I. Tsaousis European&Journalof S. Georgiades: Psychological The Greek Assessment © 2009 Personality Hogrefe 2009;Adjective Vol. & Huber 25(3):164–174 Ch Publishers ecklist
Development and Psychometric Properties of the Greek Personality Adjective Checklist (GPAC) Ioannis Tsaousis1 and Stelios Georgiades2 1
University of Crete, Department of Psychology, Greece McMaster University, Department of Psychiatry and Behavioral Neurosciences, Hamilton, Ontario, Canada
2
Abstract. This study presents the development and the psychometric properties of the Greek Personality Adjective Checklist (GPAC), a new instrument assessing personality in the Greek population. The GPAC is based on the lexical hypothesis tradition and its theoretical framework was derived from the emic study of the Greek personality lexicon (Saucier, Georgiades, Tsaousis, & Goldberg, 2005). It consists of 94 adjectives measuring six concrete dimensions: Even Temper, Introversion/Melancholia, Prowess/Heroism, Agreeableness/Positive Affect, Conscientiousness, and Negative Valence/Honesty. Results from exploratory and confirmatory factor analyses provided support for a six-factor solution for the structure of Greek personality. Additional results provided empirical evidence for the reliability of the GPAC. The Cronbach’s α coefficients for the six scales ranged between .85 and .95. The test-retest correlation coefficients ranged between .71 and .85. Finally, preliminary results provided evidence of construct validity based on convergence correlations with other personality measures such as the Traits Personality Questionnaire 5 (TPQue5), the Big Five Inventory (BFI), and the 50 Big Five Factor Markers (50 BFFM), as well as other criterion personality measures such as the Eysenck Personality Questionnaire (EPQ). It is concluded that the GPAC is a reliable and valid measure, useful for the assessment of normal personality in the Greek population. Keywords: GPAC, personality checklist, lexical approach, emic approach, six-factor model
Introduction In recent years, researchers studying the lexical theory of personality have used the emic/indigenous (as opposed to the etic/imported) approach in an attempt to identify indigenous dimensions of personality in various sociocultural environments. It is argued that the emic approach is more appropriate and effective than the etic approach, since it explores personality from within the cultural system, and allows for determination of culture-specific aspects of personality within a language framework (Benet-Martinez, 2007). Apart from the published Dutch and German lexically based factor structures (Brokken, 1978; Ostendorf, 1990), there are also English, Italian, French, Polish, Czech, Hungarian, Spanish, Chinese, Korean, Japanese, Filipinese, Belgian, Hebrew, Greek, Estonian, Finnish, Croatian, and Turkish lexical taxonomic works (see Saucier & Goldberg, 2001 for a review). Recently, Saucier, Georgiades, Tsaousis, and Goldberg (2005), using an emic approach, reported findings regarding the structure of the Greek personality lexicon. That study produced the following six factors: Even Temper, Introversion/Melancholia, Prowess/Heroism, Agreeableness/Positive Affect, Conscientiousness, and Negative Valence/Honesty. Factor I, Even Temper, is defined by terms such as nervous, impatient, abrupt, aggressive, and touchy, and seems to resemble the Big Five Neuroticism factor. Factor II, Introversion/Melancholia, is defined by terms such as depressive, European Journal of Psychological Assessment 2009; Vol. 25(3):164–174 DOI 10.1027/1015-5759.25.3.164
moody, sad, pessimistic, insecure, and loner, and seems to correspond with the opposite pole of the Big Five factor Extraversion. Factor III, Prowess/Heroism, is defined by terms such as genius, inventive, multitalented, agonistic, and pioneering, and seems to resemble the Big Five Intellect/Openness to Experience factor. Factor IV, Agreeableness/Positive Affect, is defined by terms such as kind, tender, friendly, and kind-hearted, and seems to resemble the cross-language variant of Agreeableness, although it includes some adjectives that correspond to the positive-affect domain (e.g., affectionate, smiling, cheerful). Factor V, which includes terms such as untidy, lazy, organized, and hard working, was labeled Conscientiousness in view of its similarity to the Big Five factor by that name. Finally, Factor VI, Negative Valence/Honesty, is defined by terms such as immoral, dishonest, inhuman, and violent, and seems to correspond with the sixth factor, Negative Valence, from the Tellegen and Waller (1987) Big Seven model. From the aforementioned it could be argued that the Greek lexical factor solution seems to share common conceptual characteristics but does not correspond isomorphically to the Big Five model, one of the most popular personality models (e.g., Goldberg, 1990; Costa & McCrae, 1992). The major difference is the emergence of a sixth factor (Negative Valence), which is not included in the Big Five model. In addition, some factors are compounded by two different constructs such as the Introversion/Melancholia, Prowess/Hero© 2009 Hogrefe & Huber Publishers
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
ism, and Agreeableness/Positive Affect dimensions. These factors contain terms that correspond to their Big Five counterparts (i.e., Extraversion, Openness to Experience/Intellect, and Agreeableness), but also capture other constructs that broaden the conceptual meaning of the dimension. These differences can be partly explained by the fact that methods used in this particular study are somewhat different from those used in previous investigations conducted in other languages (e.g., the inclusion, apart from adjective terms, of evaluative terms such as “stupid,” “evil,” “crazy,” etc). These differences may also reflect specific cultural characteristics of the structure of Greek personality. For example, the term “goodhearted” in the Greek language is closely related to the term “amusing,” and the term “pessimistic” is related to the term “loner.” The current study tested the psychometric properties of the GPAC, a newly developed instrument designed to measure the six-factor structure derived from the Greek personality lexicon (Saucier et al., 2005). More specifically, the objectives of the present study were fourfold: (1) to examine whether the six-factor solution could be replicated with a new sample via the use of exploratory factor analysis (EFA). We hypothesized that all GPAC items would load on the indented factor according to the theoretical framework used (i.e., the six-factor solution). (2) To further establish this six-factor model via the use of confirmatory factor analysis (CFA) by comparing it to alternative models (one-, two-, three-, four-, and five-factor models). We hypothesized that the six-factor model would be superior to the alternative models tested. (3) To examine and report the reliability of the GPAC. Using two different reliability indices (i.e., test-retest and Cronbach’s α coefficients), we evaluated whether the GPAC is a reliable measure of personality. (4) To provide evidence regarding the convergent and discriminant validity of the GPAC subscales. Because of the relative conceptual resemblance among the GPAC scales and the typical Big Five measures, we hypothesized that all GPAC subscales would correlate moderately with their counterparts from the classic Big Five measures, such as the 50 Big Five Factor Markers (BFFM; Goldberg, 1999), the Big Five Inventory (BFI; Benet-Martinez & John, 1998), and the Traits Personality Questionnaire (TPQue5; Tsaousis & Kerpelis, 2004), as well as with scales from other theoretical frameworks, such as the Eysenckian model (Eysenck & Eysenck, 1975).
Method Participants and Procedure Three different samples were used in this study. The first sample consisted of 595 students (37.5% male) with a mean age of 28.29 years (SD = 10.51). All participants filled out the GPAC as partial fulfilment of a research participation option offered through a course in psychometrics. This sample was used to investigate the factor structure of the © 2009 Hogrefe & Huber Publishers
165
GPAC. The second sample consisted of 84 students (72.5% female) with a mean age 30.73 years (SD = 9.70) who voluntarily participated in this study. This sample was used to examine the test-retest reliability of the GPAC, by having participants complete the GPAC twice, with an interval of 4 weeks between administrations. The third sample included 307 students (35.2% male) who voluntarily participated in exchange for course credits in three psychology courses. The mean age for this sample was 25.87 (SD = 8.90). This sample was used to examine the convergent and discriminant validity of the GPAC. All participants were asked to complete the GPAC, as well as a booklet containing four additional personality measures during a 3-hour psychology course, with breaks in between administrations.
Measures The Greek Personality Adjective Checklist (GPAC) The initial version of the GPAC consisted of 96 adjectives that were derived from the Greek lexical study (Saucier et al., 2005). The GPAC has six scales that correspond to the six principal personality dimensions derived from that study. The six scales are labeled Negative Valence/Honesty, Agreeableness/Positive Affect, Prowess/Heroism, Introversion/Melancholia, Even Temper, and Conscientiousness. The GPAC uses a 5-point response scale for the degree to which each adjective is representative (or not) of the personality of the person responding (1 not representative at all and 5 very representative). Apart from the GPAC, the following measures were used to test the convergent and discriminant validity of the GPAC scales:
Traits Personality Questionnaire 5 (TPQue5; Tsaousis & Kerpelis, 2004) The TPQue5 is a measure of the Big Five model based on Costa & McCrae’s theoretical framework, which has been developed and validated specifically for use with Greek adults. It is a short version (101 items) of the TPQue (Tsaousis, 1999) comprising scales of Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness as well as a lie and a social desirability scale. The TPQue5 has shown acceptable reliability and validity data (Tsaousis & Kerpelis, 2004).
The Big Five Inventory (BFI; Benet-Martinez & John, 1998) The BFI is a 44-item self-report measure of personality based on the five-factor model. The BFI has been used effectively across cultures and languages (Benet-Martinez & John, 1998). It was adapted in the Greek language for the European Journal of Psychological Assessment 2009; Vol. 25(3):164–174
166
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
purposes of a previous study, and showed acceptable psychometric properties (Tsaousis & Kerpelis, 2004).
50 Big Five Factor Markers (BFFM; Goldberg, 1999) The 50 BFFM is a broad-bandwidth personality inventory that contains items to measure the higher-order domains of the Big Five model. It has been derived from the International Personality Item Pool (IPIP) scientific collaboratory, and consists of 50 items. The BFFM possesses good psychometric properties for the Greek population (Vakola, Tsaousis, & Georgiades, 2006).
The Eysenck Personality Questionnaire (EPQ; Eysenck & Eysenck, 1975) This is an 84-item questionnaire that is based on the theoretical model proposed by Eysenck (1975). The EPQ measures three basic personality dimensions: Psychoticism (P), Extraversion (E), and Neuroticism (N), plus a lie or social desirability responding (L). It has been standardized into Greek (Demetriou, 1986).
a personality context (e.g., we do not expect people who demonstrate prowess in heroism to be described as cowards). At the end of this study, the total number of items composing the GPAC was 94. Following Digman’s (1997) suggestion for higher-order examination of the factor solutions derived from lexical studies, the relationships between the six GPAC dimensions were also examined by subjecting the extracted factors to a second-order factor analysis. This analysis revealed two distinct factors that explained approximately 60% of the total variance. The first factor, labeled Social Propriety, explained 40.03% of the total variance and consists of the scales Even Temper, Negative Valence/Honesty, and Conscientiousness. The second factor, labeled “Dynamism,” explains 19.72% of the total variance and consists of the scales Prowess/Heroism, Agreeableness/Positive Affect, and Introversion/Melancholia. These results are consistent with findings from other theoretical and empirical studies within the lexical tradition (Di Blas & Forzi, 1999; Hogan, 1983; Paulhus & John, 1998), which suggest two distinct higher-order dimensions: “getting ahead” (Dynamism) and “getting along” (Social Propriety).
Confirmatory Factor Analysis
In order to investigate the robustness of the factor structure of the GPAC, the data were subjected to maximum-likelihood factor analysis (FA) with oblique direct oblimin rotation. We preferred the use of maximum-likelihood method instead of the most commonly known principal components analysis, since the purpose of the former is to understand the latent factors that account for the shared variance among items, and as a result, it is more closely aligned with the development of new scales (Worthington & Whittaker, 2006). Since we already knew from the Greek lexical study that the six factors were intercorrelated (Saucier et al., 2005), we chose to use the oblique rotation method. Visual examination of the scree plot suggested a six-factor solution. The corresponding eigenvalues for the first eight factors were 19.49, 8.61, 6.20, 4.30, 3.34, 2.75, 1.84, and 1.55, and the total variance explained by the Greek solution (obtained before rotation) was 48%. Results from the factor analysis are presented in Table 1. As can be seen, most of the items had their highest loading on the targeted factor. Only two items did not load on the factor expected according to the theory (Items 19 cunning, and 92 carefree) and, thus, were excluded from the final version of the GPAC. Three items presented high secondary loadings (> .40). Item 52, human, had a large negative loading on Negative Valence (–.45), Item 22, useless, had a positive loading on Conscientiousness (.44), and Item 75, coward, had a large negative loading on Prowess/Heroism (–.43). These secondary loading were meaningful and justifiable within
To investigate the adequacy of fit of the six-factor solution previously defined via EFA, we ran a CFA. To examine how closely the suggested model fit the data, we calculated several fit indices. It is suggested that it is necessary to use at least four fit indices to build an overall understanding of fit to the measurement model (Griffin, 2005); model fit is a multifaceted concept and no fit indices in isolation should be considered. In this study, five fit indices were examined for each model: (a) the chi-square/degrees of freedom (χ2/df) and the goodness-of-fit index (GFI), which can be classified as absolute goodness-of-fit indices, (b) the comparative fit index (CFI) and the Tucker-Lewis index (TLI), which can be classified as incremental or comparative indices of fit, and (c) the Root Mean Square Error of Approximation index (RMSEA). A χ2/df value less than 3.0; GFI, CFI, TLI values higher than 0.90; and a RMSEA value less than .05 indicate a model with a good fit (Hu & Bentler, 1999). Moreover, for comparisons of alternative models the χ2 difference test and the Akaike Information Criterion (AIC; Akaike, 1987) were used. The units of analysis used in this study were not the original items included in the scale; instead we preferred to use parcels of items by combining items into small groups within the GPAC scales. Compared with item-level data, models based on parcelled data (a) are more parsimonious (i.e., have fewer estimated parameters both locally in defining a construct and globally in representing an entire model), (b) have fewer chances for residuals to be correlated or dual loadings to emerge (both because fewer indicators are used and because unique variances are smaller), and (c) lead to reductions in various sources of sampling error (MacCallum, Widaman, Zhang, & Hong 1999). Furthermore, one of the principle
European Journal of Psychological Assessment 2009; Vol. 25(3):164–174
© 2009 Hogrefe & Huber Publishers
Results
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
167
Table 1. Factor structure of the GPAC 6-factor model (N = 595) Item no.
Description
A/PA
NV/H
C
I/M
P/H
36
Big-hearted
.75
–.33
–.13
ET .08
–.01
.27
47
Pleasant
.74
–.24
–.11
.00
–.32
.32
44
Cheerful
.74
–.21
–.01
.07
–.24
.26
74
Warm
.73
–.31
–.19
.06
–.15
.31
91
Cordial
.73
–.35
–.15
.08
–.07
.29
26
Kind-hearted
.72
–.31
–.09
.05
.07
.16
35
Sweet
.70
–.26
–.11
.09
–.14
.24
66
Adorable
.70
–.15
–.10
.04
–.22
.35
29
Big-hearted
.69
–.30
–.12
.06
.03
.27
33
Open-hearted
.69
–.27
–.07
.01
–.10
.17
18
Lovable
.69
–.21
–.12
.04
–.23
.28
28
Friendly
.68
–.32
–.09
.01
–.17
.23
79
Cheerful
.67
–.22
–.14
.05
–.35
.26
14
Kind
.66
–.39
–.18
–.01
.07
.19
21
Good-natured
.66
–.24
–.08
.10
–.08
.16
41
Likable
.66
–.24
–.13
.01
–.19
.26
51
Unselfish
.65
–.26
–.16
.02
–.06
.35
38
Affectionate
.63
–.32
–.14
.05
–.02
.15
2
Hospitable
.61
–.34
–.13
.05
–.15
.17
52
Human
.60
–.45
–.20
.04
.04
.14
40
Tender
.60
–.30
–.16
.04
–.03
.14
11
Amusing
.60
–.10
–.10
–.15
–.36
.34
69
Smiling
.60
–.21
–.10
.09
–.34
.18
76
Accommodating
.59
–.36
–.24
.08
.06
.16
10
Cute
.59
–.16
–.04
–.03
–.12
.22
88
Sympathetic
.59
–.36
–.18
.13
.08
.09
49
Demonstrative
.46
–.08
–.05
–.19
–.25
.24
20
Cold
–.41
–.38
–.11
.17
–.33
–.03
65
Immoral
.30
.77
.26
–.19
–.02
.07
54
Abject
.31
.75
.29
–.07
.12
–.06
50
Corrupt
.28
.73
.30
–.17
.02
.04
17
Vulgar
.17
.70
.20
–.23
.02
.06
9
Disgusting
.22
.70
.18
–.17
.05
–.03
59
Immodest
.23
.69
.24
–.17
.01
–.03
32
Paranoid
.28
.68
.31
–.25
.14
–.01
5
Perverse
.18
.67
.21
–.24
–.02
.10
48
Dishonorable
.29
.67
.25
–.13
.15
–.08
68
Unpleasant
.36
.67
.25
–.07
.14
–.10
15
Merciless
.33
.67
.18
–.10
.12
.02
34
Gross
.17
.66
.30
–.29
.01
.08
63
Inhuman
.34
.65
.13
–.13
.04
.05
30
Unacceptable
.31
.65
.39
–.15
.24
–.09
37
Calumniator
.31
.64
.18
–.09
.16
–.12
13
Dishonest
.31
.60
.20
.00
.08
–.05
23
Violent
.22
.60
.23
–.37
.02
.13
16
Pathetic
.28
.59
.24
.04
.32
–.11
31
Intrigueful
.23
.58
.18
–.30
.09
.07
22
Useless
.33
.54
.44
.04
.27
–.15
© 2009 Hogrefe & Huber Publishers
European Journal of Psychological Assessment 2009; Vol. 25(3):164–174
168
Item no. 1
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
Description
A/PA
NV/H
C
ET
I/M
P/H
Barbarous
.19
.53
.22
–.31
–.03
.14
25
Stupid
.23
.46
.29
.05
.28
–.28
87
Loser
.30
.43
.35
.03
.34
–.26
90
Unscheduled
.09
.28
–.84
–.15
.16
–.08
89
Neglectful
.17
.36
–.78
–.20
.12
–.13
24
Disorganized
.04
.22
–.78
–.11
.18
–.11
95
Organized
.16
.14
.78
–.08
.08
–.27
45
Untidy
.03
.27
–.76
–.16
.17
–.04
93
Inconsistent
.15
.31
–.73
–.16
.08
–.07
46
Lazy
.20
.42
–.67
–.23
.18
–.20
43
Absent-minded
–.01
.17
–.61
–.20
.30
–.15
96
Hard-working
.30
.25
.57
–.09
.12
–.34
78
Nervous
.05
.16
.14
–.80
.08
–.04
42
Touchy
.02
.10
.16
–.78
.07
.03
8
Touchy
.08
.16
.07
–.78
.06
.06
81
Abrupt
.12
.24
.20
–.71
.16
.04
80
Reactive
.00
.19
.20
–.68
.08
.07
77
Aggressive
.19
.35
.07
–.65
.02
.13
85
Brawling
.14
.33
.11
–.62
–.06
.09
83
Neurotic
.18
.35
.21
–.59
.20
.01
84
Inpatient
–.08
.07
.25
–.52
.15
.01
82
Fretful
.10
.12
.24
–.48
.31
–.09
86
Calm
.12
.04
.14
.35
–.05
–.25
70
Melancholic
.05
.15
.19
–.22
.73
–.12
67
Depressive
.17
.24
.26
–.22
.72
–.14
72
Moody
.31
.29
.28
–.19
.72
–.25
71
Pessimistic
.20
.15
.29
–.13
.70
–.31
73
Loner
.30
.18
.11
–.14
.65
–.09
94
Sad
.23
.31
.26
–.22
.65
–.19
75
Cowardly
.16
.15
.26
.07
.56
–.43
39
Hesitant
.00
–.01
.24
.05
.55
–.25
7
Insecure
–.01
.02
.20
–.08
.48
–.25
27
Taciturn
.26
.03
.02
.16
.44
–.07
62
Genius
–.27
.01
–.10
.01
–.13
.75
56
Ingenious
–.28
–.01
–.16
.02
.19
.74
58
Pioneering
–.28
.02
–.18
–.04
–.22
.74
60
Inventive
–.22
–.06
–.13
.02
–.11
.72
53
Inventive
–.19
–.03
–.13
–.02
–.09
.72 .70
4
Multi-talented
–.26
.01
–.08
–.08
.16
Talented
–.34
–.05
–.07
–.10
–.15
.67
6
Genius
–.22
.12
–.04
–.05
–.16
.63
57
Fighter
–.30
.07
–.29
–.22
–.32
.61 .58
12
3
Dynamic
–.29
–.08
–.30
–.22
–.39
55
Successful
–.32
–.11
–.35
.09
–.22
.57
64
Agonistic
–.23
–.09
–.24
–.20
–.21
.52
61
Autonomous
–.13
–.02
–.18
–.04
–.09
.41
Variance Ex20.73 9.16 6.59 4.58 3.56 2.93 plained Note. In cases in which an English word is repeated, two Greek words are best translated into the same English word. Secondary loadings higher than .40 are presented in boldface. A/PA = Agreeableness/Positive Affect, NV/H = Negative Valence/Honesty, C = Conscientiousness, ET = Even Temper, I/M = Introversion/Melancholia, P/H = Prowess/Heroism. European Journal of Psychological Assessment 2009; Vol. 25(3):164–174
© 2009 Hogrefe & Huber Publishers
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
169
parcel 1
e1 1
parcel 2 Agreeableness
e2 1
parcel 3
1
e3 1
parcel 4
e4 1
parcel 5
e5 1
parcel 6
e6 1
parcel 7
e7 1
parcel 8
e8 1
Negative Valence
1
parcel 9
e9 1
parcel 10
e10 1
parcel 11
e11 1
parcel 12
e12 1
parcel 13 Prowess/Heroism
e13 1
parce l14 1
e14 1
parcel 15
e15 1
parcel 16
e16 1
parcel 17 Even Temper
e17 1
parcel 18 1
e18 1
parcel 19
e19 1
parcel 20
e20 1
parcel 21
e21 1
Introversion/Melancholia
parcel 22 1
parcel 23 parcel 24 parcel 25
Conscientiousness
parcel 26 1
e22 1
parcel 27
e23 1 e24 1 e25 1 e26 1 e27 1
parcel 28
e28 1
Figure 1. Hypothesized model of the factorial structure of the Greek Personality Adjective Checklist. Parcel1–parcel28 = parcels of GPAC items. e1–e28 = measurement error for parcels 1–28, respectively. weaknesses of item-level factor analysis may be partially overcome using item parcels. More precisely, item parcels may be used to overcome the inadequacies presented by the assumption that the observed variables are continuously measured, interval-level data (Panter, Swygert, Dahlstrom, & Tanaka, 1997). Besides the above mentioned advantages, this method also faces a potential risk because of the way participants might perceive and, consequently, react to each parcel. © 2009 Hogrefe & Huber Publishers
In particular, it is possible that participants might be inclined to follow their own rules (i.e., react to only part of the parcel) and not the intentions of the investigators. The complete confirmatory six-factor model is presented in Figure 1. Furthermore, we were interested in examining whether the six-factor solution was significantly better than other solutions reported in the Saucier et al. (2005) study. Thus, structural models of one-, two-, three-, four-, five-, and sixEuropean Journal of Psychological Assessment 2009; Vol. 25(3):164–174
170
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
Table 2. Goodness-of-fit indices of the hypothesized latent-factor models of the GPAC Model
χ2
df
χ2/df
GFI
TLI
CFI
RMSEA
AIC
Δχ2
Model I
4192.30
350
11.98
.49
.33
.38
.14
4304.30
.001
Model II
1755.20
349
5.03
.80
.86
.87
.08
1869.20
.001
Model III
2008.75
347
5.79
.77
.84
.85
.09
2126.75
.001
Model IV
1731.11
344
5.03
.81
.86
.88
.08
1855.11
.001
Model V
1339.84
340
3.94
.84
.90
.91
.07
1471.84
.001
Model VI
996.92
335
2.98
.89
.94
.95
.05
1138.92
Note. Model I = one-factor solution, Model II = two-factor solution, Model III = three-factor solution, Model IV = four-factor solution, Model V = five-factor solution, Model VI = six-factor solution, GFI = goodness-of-fit index; TLI = Tucker-Lewis index; CFI = comparative fit index; RMSEA = root-mean-square error approximation; AIC = Akaike’s information criterion.
Table 3. Correlation coefficients of the GPAC scales with various criterion scales Criterion scales
A/PA
NV/H
.95
–.46**
P/H
ET
I/M
C
The GPAC scales (N = 595) Agreeableness/positive affect Negative valence/honesty
.93
Prowess/heroism
.43**
.15**
–.31**
.21*
–.09
–.34**
.28**
–.43**
.89
.02
–.38**
.24**
.87
–.24**
.29**
.85
–.33**
Even temper Introversion/melancholia Conscientiousness
.89
The TPQue5 (N = 307) Agreeableness Neuroticism
.43**
–.29**
.02
.36**
–.13*
.09
–.10
.12*
–.34**
–.34**
.42**
–.14*
Openness to experience
.08
.00
.11
.06
.03**
–.14*
Extraversion
.39**
–.17**
.52**
.00
–.51**
.10
Conscientiousness
.14
–.31**
.26**
.28**
–.18**
.64**
.57**
–.42**
.09
.21**
–.21**
.16**
.06
–.14*
–.36**
.50**
.05
The BFFM (N = 307) Agreeableness Neuroticism
–.07
Intellect
.25**
–.12*
.56**
.00
–.20**
.09
Extraversion
.36**
–.07
.40**
.05
–.56**
.02
Conscientiousness
.13*
–.31**
.16**
.23**
–.09
.67**
.49**
–.48**
.05
.40**
–.23**
.05
–.21**
–.45**
.36**
The BFI (N = 307) Agreeableness Neuroticism
–.11
.26** –.11
Openness to experience
.18**
.01
.55**
.04
–.13*
–.09
Extraversion
.32**
–.06
.43**
–.17**
–.55**
.00
Conscientiousness
.15*
–.34**
.26**
.21**
–.23**
.68**
.39**
–.12*
.40**
–.55**
.05
The EPQ (N = 307) Extraversion
–.01
Neuroticism
–.11
.15**
–.22**
–.35**
.54**
–.21**
Psychoticism
–.36**
.51**
.04
–.36**
.17**
–.37**
Note. Values in the main diagonal of the upper part of the table represent the α reliabilities of the GPAC scales. TPQue5 = Traits Personality Questionnaire 5, BFI = Big Five Inventory, BFFM = Big Five Factor Markers, EPQ = Eysenck Personality Questionnaire. A/PA = Agreeableness/Positive affect, NV/H = Negative valence/Honesty, C = Conscientiousness, ET = Even Temper, I/M = Introversion/Melancholia, P/H = Prowess/Heroism. *p < .05, **p < .01. European Journal of Psychological Assessment 2009; Vol. 25(3):164–174
© 2009 Hogrefe & Huber Publishers
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
factor solutions were defined and tested against each other. Particularly, Model I specifies a single-factor model, one latent variable containing all 94 items, and was labeled Evaluation. Model II specifies two correlated factors labeled Morality/Social Propriety and Dynamism. Model III specifies three factors; the two from Model II plus one more, labeled Sensitivity. Model IV specifies four factors labeled Morality/Social Propriety, Sensitivity, Positive Affect/Sociability, and Prowess/Heroism. Model V specifies five factors labeled Negative Valence/Honesty, Even Temper, Introversion/Melancholia, Agreeableness/Positive Affect, and Prowess/Heroism. Finally, Model VI specifies six factors; the five from Model V plus one more, labeled Conscientiousness. Table 2 presents the fit statistics for the sequence of hierarchically nested models for the GPAC. Results in Table 2 suggest that the six-factor solution represents an adequate explanation of the data; χ2 (335, N = 595) = 996.92, χ2/df = 2.98, GFI = 0.89, CFI = 0.95, TLI = 0.94, RMSEA = 0.05). By examining the fit indices of the alternative models, one could conclude that the six-factor model describes the data better, since its values are closer to the values that declare adequate fit. Furthermore, the difference among the χ2 values for Model VI and all the other models (all are significant at a 0.01 level, in favor of Model VI) as well as the AIC fit index (AIC = 1138.92), provide further support for the superiority of Model VI over all other models.
Reliability Cronbach’s α and test-retest indices were used as reliability estimates of the GPAC scales. The internal consistency estimates ranged between .85 (Introversion/Melancholia) and .95 (Agreeableness/Positive Affect), with a mean α reliability of .90. α reliabilities for each GPAC scale can be found in the main diagonal of the upper part of Table 3. Similarly, all test-retest reliability coefficients were above the minimum requirement value of 0.70 (Nunnally & Bernstein, 1994). In particular, the test-retest estimates of Negative Valence/Honesty was .74, Agreeableness/Positive Affect .71, Prowess/Heroism .84, Introversion/Melancholia .78, Even Temper .77, and Conscientiousness .85, with a mean test-retest reliability of .78.
Convergent and Discriminant Validity The intercorrelations among the scales of the GPAC are presented in Table 3 (upper top). As can be seen, most of the GPAC scales were moderately intercorrelated, a result that demonstrates that the six personality dimensions are not independent of each other. The other part of Table 3 presents the correlations between the GPAC scales and the corresponding scales from other personality measures. It was hypothesized that all GPAC subscales would correlate at a moderate-to-high level with similar constructs from the © 2009 Hogrefe & Huber Publishers
171
lexical tradition, such as the Big Five, as well as from other theoretical frameworks, such as the Eysenckian model. Particularly, we predicted the following: (a) moderately high convergent correlations between the GPAC scales of Agreeableness and Conscientiousness and their Big Five counterparts as well as the Eysenckian Psychoticism subscale (Eysenck, 1992); (b) convergent correlations between GPAC Prowess/Heroism and Big Five Intellect/Openness to Experience; (c) the GPAC Introversion/Melancholia scale would correlate negatively with the corresponding Big Five and Eysenckian Extraversion factor; (d) convergent correlations among GPAC Even Temper and the Big Five and Eysenckian Neuroticism dimension; and (e) the GPAC Negative Valence/Honesty subscale would be negatively correlated with the Big Five Agreeableness dimension and positively correlated with the Eysenckian Psychoticism Factor. As can be seen from the results reported in Table 3, GPAC Agreeableness/Positive Affect was positively correlated with its counterparts from the TPQue5 (.43), BFFM (.57), and BFI (.49). It was also positively correlated with the Extraversion subscale of the above instruments (.39, .36, and .32, respectively) as well as with its EPQ counterpart (.39). Finally, it was negatively correlated with the EPQ Psychoticism scale (–.36). Furthermore, the GPAC Conscientiousness subscale was positively correlated with its TPQue5 (.64), BFFM (.67), and BFI (.68) counterparts, and negatively with the EPQ Psychoticism subscale (–.37). Additionally, the GPAC Introversion/Melancholia subscale was negatively correlated with the TPQue5, BFFM, BFI, and EPQ Extraversion subscales (–.51, –.56, –.55, and .55, respectively) as well as with their Neuroticism counterparts (.42, .50, .36, and .54, respectively, because of the Melancholia component of the scale. From the results presented in Table 3 it can also be seen that the GPAC Even Temper subscale was negatively correlated with the TPQue5 (–.34), BFFM (–.36), BFI (–.45), and EPQ (–.35) Neuroticism subscales. It was also positively correlated with the Agreeableness subscales of the three Big Five instruments (.36, .21, and .40, respectively), and negatively correlated with the EPQ Psychoticism subscale (–.36). Table 3 also reports positive correlations for the GPAC Prowess/Heroism subscale and BFFM Intellect scale (.56) and BFI Openness to Experience (.55). However, there was no relationship between the GPAC Prowess/Heroism subscale and the TPQue5 Openness to Experience scale. The GPAC Prowess/Heroism was also positively correlated with the Extraversion subscales of the TPQue5 (.52), BFFM (.40), BFI (.43), and EPQ (.40) because of the Heroism component of the scale. Finally, the GPAC Negative Valence subscale was negatively correlated with the TPQue5 (–.29), BFFM (–.42), and BFI (–.48) Agreeableness subscales, as well as with the TPQue5 (–.31), BFFM (–.31), and BFI (–.34) Conscientiousness subscales. The GPAC Negative Valence was also positively correlated with EPQ Psychoticism subscale (.51). Overall, the above findings support both convergent and European Journal of Psychological Assessment 2009; Vol. 25(3):164–174
172
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
discriminant validity for the GPAC scales. Where the first is corroborated by high correlations with scales measuring similar traits, the second is corroborated by the absence of high correlations with dimensions tapping different aspects of personality. Indeed, none of the GPAC scales showed correlations higher than .29 with constructs that, according to the theory, were not expected to be correlated with.
The aim of this study was to evaluate the psychometric properties of the GPAC, a new instrument developed based on the lexical hypothesis tradition. More specifically, the GPAC was developed based on the theoretical framework derived from the emic taxonomic study of the Greek personality lexicon (Saucier et al., 2005). The GPAC consists of 94 adjectives measuring six concrete dimensions: Even Temper, Introversion/Melancholia, Prowess/Heroism, Agreeableness/Positive Affect, Conscientiousness, and Negative Valence/Honesty. Results from EFA replicated the factor structure obtained from the Saucier et al. (2005) study. Indeed, a clear six-factor solution, explaining 48% of the total variance, emerged, with all items having their largest loadings on the intended factors. Similarly, results from CFA further established the latent six-factor structure of the GPAC (all goodness-of-fit indices used reflected a good fit for the data). Moreover, this solution was found to be superior to other alternative models (one, two, three, four, and five factors). It should also be noted that the twofactor model (i.e., Dynamism vs. Social Propriety), proposed by Saucier et al. (2005) as an alternative model of personality, did not provide a satisfactory account for the current data. All GPAC scales demonstrated high internal consistency, indicating that they are quite homogeneous in their measurement. Results from this study also support the temporal stability of the GPAC as test-retest data (over a 4-week period) suggest that it can be rated reliably. Finally, evidence provides support for good convergent and discriminant validity of all GPAC scales, suggesting that the checklist captures a fairly broad range of personality-related constructs. As expected, all GPAC scales showed a moderately high pattern of correlations with similar constructs from the trait-lexicon tradition (i.e., Big Five). Specifically, there were moderately high convergent correlations between the GPAC scales of Agreeableness and Conscientiousness with their Big Five counterparts, providing support for the resemblance of these constructs. Similarly, the convergent correlations between GPAC Prowess/Heroism and Big Five Intellect/Openness to Experience were moderately high, suggesting a considerable overlap between the two constructs. The GPAC Introversion/Melancholia scale showed moderately high negative correlation with the corresponding Big Five and Eysenckian Extraversion factor, suggesting that this construct lies at the opposite pole of the
classic Extraversion factor. In addition, the correlations involving the GPAC Even Temper with the Big Five and Eysenckian Neuroticism dimension were moderate, suggesting a similar but not perfectly overlapping definition of these constructs. Interestingly enough, results regarding the GPAC Negative Valence/Honesty scale were not very clear, since neither of the Big Five nor Eysenck’s measures that were used contained a direct counterpart of this construct. In particular, we found a moderate negative correlation with the Big Five Agreeableness and Conscientiousness dimensions, and a moderately high positive correlation with Eysenck’s Psychoticism. Significant negative but low correlations were found between Negative Valence and Extraversion, positive but low with Neuroticism, and no relationship with Openness to Experience. These results are in line with those reported by some previous studies. Particularly, Simms (2007), using Tellegen, Grove, and Waller’s (1991) Inventory of Personal Characteristics (IPC-7), an instrument that measures their Big-Seven model, found that Negative Valence is negatively and moderately correlated with both Agreeableness (–.43) and Conscientiousness (–.31), negatively but low with Extraversion (–.27) and Openness to Experience (–.17), and positively but low with Neuroticism (.19). In another study, Church, Katigbak, and Reyes (1998) presented results from two independent studies and also noted moderate negative correlations between Negative Valence and Agreeableness (–.24 and –.42, respectively), negative but low correlations with Conscientiousness (–.15 and –0.09, respectively), Extraversion (–.18 and –.24, respectively), and Neuroticism (–.15 and –.10), and moderate negative correlation with Openness to Experience (–.43 and –.32, respectively).On the other hand, there are studies that report no relationship of this construct with the above mentioned Big Five dimensions except of a moderate negative correlation (–.25) with Neuroticism (Katigbak, Church, Guanzon-Lapena, Carlota, & del Pilar, 2002). Another issue that deserves to be discussed is the incremental validity of the GPAC as a new personality measure in the Greek culture. The question that needs to be answered is “Why would someone use the GPAC instead of other established lexical personality measures currently available in the Greek language (e.g., TPQue, BFI, HEXACO, etc.)?” The fact that the GPAC was developed based on a theoretical and procedural framework derived from the emic (indigenous) study of the Greek personality structure makes it more appropriate in identifying culture-specific aspects of a construct (i.e., it is ecologically valid). All other personality measures currently available in the Greek language were developed using the etic (i.e., imported) approach, which may distort the meaning of constructs in some cultures or overlook their culture-specific (emic) aspects (Benet-Martinez, 2007). Furthermore, as demonstrated in the current study, the GPAC has sound psychometric properties. One possible limitation of the GPAC is the lack of re-
European Journal of Psychological Assessment 2009; Vol. 25(3):164–174
© 2009 Hogrefe & Huber Publishers
Discussion
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
verse-keyed items in its scales. Although this was not done intentionally, we believe it might reflect a certain natural bias evident in the original item pool of the Greek personality taxonomy (see Saucier et al., 2005). Another possible explanation is that negative descriptors are captured by the distinct Negative Valence dimension from the Greek taxonomy. As a result, we see a reduced representation of this kind of term in the remaining GPAC scales. At the same time, Benet-Martinez and Waller (2002) note that the predominance of negative, or in the case of the present study, positive descriptors in a sample is potentially informative about specific sociocognitive processes involved in personality perception. It is possible that these processes differ across languages/cultures, therefore, we propose that future research should compare the taxonomic representation of positive vs. negative descriptors across cultures and/or languages. In this paper we provided empirical evidence for the sound psychometric properties of the GPAC, a new instrument developed based on the lexical hypothesis tradition. More specifically, our results demonstrate that the GPAC has a clear six-factor structure, strong reliability indices, and sufficient construct validity. The GPAC, which was developed using an emic, culture-specific approach, is a promising and useful measure for the assessment of normal personality in the Greek language.
References Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317–332. Benet-Martinez, V. (2007). Cross-cultural personality research: Conceptual and methodological issues. In R.W. Robins, R.C. Fraley, & R. Krueger (Eds.), Handbook of research methods in personality psychology (pp. 170–189). New York: Guilford Press. Benet-Martinez, V., & John, O.P. (1998). Los Cinco Grandes across cultures and ethnic groups: Multitrait-multimethod analyses of the Big Five in Spanish and English. Journal of Personality and Social Psychology, 75, 729–750. Benet-Martinez, V., & Waller, N.G. (2002). From adorable to worthless: Implicit and self-report structure of highly evaluative personality descriptors. European Journal of Personality, 16, 1–41. Brokken, F.B. (1978). The language of personality. Meppel, The Netherlands: Krips. Church, A.T., Katigbak, M.S., & Reyes, J.A.S. (1998). Further exploration of Filipino personality structure using the lexical approach: Do the Big-Five or Big-Seven dimensions emerge? European Journal of Personality, 12, 249–269. Costa Jr., P.T., & McCrae, R.R. (1992). NEO PI-R. Professional manual. Odessa, FL: Psychological Assessment Resources. Demitriou, E.C. (1986). To Erwtvmatolígio Proswpikítvta™ EPQ (Eysenck Personality Questionnaire): Stâ®misv ston Ellvnikí Plv®usmí, enŒliko kai paidikí [The Eysenck Personality Questionnaire (EPQ): The validity of the Greek adult and junior version]. Egkìfalo™, 23, 41–54. © 2009 Hogrefe & Huber Publishers
173
Di Blas, L., & Forzi, M. (1998). An alternative taxonomic study of personality-descriptive adjectives in the Italian language. European Journal of Personality, 12, 75–101. Digman, J.M. (1997). Higher-order factors of the Big Five. Journal of Personality and Social Psychology, 73, 1246–1256. Eysenck, H.J. (1992). Four ways five factors are not basic. Personality and Individual Differences, 13, 667–673. Eysenck, H.J., & Eysenck, S.B.G. (1975). Manual of the Eysenck Personality Questionnaire. London: Hodder & Stoughton. Goldberg, L.R. (1990). An alternative “description of personality”: The Big-Five factor structure. Journal of Personality and Social Psychology, 59, 1216–1229. Goldberg, L.R. (1999). A broad-bandwidth, public-domain, personality inventory measuring the lower-level factors of several five-factor models. In I. Mervielde, I. Deary, F. De Fruyt, & F. Ostendorf (Eds.), Personality psychology in Europe (Vol. 7, pp. 7–28). Tilburg, The Netherlands: Tilburg University Press. Griffin, P. (2005). Developing a measure of wealth for primary student families in a developing country: Comparison of two methods of psychometric calibration. Studies in Educational Evaluation, 31, 192–206. Hogan, R. (1983). A socioanalytic theory of personality. In M.M. Page (Ed.), Nebraska Symposium on Motivation (pp. 336–355). Lincoln: University of Nebraska Press. Hu, L., & Bentler, P.M. (1999). Cut-off criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modelling, 6, 1–55. Katigbak, M.S., Church, T., Guanzon-Lapena, M.A., Carlota, J.A., & del Pilar, G.H. (2002). Are indigenous personality dimensions culture specific? Philippine inventories and the fivefactor model. Journal of Personality and Social Psychology, 82, 89–101. MacCallum, R.C., Widaman, K.F., Zhang, S., & Hong, S. (1999). Sample size in factor analysis. Psychological Methods, 4, 84–99. Nunnally, J.C., & Bernstein, I.H. (1994). Psychometric Theory (3rd ed.). New York: McGraw-Hill. Ostendorf, F. (1990). Sprache und Personlichkeitsstruktur: zur Validität des Fünf-Faktoren-Modells der Persönlichkeit [Language and personality structure: Toward the validity of the five-factor model of personality]. Regensburg: S. Roderer. Panter, A.T., Swygert, K.A., Dahlstrom, W.G., & Tanaka, J.S. (1997). Factor-analytic approaches to personality item-level data. Journal of Personality Assessment, 68, 561–589. Paulhus, D.L., & John, O.P. (1998). Egoistic and moralistic biases in self-perception: The interplay of self-descriptive styles with basic traits and motives. Journal of Personality, 66, 1025–1060. Saucier, G., Georgiades, S., Tsaousis, I., & Goldberg, L.R. (2005). The factor structure of Greek personality adjectives. Journal of Personality and Social Psychology, 88, 856–875. Saucier, G., & Goldberg, L.R. (2001). Lexical studies of indigenous personality factors: Premises, products, and prospects. Journal of Personality, 69, 847–879. Simms, L.J. (2007). The Big Seven model of personality and its relevance to personality pathology. Journal of Personality, 75, 65–94. Tellegen, A., Grove, W., & Waller, N.G. (1991). Inventory of Personal Characteristics #7. Unpublished measure, University of Minnesota. Tellegen, A., & Waller, N.G. (1987, August). Reexamining basic dimensions of natural language trait descriptors. Paper preEuropean Journal of Psychological Assessment 2009; Vol. 25(3):164–174
174
I. Tsaousis & S. Georgiades: The Greek Personality Adjective Checklist
sented at the 95th Annual Convention of the American Psychological Association, New York. Tsaousis, I. (1999). The Trait Personality Questionnaire: A Greek measure for the five-factor model. Personality and Individual Differences, 26, 262–274. Tsaousis, I., & Kerpelis, P. (2004). The Traits Personality Questionnaire 5 (TPQue5): Psychometric properties of a shortened version of a Big Five measure. European Journal of Psychological Assessment, 20, 180–191. Vakola, M., Tsaousis, I., & Georgiades, S. (2006). The psychometric properties of the BFFM questionnaire with a Greek sample. Unpublished manuscript, Athens University of Economics and Business. Worthington, R.L., & Whittaker, T.A. (2006). Scale development
European Journal of Psychological Assessment 2009; Vol. 25(3):164–174
research: A content analysis and recommendations for best practices. The Counseling Psychologists, 34, 806–838.
Ioannis Tsaousis University of Crete Department of Psychology Gallos Campus – 74100 Rethymno Greece Greece E-mail
[email protected]
© 2009 Hogrefe & Huber Publishers