ArticlePDF Available

Abstract and Figures

We assessed the factor structure and psychometric properties of the Emotionality, Activity, and Sociability (EAS) Temperament Survey (Buss & Plomin, 1984) for adults using a longitudinal sample of adult women. The stability estimates of the EAS instrument were assessed over a period of 3 years. The results indicated an acceptable fit for the basic theoretical EAS model, implying that the scale is functioning satisfactory. However, the results also suggest that the measure could be improved. Across time, latent stability factors explained within-scale covariances. Both latent stability factors and time-specific factors accounted for cross-sectional covariances between subscales. Additional research is warranted to guide the further development of the EAS model.
Content may be subject to copyright.
This article was downloaded by: [University of Oslo]
On: 13 August 2015, At: 06:28
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place,
London, SW1P 1WG
Journal of Personality Assessment
Publication details, including instructions for authors and subscription information:
Temperament in Adults-Reliability, Stability, and Factor
Structure of the EAS Temperament Survey
Ane Naerde , Espen Roysamb & Kristian Tambs
Published online: 10 Jun 2010.
To cite this article: Ane Naerde , Espen Roysamb & Kristian Tambs (2004) Temperament in Adults-Reliability, Stability,
and Factor Structure of the EAS Temperament Survey, Journal of Personality Assessment, 82:1, 71-79, DOI: 10.1207/
To link to this article:
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
Temperament in Adults—Reliability, Stability,
and Factor Structure of the EAS Temperament Survey
Ane Nærde, Espen Røysamb, and Kristian Tambs
Division of Epidemiology
Norwegian Institute of Public Health
Oslo, Norway
We assessed the factor structure and psychometric properties of the Emotionality, Activity, and
Sociability (EAS) Temperament Survey (Buss & Plomin, 1984) for adults using a longitudinal
sample of adult women. The stability estimates of the EAS instrument were assessed over a pe
riod of 3 years. The results indicated an acceptable fit for the basic theoretical EAS model, im-
plying that the scale is functioning satisfactory. However, the results also suggest that the mea-
sure could be improved. Across time, latent stability factors explained within-scale
covariances. Both latent stability factors and time-specific factors accounted for cross-sectional
covariances between subscales. Additional research is warranted to guide the further develop-
ment of the EAS model.
There is a general agreement that temperament refers to the
stylistic component of an individual’s mental or behavioral
repertoire (Buss & Plomin, 1984). Temperament theories dif-
fer regarding the definition of temperament, the number and
structure of temperament traits, and the importance of spe
cific factors determining individual differences in tempera
ment. Although much research has focused on temperament
in relation to infant behavior, there are fewer studies on tem
perament in adults.
The theory of temperament on which the Emotionality,
Activity, and Sociability (EAS) Temperament Survey (Buss
& Plomin, 1984) is based regards temperament as a subclass
of personality traits characterized by appearance during the
1st year of life, persistence later in life, and a high contribu
tion of heredity (Buss & Plomin, 1984). The three personal
ity traits that meet these criteria are emotionality, activity,
and sociability, from which are derived the acronym EAS
(Buss & Plomin, 1984). Emotionality is defined as primor
dial distress, which is assumed to differentiate into fear and
anger during the first 6 months of life. Activity is defined as
the sheer expenditure of physical energy, and sociability is
defined as a preference for being with others rather than be
ing alone. Emotionality, activity, and sociability are found in
various forms in almost every model of temperament (Rutter,
1989). (For a more detailed description of the components of
the three temperaments as well as the motivational aspects of
each, see Buss, 1989; Buss & Plomin, 1984, 1986). The EAS
theory was introduced by Buss and Plomin in 1975 and origi-
nally included four temperaments—emotionality, activity,
sociability, and impulsivity (EASI). The items relating to
impulsivity were eventually omitted because the evidence
for its inheritance was mixed, and also, this factor did not re
liably emerge from factor analytic studies (Buss & Plomin,
1975; Rowe & Plomin, 1977). Two instruments were origi
nally developed by the use of exploratory factor analysis: a
temperament survey for children (parental ratings) and a
self-report inventory for adults. The instruments have under
gone several revisions (EASI I, EASI II, EASI III). The final
adult EAS version (Buss & Plomin, 1984) is a 20-item scale
measuring each of the following traits: activity, sociability,
and the three subdimensions anger, distress, and fearfulness,
being components of emotionality.
A considerable amount of research has been conducted
with the adult EAS scale and its variants (Angleitner &
Ostendorf, 1994; Digman, 1989, 1990; John, 1989; McCrae
& Costa, 1985; Plomin, Pedersen, McClearn, Nesselroade, &
Bergman, 1988; Ruch, Angleitner, & Strelau, 1991; Windle,
1989a, 1989b). Published studies on the scales’
psychometric properties are surprisingly scarce apart from
those in which the scales were initially derived. We have
only been able to identify one study reporting on the
psychometric properties of the adult EASI–III scale
Copyright © 2004, Lawrence Erlbaum Associates, Inc.
Downloaded by [University of Oslo] at 06:28 13 August 2015
(Braithwaite, Duncan-Jones, Bosly-Craft, & Goodchild,
1984). Thus, for the final EAS version of the adult tempera
ment survey, there is no available psychometric information
other than the original statistics reported by Buss and Plomin
(1984). Moreover, common to all the EASI/EAS versions is
that the data used by Buss and Plomin (1975, 1984) to derive
the scales are based on nonrandom and nonrepresentative
samples consisting of college students. Also, previous stabil
ity estimates of the EAS scale over time are based on test–re
test correlations only, and there is a lack of studies applying
more refined methods such as structural equation programs.
The EAS theory represents one of several theoretical ap
proaches to temperament. We necessarily need to limit our
coverage of alternative models. A previous roundtable dis
cussion of four different approaches to the concept of tem
perament (including the EAS model) illustrates some links
and divergences among different theories (Goldsmith et al.,
The purpose of this study is twofold. First, we assessed the
factor structure and psychometric properties of the EAS
Temperament Survey for adults. Second, we assessed the
stability estimates of the EAS instrument over a period of 3
years. More specifically, the purpose of the study was to (a)
test the fit of the factor structure presented by Buss and
Plomin (1984), including a five-factor and a three-factor so-
lution; (b) estimate the correlations between factors when
controlling for measurement errors; (c) investigate whether
test–retest correlations are due to occasion-specific latent
stability factors or due to transmission of effects from one
time point to another; and (d) explore how interfactor corre-
lations observed at any point in time are due to covariances
between the stable or the time-specific parts of the factors.
The data were derived from a longitudinal epidemiological
survey of mothers and their preschool children to investigate
early childhood behavior problems (Mathiesen & Sanson,
2000). Participants were recruited from a nationwide, manda
tory childpublic health program that includes about 95% of all
families with preschool children in Norway. Most often the
mothers, and ina few cases, thefathersor both parents,accom
pany their child at the health examinations. This study utilizes
only data from the mothers. The participants completed a
questionnaire when their children were 1.5 years, 2.5 years,
and 4 years old. The details concerningthe participation rate at
each time point have been presented previously (Nærde,
Tambs,Mathiesen, Dalgard, & Samuelsen, 2000).Thepartici
pation rate was 87% at Time 1 (T1; n = 939; 1,081 eligible par
ents), 74% at T2 (n = 804; 1,087 eligible parents) and 72% at
T3 (n = 759; 1,059 eligible parents). Altogether, 63% of the
families participated at all three occasions (n = 682).
Demographic information for the nonparticipants was ob
tained from records maintained at the child health clinic
(mothers’ age, number of childreninthe family, marital status,
mothers’ education, and employment status). Attrition analy
ses showed no significant differencesbetweentheparticipants
and the nonparticipants regarding these variables. The age of
the participating mothers ranged from 19 to 46 years, with a
mean age of 30 years(SD = 4.7). Forty-nine percentofthe fam
ilies had only 1 child, 37% had 2, and 15% had 3 to10 children.
Most mothers (91%) were living with a partner.For amore de
tailed account of the demographic characteristics of the sam
ple, see Mathiesen, Tambs, and Dalgard (1999). The
demographic profile of the sample did not vary significantly
across time points. Comparisons on the demographic vari
ables between the participants with complete data (n = 682)
and the participants who dropped out at either T2 or T3 (n =
262) revealed two significant differences between the groups
(Nærde, Tambs, & Mathiesen, 2001). The dropouts were sig
nificantly younger than the participants (28 years vs. 30
years), t(942) = 3.24, p = .001. Also, significantly fewer of the
dropouts were employed compared to the participants (56%
vs. 67%; p = .002). Finally, there were no significant differ
ences between the participants with complete versus incom-
plete data regarding the EAS measures.
Temperament was assessed by the EAS Temperament
Survey for adults (Buss & Plomin, 1984). This is the newest
version of the instrument and has 4 items corresponding to
each of the five subscales. Each of the items was rated on a
Likert scale ranging from 1 (not characteristic or typical of
yourself)to5(very characteristic or typical of yourself). The
scores from the questions belonging to each of the five scales
were added and each scale score was divided by 4 (number of
items per scale). The 20 EAS items are listed in Table 1. The
standard EAS measure was translated from English to Nor
wegian by a British psychologist who had spent several years
working at a Norwegian research center. To assure that the
meaning of the original items were retained, a blind back
translation from Norwegian to English was completed by a
Canadian researcher working at the same center. The
back-translated version had slight differences in meaning for
three of the EAS questions. These differences were discussed
among the two translators and five Norwegian researchers
until consensus was reached regarding Norwegian equiva
lents for the original items.
Statistical Analyses
Structural equation modeling (SEM) was used to evaluate
various aspects of the EAS. In general, the application of
SEM involves several advantages as compared to more tradi
tional multivariate methods. First, SEM enables the simulta
neous test of a number of interrelations between variables.
Downloaded by [University of Oslo] at 06:28 13 August 2015
Second, SEM analyses provide fit measures for an entire
model, in addition to regular path coefficients and multiple
Rs. These fit measures reflect how well the proposed model
recaptures the variance–covariance structure of the data un
der analysis. Thus, a theoretically based model can be tested
against the data. Third, random measurement errors that are
typically comprised in ordinary sum-score indexes can be
eliminated by modeling latent factors (Bollen, 1989; Hoyle,
1995; Jöreskog & Sörbom, 1993; Loehlin, 1998). Due to
skewness in the data, the analyses were based on polychoric
correlations and asymptotic covariance matrices computed
with Prelis (Jöreskog & Sörbom, 1993).
In these analyses, SEM was first used to perform confir
matory factor analyses (CFA) in which the theoretical
five-dimensional EAS model (activity, sociability, anger,
distress, and fearfulness) was tested against data at each of
the three time points. In the next step, we tested for the pres
ence of higher order factors corresponding with the
three-dimensional EAS model (emotionality, activity, and
sociability) and tested whether this model fitted the data
better than the five-dimensional model. We adopted a strat
egy of analysis involving first the testing of the basic theoret
ical model at each of the time points. The basic model
comprised a simple structure, allowing for no cross-loadings
and no correlated errors. Further, to identify covariances not
accounted for by this model and investigate to what extent
the model could be improved, we tested alternative models
by allowing cross-loadings based on the modification in-
dexes provided.
In the next set of analyses, a longitudinal model was ana-
lyzed using SEM and the EQS computer program (Bentler,
This model investigated the nature of the cross-time,
same-trait covariances as well as the same-time, cross-trait
covariances within the EAS. These analyses were based on
the variance–covariance matrices of the sum-score indexes
of each of the five temperament dimensions.
The strategy of
analysis involved testing alternative hypotheses by relaxing
sets of constraints in a stepwise fashion. We elaborate on the
specific constraints in the Results section.
The overall fit of the models was assessed with the good
ness of fit index (GFI), comparative fit index (CFI) and the
root mean square error of approximation (RMSEA). There is
not full consensus about the selection of a standard for deter
mining “adequate” fit. Thus, we chose to use standards com
monly accepted today. An adequate model should have a GFI
or a CFI of at least 0.90 (Hoyle, 1995). Moreover, a RMSEA
index lower than 0.05 indicates a very good fit, whereas one
up to 0.08 indicates a reasonably good fit (Browne &
Cudeck, 1993; Jöreskog & Sörbom, 1993). Chi-square statis
tics were used to compare alternative models and to develop
modifications to the basic models (Bentler, 1995; Bollen,
1989; Hoyle, 1995; Jöreskog & Sörbom, 1993).
Descriptive Statistics for the 20 Items
and the Five Subscales of the EAS
Descriptive statistics for the 20 items of the EAS scale at T1
(means and standard deviations) and polychoric test–retest
correlations are summarized in Table 1. The single item
test–retest correlations pooled across the 20 items were .62
for T1 to T2, .63 for T2 to T3, and .58 for T1 to T3.
Descriptive statistics for the five subscales of the EAS
(means and standard deviations) and Cronbach alpha reli-
ability estimates for the subscales at T1, T2, and T3 are pre-
sented in Table 2. The means for each of the subscales did not
change significantly over time. The Cronbach alpha reliabil-
ity ranged from 0.53 to 0.75. The skewness for the five
subscales at the three time points ranged from –0.76 to 0.51,
whereas the kurtosis ranged from –0.22 to 0.58.
The test–retest correlations between the EAS subscales at
T1, T2, and T3 and pooled Pearson intercorrelations (at T1,
T2, and T3) between the five dimensions are listed in Table
3. The test–retest correlations ranged from .61 to .72 and did
not change significantly over time. The intercorrelations be
tween the temperament dimensions ranged from –.12 to .61.
Factor Structure of the EAS
The basic five-dimensional model yielded an acceptable fit.
Based on the T1 data, χ
(160, N = 921) = 856.70; GFI = 0.96;
CFI=0.94;and RMSEA = 0.069. A similar fitwasobtainedfor
the T2 data, χ
(160, N = 784)= 879.85; GFI =0.95;CFI = 0.94;
RMSEA=0.076.Likewise, the fitatT3wasχ
(160,N =736)=
840.66; GFI = 0.95; CFI = 0.94; and RMSEA = 0.076.
To obtain factor loadings based on all three time points, a
three-group model was tested that included one group for
each point of time and in which all parameters were con
The reason for switching from LISREL to EQS was that inasmuch
as the first set of analyses was based on categorical and skewed data,
the use of polychoric correlations, asymptotic covariance matrices,
and weighted least squares estimation would be more appropriate than
applying ordinary covariance matrices and maximum likelihood esti
mation. Because Lisrel and Prelis are better developed to handle these
situations, we used Lisrel for the first set of analyses and then returned
to our main SEM program, namely EQS, for the remaining analyses
involving sum-score indexes as observed variables.
Regarding the choice of the specific type of association matrix
used, polychoric correlations were used in the first set of analyses in
which the observed variables were categorical and skewed and the
analyses were cross-sectional. Further, in the remaining analyses
based on sum-score indexes as observed variables, we used vari
ance–covariance matrices, which in general comprise more infor
mation than ordinary correlation matrices and typically are prefera
ble in longitudinal analyses.
By pooled we mean the average of the z transforms of the corre
lation values transformed back to the corresponding correlation
Downloaded by [University of Oslo] at 06:28 13 August 2015
strained to be equal across groups. Due to the lack of inde
pendence between observations, the fit statistics for such a
model are not valid; however, the parameter estimates repre
sent the common structure across time. Figure 1 shows factor
loadings and correlations based on the three-group model. As
can be seen, factor loadings ranged from 0.17 to 0.72. With
the exception of two items (i.e., Item 19 on fearfulness and
Item 5 on anger), all loadings were 0.40 or higher. Although
the general picture involves evidence of an acceptable fit and
an acceptable level of factor loadings, we also found indica
tions of less than optimal aspects of the scale.
The factor correlations ranged from .11 to .96 for the
five-dimensional model. In particular, note the high corre
lation between distress and fearfulness, suggesting that
when random measurement error is eliminated, these two
factors are practically representing the same phenomenon.
We tested this notion specifically by constraining the factor
correlation to unity. At T1 and T2 this model did not fit
worse than the original model in which the parameter was
freely estimated, ∆χ
(1, N = 921) = 2.32, ns (T1); ∆χ
(1, N
= 784) = 1.19, ns (T2). At T3, there was a slight but signifi
cant reduction in fit, ∆χ
(1, N = 736) = 6.29, p < .05. In to
tal, these analyses support the notion of the Distress and
Fearfulness scales in the EAS as measures of virtually the
same underlying dimension.
In the next set of analyses, a higher order,
three-dimensional EAS model was tested against the data at
each of the three time points. This model also yielded accept-
able fit. Based on the T1 data, χ
(164, N = 921) = 887.42;
GFI = 0.96; CFI = 0.93; and RMSEA = 0.069. At T2, a simi-
lar fit was obtained, χ
(164, N = 784) = 906.17; GFI = 0.95;
CFI = 0.93; and RMSEA = 0.076. Likewise, the fit at T3 was
(164, N = 736) = 867.58; GFI = 0.95; CFI = 0.94; and
RMSEA = 0.076. The three-dimensional model did, how-
ever, fit significantly worse than the five-dimensional model
at all time points, ∆χ
(4, N = 921) = 30.72, p < .001 (T1);
(4, N = 784) = 26.32, p < .001 (T2); and ∆χ
(4, N = 736) =
26.92, p < .001 (T3). For this reason, we did not perform any
further analyses with the three-dimensional model.
The next step in the confirmatory analyses was to mod
ify the five-dimensional model by relaxing the constraints
of no cross-loadings. This model allowed for the estimation
of the parameters that would yield the highest contribution
to decrease in chi-square. In a step-wise manner, parame
ters were freed one at a time until no significant improve
ments could be found. No restrictions were applied
regarding which factor loaded on which item, and thereby,
all potential cross-loadings could be estimated. These anal
yses were performed on the T1 sample. A total of 16 signif
icant cross-loadings were identified, and the model yielded
good fit, χ
(144, N = 921) = 400.82; GFI = 0.98; CFI =
0.98; RMSEA = 0.044. However, when such modifications
are performed ad hoc, there is a risk of capitalizing on
chance characteristics. To avoid the time-specific charac
teristics and to identify only the cross-loadings that were
present over time, this modified model was then tested on
the T2 data. Four of the 16 cross-loadings from T1 were no
longer significant and were hence omitted. Thereafter, this
model was tested on the T3 data, and another 4
cross-loadings were found to be insignificant. After testing
Descriptive Statistics (T1 Only) and
Polychoric Test–Retest Correlations for the
20 Items of the Emotionality, Activity,
Sociability Temperament Survey
Item M SD
1. I like to be with people (Soc) 4.09 0.80 .71
2. I usually seem to be in a hurry (Act) 3.39 0.79 .63
3. I am easily frightened (E–f) 2.16 1.10 .67
4. I frequently get distressed (E–d) 2.77 0.91 .63
5. When displeased, I let people know
it right away (E–a)
3.56 0.91 .66
6. I am something of a loner (rev., Soc) 1.90 1.08 .70
7. I like to keep busy all the time (Act) 3.16 0.85 .61
8. I am known as hot blooded and
quick-tempered (E–a)
3.01 1.13 .77
9. I often feel frustrated (E–d) 2.67 0.99 .60
10. My life is fast paced (Act) 2.99 1.01 .62
11. Everyday events make me troubled
and fretful (E–d)
2.43 1.05 .53
12. I often feel insecure (E–f) 2.70 0.99 .66
13. There are many things that annoy me
2.99 0.90 .56
14. When I get scared, I panic (E–f) 1.80 1.05 .55
15. I prefer working with others rather
than alone (Soc)
3.47 0.94 .62
16. I get emotionally upset easily (E–d) 2.89 0.99 .58
17. I often feel as if I’m bursting with
energy (Act)
3.24 0.84 .56
18. It takes a lot to make me mad (rev.,
3.12 0.88 .67
19. I have fewer fears than most people
my age (rev., E–f)
2.95 1.01 .39
20. I find people more stimulating than
anything else (Soc)
3.26 0.81 .66
Note. T = time; Soc = sociability; Act = Activity; E–f = Emotionality–
fearfulness; E–d = Emotionality–distress; E–a = Emotionality–anger; rev. =
reversed item.
Pooled values based on estimates from T1 to T2 and T2 to T3.
Descriptive Statistics and Chronbach Alpha
for the Five Emotionality, Activity, Sociability,
Temperament Survey at T1, T2, and T3
Mean SD α
Dimension T1
T2b T3
T1 T2 T3 T1 T2 T3
Distress 2.36 2.33 2.32 0.73 0.72 0.69 .73 .75 .74
Fearfulness 2.25 2.22 2.19 0.63 0.60 0.57 .57 .55 .56
Anger 2.99 2.94 2.92 0.70 0.69 0.66 .57 .62 .58
Activity 3.01 3.10 3.16 0.69 0.69 0.70 .65 .67 .72
Sociability 3.74 3.76 3.73 0.58 0.60 0.62 .53 .61 .65
Note. T = time.
N = 921.
N = 784.
N = 736.
Downloaded by [University of Oslo] at 06:28 13 August 2015
across time, a total of 8 cross-loadings were found to be
significant at all three time points, whereas the remaining
cross-loadings appeared to be due to chance characteristics
specific to single time points. The final five-dimensional
model with the remaining 8 cross-loadings yielded good fit.
Based on the T1 data, χ
(152, N = 921) = 496.15; GFI =
0.98; CFI = 0.97; and RMSEA = 0.050. At T2, a similar fit
was obtained, χ
(152, N = 784) = 448.16; GFI = 0.98; CFI
= 0.98; and RMSEA = 0.050. Likewise, the fit at T3 was
(152, N = 736) = 455.47; GFI = 0.97; CFI = 0.97; and
RMSEA = 0.052.
The factor loadings of the final three-group model ranged
from 0.20 to 0.85, with a total of eight cross-loadings. Partic
ularly noteworthy is Item 13 loading only 0.28 on the original
Anger factor but loading 0.53 on the Fearfulness factor. Item
2 loaded 0.51 on the original activity dimension as well as
0.31 on the sociability dimension, whereas Item 5 loaded
0.44 on the original anger dimension and –0.35 on the fear
dimension. Note also that three items in the Emotionality fac
tor (4, 18, and 19) all load negatively on the Activity factor.
For example, a high score on Item 19 appears to indicate high
distress and low activity rather than a pure representation of
level of distress only.
Longitudinal Analyses
The first model to be tested comprised a latent stability factor
for each temperament. For example, a latent activity factor
was modeled that yielded equally strong influence on the ac
tivity indexes at each time point and which accounted for all
the covariance between the three activity measures. The five
latent stability factors were allowed to correlate, but the re
sidual variances for each measure were constrained to be
uncorrelated. Figure 2 shows the tested model, with the solid
lines indicating the free parameters in the first model.
model yielded acceptable yet not very high fit measures,
Test–Retest Correlations for the Five Emotionality, Activity, Sociability, Temperament Survey (for T1, T2,
and T3) and Pooled Intercorrelations (for T1, T2, and T3) Between the Dimensions
T1 to T2 T2 to T3 T1 to T3
Correlation Matrix
Dimension Fearfulness Anger Activity Sociability
Distress .66 .66 .62 .61 .52
.30 .37 .07 –.08 –.12 –.04
Fearfulness .62 .68 .62 .10 .17 –.11 –.02 –.12 –.06
Anger .73 .76 .71 .28 .05 –.01 –.06
Activity .67 .68 .65 .28 .21
Sociability .66 .71 .65
Note. Intercorrelations from Buss and Plomin (1984) are underlined (n = 220). All correlations were significant at the .01 level (two-tailed). T = time.
FIGURE 1 Confirmatory factor analysis of the five-dimensional Emotionality, Activity, and Sociability Temperament Survey; standardized coeffi-
cients. Soc = sociability; Act = activity; Fear = fearfulness; Ang = anger; and Dis = distress.
Correlations between latents are not shown in the figure due to
visual complexity but are reported in Table 4. Also note that the fig
ure depicts standardized coefficients. In the modeling analyses, the
factor loadings (the three time points on the latent stability factor)
were constrained to be equal in terms of unstandardized values. As
expected, when standardizing, the loadings varied slightly due to
differences in variances. Moreover, whereas a three-indicator factor
model without constraints would only be just identifiable and thus
unable to provide a true test of fit, by constraining loadings to be
equal, overidentified models are tested.
Downloaded by [University of Oslo] at 06:28 13 August 2015
(90, N = 682) = 466.11; GFI = 0.92; CFI = 0.93; and
RMSEA = 0.078.
The first model involved the rather strict assumption of no
direct transmission of effects over time. An alternative model
would include direct effects across time within each temper
ament. We applied the Lagrange Multiplier Test (LM;
Bentler, 1995) to identify potential direct effects. This test
evaluates the effect of adding free parameters to a restricted
model (i.e., reducing the restrictions of the model) and the
LM statistics can be interpreted as an approximate decrease
in model goodness-of-fit chi-square resulting from freeing
previously fixed parameters and from eliminating equality
restrictions (Bentler, 1995). Among the 10 possible direct ef
fects, only one path yielded a significantly lower chi-square;
that is the effect of anger at T1 on anger at T2 (depicted as
dotted arrow in Figure 2). The fit was χ
(89, N = 682) =
451.30; GFI = 0.92; CFI = 0.94; and RMSEA = 0.077. Ex
cept for this single freed parameter, all the cross-time
covariances between the same temperaments appeared to be
accounted for by latent stability factors.
The next step involved investigating cross-trait residual
correlations within each time point. Whereas the first model
assumed that all cross-trait correlations could be accounted
for by correlations between the latent stability factors, we
also wanted to identify potential time-specific correlations
across traits. Again, we used the LM Test to identify which
parameters, if freed, would contribute to significantly better
fit. A total of 10 residual correlations were identified indicat
ing time-specific correlations between Sociability and Activ
ity, and between the three Emotionality dimensions
(depicted as dotted curves in Figure 2). Estimation of these
parameters strongly contributed to improved fit, and this fi
nal model accounted for the observed data structure to a high
extent, χ
(79, N = 682) = 137.53; GFI = 0.97; CFI = 0.99; and
RMSEA = 0.033. See Figure 2 for parameter estimates.
Thus, it appeared that from one time point to another, there
was a clear codevelopment among the measures, for exam
ple, a relative change in fearfulness from T1 to T2 (indicated
by the residual time-specific variance at T2) occurs in paral
lel with a corresponding change in distress.
The correlations between the latent factors are not in
cluded in Figure 2 but are presented in Table 4. The values
vary considerably; of particular note is that the correlation
between the stable fearfulness factor and the stable distress
factor is .73. Using regular path analytic tracing rules, we de-
composed the total fearfulness–distress correlation into
amounts due to stable and time-specific correlations. Aver-
aging across the three time points, this calculation estimated
that 77% of the observed correlations between the fearful-
ness and distress indexes are attributable to the correlation
between the stable fearfulness and distress factors, and the
remaining 23% of the correlations are accounted for by
time-specific correlations.
In summary, the results show that latent stable factors ac-
count for the covariances within each temperament measure
across time. Furthermore, the covariances between the differ
ent temperament indexes were both due to correlations be
tween the latent stability factors and correlations between the
The purpose of this study was to examine the psychometric
properties of the newest version of Buss and Plomin’s (1984)
FIGURE 2 Longitudinal model of the five-dimensional Emotion-
ality, Activity, Sociability Temperament Survey with standardized
parameters. Solid lines indicate basic model; dotted lines show
added parameters contributing to increased fit. χ
(79, N = 682) =
137.53; GFI = 0.97; CFI = 0.99; and RMSEA = 0.033. Rectangular
boxes represent observed sum-score indexes at Time 1 (T1), T2, and
T3; ellipses represent latent stability factors. Soc = sociability; Act =
activity; Fear = fearfulness; Ang = anger; Dis = distress.
Estimated Correlations Between the Latent
Stable Factors of the Emotionality, Activity,
Sociability, Temperament Survey
Factor 1 2 3 4 5
1. Sociability
2. Activity .35*
3. Fearfulness –.15* –.19*
4. Anger .00 .37* .11*
5. Distress –.17* .04 .73* .33*
*p < .05.
Downloaded by [University of Oslo] at 06:28 13 August 2015
EAS Temperament Survey for Adults. The lack of available
psychometric information for the different EAS versions is
surprising given their widespread use and popularity. Al
though it was pointed out almost 20 years ago that one of the
weaknesses with the EAS measure was that the relative new
ness of the instrument resulted in a paucity of information
about its psychometric properties (Braithwaite et al., 1984),
this picture is not very different today. To date, there have
been no published CFAs of the EAS scale. Our study pro
vides some of this missing information by taking advantage
of the implicit strength of more advanced methods such as
SEM as well as contributing with longitudinal data.
Comparisons between the results from our study and the
results from other studies employing the EAS Temperament
Survey for Adults are complicated for several reasons. First,
most other studies have used earlier versions of the EAS
(e.g., Angleitner & Ostendorf, 1994; Braithwaite et al., 1984;
Ruch et al., 1991; Strelau, 1991; Windle, 1989b). Second, the
original studies by Buss and Plomin (1984) are based on
small samples that except for the 2-week test–retest data pro
vided data from one point in time only.
The means and standard deviations for the five EAS
subscales in our study based on indexes of the summed
scores are in general in accordance with the ones presented
by Buss and Plomin (1984). Our 1-year test–retest correla-
tions are generally lower (on average, 0.15) than those based
on a 2-week period (which ranged from 0.75 to 0.85) as re-
ported by Buss and Plomin (1984). This result was expected
considering the different time lags in the two studies. An ear-
lier psychometric study of the EAS Temperament Survey for
children (Mathiesen & Tambs, 1999) based on the same lon-
gitudinal survey as this study reported high stability esti-
mates (mean value corrected for measurement error was 0.81
for children from 18 to 30 months of age, 0.79 for the period
30 to 50 months, and 0.68 for the period 18 to 50 months).
There is hardly any substantial difference between the
intercorrelations reported by Bussand Plomin (1984)and ours
except for the estimates for the anger and the activity dimen
sion. Assuming that our pooled correlations have the same
standard errors as each of the observed correlations (which isa
conservative approximation), calculations show a highly sig
nificant difference between the two studies for the correlation
between anger and activity. It is likely that these correlations
between the dimensions are underestimated because they
were assessed directly between indexes somewhat contami
nated by measurement error. Applying methods that correct
for attenuations of correlations due to measurement errors
provide more correct estimates, and our results from the SEM
analyses do imply higher intercorrelations. Several of the ear
lier EAS studies are inconsistent in that they apply varimax ro
tation while at the same time reporting fairly high correlations
between the different dimensions (Angleitner & Ostendorf,
1994; Braithwaite et al., 1984; Buss & Plomin, 1984; Ruch et
al., 1991). Our results imply that all the dimensions in the tem
perament of emotionality (distress, fearfulness, and anger) are
highly intercorrelated. In particular, we found an almost
perfect correlation between fearfulness and distress. Thus, se
rious doubts can be raised as to whether these two scales are
measuring empirically distinct constructs.
Due to the low number of items, the estimates of internal
consistency for the five EAS scales are only moderate. The
highest alpha coefficients across the three time points were
found for the Distress scale (an average of .74), whereas the
lowest occur for the Fearfulness scale (an average of .56).
Buss and Plomin (1984) presented no corresponding infor
mation on the reliability of the EAS temperament scales. The
reliability estimates reported by Strelau (1991) on the total
EAS scale is in accordance with our results. Moreover, stud
ies on earlier versions of the EAS (EASI–II and EASI–III)
found moderate to low reliability coefficients (Angleitner &
Ostendorf, 1994; Braithwaite et al., 1984; Ruch et al., 1991).
It is interesting to note that the dimensions with the weakest
alpha reliability coefficients constitute two of the three
subdimensions representing the temperament emotionality
(fearfulness and anger). Our test–retest correlations, which
can be interpreted as reliability estimates, ranging from .62 to
.76, suggest that the alpha values slightly underestimate the
Generally, the main finding from the CFAs suggests an
acceptable fit for the basic theoretical five-dimensional EAS
model and the results presented herein suggest that the scale
is functioning satisfactorily. However, although the model fit
is acceptable, it is nevertheless far from perfect and could be
improved. The higher order, three-dimensional model yields
acceptable but significantly poorer fit. Particularly notewor-
thy is the unexpected result for the factor loadings of the final
three-group, five-dimensional model in which Item 13
(“There are many things that annoy me”) loads much higher
on the fearfulness dimension than on the original anger di
mension. We can only speculate about the reason for this re
sult. Of course it might be related to the actual translation of
the measure. The word annoy was translated into a Norwe
gian word, which is synonymous with irritated in English.
Although we do think that the word irritated has a clear con
notation of anger, it might well be that the discrepancy in
meaning between annoyed and irritated is a contributing fac
tor to our finding. There might also be other explanations for
this result such as cultural differences regarding the expres
sion of emotions. Through SEM analyses, we are able to
identify more reliably the poorly functioning items than has
been done in earlier studies. Accordingly, if the results from
this study can be replicated, a modification of the scale might
be advisable.
The longitudinal analyses based on EQS are performed to
investigate the nature of stability and change of the EAS
scale. Although earlier studies have presented test–retest
scores, we investigated these scores further. First we ex
plored, for each of the five temperaments, whether there was
a stable factor explaining the correlations across time. Our
results do indeed imply that the test–retest correlations are
Downloaded by [University of Oslo] at 06:28 13 August 2015
due to a stable factor affecting the measurements at each
point in time. Next, we tested whether there were
time-specific correlations between the different tempera
ments. The results suggest several such correlations, particu
larly for the emotionality dimension. More specifically, the
relative change in fearfulness coevolves with the relative
change in distress and anger. However, path analyses show
that the larger part of covariation between the emotionality
scores was due to stable factors, and only approximately one
fourth was due to time-specific factors.
A few caveats are warranted regarding these findings.
First, the sample includes women only. A balanced number
of male and female participants would have permitted as
sessment of gender differences and similarities for the con
structs measured. Nevertheless, it has been claimed that the
factor structure of the EAS is similar for women and men
(Plomin et al., 1988), and Buss and Plomin (1984) reported
only small differences between women and men regarding
the results from factor analysis. No strong gender differ
ences were found in the earlier mentioned study on the
EAS questionnaire for children (Mathiesen & Tambs,
1999). Nonetheless, future studies would clearly benefit
from including both sexes. Also, our sample comprised
women between the ages of 19 and 46, with a mean age of
30 years (SD = 4.7). It would be advantageous to include
samples with a broader age range.
This study has provided information on the psychometric
properties of the EAS Temperament Survey for adults. In
general, our results confirm the structure and stability of the
EAS. They also indicate that there is scope for improve-
ments. Additional research is warranted to guide the further
development of the EAS model.
This research was supported by grants from the Norwegian
Council for Mental Health and by the Norwegian Research
Council. We thank Kristin Schjelderup Mathiesen for gen
erously giving us access to the data and Jennifer Harris for
offering advice about the language. We are also grateful to
the mothers who participated in the study and the public
health nurses who assisted in extensive work related to data
Angleitner, A., & Ostendorf, F. (1994). Temperament and the big five factors
of personality. In C. F. Halverson, G. A. Kohnstam, & R. P. Martin (Eds.),
The developing structure of temperament and personality from infancy to
adulthood (pp. 69–90). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Bentler, P. M. (1995). EQS structural equations program manual. Encino,
CA: Multivariate Software.
Bollen, K. A. (1989). Structural equations with latent variables. New York:
Braithwaite, V., Duncan-Jones, P., Bosly-Craft, R., & Goodchild, M. (1984).
A psychometric investiagtion of the usefulness of the EASI–III tempera
ment survey in the Australian general population. Australian Journal of
Psychology, 36, 85–95.
Browne, M. W., & Cudeck, R. (1993). Single sample cross-validation indi
ces for covariance structures. Multivariate Behavioral Research, 10,
Buss, A. (1989). Temperament as personality traits. In G. A. Kohnstamm, J.
E. Bates, & M. K. Rothbart (Eds.), Temperament in childhood (pp.
49–58). New York: Wiley.
Buss, A. H., & Plomin, R. (1975). A temperament theory of personality de
velopment. New York: Wiley.
Buss, A. H., & Plomin, R. (1984). Temperament: Early developing person
ality traits. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Buss, A. H., & Plomin, R. (1986). The EAS approach to temperament. In R.
Plomin, & J. Dunn (Eds.), The study of temperament: Changes, continui
ties and challenges (pp. 67–79). Hillsdale, NJ: Lawrence Erlbaum Asso
ciates, Inc.
Digman, J. M. (1989). Five robust trait dimensions: Development, stability
and utility. Journal of Personality, 57, 195–214.
Digman, J. M. (1990). Personality structure: Emergence of the five-factor
model. Annual Review of Psychology, 41, 417–440.
Goldsmith, H. H., Buss, A. H., Plomin, R., Rothbart, M. K., Thomas, A.,
Chess, S., et al. (1987). Roundtable: What is temperament? Four ap
proaches. Child Development, 58, 505–529.
Hoyle, R. H. (1995). Structural equation modeling. Concepts, issues, and
applications. Thousand Oaks, CA: Sage.
John, O. P. (1989). Towards a taxonomy of personality descriptors. In D. M.
Buss & N. Cantor (Eds.), Personality psychology: Recent trends and
emerging directions (pp. 261–271). New York: Springer.
Jöreskog, K. G., & Sörbom, D. (1993). LISREL 8 user’s reference guide.
Chicago: SSI.
Loehlin, J. C. (1998). Latent variable models. An introduction to factor,
path, and structural analysis. Mahwah, NJ: Lawrence Erlbaum Associ-
ates, Inc.
Mathiesen, K. S., & Sanson, A. (2000). Dimensions of early childhood be-
havior problems: Stability and predictors of change from 18 to 30 months.
Journal of Abnormal Child Psychology, 28, 15–31.
Mathiesen, K. S., & Tambs, K. (1999). The EAS Temperament
Questionnaire—Factor structure, age trends, reliability, and stability in a
Norwegian sample. Journal of Child Psychology and Psychiatry and
Allied Sciences, 40, 431–439.
Mathiesen, K. S., Tambs, K., & Dalgard, O. S. (1999). The influence of so
cial class, strain and social support on symptoms of anxiety and depres
sion in mothers of toddlers. Social Psychiatry and Psychiatric Epidemiol
ogy, 34, 61–72.
McCrae, R. R., & Costa, P. T. (1985). Openness to experience. In R. Hogan
& W. H. Jones (Eds.), Perspectives in personality (Vol. 1, pp. 145–172).
Greenwich, CT: JAI.
Nærde, A., Tambs, K., & Mathiesen, K. S. (2001). Child related strain and
maternal mental health: A longitudinal study. Acta Psychiatrica
Scandinavica, 105, 301–309.
Nærde, A., Tambs, K., Mathiesen, K. S., Dalgard, O. S., & Samuelsen, S. O.
(2000). Symptoms of anxiety and depression among mothers of preshool
children: Effects of chronic strain related to children and child
care-taking. Journal of Affective Disorders, 58, 181–199.
Plomin, R., Pedersen, N. L., McClearn, G. E., Nesselroade, J. R., &
Bergeman, C. S. (1988). EAS temperament during the last half of the life
span: Twins reared apart and twins reared together. Psychology and
Aging, 3, 43–50.
Rowe, D. C., & Plomin, R. (1977). Temperament in early childhood. Journal
of Personality Assessment, 41, 150–156.
Ruch, W., Angleitner, A., & Strelau, J. (1991). The Strelau temperament in
ventory–Revised (STI–R): Validity studies. European Journal of Person
ality, 5, 287–308.
Rutter, M. (1989). Temperament: Conceptual issues and clinical implica
tions. In G. A. Kohnstamm, J. E. Bates, & M. K. Rothbart (Eds.), Temper
ament in childhood (pp. 463–482). Chichester, England: Wiley.
Downloaded by [University of Oslo] at 06:28 13 August 2015
Strelau, J. (1991). Individual differences in temperament: An international
perspective. In J. G. Adair, D. Belanger, & K. L. Dion (Eds.), Advances in
psychological science, Vol. 1: Social, personal and cultural aspects (pp.
33–50). Hove, England: Psychology Press.
Windle, M (1989a). Predicitng temperament–mental health relationships: A
covariance structure latent variable analysis. Journal of Research in Per
sonality, 23, 118–144.
Windle, M. (1989b). Temperament and personality: An exploratory,
interinventory study of the DOTS–R, EASI–II, and EPI. Journal of Per
sonality Assessment, 53, 487–501.
Ane Nærde
Division of Epidemiology
Norwegian Institute of Public Health
P.O. Box 4404 Nydalen
N0–0403 Oslo, Norway
Received September 25, 2002
Revised May 28, 2003
Downloaded by [University of Oslo] at 06:28 13 August 2015
... Past research on sociability has provided some understanding of how this construct relates to psychological, interpersonal, and health functioning across different developmental periods (Chen et al., 2018;Cohen, 2004;Poole, Van Lieshout, & Schmidt, 2017). Yet, it is unclear whether the structure of the sociability construct as measured through multi-item surveys in these varied studies was equivalent across age and sex and whether different respondents understood the survey items in the same way as to allow for unbiased comparisons among different groups (Mathiesen & Tambs, 1999;Naerde, Røysamb, & Tambs, 2004;Walker, Ammaturo, & Wright, 2017). Furthermore, it appears average levels of sociability across the life course have been left mostly uncharted despite a lifespan perspective suggesting that average levels vary alongside maturational changes and life transitions (Baltes, Lindenberger, & Staudinger, 2006;Roberts & Nickel, 2017); understanding variance in average levels of sociability is likely a fundamental step to unraveling the relation between sociability and individual differences over the life course. ...
... While measuring the same trait over such a diverse age range is challenging, we hoped to surmount the obstacle of differing temperament/personality theoretical assumptions by using sociability items from the well-established self-report sociability subscale of the Cheek and Buss Shyness and Sociability Scale -CBSS - (Cheek & Buss, 1981) and the equally well-known parent-report sociability subscale of the Colorado Children's Temperament Inventory -CCTI - (Buss & Plomin, 1984). These two measures in tandem are well suited to studying sociability across the lifespan; the CBSS and CCTI sociability subscales were developed by the same researchers to study sociability from early childhood to older ages with item content that is almost indistinguishable except for some age-related wording (Table S1), are comparable by definition and theoretical underpinnings (Buss & Plomin, 1984), and have been established as reliable and valid indicators of sociability (Mathiesen & Tambs, 1999;Naerde et al., 2004). ...
Although sociability is a fundamental dimension of temperament and personality, few studies have examined it over the lifespan. In this study, sociability was measured across ages 3 to 86 years after assessing for measurement invariance through the multigroup confirmatory factor framework and the more recent alignment method to ensure meaningful differences were assessed between different age groups. Using a repeated cross-sectional design, separate adult (N = 1366, ages 17–86 years) and child/adolescent (N = 543, ages 3–16 years) datasets were created to improve research validity across two different but comparable sociability scales. The findings indicated that there was measurement invariance across adult age groups, but not among child/adolescent age groups. Average levels of sociability followed a significant nonlinear trend (quadratic) across the adult lifespan. Measurement invariance was found across sex for both adult and child/adolescent samples. In adults, females had higher average levels of sociability than males, whereas in children/adolescents, females and males did not differ in mean-levels of sociability. We discuss potential explanations for the quadratic nature of sociability across the adult lifespan, the theoretical implications of these results to understanding personality development, and the methodological issues encountered in studying lifespan differences in sociability from early childhood to senescence.
... The reliability and construct validity of the EAS have been confirmed also previously (e.g. Naerde et al., 2004). ...
... Finally, all the scales that we used in this study have been widely used and validated (e.g. Mackinnon et al., 1999;Naerde et al., 2004;Thompson, 2007). ...
We investigated the associations of individual’s compassion for others with his/her affective and cognitive well-being over a long-term follow-up. We used data from the prospective Young Finns Study (N = 1312‒1699) between 1997‒2012. High compassion was related to higher indicators of affective well-being: higher positive affect (B = 0.221, p < .001), lower negative affect (B = −0.358, p < .001), and total score of affective well-being (the relationship of positive versus negative affect) (B = 0.345, p < .001). Moreover, high compassion was associated with higher indicators of cognitive well-being: higher social support (B = 0.194, p < .001), life satisfaction (B = 0.149, p < .001), subjective health (B = 0.094, p < .001), optimism (B = 0.307, p < .001), and total score of cognitive well-being (B = 0.265, p < .001). Longitudinal analyses showed that high compassion predicted higher affective well-being over a 15-year follow-up (B = 0.361, p < .001) and higher social support over a 10-year follow-up (B = 0.230, p < .001). Finally, compassion was more likely to predict well-being (B = [−0.076; 0.090]) than vice versa, even though the predictive relationships were rather modest by magnitude.
... In this study, between-person internal consistency (α) was acceptable at .73 on average across all waves. The scale items have also demonstrated acceptable twoweek retest-reliability (r = .53-.63) and construct validity (Naerde, Roysamb, & Tambs, 2004). Within-person α across waves (.71) was also good herein. 1 Verbal WM was assessed using the backward digit span test (DST) (Finkel & Pedersen, 2004). ...
Full-text available
Background Trait negative affect (NA) is a central feature of anxiety and depression disorders. Neurocognitive and scar models propose that within‐person increase in NA across one period of time relates to a decline in cognitive functioning at a future period of time and vice versa. Yet, there has been little research on whether a within‐person change in trait NA across one time‐lag precedes and is associated with a change in cognition across a future time lag and vice versa. Due to a growing aging population, such knowledge can inform evidence‐based prevention. Methods Participants were 520 dementia‐free community‐dwelling adults (mean age = 59.76 years [standard deviation = 8.96], 58.08% females). Trait‐level NA (negative emotionality scale), spatial cognition (block design and card rotations), verbal working memory (WM; digit span backward), and processing speed (symbol digit modalities) were assessed at five time points (waves) across 23 years. Bivariate dual latent change score (LCS) approaches were used to adjust for regression to the mean, lagged outcomes, and between‐person variability. Results Unique bivariate LCS models showed that within‐person increase in trait NA across two sequential waves was related to declines in spatial cognition, verbal WM, and processing speed across the subsequent two waves. Moreover, within‐person reductions in spatial cognition, verbal WM, and processing speed across two sequential waves were associated with future increases in trait NA across the subsequent two waves. Conclusions Findings concur with neurobiological and scar theories of psychopathology. Furthermore, results support process‐based emotion regulation models that highlight the importance of verbal WM, spatial cognition, and processing speed capacities for downregulating NA.
... The sample is composed of participants aged 70 years and older (mean age = 81.49(5.24)), and 59.2% are female (Katz et al., 2012;Mathiesen & Tambs, 1999;Naerde, Roysamb, & Tambs, 2004). The current analysis used 1530 participants with the requisite data. ...
This study examined the Big Five personality traits as predictors of mortality risk, and smoking as a mediator of that association. Replication was built into the fabric of our design: we used a Coordinated Analysis with 15 international datasets, representing 44,094 participants. We found that high neuroticism and low conscientiousness, extraversion, and agreeableness were consistent predictors ofmortality across studies. Smoking had a small mediating effect for neuroticism. Country and baseline age explained variation in effects: studies with older baseline age showed a pattern of protective effects (HR<1.00) for openness, and U.S. studies showed a pattern of protective effects for extraversion. This study demonstrated coordinated analysis as a powerful approach to enhance replicability andreproducibility, especially for aging-related longitudinal research.
... The sample is composed of participants aged 70 years and older (mean age = 81.49(5.24)), and 59.2% are female (Katz et al., 2012;Mathiesen & Tambs, 1999;Naerde, Roysamb, & Tambs, 2004). The current analysis used 1530 participants with the requisite data. ...
This study assessed change in the Big Five personality traits. We conducted a coordinated integrative data analysis (IDA) using data from 14 studies including 47,190 respondents to examine trajectories of change in the traits of neuroticism, extraversion, openness, conscientiousness, and agreeableness. Coordinating models across multiple study sites, we fit nearly identical multi-level linear growth curve models to assess and compare the extent of trait change over time. Quadratic change was assessed in 8 studies with four or more measurement occasions. Across studies, the linear trajectory models revealed stability for agreeableness and decreases for the other four five traits. The non-linear trajectories suggest a U-shaped curve for neuroticism, and an inverted-U for extraversion. Meta-analytic summaries indicate that the fixed effects are heterogeneous, and that the variability in traits is partially explained by baseline age and country of origin. We conclude from our study that neuroticism, extraversion, conscientiousness, and openness go down over time, while agreeableness remains relatively stable.
... The EAS includes subscales for activity, sociability, shyness, and emotionality (five items in each subscale). The EAS has demonstrated predictive validity in longitudinal studies [10,66] and has been used in community 1 3 [51,66] and clinical [42,58] samples. Cronbach's alphas for these subscales in the present study were 0.71 (activity), 0.58 (sociability), 0.62 (shyness), and 0.84 (emotionality). ...
Full-text available
Selective mutism (SM) is a stable, debilitating psychiatric disorder in which a child fails to speak in most public situations. Considerable debate exists as to the typology of this population, with empirically-based studies pointing to possible dimensions of anxiety, oppositionality, and communication problems, among other aspects. Little work has juxtaposed identified symptom profiles with key temperamental and social constructs often implicated in SM. The present study examined a large, diverse, non-clinical, international sample of children aged 6-10 years with SM to empirically identify symptom profiles and to link these profiles to key aspects of temperament (i.e., emotionality, shyness, sociability, activity) and social functioning (i.e., social problems, social competence). Exploratory and confirmatory factor analysis revealed anxiety/distress, oppositionality, and inattention domains. In addition, latent class analysis revealed nuanced profiles labeled as (1) moderately anxious, oppositional, and inattentive, (2) highly anxious, and moderately oppositional and inattentive, and (3) mildly to moderately anxious, and mildly oppositional and inattentive. Class 2 was the most impaired group and was associated with greater emotionality, shyness, and social problems. Class 3 was the least impaired group and was associated with better sociability and social competence and activity. Class 1 was largely between the other classes, demonstrating less shyness and social problems than Class 2. The results help confirm previous findings of anxiety and oppositional profiles among children with SM but that nuanced classes may indicate subtle variations in impairment. The results have implications not only for subtyping this population but also for refining assessment and case conceptualization strategies and pursuing personalized and perhaps less lengthy treatment.
... The sample is composed of participants aged 70 years and older (mean age = 81.49(5.24)), and 59.2% are female (Katz et al., 2012;Mathiesen & Tambs, 1999;Naerde, Roysamb, & Tambs, 2004). The current analysis used 1530 participants with the requisite data. ...
This study examined the Big Five personality traits as predictors of mortality risk, and smoking as a mediator of that association. Replication was built into the fabric of our design: we used a Coordinated Analysis with 15 international datasets, representing 44,094 participants. We found that high neuroticism and low conscientiousness, extraversion, and agreeableness were consistent predictors of mortality across studies. Smoking had a small mediating effect for neuroticism. Country and baseline age explained variation in effects: studies with older baseline age showed a pattern of protective effects (HR<1.00) for openness, and U.S. studies showed a pattern of protective effects for extraversion. This study demonstrated coordinated analysis as a powerful approach to enhance replicability and reproducibility, especially for aging-related longitudinal research.
... There are ample worldwide reports of the EAS temperament survey with good psychometric properties (e.g., Bobes Bascarán et al. 2011;Naerde et al. 2004;Spence et al. 2013;Gasman et al. 2002;Mathiesen and Tambs 1999;Boer and Westenberg 1994). ...
Introduction: It is argued that all personality pathology represents the final emergent product of a complex interaction of underlying neurobehavioral systems, which are reflected in personality factors, in conjunction with environmental inputs. Neurobehavioral systems manifest themselves in dispositional temperament and personality processes. Environmental inputs include, obviously, interpersonal relationships (e.g., parenting, social, and mentoring relations) as well as other factors such as abuse, neglect, and/or environmental insults (e.g., economic hardship, deprivation). Narcissistic personality disorder (NPD) is hypothesized to reflect both dispositional and environmental inputs to its pathogenesis. Temperament and personality-based theorizing regarding NPD proposes high dispositional levels of anger and related temperament features that could shape early development and subsequent NPD. Many classic theorists (e.g., Freud, Kernberg, Kohut, Miller) have also proposed that profound parenting failures are implicated in the emergence of NPD, each suggesting some failure in proper engagement and responsivity with the developing child. Such a failure in parenting can be thought of as reflecting diminished proximal process engagement with the developing child. Method: Using data from the Longitudinal Study of Personality Disorders, the present study examines both proximal process and temperament factors in relation to clinically significant NPD features from a prospective perspective. Results: Results suggest that both proximal process and temperament (notably anger) factors independently predict the level of NPD features over time. Conclusion: Both interpersonal relationships and temperament should be considered in models of etiology of NPD, it is not just one or the other.
Full-text available
This exploratory study investigated the interinventory relations of constructs measured by the Revised Dimensions of Temperament Survey (DOTS-R), the Emotionality, Activity, Sociability, Impulsivity (EASI-II) temperament measure, and Eysenck's Personality Inventory (EPI). The zero-order correlational data collected from 153 college students provided concurrent validity for the DOTS-R attributes in relation to the EASI-II and EPI traits. Neuroticism was negatively correlated with the DOTS-R attributes of (low) distractibility, approach-withdrawal, flexibility-rigidity, mood quality, and two rhythmicity dimensions; extraversion was positively correlated with activity level-general approach-withdrawal, flexibility-rigidity, and mood quality. Moderate-to-high correlations were found between similarly labeled attributes of the three inventories and low correlations were generally found between dissimilarly labeled attributes. Multiple regression analysis, used to determine the degree of independence/redundancy among similarly labeled dimensions of the three measurement instruments, indicated a moderate degree of convergence among some of the attributes of the three measures.
Full-text available
The psychometric properties of Buss and Plomin's EASI-III Temperament Survey were examined for a random sample of 290 respondents from the Australian general population. These data support the use of the emotionality, activity and sociability scales with some minor modifications, but cast doubt on the impulsivity scale. Not only did impulsivity emerge as a multidimensional construct, but its components related to other temperaments in markedly different ways. The preliminary analyses also suggested that the EASI-III could be used to measure other constructs, the most important of which are neuroticism and extraversion. The advantages offered by the EASI-III over currently available instruments are discussed: in particular, the simplicity and clarity of the items and the well-articulated sampling framework for their selection.
Personality psychology has not yet established a generally accepted, systematic framework for distinguishing, ordering, and naming individual differences in people’s behavior and experience. Such a systematic framework is generally called a taxonomy. In biology, for example, the Linnean taxonomy established an orderly classification of plants and animals and a standard nomenclature. The availability of this initial taxonomy has been a tremendous asset for biologists: it has permitted researchers to study specified classes of instances instead of examining separately every individual instance, and it has served to facilitate the communication and accumulation of empirical findings about these classes and their instances.
Model Notation, Covariances, and Path Analysis. Causality and Causal Models. Structural Equation Models with Observed Variables. The Consequences of Measurement Error. Measurement Models: The Relation Between Latent and Observed Variables. Confirmatory Factor Analysis. The General Model, Part I: Latent Variable and Measurement Models Combined. The General Model, Part II: Extensions. Appendices. Distribution Theory. References. Index.
This book introduces multiple-latent variable models by utilizing path diagrams to explain the underlying relationships in the models. This approach helps less mathematically inclined students grasp the underlying relationships between path analysis, factor analysis, and structural equation modeling more easily. A few sections of the book make use of elementary matrix algebra. An appendix on the topic is provided for those who need a review. The author maintains an informal style so as to increase the book's accessibility. Notes at the end of each chapter provide some of the more technical details. The book is not tied to a particular computer program, but special attention is paid to LISREL, EQS, AMOS, and Mx. New in the fourth edition of Latent Variable Models: * a data CD that features the correlation and covariance matrices used in the exercises; * new sections on missing data, non-normality, mediation, factorial invariance, and automating the construction of path diagrams; and * reorganization of chapters 3-7 to enhance the flow of the book and its flexibility for teaching. Intended for advanced students and researchers in the areas of social, educational, clinical, industrial, consumer, personality, and developmental psychology, sociology, political science, and marketing, some prior familiarity with correlation and regression is helpful. © 2004 by Lawrence Erlbaum Associates, Inc. All rights reserved.