.

    Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO PI-RTM) and NEO Five-Factor

    Inventory (NEO-FFI). Odessa: Psychological Assessment Resources, Inc.

    Digman, J. M. (1990). Personality structureemergence of the 5-Factor Model. Annual Review of Psychology, 41, 417

    440.

    DiLalla, D. L., Gottesman, I. I., Carey, G., & Bouchard, T. J. Jr. (1999). Heritability of MMPI Harris-Lingoes and

    Subtle-Obvious subscales in twins reared apart. Assessment, 6(4), 353366.

    Earleywine, M. (1993). The examination of alternative models enhances confirmatory factor-analyses. Personality and

    Individual Differences, 15(5), 593594.Earleywine, M., Finn, P. R., Peterson, J. B., & Pihl, R. O. (1992). Factor structure and correlates of the Tridimensional

    Personality Questionnaire. Journal of Studies on Alcohol, 53(3), 233238.

    Ebstein, R. P., Novick, O., Umansky, R., Priel, B., Osher, Y., Blaine, D., Bennett, E. R., Nemanov, L., Katz, M., &

    Belmaker, R. H. (1996). Dopamine D4 receptor (D4DR) exon III polymorphism associated with the human per-

    sonality trait of novelty seeking. Nature Genetics, 12, 7880.

    Eysenck, H. J., & Eysenck, S. B. G. (1964). Manual of the Eysenck Personality Inventory. London: University of

    London Press.

    Eysenck, S. B. G., Eysenck, H. J., & Barrett, P. (1985). A revised version of the psychotocism scale. Personality and

    Individual Differences, 6, 2129.

    Francis, L. J. (1993). The dual nature of the Eysenckian Neuroticism Scalesa question of sex-differences. Personality

    and Individual Differences, 15(1), 4359.

    Giancola, P. R., Zeichner, A., Newbolt, W. H., & Stennett, R. B. (1994). Construct-validity of the dimensions ofCloninger Tridimensional Personality Questionnaire. Personality and Individual Differences, 17(5), 627636.

    Heath, A. C., Eaves, L. J., & Martin, N. G. (1989). The genetic-structure of personality. 3. Multivariate genetic item

    analysis of the EPQ Scales. Personality and Individual Differences, 10(8), 877888.

    Jang, K. L., McCrae, R. R., Angleitner, A., Riemann, R., & Livesley, W. J. (1998). Heritability of facet-level traits in a

    cross-cultural twin sample: support for a hierarchical model of personality. Journal of Personality & Social Psychol-

    ogy, 74(6), 15561565.

    Jang, K. L., Vernon, P. A., & Livesley, W. J. (2001). Behavioural-genetic perspectives on personality function.

    Canadian Journal of Psychiatry-Revue Canadienne De Psychiatrie, 46(3), 234244.

    Jorm, A. F. (1987). Sex differences in neuroticism: a quantitative synthesis of published research. Australian & New

    Zealand Journal of Psychiatry, 21(4), 501506.

    1416 M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    15/16

    Katigbak, M. S., Church, A. T., & Akamine, T. X. (1996). Cross-cultural generalizability of personality dimensions:

    relating indigenous and imported dimensions in two cultures. Journal of Personality & Social Psychology, 70(1), 99

    114.

    Kleifield, E. I., Sunday, S., Hurt, S., & Halmi, K. A. (1994). The tridimensional personality questionnaire: anexploration of personality traits in eating disorders. Journal of Psychiatric Research, f5, 413423.

    Kline, P. (1993). The handbook of psychological testing. London: Routledge.

    Le Bon, O., Staner, L., Tecco, J., Pull, C., & Pelc, I. (1998). Tridimensional personality questionnaire (TPQ): validation

    of a French-speaking control population. Encephale-Revue De Psychiatrie Clinique Biologique et Therapeutique,

    24(1), 4045.

    Lepine, J. P., Pelissolo, A., Teodorescu, R., & Teherani, M. (1994). Psychometric properties of the French version of

    the Tridimensional Personality Questionnaire (TPQ). Encephale-Revue De Psychiatrie Clinique Biologique et Ther-

    apeutique, 20(6), 747753.

    Malhotra, A. K., Virkkunen, M., Rooney, W., Eggert, M., Linnoila, M., & Goldman, D. (1996). The association

    between the dopamine D4 receptor (D4DR) 16 amino acid repeat polymorphism and novelty seeking. Molecular

    Psychiatry, 1, 388391.

    McCrae, R. R., Costa, P. T., Ostendorf, F., Angleitner, A., Hrebickova, M., Avia, M. D., Sanz, J., Sanchez-Bernardos,

    M. L., Kusdil, M. E., Woodfield, R., Saunders, P. R., & Smith, P. B. (2000). Nature over nurture: temperament,personality, and life span development. Journal of Personality and Social Psychology, 78(1), 173186.

    McCrae, R. R., Jang, K. L., Livesley, W. J., Riemann, R., & Angleitner, A. (2001). Sources of structure: genetic,

    environmental, and artifactual influences on the covariation of personality traits. Journal of Personality, 69(4), 511

    535.

    Ostendorf, F., & Angleitner, A. (1994). A comparison of different instruments proposed to measure the Big-5.

    European Review of Applied Psychology-Revue Europeenne De Psychologie Appliquee, 44(1), 4553.

    Otter, C., Huber, J., & Bonner, A. (1995). Cloninger Tridimensional Personality Questionnairereliability in an

    English sample. Personality and Individual Differences, 18(4), 471480.

    Parker, J. D. A., Bagby, R. M., & Joffe, R. T. (1996). Validation of the biosocial model of personality: confirmatory

    factor analysis of the tridimensional personality questionnaire. Psychological Assessment, 8(2), 139144.

    Raykov, T. (1993). Confirmatory factor analysis of the Tridimensional Personality Questionnaire: comments on

    Bagby, Parker and Joffe. Personality and Individual Differences, 15, 729731.Reich, J., Noyes Jr., R., Hirschfeld, R., Coryell, W., & OGorman, T. (1987). State and personality in depressed and

    panic patients. American Journal of Psychiatry, 144(2), 181187.

    Richter, M. A., Summerfeldt, L. J., Joffe, R. T., & Swinson, R. P. (1996). The Tridimensional Personality Ques-

    tionnaire in obsessive-compulsive disorder. Psychiatry Research, 65(3), 185188.

    Roger, D., & Morris, J. (1991). The internal structure of the EPQ Scales. Personality and Individual Differences, 12(7),

    759764.

    Sher, K. J., Wood, M. D., Crews, T. M., & Vandiver, P. A. (1995). The Tridimensional Personality Questionnaire

    reliability and validity studies and derivation of a short-form. Psychological Assessment, 7(2), 195208.

    Stallings, M. C., Hewitt, J. K., Cloninger, C. R., Heath, A. C., & Eaves, L. J. (1996). Genetic and environmental

    structure of the Tridimensional Personality Questionnaire: three or four temperament dimensions? Journal of

    Personality and Social Psychology, 70(1), 127140.

    Starcevic, V., Uhlenhuth, E. H., Fallon, S., & Pathak, D. (1996). Personality dimensions in panic disorder and

    generalized anxiety disorder. Journal of Affective Disorders, 37, 7579.

    Stewart, M. E., Deary, I. J., & Ebmeier, K. P. (2002). Neuroticism as a predictor of mood change: the effects of

    tryptophan depletion. British Journal of Psychiatry, 181(SEPT), 242247.

    Waller, N. G., Lilienfeld, S. O., Tellegen, A., & Lykken, D. T. (1991). The Tridimensional Personality Questionnaire

    structural validity and comparison with the Multidimensional Personality Questionnaire. Multivariate Behavioral

    Research, 26(1), 123.

    Weyers, P., Krebs, H., & Janke, W. (1995). Reliability and construct-validity of the German version of Cloningers

    Tridimensional Personality Questionnaire. Personality and Individual Differences, 19(6), 853861.

    Yang, K. S., & Bond, M. H. (1990). Exploring implicit personality theories with indigenous or imported constructs: the

    Chinese case. Journal of Personality & Social Psychology, 58(6), 10871095.

    M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418 1417

  • 8/2/2019 Stewart (2004) PAID Tpq Structure

    16/16

    Zohar, A. H., Lev-Ari, L., Benjamin, J., Ebstein, R., Lichtenberg, P., & Osher, Y. (2001). The psychometric properties

    of the Hebrew version of Cloningers Tridimensional Personality Questionnaire. Personality and Individual Differ-

    ences, 30, 12971309.

    Zuckerman, M., Kuhlman, D. M., & Camac, C. (1988). What lies beyond E and N? Factor analyses of scales believedto measure basic dimensions of personality. Journal of Personality and Social Psychology, 54, 96107.

    1418 M.E. Stewart et al. / Personality and Individual Differences 36 (2004) 14031418