Personality and Individual Differences 43 (2007) 693–702

www.elsevier.com/locate/paid

A psychometric investigation of two self-report measures

of emotional expressivity
a,*
Jennifer L. Dobbs , Denise M. Sloan b, Andrew Karpinski a

a
Department of Psychology, Temple University, Weiss Hall, Philadelphia, PA 19122, United States
b
National Center for PTSD, VA Boston Healthcare System, Boston, MA 02130, United States

Received 20 August 2006; received in revised form 13 December 2006

Available online 26 March 2007

Abstract

Two commonly used self-report measures of emotional expressivity were evaluated in this study, the
emotional expressivity scale (EES; Kring, Smith, & Neale, 1994) and the Berkeley expressivity question-
naire (BEQ; Gross & John, 1995). The EES is intended to capture emotion expressivity in general, reflecting
a one-factor model. The BEQ is intended to capture the general domain of emotional expressivity by assess-
ing internal emotional experience and the expression of positive and negative emotions, reflecting a hierar-
chical three-factor model. The proposed models of the EES and the BEQ were examined using exploratory
and confirmatory factor analysis. Results support the hypothesized factor structure of the EES, but do not
support the BEQ.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Emotional expression; Emotion; Confirmatory factor analysis; Exploratory factor analysis

0. Introduction

Emotional expression is of vital importance to adaptive human functioning. Expressing emo-

tion has been shown to benefit physical health in the general population (Sloan & Marx, 2004),

*
Corresponding author. Tel.: +1 925 330 0472; fax: +1 215 204 5539.
E-mail address: jdobbs@temple.edu (J.L. Dobbs).

0191-8869/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.paid.2007.01.010
694 J.L. Dobbs et al. / Personality and Individual Differences 43 (2007) 693–702

as well as specific populations, such as women diagnosed with breast cancer (Stanton et al., 2000)
and older adults (Shaw et al., 2003). Emotional expression also plays a central role in psychopa-
thology, including depression (e.g., Sloan, Strauss, & Wisner, 2001), schizophrenia (e.g., Earnst &
Kring, 1999), and borderline personality disorder (Herpertz et al., 2001).
Given the critical role of emotional expression in adaptive functioning, it is imperative that we
have psychometrically strong measures for assessing emotional expression. A variety of methods
have been developed to assess emotional expression.
Observational coding systems and psychophysiology, such as facial electromyography activity
(EMG), are commonly used methods to assess emotional expression. Although there are several
advantages to these methods, a major limitation is that it is not feasible to use these methods out-
side the laboratory, which raises ecological validity concerns.
Self-report is a third method for assessing emotional expression, which allows for a broad
assessment of emotional expression. In contrast to observational and psychophysiological meth-
ods, self-report measures can be designed to assess emotional expression both within and outside
the laboratory, which allows for assessment of emotional expression in general as well as in re-
sponse to more specific situations or contexts. Furthermore, self-report measures permit assess-
ment of emotionally expressive behavior in a variety of different modalities, such as facial
expressions, body gestures, and vocal inflection. Despite these advantages, the self-report method
has several drawbacks. Self-report measures require the ability to accurately reflect on and report
one’s emotional expressiveness. Also, as self-report relies on recall, this method of assessment may
reflect a unique pattern of reporting biases (Robinson & Clore, 2002). Given the potential pitfalls
of self-reporting and the crucial role emotional expression plays in adaptive functioning, it is
essential that we thoroughly examine the reliability and validity of self-report measures of emo-
tional expressivity.
Two commonly used measures of self-reported emotional expressivity are the emotional
expressivity scale (EES; Kring, Smith, & Neale, 1994) and the Berkeley expressivity question-
naire (BEQ; Gross & John, 1995). The EES (e.g., Marx & Sloan, 2002; Shaw et al., 2003)
and the BEQ (e.g., Mennin, Heimberg, Turk, & Fresco, 2005; Tsai, Levenson, & Carstensen,
2000) are frequently used to assess emotional expressivity in both normal and psychopatholog-
ical populations. The EES is intended to measure emotional expressivity in general and is not
intended to assess the content of the expressed emotions (i.e., discrete emotions or valence).
Emotional expression is conceptualized as the behavioral adjustments – facial, verbal, or other-
wise – resulting from emotion. The BEQ is intended to measure emotional expression by assess-
ing both internal emotional experience as well as outward displays of emotional expression.
Emotional expression is comprised of the expression of positive and negative emotions. Taken
together, internal emotional experience, expression of positive emotions, and expression of neg-
ative emotions are intended to reflect emotional expressivity in general. Similar to the EES,
emotional expressions are defined by the BEQ as facial, verbal, or behavior changes attributable
to an emotional experience.
One method to evaluate whether a measure is assessing its intended dimensions is through the
use of factor analysis. Two types of factor analysis are commonly used to assess construct valid-
ity, exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). EFA does not
require an a priori theory about the underlying factor structure. Consequently, EFA is useful
at the beginning stages of scale construction, as the results of the EFA are not based on theory
 J.L. Dobbs et al. / Personality and Individual Differences 43 (2007) 693–702 695

and are therefore not biased by expectations of a given structure. In contrast, CFA requires a
priori theory of the underlying factor structure, and is therefore useful in the later stages of
scale development as a direct test of a hypothesized factor structure.
The purpose of the present study is to examine the factor structures of the EES and the BEQ.
The EES is intended to assess emotional expressivity in general and EFA conducted in two studies
supports a one-factor model (Gross & John, 1998; Kring et al., 1994). However, a CFA of the
EES has not been conducted. In the development of the BEQ, Gross and John (1995) performed
an EFA to assess its factor structure. Consistent with the purported intentions of the measure,
EFAs in two studies suggested a hierarchical three-factor structure (Gross & John, 1995, 1997).
Gross and John (1997) also conducted a CFA and they reported that a hierarchical three-factor
structure fit the data better than either an orthogonal three-factor structure or a one-factor struc-
ture. However, no goodness of fit indices were reported for the CFA. Therefore, it is unclear if the
hierarchical three-factor structure provided an adequate fit to the data.
In this study we expected that an EFA of the EES would reflect a one-factor model, replicating
previous findings (Gross & John, 1998; Kring et al., 1994) and a CFA of the EES would further
support the one-factor model. In addition, it was predicted that an EFA and a CFA would rep-
licate Gross and John’s (1995, 1997) findings that the BEQ reflects a hierarchical model comprised
of three factors subsumed under a general expressivity domain.

1. Method

1.1. Participants

The sample consisted of 365 students (264 female) from Temple University. Students partici-
pated in exchange for research course credit required as part of an introductory psychology
course. Participants ranged in age from 17 to 52 years (M = 19.51, SD = 0.17). The ethnic com-
position of the sample was consistent with the general student population at the university (61.1%
Caucasian, 22.7% African–American, 7.7% Asian–American, and 7.7% other).

1.2. Measures

The EES (Kring et al., 1994) is a 17-item measure of emotional expressivity in which partici-
pants rate their responses on a 6-point Likert-type scale (1 = never true to 6 = always true).
The EES has been shown to have high internal consistency and test retest reliability (Kring
et al., 1994). The EES has also been shown to have convergent and discriminant validity based
on both self-report, other report, and observational methods of assessment (Kring et al., 1994).
The BEQ (Gross & John, 1995) is a 16-item measure of emotional expressivity with items rated
on a 7-point Likert-type scale (1 = strongly disagree to 7 = strongly agree). The BEQ is com-
prised of three subscales: impulse strength, negative expressivity, and positive expressivity, sub-
sumed under the domain of general expressivity. The impulse strength subscale assesses the
strength of the individual’s emotional expression. The negative expressivity subscales consists
of 6 items and the positive expressivity subscale consists of 4 items. The BEQ total score has
shown high internal consistency and good test retest reliability (Gross & John, 1997). Convergent
696 J.L. Dobbs et al. / Personality and Individual Differences 43 (2007) 693–702

and discriminant validity for each of the subscales has been demonstrated (Gross & John, 1997).
It should be noted that there have been only a handful of studies examining the validity of the
BEQ and EES.

1.3. Procedure

After informed consent was obtained, participants were asked to complete a demographic ques-
tionnaire (e.g., age, racial background), the EES and the BEQ. Participants completed the ques-
tionnaires in a group setting. Other questionnaires were completed as well, but they are not
reported here. The study was approved by the university Institutional Review Board.

1.4. Data analysis plan

Previous research has found that the EES reflects a one-factor model (Gross & John, 1998;
Kring et al., 1994) and the BEQ reflects a hierarchical three-factor model (Gross & John, 1995,
1997). To examine whether these findings were replicated with use of the current sample we con-
ducted principal components EFA of the EES and the BEQ. In cases where EFA results ade-
quately replicated previous findings, we conducted CFA to further examine the factor structure
of the EES and the BEQ.
In order to examine the hypothesized factor structures of the EES and the BEQ, we conducted a
maximum likelihood CFA of each of the hypothesized models for each measure using AMOS 5.0
(Arbuckle, 2002). All factors were allowed to freely correlate and all items were allowed to load
only on their hypothesized factor. Three fit indices were examined to assess the fit of the model:
the Tucker–Lewis index (TLI), the comparative fit index (CFI), and the root-mean-square error of
approximation (RMSEA). Using conservative criteria, Hu and Bentler (1999) established that a
good fit between the hypothesized model and the observed data is indicated by TLI > .95,
CFI > .95, and RMSEA 6 .06. Thus, these criteria were used to determine an adequate fit. In
cases where the models did not yield an adequate fit to the data, model modification procedures
were used in accordance with recommendations by (Hatcher, 1994; Hopko et al., 2003).

2. Results

All items in the EES met univariate normality (coefficients for skewness and kurtosis < 1). The
mean total score of the EES was 63.91 (.74). The a-reliability coefficient for the EES was .93. For
the BEQ, all items met univariate normality (coefficients for skewness and kurtosis < 1) with the
exception of items 1, 2, 4, 6, and 8, which were not normally distributed. The mean total score of
the BEQ was 4.81 (.04). The a-reliability coefficient for the BEQ was .82. The a-reliability coeffi-
cient for the BEQ subscales were as follows: impulse strength, a = .78; positive expression a = .63;
and negative expression a = .63. Intercorrelations between the BEQ subscales were as follows: im-
pulse strength and positive expression, r = .47; impulse strength and negative expression, r = .41;
and positive expression and negative expression, r = .47. Intercorrelations found in this study
were similar to Gross and John (1995, r = .50–.62). Reliability analyses showed that the negative
expression subscale reliability would be improved with the removal of item 8 as the item deleted
 J.L. Dobbs et al. / Personality and Individual Differences 43 (2007) 693–702 697

a-reliability coefficient was .66. The mean scores for the EES and the BEQ are consistent with
those found by Gross and John (1995) and Kring et al. (1994).

2.1. Factor structure of the EES

In regards to the results of the EFA on the EES, a scree test (Cattell, 1966) suggested a one-
factor solution. The first factor accounted for 47.30% of the variance and subsequent factors ac-
counted for 8.03% and 6.66% of the remaining variance. Additionally, all items but one (item 12)
loaded most strongly on the first factor. Taken together, this provides very strong support for a
one factor solution. Item loadings for the one-factor solution of the EES are shown in Table 1.
These results are also consistent with Gross and John (1998) who reported that an unrotated
one-factor model of the EES accounted for 47% of the variance. Unfortunately, factor loadings
were not reported by Gross and John (1998) so further examination of the replicability of these
findings is not possible.
To further examine the factor structure of the EES we conducted a CFA of the proposed one-
factor model of the EES. Reverse scored items were not correlated due to the large number of
such items (11 reverse scored items). CFA of the EES revealed that the model was not an accept-
able fit for the data (X2 (119) = 639, CFI = .84, TLI = .82, RMSEA = .11). To ensure that the
inadequate fit of the model was not due to a violation of multivariate normality, steps were taken
to improve multivariate normality including transformations and combinations of several items.
Despite these efforts, none of the CFAs of the one-factor model provided an acceptable fit to the
data (CFIs < .95, TLIs < .89, RMSEAs > .12).
Inadequate fit of the one-factor model may be due to a small number of ‘‘bad’’ items. To
explore this possibility, we removed the item with lowest factor loading and conducted a CFA

Table 1
Factor loadings from the exploratory factor analysis of the EES
EES items Factor loading
5. I keep my feelings to myself (R) .81
14. Even if I am feeling very emotional, I do not let others see my feelings (R) .79
7. I display my emotions to other people .78
17. I hold my feelings in (R) .78
11. I am not very emotionally expressive (R) .77
15. I think of myself as emotionally expressive .77
1. I do not express my emotions to other people (R) .75
2. Even when I am experiencing strong feelings, I do not express them outwardly (R) .74
9. I do not like to let other people see how I am feeling (R) .69
6. Other people are not easily able to observe what I am feeling (R) .69
4. People can ‘‘read’’ my emotions .67
8. People think of me as an unemotional person (R) .66
3. Other people believe me to be very emotional .60
10. I cannot hide the way I am feeling .58
13. I am able to cry in front of other people .53
16. The way I feel is different from how others think I feel (R) .52
12. I am often considered indifferent by others (R) .42
Note: EES = Emotional expressivity scale; (R) = reversed scored item.
698 J.L. Dobbs et al. / Personality and Individual Differences 43 (2007) 693–702

of the one-factor model on this reduced EES. If an adequate fit was not obtained, we repeated this
procedure, once again eliminating the item with the lowest factor loading. Despite removing up to
6 items, CFAs of the one-factor solution of the EES all failed to provide an adequate fit to the
data (CFIs < .92, TLIs < .89, RMSEAs > .11).
Given the poor fit of the one-factor model, we engaged in model modification procedures in the
attempt to find a model that would more adequately fit the data. Modification procedures were
consistent with the algorithm described above. The obtained t statistics of the unstandardized
path coefficients of all the items for the EES were significant at p < .05 level, thus, all items were
retained. Next, the modification indexes were evaluated. Parameter estimates were added in
descending order of the related modification indexes until an adequate fit to the data was
achieved. Twenty-one parameter estimates correlating error variances were added to the model
to achieve an adequate fit (X2 (98) = 230.5; CFI = .96, TLI = .95, RMSEA = .06); this model
fit significantly better than the unmodified one factor model (X2 (21) = 408.5, p < .001). All added
correlations were theoretically and statistically substantive in that only error variances of concep-
tually similar items were correlated (e.g., ‘‘I keep my feeling to myself.’’ ‘‘I hold my feelings in.’’).

2.2. Factor structure of the BEQ

To examine whether previous EFA results indicating that the BEQ reflected a three-factor
model (Gross & John, 1995, 1997) were replicated, we conducted an EFA with varimax rotation
on the BEQ. This analysis revealed a three-factor solution, accounting for 48.6% of the variance.
A comparison of the factor loadings obtained here to the factor loadings obtained by Gross and
John (1997) revealed considerable consistency (see Table 2).
To continue to explore the factor structure of the BEQ, we conducted a CFA of the hypothe-
sized models of the BEQ. The error variances of the reverse scored items were correlated. Given
the small number of reverse scored items and the similarity in presentation of these items, the the-
oretical support for correlating the error variances outweighs the potential inflation of the type 2
error. Neither the one factor model (X2 (101) = 564.1; CFI = .70, TLI = .64, RMSEA = .11), the
three factor orthogonal model (X2 (101) = 620.9; CFI = .66, TLI = .60, RMSEA = .12), nor the
three factor hierarchical model (X2 (98) = 435.7; CFI = .78, TLI = .73, RMSEA = .10) yielded
an acceptable fit to the data. Consistent with Gross and John (1997), the three-factor hierarchical
model yielded the best fit to the data, however, this model did not yield an adequate fit to the data
according to Hu and Bentler’s (1999) criteria. To ensure that the poor fit of the model was not due
to a violation of multivariate normality, steps were taken to obtain multivariate normality includ-
ing transformations and combinations of individual items. CFA of the hierarchical three-factor
model did not yield an adequate fit to the data (CFI < .885, TLI < .847, RMSEA > .083).
We conducted a series of CFAs to examine whether the poor fit obtained by the CFA might be
due to poor fitting items. However, CFAs of the hierarchical three-factor solution of the BEQ
with up to two poor fitting items removed continued to yield an inadequate fit to the data
(CFIs < .86, TLIs < .83, RMSEAs > .08).
Given the inadequate fit of the hypothesized models of the BEQ, we conducted model modifi-
cation procedures to the hierarchical three-factor model of the BEQ. Modification procedures
were consistent with the algorithm described above. The obtained t statistics of the unstandard-
ized path coefficients of all the items for the BEQ were significant at p < .05, so all items were
 J.L. Dobbs et al. / Personality and Individual Differences 43 (2007) 693–702 699

Table 2
Varimax-rotated factor loadings from the exploratory factor analysis of the BEQ
BEQ Items Current study Gross and John, 1997
Negative Positive Impulse
expression expression strength
Negative expression
16. What I am feeling is written all over my face .72 .16 .22 .65
13. Whenever I feel negative emotions, people can .64 .01 .12 .66
easily see exactly what I am feeling
9. No matter how nervous or upset I am, I tend to .58 .06 .05 .67
keep a calm exterior (R)
5. It is difficult for me to hide my fear .53 .01 .24 .51
3. People often do not know what I am feeling (R) .52 .46 .11 .61
8. I have learned it is better to suppress my anger .18 .38 .05 .39
than to show it (R)
Positive expression
6. When I am happy my feelings show .08 .77 .18 .77
1. Whenever I feel positive emotions, people can .13 .70 .11 .68
easily see what I am feeling
4. I laugh out loud when someone tells me a joke .21 .64 .18 .66
that I think is funny
10. I am an emotionally expressive person .53 .38 .39 .38
Impulse strength
15. I experience my emotions very strongly .19 .16 .74 .74
7. My body reacts very strongly to emotional .23 .22 .69 .65
situations
11. I have strong emotions .15 .32 .68 .65
14. There have been times when I have not been able .10 .11 .67 .68
to stop crying even though I tried to stop
2. I sometimes cry during sad movies .03 .02 .66 .62
12. I am sometimes unable to hide my feelings, even .48 .06 .52 .53
though I would like to
Note: BEQ = Berkeley expressivity questionnaire; (R) = reversed scored item.

retained. Next, the modification indexes were evaluated. Parameter estimates were added in
descending order of the related modification indexes until an adequate fit to the data was
achieved. Per the modification indexes, item 12 was permitted to load on the negative expression
factor in addition to the impulse strength factor. Additionally, seven parameter estimates corre-
lating item error variances and four parameter estimates correlating item error variances with
factor error variances were added to the model to achieve an adequate fit (X2 (84) = 139.8;
CFI = .96, TLI = .95, RMSEA = .04). Given that the modification indexes suggest that item
12 load on multiple factors, error variances of items loading on different factors be correlated
(e.g., ‘‘I am an emotionally expressive person.’’ ‘‘There have been times when I have not been able
to stop crying even though I tried to stop.’’), and error variances of certain items be correlated
with factor error variance (e.g., ‘‘I have learned it is better to express my anger than to show
it.’’), the recommended model modifications were not theoretically or statistically supported.
700 J.L. Dobbs et al. / Personality and Individual Differences 43 (2007) 693–702

Additional model modifications were conducted to allow for all possible theoretically and statis-
tically supported modifications, but these modifications also failed to yield an adequate fit to the
data (X2 (68) = 170.0; CFI = .93, TLI = .88, RMSEA = .06). Therefore, as an adequate fit to the
data was achieved only through modifications that are not statistically or theoretically supported,
support for the hierarchical three-factor structure of the BEQ was not obtained.

3. Discussion

The model modifications required to achieve an adequate fit for the EES were both conceptu-
ally and statistically supported. We found that all of the items of the EES significantly loaded onto
the single general expressivity factor and the modifications were statistically supported in that they
consisted entirely of correlating error variances. The items of the correlated error variances were
conceptually similar in that they often assessed same behavior yet were negatively phrased (e.g.,
‘‘People think of me as an unemotional person.’’). Given the strong conceptual support for cor-
relating such items, we believe there is sufficient support for the modifications conducted in this
study. Nonetheless, these findings should be interpreted with caution. Many researchers have
questioned the validity of findings based on the incorporation of modification indices. Such post
hoc modifications capitalize on chance, thereby limiting the generalizability of the findings (Mac-
Callum, 1995). Additionally, some researchers (e.g., Cortina, 2002) have been critical of a large
degree of correlated error variances as model fit improves by adding additional constraints to
the model.
In contrast to our findings with the EES, we did not find support for the hypothesized factor
structure of the BEQ. Consistent with Gross and John (1997), we did find that the hierarchical
three-factor solution fit the data better than either the three-factor orthogonal solution or the
one-factor solution. However, we did not find that the hierarchical three-factor solution yielded
an adequate fit to the data. The model modifications necessary to achieve an adequate fit were
neither theoretically nor statistically supported. Some of the recommended modifications con-
sisted of correlating item error variances. Although the items of the correlated error variance were
conceptually consistent, the items often loaded on different factors. Correlating such items would
violate the hypothesized model structure, consequently such modifications are not theoretically
supported. Additionally, the modifications indices suggested that item 12 loaded on multiple fac-
tors, which is also theoretically inconsistent with the hypothesized model. Moreover, modification
indices suggested that several of the item error variances were correlated with the positive expres-
sion factor’s error variance, and this modification would not be statistically supported. Given that
the EFA findings in this study and in other studies (Gross & John, 1995, 1997), suggest a three-
factor structure of the BEQ, it is surprising that the CFA was not able to confirm this model. One
explanation for the inconsistency may be the differences in criteria utilized by EFA and CFA.
Whereas EFA assesses variance accounted for by the factors, CFA evaluates unaccounted for var-
iance (Floyd & Widaman, 1995). EFA’s conducted in several studies found that the factor struc-
ture of the BEQ accounted for no more than 51% of the total variance. Streiner (1994) has
suggested that factor structures account for at least 50% of the total variance and some research-
ers have suggested as much as 80% (e.g., Floyd & Widaman, 1995). Although the hierarchical
three-factor structure of the BEQ accounts for a significant amount of the total variance, there
 J.L. Dobbs et al. / Personality and Individual Differences 43 (2007) 693–702 701

may still be a significant portion of the variance unaccounted for, which results in the inadequate
fit to the data using CFA procedures.
The relatively small amount of total variance accounted for by the factor structure of the BEQ
calls into question the importance of assessing commonalities between variables compared with
assessing the variance associated with the specific variables (Floyd & Widaman, 1995). Conse-
quently, emotional expressivity, as measured by the BEQ, may be driven more by characteristics
specific to the discrete emotions expressed rather than their shared features (i.e., valence). This
argument is supported by the low a-reliability coefficients found for both the positive expression
(a = .63) and negative expression (a = .63) subscales of the BEQ, though the small number of
items likely contributed to the low alpha values. Furthermore, the item deleted a-reliability coef-
ficient of item 8 (‘‘I have learned it is better to suppress my anger than to show it.’’) is higher than
the a-reliability coefficient of the negative expression subscale, which indicates that the reliability
of the subscale would improve if item 8 were removed. Taken together, it appears that the positive
and negative expression subscales of the BEQ do not fully capture the expression of positive and
negative emotions. Other researchers have also raised concerns about the negative expression sub-
scale of the BEQ. Trierweiler, Eid, and Lischetzke (2002) found that although most emotions were
highly correlated with their respective factors of the BEQ, the emotions of anger and shame were
poorly represented in the negative expression subscale of the BEQ. Thus, emotional expressivity
may be better captured by assessing the expression of discrete emotions rather than assessing
valence.
As noted above, using a self-report approach to assess emotional expressivity has limitations,
namely that this approach assumes an individual can accurately reflect upon and report on their
emotionally expressive behavior. As each assessment method of emotional expression has specific
limitations, it may be best to incorporate several methods of assessment (e.g., observation and
self-report) in order to most accurately capture emotional expression. Indeed, emotion is mani-
fested by multiple components (subjective, behavioral, and physiological) and evaluating of each
of these components would likely provide the most accurate assessment approach.

References

Arbuckle, J. L. (2002) (Version 5). AMOS for Windows. Analysis of moment structures. Chicago, IL: SmallWaters.
Cattell, R. B. (1966). The scree test for the number of factors. Multivariate Behavioral Research, 1, 245–276.
Cortina, J. M. (2002). Big things have small beginnings: An assortment of ‘‘minor’’ methodological misunderstandings.
Journal of Management, 28, 339–362.
Earnst, K. S., & Kring, A. M. (1999). Emotional responding in deficit and non-deficit schizophrenia. Psychiatry
Research, 88, 191–207.
Floyd, F. J., & Widaman, K. F. (1995). Factor analysis in the development and refinement of clinical assessment
instruments. Psychological Assessment, 7, 286–299.
Gross, J. J., & John, O. P. (1995). Facets of emotional expressivity: Three self-report factors and their correlates.
Personality and Individual Differences, 19, 555–568.
Gross, J. J., & John, O. P. (1997). Revealing feelings: Facets of emotional expressivity in self-reports, peer ratings, and
behavior. Journal of Personality and Social Psychology, 72, 435–448.
Gross, J. J., & John, O. P. (1998). Mapping the domain of expressivity: Multimethod evidence for a hierarchical model.
Journal of Personality and Social Psychology, 74, 170–191.
Hatcher, L. (1994). A step-by-step approach to using the SAS system for factor analysis and structural equation modeling.
Cary, NC: SAS Institute, Inc.
702 J.L. Dobbs et al. / Personality and Individual Differences 43 (2007) 693–702

Herpertz, S. C., Werth, U., Lukas, G., Quanaibi, M., Schuerkens, A., Kunert, H., et al. (2001). Emotion in criminal
offenders with psychopathy and borderline personality disorder. Archives of General Psychiatry, 58, 737–745.
Hopko, D. R., Stanley, M. A., Reas, D. L., Wetherell, J. L., Beck, J. G., Novy, D. M., et al. (2003). Assessing worry in
older adults: Confirmatory factor analysis of the Penn State Worry Questionnaire and psychometric properties of an
abbreviated model. Psychological Assessment, 15, 173–183.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria
versus new alternatives. Structural Equation Modeling, 6, 1–55.
Kring, A. M., Smith, D. A., & Neale, J. M. (1994). Individual difference in dispositional expressiveness: Development
and validation of the Emotional Expressivity Scale. Journal of Personality and Social Psychology, 66, 934–949.
MacCallum, R. C. (1995). Model specification: Procedures, strategies, and related issues. In R. H. Hoyle (Ed.).
Structural equation modeling: Concepts, issues, and applications (pp. 16–36). Thousand Oaks, CA: Sage Publications,
Inc.
Marx, B. P., & Sloan, D. M. (2002). The role of emotion in the psychological functioning of adult survivors of
childhood sexual abuse. Behavior Therapy, 33, 563–577.
Mennin, D. S., Heimberg, R. G., Turk, C. L., & Fresco, D. M. (2005). Preliminary evidence for an emotion
dysregulation model of generalized anxiety disorder. Behaviour Research and Therapy, 43, 1281–1310.
Robinson, M. D., & Clore, G. L. (2002). Belief and feeling: Evidence for an accessibility model of emotional self-report.
Psychological Bulletin, 128, 934–960.
Shaw, W. S., Patterson, T. L., Semple, S. J., Dimsdale, J. E., Ziegler, M. G., & Grant, I. (2003). Emotional
expressiveness, hostility, and blood pressure in a longitudinal cohort of Alzheimer caregivers. Journal of
Psychosomatic Research, 54, 293–302.
Sloan, D. M., & Marx, B. P. (2004). Taking pen to hand: Evaluating theories underlying the written emotional
disclosure paradigm. Clinical Psychology: Science and Practice, 11, 121–137.
Sloan, D. M., Strauss, M. E., & Wisner, K. L. (2001). Diminished response to pleasant stimuli by depressed women.
Journal of Abnormal Psychology, 110, 488–493.
Stanton, A. L., Danoff-Burg, S., Cameron, C. L., Bishop, M., Collins, C. A., Kirk, S. B., et al. (2000). Emotionally
expressive coping predicts psychological and physical adjustment to breast cancer. Journal of Consulting and Clinical
Psychology, 68, 875–882.
Streiner, D. L. (1994). Figuring out factors: The use and misuse of factor analysis. Canadian Journal of Psychiatry, 39,
135–140.
Trierweiler, L. I., Eid, M., & Lischetzke, T. (2002). The structure of emotional expressivity: Each emotion counts.
Journal of Personality and Social Psychology, 82, 1023–1040.
Tsai, J. L., Levenson, R. W., & Carstensen, L. L. (2000). Autonomic, subjective, and expressive responses to emotional
films in older and younger Chinese Americans and European Americans. Psychology and Aging, 4, 684–693.