Running head: BODY APPRECIATION SCALE-2
Factor Structure and Psychometric Properties of a Romanian Translation of the Body
Viren Swami1-2, David Barron2, Otilia Tudorel3, Cosmin Goian4, & Mona Vintila3
1 Department of Psychology, Anglia Ruskin University, Cambridge, UK
2 Centre for Psychological Medicine, Perdana University, Serdang, Malaysia
3 Department of Psychology, West University of Timișoara, Timișoara, Romania
4 Department of Social Work, West University of Timișoara, Timișoara, Romania
Address correspondence to: Prof. Viren Swami, Department of Psychology, Anglia Ruskin
University, East Road, Cambridge, Cambridgeshire CB1 1PT, UK. Email:
We examined the psychometric properties of a Romanian translation of the 10-item Body
Appreciation Scale-2 (BAS-2). A total of 453 university students from Romania completed
the BAS-2, along with measures of disordered eating, self-esteem, satisfaction with life, and
subjective happiness. In addition, a separate sample of university students (N = 109)
completed only the BAS-2 at two time-points three weeks apart. Principal-axis factor analysis
indicated that BAS-2 scores had a one-dimensional factor structure in both women and men.
Confirmatory factor analysis indicated that this factor structure had adequate fit, but
invariance across sex was not supported. Further analyses indicated that BAS-2 scores
evidenced internal consistency, convergent validity, and test-retest reliability in both women
and men. These results suggest that BAS-2 scores reduce to one dimension in Romanian
adults, but the lack of sex invariance may indicate that the same latent construct is not being
measured in women and men.
Keywords: Body appreciation; Positive body image; Scale translation;
Body Appreciation Scale-2
Over the past decade, researchers have given sustained attention to the construct of
positive body image, signalling an important shift in the field from a primary focus on
pathology to a comprehensive account of the body image concept (e.g., Smolak & Cash,
2011; Tylka, 2012). One particular facet of positive body image that has benefitted from this
attention is body appreciation, which Tylka and Wood-Barcalow (2015a, p. 53) defined as
“accepting, holding favorable opinions toward, and respecting the body, while also rejecting
media-promoted appearance ideals as the only form of human beauty.” To measure this facet,
Avalos, Tylka, and Wood-Barcalow (2005) developed a 13-item measure, the Body
Appreciation Scale (BAS), and its scores were found to have a one-dimensional factor
structure in college women in the United States (U.S.). In a later study, Tylka (2013) reported
that this one-dimensional factor structure was invariant across U.S. college women and men.
Studies using factor analysis have supported the one-dimensional nature of BAS
scores in a community sample of Austrian adults (Swami, Stieger, Haubner, & Voracek,
2008), Spanish adolescents (Lobera & Rios, 2011), and Turkish college women (Swami,
Özgen, Gökçen, & Petrides, 2015; but see Bakalim & Tasleden-Karçkay, 2016). However, in
an early study with Malaysian women, Swami and Chamorro-Premuzic (2008) reported,
using confirmatory factor analysis (CFA), that the one-dimensional factor structure of BAS
scores had poor fit. Using principal-axis factor analysis, these authors instead extracted a
two-dimensional structure consisting of a primary factor tapping what they termed “general
body appreciation” and a secondary factor that was said to measure “body image
investment.” Later studies also supported a two-factor structure of BAS scores across distinct
linguistic and national groups (for reviews, see Swami, 2017; Webb, Wood-Barcalow, &
Tylka, 2015), including in Brazil (Ferreira, Neves, & Tavares, 2014; Swami et al., 2011),
Indonesia (Swami & Jaafar, 2012), South Korea (Swami, Hwang, & Jung, 2012), Poland
(Taylor, Szpakowska, & Swami, 2013), Zimbabwe (Swami, Mada, & Tovée, 2012), Hong
Kong (Ng, Barron, & Swami, 2015), and Iran (Atari, Akbari-Zardkhaneh, Mohammadi, &
In addition to these discrepant factor structures, the BAS was also developed prior to
sustained focus on the concept of positive body image (e.g., Coppola, Dimler, Letendre, &
McHugh, 2017; Holmqvist & Frisén, 2012; McHugh, Coppola, & Sabiston, 2014; Wood-
Barcalow, Tylka, & Augustus-Horvath, 2010). In particular, some items of the BAS reflected
the antiquated belief that positive body image was equivalent to the absence of indices of
negative body image. Given these issues, Tylka and Wood-Barcalow (2015b) developed a
revised version of the BAS. Specifically, they deleted one sex-specific item, two items that
represent low levels of negative body image, and two items that are inconsistent with current
conceptualizations of positive body image (i.e., containing language that assumes bodies are
flawed). They developed additional items that would better reflect positive body image and,
alongside the eight remaining original items, conducted a factor analysis, in which items were
retained that had high factor loadings and low item redundancy. This revised measure, known
as the Body Appreciation Scale-2 (BAS-2), consists of 10 items (five original BAS items and
five newly developed items), and scores on the scale have been found to have a one-
dimensional factor structure in college and community adults from the U.S. (Tylka & Wood-
As with the BAS, scholars have also been quick to examine the factor structure of
scores derived from translations of the BAS-2. Thus, using principal-axis factor analysis, the
one-dimensional factor structure of BAS-2 scores has been supported with college samples
from Hong Kong (Swami & Ng, 2015), the Netherlands (Alleva, Martijn, Veldhuis, & Tylka,
2016), and Iran (Atari, 2016), as well as a community sample from Serbia (Jovic, Sfroza,
Jovanovic, & Jovic, 2016). In addition, using CFA, the one-dimensional factor structure of
Body Appreciation Scale-2
the BAS-2 has been confirmed in a mixed college staff-and-student sample from mainland
China (Swami, Ng, & Barron, 2016), in a college sample from France (Kertechian & Swami,
2017), and in community samples from Spain (Swami, García, & Barron, 2017) and Poland
(Razmus & Razmus, 2017). As in the parent study (Tylka & Wood-Barcalow, 2015b), these
CFA-based studies have also reported that BAS-2 scores are invariant across participant sex,
evidence internal consistency, and reveal patterns of convergent validity. In addition, most
studies with relevant data have reported that men have significantly higher BAS-2 scores than
women, with small-to-moderate effect size (ds = 0.15-0.58; Atari, 2016; Kertechian &
Swami, 2017; Swami & Ng, 2015; Tylka & Wood-Barcalow, 2015), although two studies
have failed to find significant sex differences (Swami et al., 2016, 2017). Taken together, the
available evidence suggests that BAS-2 scores have a one-dimensional factor structure across
cultural groups, with good psychometric properties.
The Romanian Context
Here, we examined the factor structure and psychometric properties of a Romanian
(limba română or лимба ромынэ in Moldovan Cyrillic) translation of the BAS-2. We believe
this is important for several reasons. First, Romanian is spoken by around 24 million people
as a first language, mainly in Romania and Moldova (where it has official status), as well as
by several million more as a second language (European Commission, 2012). The availability
of a translation of the BAS-2 would, therefore, provide an important tool for the assessment
of positive body image in relatively neglected linguistic groups within the body image
literature. To be sure, a Romanian translation of the BAS appears to have been prepared
(Dumitrescu, Zetu, Teslaru, Dogaru, & Dogaru, 2008), but the authors neglected to report on
the translational procedures. Just as important, the authors of this translation also did not
examine the factor structure of Romanian BAS scores, instead assuming that these scores
were one-dimensional. This is problematic because the assumed one-dimensional structure
may not be appropriate for these data and may result in spurious associations with other
Second, investigating body image issues in Romania is important in its own right
because of the unique cultural and historical trajectories of its population. For example, it has
been suggested that, prior to the adoption of free market economics in Eastern European
nations, Communism may have protected women in the region from body image disturbances
through a rejection of objectification and a promotion of egalitarianism (Catina & Joja, 2001;
Haavio-Mannila & Purhonen, 2001). Although this claim is contested (Eisler, 2001), what is
clear is that the introduction of a market economy brought a raft of social and cultural
changes (e.g., an influx of Western media emphasising the thin ideal and an ethos of working
on the body as necessary) that may have contributed to a rise in body image disturbance in
women (e.g., Catina & Joja, 2001; Frederick, Forbes, & Berezovskaya, 2008). Conversely,
however, few studies have specifically focused on issues of positive body image in these
nations (but see Taylor et al., 2013). In addition, body image research in Romania specifically
remains piecemeal and hampered by a lack of psychometrically sound assessment tools,
although it is clear that the country is experiencing nutrition and physical activity transitions
with increased intake of foods rich in carbohydrates and saturated fats, and an increase in
sedentary lifestyles (Mocanu, 2013; Ulijaszek & Koziel, 2007).
The Present Study
As a contribution to the literature, therefore, we prepared a Romanian translation of
the BAS-2 and examined its factor structure and psychometric properties in a sample of
Romanian adults. First, we began by examining the factor structure of the Romanian BAS-2
using both principal-axis factor analysis and CFA, and based on the available literature we
expected to find evidence of a one-dimensional structure with all 10 items. Second, we
assessed the scale’s internal consistency, expecting that scores would demonstrate internal
Body Appreciation Scale-2
reliability. Third, we examined whether BAS-2 scores were invariant across sex,
hypothesising that they would be. Based on our expectation of sex invariance, we also
hypothesised that men would have significantly higher body appreciation than women.
Fourth, we examined the test-retest reliability of BAS-2 scores after three weeks,
hypothesising that scores would demonstrate stability across time. Finally, we conducted a
preliminary examination of the convergent validity of BAS-2 scores by examining
associations with scores from measures of disordered eating, self-esteem, satisfaction with
life, subjective happiness, and body mass index (BMI). Evidence of convergent validity
would be provided by negative associations with disordered eating and BMI, and positive
associations with self-esteem, satisfaction with life, and subjective happiness (cf. Tylka &
Main sample. Participants of the main study were 212 female and 241 male students
from a university in Timiș County in the west of Romania. Participants ranged in age from 17
to 61 years (M = 23.57, SD = 7.86) and in self-reported BMI from 13.98 to 37.11 kg/m2 (M =
22.18, SD = 3.46). Most participants were enrolled on undergraduate programmes (72.2%;
Masters programmes = 16.1%, other = 11.7%) and self-reported as being of Romanian
ethnicity (97.3%). In terms of religion, 68.1% reported being Eastern Orthodox, 8.4%
Catholic, 8.0% atheist, and the remainder reported being of some other religious affiliation.
These participants completed all the measures listed below.
Test-retest sample. The test-retest sample consisted of 57 female and 52 male
university students from the same university in the west of Romania. All participants were of
Romanian ethnicity and ranged in age from 18 to 57 years (M = 25.02, SD = 8.91). No further
demographics were collected, and these participants only completed the BAS-2 at two time-
points three weeks apart.
Body appreciation. Participants completed a Romanian translation of the BAS-2
(Tylka & Wood-Barcalow, 2015b). The scale was translated from English into Romanian
using the parallel back-translation procedure (Brislin, 1986). First, a bilingual individual
unaffiliated with the study translated the scale from English to Romanian, while a second
individual translated this version back into English. Next, the items obtained were assessed
by a committee consisting of the individuals who participated in the translation process, the
third to fifth authors, and two psychology professors who settled minor discrepancies in the
translation through consensus. All items are presented in Table 1 and were rated on a 5-point
scale, ranging from 1 (Never) to 5 (Always).
Symptoms of disordered eating. The questionnaire included the Body Image
Screening Questionnaire for Eating Disorder Early Detection (BISQ; Jenaro, Flores, Tomsa,
& Bermejo, 2011; Romanian translation: Tomsa, Istfan, Jenaro, Flores, & Bermejo, 2012).
This is a 24-item scale that measures symptomatology of disordered eating along five
dimensions, namely bulimia, anorexia, orthorexia, perception of obesity, and vigorexia (i.e.,
muscle dysmorphia”. Items were rated on a 6-point scale (1 = Never, 6 = Always). Although a
Romanian translation of this scale has been prepared (Tomsa et al., 2012), the authors failed
to factor analyse scores derived from the scale and reported that internal consistency
coefficients for all but one of the subscales were less than adequate. However, total scores
were reported to yield evidence of internal consistency (Cronbach α = .85) and discriminant
validity. In the present study, we therefore used an overall BISQ score, computed as the mean
of all items following reverse-coding of relevant items. Higher scores on this scale reflect
Body Appreciation Scale-2
greater degree of eating disorder symptomatology (Tomsa et al., 2012). Cronbach’s alpha for
the overall score in the present study was .82 for women and .80 for men.
Self-esteem. We included the Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965;
Romanian translation: Schmitt & Allik, 2005), a 10-item measure of global self-evaluations
of worth as a human being. All items were rated on a 4-point scale (1 = Definitely disagree, 4
= Definitely agree). An overall score was computed as the mean of all 10 items, following
reverse-coding of 5 items. Higher scores on this scale reflect greater self-esteem. The
Romanian version of the RSES has evidenced factorial validity (Schmitt & Allik, 2005) and
patterns of convergent validity (e.g., Sava, Maricuțoiu, Rusu, Macsinga, & Vîrgā, 2011). In
the present study, Cronbach α for this scale was .87 for women and .84 for men.
Satisfaction with life. The survey package included the 5-item Satisfaction With Life
Scale (SWLS; Diener, Emmons, Larsen, & Griffin, 1985; Romanian translation: Lambru et
al., 2012), which taps global assessments of life satisfaction. Items on this scale were rated on
a 7-point scale (1 = Strongly disagree, 7 = Strongly agree). An overall score was computed at
the mean of all five items, so that higher scores reflect greater life satisfaction. Scores on the
Romanian version of the scale have been shown to have a one-dimensional factor structure
and evidence patterns of convergent validity (Lambru et al., 2012). In the present study,
Cronbach’s alpha was .85 for women and .86 for men.
Subjective happiness. We also included the Subjective Happiness Scale (SHS;
Lyubomirsky & Lepper, 1999), a 4-item measure that assesses the degree to which
respondents consider themselves to be happy or unhappy persons. Items on the scale were
rated on a 7-point scales with various anchors. Because a Romanian translation of this scale
was not available at the time of the study, we prepared such a translation using the back-
translation technique (Brislin, 1986), as described above. To examine the factor structure of
scores on this translation of the scale, we computed a principal-axis EFA with the total
sample (N = 453) using a quartimax rotation because of the expectation of a single factor. The
Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy (.77) and the significance of
Bartlett’s test of sphericity, χ2(6) = 642.81, p < .001, indicated that these data had adequate
factorability. The results of the EFA indicated that only a single factor should be extracted,
with an eigenvalue (λ) of 2.52 and explaining 63.0% of the total variance. All four items of
the SHS loaded onto this factor and had good-to-excellent factor loadings (.56-.86). These
results are consistent with other translational studies of the SHS (e.g., Swami, 2008; Swami et
al., 2009). Based on these results, we computed an overall SHS score as the mean of all four
items, with higher scores reflecting greater subjective happiness. Cronbach’s alpha was .89
for women and .92 for men.
Demographics. Participants provided their demographic details consisting of sex,
age, degree programme, ethnicity, and religion. Participants were also asked to self-reported
their height and weight, which we used to calculate BMI as kg/m2.
Ethics approval for this project was obtained from the relevant university ethics
committee. The main study was advertised as a project on health and well-being on campus
locations between January and March 2017. Individuals who agreed to participate were
provided with an information sheet that contained brief information about the projects and
provided written informed consent. They then completed an anonymous paper-and-pencil
questionnaire in a private cubicle. The questionnaire took approximately 20 minutes to
complete and the order of presentation of the scales above was semi-randomised
(demographic items were always presented last). All participation was voluntary and
participants did not receive any remuneration. Upon return of completed questionnaires,
participants were provided with written debrief information.
Body Appreciation Scale-2
The test-retest participant sample was recruited in June 2017, and participants were
only eligible if they had not taken part in the earlier study. Where participation was agreed,
participants provided written informed consent and complete the BAS-2 along with brief
demographics in a private cubicle. Three weeks after initial testing, participants were
contacted and asked to completed the BAS-2 a second time. No participant declined the
invitation. Unique codes were generated to link test-retest data and were destroyed prior to
analyses. All participation was voluntary and participants did not receive any remuneration.
At the end of the second testing session, participants were provided with written debrief
Missing data accounted for < 0.6% of the main dataset and were missing completely
at random, as determined by Little’s MCAR analyses. In this case, we used the mean
replacement technique to estimate missing values. There were no missing data in the test-
retest dataset. We used a two-step analytic strategy to examine the psychometric properties of
the Romanian BAS-2. First, we split the main dataset so that the first subsample had a
random allocation of 100 women and 100 men, and so that the second subsample included all
remaining participants (women n = 112, men n = 141). In the first subsample, we assessed the
factor structure of BAS-2 scores in women and men separately, using principal-axis factor
analysis in IBM SPSS Statistics v.20. The sample sizes for women and men in this subsample
met a conservative item-to-participant ratio of 10:1 (Tabacknick & Fidell, 2013). As in Tylka
and Wood-Barcalow (2015b), we used a quartimax rotation as we hypothesised that BAS-2
scores would have a single, orthogonal factor (Pedhazur & Schmelkin, 1990). Factor loadings
were interpreted using Tabachnick and Fidell’s (2013) recommendations, with loadings of .71
and above considered excellent, .63-.70 considered very good, .55-.62 considered good, .
33-.54 considered fair, and .32 or lower considered poor.
Next, data from the second subsample (n = 243) were subjected to CFA using
Analysis of Moment Structures (AMOS v.23). Hypothesized modelling was based on the
results of the EFA in the first subsample, as well as the expected one-dimensional factor
structure. Standard goodness-of-fit indices were selected a priori to assess the measurement
models (Hu & Bentler, 1999). The normed model chi-square (χ²normed) is reported with lower
values of the overall model χ² indicating goodness-of-fit. A χ²normed value of < 3.00 indicates
good fit (we also report the mean chi-square value, χ²M, for completeness). The Steiger-Lind
root mean square error of approximation (RMSEA) and its 90% confidence interval provide a
correction for model complexity. RMSEA values close to .06 indicate a good fit, with values
ranging to .10 representing a mediocre fit. The standardized root mean square residual
(SRMR) assesses the mean absolute correlation residual and is a badness-of-fit index: the
smaller the SRMR, the better the model fit. A cut-off value for SRMR is recommended to be
close to or < .09. The comparative fit index (CFI) measures the proportionate improvement in
fit by comparing a target model with a more restricted, nested baseline model. The CFI
reflects a goodness-of-fit index and is recommended to close to or > .95 for good fit, but can
be as low as .90 for adequate fit.
In this second subsample, we also examined whether BAS-2 scores were invariant
across sex at the configural (i.e., whether similar factors are measured), factor loading (i.e.,
whether the magnitude of factor loadings is the same), and intercept (i.e., whether the
intercept of the regression relating each item to its factor is the same) levels (Chen, 2007). In
both subsamples separately, and for both women and men, we examined internal consistency
by computing Cronbach’s α. For the test-retest sample (n = 109), intraclass correlation
coefficients and paired-sample t-tests were used to estimate the stability of BAS-2’s scores in
women and men separately. To examine sex differences in BAS-2 scores, we planned on
using independent-samples t-tests. Finally, we examined convergent validity by computing
Body Appreciation Scale-2
bivariate correlations between BAS-2 scores and scores on all other included measures.
Principal-Axis Factor Analysis
Female subsample. Descriptive statistics (Ms and SDs) for all BAS-2 items for the
female subsample (n = 100) are reported in Table 1. Based on item distribution, average
correlation with other items, and item-total correlations (Clark & Watson, 1995), these data
were suitable for factor analysis. The Kaiser-Meyer-Olkin (KMO) measure of sampling
adequacy (.92) and the significance of Bartlett’s test of sphericity, χ2(45) = 674.52, p < .001,
indicated that these data had adequate factorability. The results of the factor analysis
indicated that only a single factor should be extracted; this factor had an λ of 6.14 and
explained 61.4% of the total variance. All 10 items of the BAS-2 loaded onto this factor and
had good-to-excellent factor loadings (see Table 1). Cronbach’s alpha was .93 (95% CI = .
91-.95), providing evidence of internal consistency.
Male subsample. Descriptive statistics for all 10 items for men (n = 100) are reported
Table 1. These data passed standard criteria for factor analysis (Clark & Watson, 1995). The
KMO measure of sampling adequacy was .88 and Bartlett’s test of sphericity was significant,
χ2(45) = 368.06, p < .001, indicative of adequate factorability. Unlike the female sample,
however, the results of the factor analysis for men pointed to the existence of two factors with
λ > 1.0. In this case, we applied parallel analysis (Hayton, Allen, & Scarpello, 2004) to
determine the number of factors that should be extracted. Parallel analysis reduces the
likelihood of factor over-retention, which is known to affect other factor extraction criteria
(Patil, McPherson, & Friesner, 2010), and works by creating random datasets with the same
number of cases and variables as the actual dataset. Factors in the actual data are only
retained if their eigenvalues are greater than the eigenvalues from the random data (Hayton et
al., 2004). The results of parallel analysis in the present study suggested that only a single
factor should be extracted: only the first factor from the actual data had λ greater than the
criterion λ generated from the random data (i.e., 4.18 [actual data] compared to 2.99 [random
data] for the first factor, and 1.46 [actual data] compared to 2.17 [random data] for the second
factor). As such, we retained only the first factor, which explained 41.8% of the common
variance. It is important to note that although all items loaded on to the primary factor,
loadings were more variable than in the female subsample, with half the items only reaching
“fair” loadings (see Table 1). Cronbach’s alpha was .84 (95% CI = .78-.88), demonstrating
Confirmatory Factor Analysis
Hypothesized model fit. The results of the principal-axis factor analyses with the
above subsamples point to a one-dimensional factor structure for Romanian BAS-2 scores,
with all 10 items loading onto this dimension. We, therefore, examined the fit of this
hypothesised one-factor solution using CFA with the second subsample (n = 243). Skewness
and kurtosis were below critical limits for CFA (Kline, 2005), so no data were transformed
prior to analysis. The standardised estimates of factor loadings for the best-fitting model were
all very good-to-excellent (see Figure 1). Fit indices values were found to be: χ²M(35) =
213.719, p < .001, χ²normed = 6.106, p < .001, CFI = .848, RMSEA = .142 with 90% CI = .
124-.161, SRMR = .068. These results indicate that fit indices values were not found to be at
acceptable intervals. Brown (2014) recommends that error covariances should be freed if
there is a theoretical or measurement rationale for doing so. We, therefore, consulted
modification indices to free error covariances between Items #2 and #9 due to the similar
wording between these items, and between Items #4 and #7 (both items reflect positive and
unique body characteristics) and Items #1 and #5 (both items represent attentiveness to the
body). This one-dimensional structure provided an acceptable fit to the data: χ²M(32) =
98.029, p < .001, χ²normed = 3.063, p < .001, CFI = .950, RMSEA = .090 (low = .070, high = .
Body Appreciation Scale-2
111), SRMR = .049 (see Figure 1). Cronbach’s alpha was .89 (95% CI = .87-.91) for this
Sex invariance. The unconstrained model had adequate fit in the female subsample
(n = 112): χ²M(32) = 59.937, p < .001, χ²normed = 1.873, p < .001, CFI = .957, RMSEA = .089
with 90% CI = .053-.123, SRMR = .050. However, the unconstrained model had poor-to-
adequate fit in the male subsample (n = 141): χ²M(32) = 77.505, χ²normed = 2.422, CFI = .921,
RMSEA = .101 with 90% CI = .072-.130, SRMR = .062. Further subsample metrics are
reported in Table 2. Differences between the unconstrained and fully constrained model were
not significant, indicating that the structure of the model achieved factor loading invariance
across sex, Δχ²(9) = 9.986, p = .352. In addition, intercept invariance was evaluated, where
all item-factor intercepts were constrained equally across participant sex and evaluated
against the factor loading invariance model. According to the Δχ², intercept invariance was
not supported, Δχ²(10) = 38.389, p < .001. However, it has been suggested that Δχ² alone may
be an overly stringent criterion of intercept invariance (Chen, Sousa, & West, 2005); Chen
(2007) also recommends exploring practical model fit changes between intercept and
factorial models, where invariance is demonstrated if ΔCFI < -.010 and ΔRMSEA < .015 or
ΔSRMR < .030. Values in the present study were: ΔCFI = -.023, ΔRMSEA = .050, and
ΔSRMR < .001, suggestive of intercept non-invariance. Overall, these data suggest that
measurement invariance was not fully achieved, so sex differences in mean scores were not
Using the main sample (n = 453), we examined convergent validity by examining
bivariate correlations between BAS-2 scores and scores on all other measures. As reported in
Table 3, higher body appreciation was significantly and negatively correlated with symptoms
of disordered eating and BMI, and positively correlated with self-esteem, satisfaction with
life, and subjective happiness in both women and men.
For the test-retest female sample (n = 57), the intraclass correlation coefficient
between the BAS-2 scores at the first and second testing sessions three weeks apart was .82
(CI = .59-.86). Mean scores were not significantly different across the first (M = 3.90, SD =
0.71) and second testing sessions (M = 4.02, SD = 0.81), t(56) = 1.33, p = .188. In men (n =
52), the intraclass correlation coefficient between the BAS-2 scores at the two testing
sessions was .87 (CI = .78-.93). Mean scores were not significantly different across the first
(M = 3.78, SD = 0.76) and second testing sessions (M = 3.74, SD = 0.81), t(51) = 0.45, p = .
The primary aim of the present study was to examine the factor structure of a
Romanian translation of the BAS-2. Using principal-axis factors analysis with part of our
sample, we found that Romanian BAS-2 scores had a one-dimensional factor structure with
all 10 items in both women and men (although it should be noted that factor loadings for half
the items for men were fair, whereas they were good-to-excellent for women). In addition,
using CFA, we found that the one-factor structure had adequate fit to the data from part of the
total sample, albeit after freeing error covariances due to similar item wording and content
between three item pairs. In broad outline, these findings are consistent with previous studies
that have examined the factor structure of BAS-2 scores across different linguistic groups
(Alleva et al., 2016; Atari, 2016; Jovic et al., 2016; Kertechian & Swami, 2017; Razmus &
Razmus, 2017; Swami & Ng, 2015; Swami et al., 2016, 2017; Tylka & Wood-Barcalow,
Body Appreciation Scale-2
However, we also found that BAS-2 scores did not evidence configural and intercept
invariance across women and men (although factor loading invariance was supported). This is
an important finding for several reasons. First, we believe this is the first instance where the
sex invariance of BAS-2 scores has not been fully supported: previous CFA-based studies
have all reported full sex invariance (Swami et al., 2016, 2017; Tylka & Wood-Barcalow,
2015b). Second, the lack of sex invariance may indicate that the same latent construct (i.e.,
“body appreciation”) is not being measured by the BAS-2 in women and men. More
specifically, it may be that some items on the BAS-2 (Items #3, 5-7, and 10) are less pertinent
or central to the meaning of body appreciation in Romanian men. Although the reasons for
this may be difficult to ascertain without further research (particularly of a qualitative nature),
it is possible that Romanian men hold different attitudes toward the body (Cornelia, 2012),
which in turn shape their experiences of body appreciation. A third reason why this finding is
important is because the lack of invariance of sex invariance means that a direct comparison
of latent BAS-2 scores cannot be conducted.
Our results also indicated that BAS-2 scores yielded evidence of internal consistency
in women and men. In addition, and as hypothesised, we found that BAS-2 scores were
significantly and negatively correlated with symptoms of disordered eating and BMI, and
positively correlated with self-esteem, satisfaction with life, and subjective happiness.
Importantly, the strengths of these correlations were generally moderate, although the
association between body appreciation and self-esteem in women was large. Finally, our
analyses also indicated that BAS-2 scores had adequate test-retest reliability in both women
and men after three weeks. Taken together, these results provide evidence that the BAS-2 is a
valid and reliable tool for the assessment body appreciation in Romanian-speaking
populations. This is important as there are currently few of such tools available for this
linguistic group. The availability of a Romanian translation of the BAS-2 should, therefore,
facilitate research in a previously neglected population.
The present results also make contribution to two further aspects of note. First, we
found that that a Romanian translation of the SHS had a one-dimensional factor structure.
Additionally, the pattern of correlations between SHS scores and scores on measures of self-
esteem and satisfaction with life supports the convergent validity of the former. On the other
hand, we advise some caution when using the BISQ. Although this scale was developed to
measure aspects of disordered eating (Jenaro et al., 2011), evidence of adequate psychometric
properties for this scale is less than optimal. BISQ subscale scores appear to have less-than-
adequate internal consistency coefficients and, although our use of overall scores may be a
better alternative, it is vital that future research fully interrogates the construct validity of this
measure. In the present work, we were hampered by the lack of translated and validated
measures of body image and disordered eating for Romanian populations, and we suggest
that associations with BISQ scores be treated with caution until this scale had received full
Our study is also limited in a number of additional ways. First, our reliance on a
student sample means that our findings cannot be generalized to the wider Romanian
population, where the percentage of the population in higher education is proportionately
small (Eurostat, 2015). It would also be useful to extend the present study to include other
Romanian-speaking populations, such as those in Moldova or among the large Romanian
migrant diaspora elsewhere in Europe (as well as by several million more as a second
language (European Commission, 2012). Finally, future work could extend the present results
by examining associations between body appreciation and other factors, to establish the
divergent and incremental validity of the Romanian BAS-2.
Body Appreciation Scale-2
These limitations notwithstanding, the present study provides preliminary evidence to
support the factorial and convergent validity, as well as the internal consistency and test-retest
reliability, of a Romanian translation of the BAS-2. However, the lack of sex invariance
throws up practical concerns for scholars intending to use this measure. Our recommendation
is that scholars and practitioners using the Romanian BAS-2 conduct tests sex invariance in
their own datasets to help them determine whether comparisons of latent scores between
women and men will be permissible. More broadly, it is notable that the BAS-2 has been
translated into a number of different languages (Swami, 2017; Webb et al., 2015) and that
scores on the measure retain a one-factor structure across all linguistic groups that have been
investigated thus far. This is important as it may facilitate cross-cultural comparisons of BAS-
2 scores, which requires that the latent dimensionality of those scores is the same across
group. In addition, the availability of a Romanian translation of the BAS-2 provides a vital
tool for the assessment of body image in a hitherto neglected linguistic group. We are hopeful
that the availability of this measure will stimulate further research into body image in
Alleva, J. M., Martijn, C., Veldhuis, J., & Tylka, T. L. (2016). A Dutch translation and
validation of the Body Appreciation Scale-2: An investigation with female university
students in the Netherlands. Body Image, 19, 44-48. doi:10.1016/j.bodyim.2016.08.008
Atari, M. (2016). Factor structure and psychometric properties of the Body Appreciation
Scale-2 in Iran. Body Image, 18, 1-4. doi:10.1016/j.bodyim.2016.04.006
Atari, M., Akbari-Zardkhaneh, S., Mohammadi, L., & Soufiabadi, M. (2015). The factor
structure and psychometric properties of the Persian version of Body Appreciation
Scale. American Journal of Applied Psychology, 3, 62-66. doi:10.12691/ajap-3-3-3
Avalos, L. C., Tylka, T. L., & Wood-Barcalow, N. (2005). The Body Appreciation Scale:
Development and psychometric evaluation. Body Image, 2, 285-297.
Bakalim, O., & Tasleden-Karçkay, A. (2016). Body Appreciation Scale: Evaluation of the
factor structure and psychometric properties among male and female Turkish university
students. Mersin Üniversitesi Egitim Fakültesi Dergisi, 12, 410-422.
Brislin, R.W. (1986). The wording and translation of research instruments. In W. Lonner & J.
Berry (Eds.), Field methods in cross-cultural research (pp. 137-164). Beverly Hills,
Brown, T. A. (2014). Confirmatory factor analysis for applied research (2nd ed.). London:
Catina, A., & Joja, O. (2001). Emerging markets: Submerging women. In M. Nasser, M. A.
Katzman & R. A. Gordon (Eds.), Eating disorders and cultures in transition (pp. 111-
119). New York, NY: Taylor & Francis.
Body Appreciation Scale-2
Chen, F. F. (2007). Sensitivity of goodness of fit indices to lack of measurement invariance.
Structural Equation Modeling, 14, 464-504. doi:10.1207/s15328007sem1203_7
Chen, F. F., Sousa, K. H., & West, S. G. (2005). Testing measurement invariance of second-
order factor models. Structural Equation Modeling, 12, 471-492.
Clark, L. A., & Watson, D. (1995). Constructing validity: Basic issues in objective scale
development. Psychological Assessment, 7, 309-319. doi:10.1037/1040-3518.104.22.1689
Coppola, A. M., Dimler, A. J., Letendre, T. S., & McHugh, T.-L. F. (2017). “We are given a
body to walk this earth”: The body pride experience of young Aboriginal men and
women. Qualitative Research in Sport, Exercise, and Health, 9, 4-17.
Cornelia, R. (2012). Body mass index: Knowledge, practice, and health evaluation.
International Journal of Collaborative Research on Internal Medicine and Public
Health, 4, 1276-1284.
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Satisfaction with Life
Scale. Journal of Personality Assessment, 49, 71-75. doi:10.1207/
Dumitrescu, A. L., Zetu, L., Teslaru, S., Dogaru, B. C., & Dogaru, C. D. (2008). Is it an
association between body appreciation, self-criticism, oral health status, and oral
health-related behaviors? Romanian Journal of Internal Medicine, 46, 343-350.
Eisler, I. (2001). Commentary 1. In M. Nasser, M. A. Katzman & R. A. Gordon (Eds.), Eating
disorders and cultures in transition (pp. 120-123). New York, NY: Taylor & Francis.
European Commission. (2012). Europeans and their languages (Special Eurobarometer 386).
Brussels: European Commission.
Eurostat. (2015). Tertiary education statistics. Available online at:
Ferreira, L., Neves, A. N., & Tavares, M. C. G. C. F. (2014). Validity of body image scales for
Brazilian older adults. Motriz: Revista de Educação Física, 20, 359-373.
Frederick, D. A., Forbes, G. B., & Berezovskaya, A. (2008). Female body dissatisfaction and
the perceptions of the attractive female body in Ghana, the Ukraine, and the United
States. Psychological Topics, 17, 203-219.
Hayton, J. C., Allen, D. G., & Scarpello, V. (2004). Factor retention decisions in exploratory
factor analysis: A tutorial on parallel analysis. Organization Research Methods, 7, 191-
205. doi: 10.1177/1094428104263675
Holmqvist, K., & Frisén, A. (2012). “I bet they aren’t that perfect in reality”: Appearance
ideals viewed from the perspective of adolescents with a positive body image. Body
Image, 9, 388-395. doi:10.1016/j.bodyim.2012.03.007
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural Equation Modeling,
6, 1-55. doi:10.1080/10705519909540118
Jenaro, C., Flores, N., Tomsa, R., & Bermejo, M. B. (2011). Body Image Screening
Questionnaire for Eating Disorder Early Detection. Journal of Experiential
Psychotherapy, 14, 3-11.
Jovic, M., Sforza, M., Jovanovic, M., & Jovic, M. (2016). The Acceptance of Cosmetic
Surgery Scale: Confirmatory factor analyses and validation among Serbian adults.
Current Psychology, in press. doi:10.1007/s12144-016-9458-7
Body Appreciation Scale-2
Kertechian, S., & Swami, V. (2017). An examination of the factor structure and sex
invariance of a French translation of the Body Appreciation Scale-2 in university
students. Body Image, 21, 26-29. doi:10.1016/j.bodyim.2017.02.005
Kline, R. B. (2005). Principles and practices of structural equation modeling (2nd ed.). New
York, NY: Guilford Press.
Lambru, I., Butucescu, A., Uscātescu, L., Constantinescu, P.-M., Sandu, G., & Stevens, M. J.
(2012). Romanian adaptation of the Satisfaction With Life Scale. Journal of
Psychological and Educational Research, 1, 17-33.
Lobera, I. J., & Ríos, P. B. (2011). Spanish version of the Body Appreciation Scale (BAS) for
adolescents. Spanish Journal of Psychology, 14, 411-420.
Lyubomirsky, S., & Lepper, H. S. (1999). A measure of subjective happiness: Preliminary
reliability and construct validation. Social Indicators Research, 46, 137-155.
McHugh, T.-L., Coppola, A. M., & Sabiston, C. M. (2014). “I’m thankful for being Native
and my body is part of that”: The body pride experiences of young Aboriginal women
in Canada. Body Image, 11, 318-327. doi:10.1016/j.bodyim.2014.05.004
Mocanu, V. (2013). Prevalence of overweight and obesity in urban elementary school
children in northeastern Romania: Its relationship with socioeconomic status and
associated dietary and lifestyle factors. BioMed Research, 2013, article 537451.
Ng, S.-K., Barron, D., & Swami, V. (2015). Factor structure and psychometric properties of
the Body Appreciation Scale among adults in Hong Kong. Body Image, 13, 1-8.
Patil, V. H., McPherson, M. Q., & Friesner, D. (2010). The use of exploratory factor analysis
in public health: A note on parallel analysis as a factor retention criterion. American
Journal of Health Promotion, 24, 178-181. doi: 10.4278/ajhp.08033131
Pedhazur, E. J., & Schmelkin, L. P. (1991). Measurement, design, and analysis: An
integrated approach. Hillsdale, NJ: Lawrence Erlbaum.
Razmus, M., & Razmus, W. (2017). Evaluting the psychometric properties of the Polish
version of the Body Appreciation Scale-2. Body Image, in press.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton
Sava, F. A., Maricuțoiu, L. P., Rusu, S., Macsinga, I., & Vîrgā, D. (2011). Implicit and
explicit self-esteem and irrational beliefs. Journal of Cognitive and Behavioral
Psychotherapies, 11, 97-111.
Schmitt, D. P., & Allik, J. (2005). Simultaneous administration of the Rosenberg Self-Esteem
Scale in 53 nations: Exploring the universal and culture-specific features of global self-
esteem. Journal of Personality and Social Psychology, 89, 623-642. doi:10.1037/0022-
Smolak, L., & Cash, T. F. (2011). Future challenges for body image science, practice, and
prevention. In T. F. Cash, & L. Smolak (Eds.), Body image: A handbook of science,
practice, and prevention (2nd ed., pp. 471-478). New York, NY: Guilford Press.
Swami, V. (2008). Translation and validation of the Malay Subjective Happiness Scale.
Social Indicators Research, 88, 347-353. doi:10.1007/s11205-007-9195-2
Swami, V. (2017, forthcoming). Considering positive body image through the lens of culture
and minority social identities. In C. Markey, E. Daniels, & M. Gillen (Eds.), The body
positive: Understanding and improving body image in science and practice.
Cambridge: Cambridge University Press.
Body Appreciation Scale-2
Swami, V., Campana, A. N. N. B. C., Ferreira, L., Barrett, S., Harris, A. S., & Tavares, M. C.
G. C. F. (2011). The Acceptance of Cosmetic Surgery Scale: Initial examination of its
factor structure and correlates among Brazilian adults. Body Image, 8, 179-185.
Swami, V., & Chamorro-Premuzic, T. (2008). Factor structure of the Body Appreciation Scale
among Malaysian women. Body Image, 5, 409-413.
Swami, V., García, A. A., & Barron, D. (2017). Factor structure and psychometric properties
of a Spanish translation of the Body Appreciation Scale-2 (BAS-2). Body Image, 22,
Swami, V., Hwang, C.-S., & Jung, J. (2012). Factor structure and correlates of the Acceptance
of Cosmetic Surgery Scale among South Korean university students. Aesthetic Surgery
Journal, 32, 220-229. doi:10.1177/1090820X11431577
Swami, V., & Jaafar. J. L. (2012). Factor structure of the Body Appreciation Scale among
Indonesian women and men: Further evidence of a two-factor solution in a non-Western
population. Body Image, 9, 539-542. doi:10.1016/j.bodyim.2012.06.002
Swami, V., Mada, R., & Tovée, M. J. (2012). Weight discrepancy and body appreciation of
Zimbabwean women in Zimbabwe and Britain. Body Image, 9, 559-562.
Swami, V., & Ng, S.-K. (2015). Factor structure and psychometric properties of the Body
Appreciation Scale-2 among adults in Hong Kong. Body Image, 15, 68-71.
Swami, V., Ng, S.-K., & Barron, D. (2016). Translation and psychometric evaluation of a
Standard Chinese version of the Body Appreciation Scale-2. Body Image, 18, 23-26.
Swami, V., Özgen, L., Gökçen, E., & Petrides, K. V. (2015). Body image among female
university students in Turkey: Concurrent translation and validation of three body
image measures. International Journal of Culture and Mental Health, 8, 176-191.
Swami, V., Stieger, S., Haubner, T., & Voracek, M. (2008). German translation and
psychometric evaluation of the Body Appreciation Scale. Body Image, 5, 122-127.
Swami, V., Stieger, S., Voracek, M., Dressler, S. G., Eisma, L., & Furnham, A. (2009).
Psychometric evaluation of the Tagalog and German Subjective Happiness Scales and a
cross-cultural comparison. Social Indicators Research, 93, 393-406.
Tabachnick, B. G., & Fidell, L. S. (2013). Using multivariate statistics (6th ed.). New York,
Taylor, D., Szpakowska, I., & Swami, V. (2013). Weight discrepancy and body appreciation
among women in Poland and Britain. Body Image, 10, 628-631.
Tomsa, R., Istfan, N., Jenaro C., Flores, N., & Bermejo, M. B. (2012). Body Image Screening
Questionnaire for eating disorder early detection: A Romanian replication. Procedia –
Social and Behavioral Sciences, 33, 423-427. doi:10.1016/j.sbspro.2012.01.156
Tylka, T. L. (2012). Positive psychology perspectives on body image. In T. F. Cash (Ed.),
Encyclopedia of body image and human appearance, Volume 2 (pp. 657-663). London:
Tylka, T. L. (2013). Evidence for the Body Appreciation Scale’s measurement
equivalence/invariance between U.S. college women and men. Body Image, 10, 415-
Body Appreciation Scale-2
Tylka, T. L., & Wood-Barcalow, N. L. (2015a). What is and what is not positive body image?
Conceptual foundations and construct definition. Body Image, 14, 118-129.
Tylka, T. L., & Wood-Barcalow, N. L. (2015b). The Body Appreciation Scale-2: Item
refinement and psychometric evaluation. Body Image, 12, 53-67.
Ulijaszek, S. J., & Koziel, S. (2007). Nutrition transition and dietary energy availability in
Eastern Europe after the collapse of communism. Economics and Human Biology, 5,
Webb, J. B., Wood-Barcalow, N. L., & Tylka, T. L. (2015). Assessing positive body image:
Contemporary approaches and future directions. Body Image, 14, 130-145.
Wood-Barcalow, N. L., Tylka, T. L., & Augustus-Horvath, C. L. (2010). “But I like my
body”: Positive body image characteristics and a holistic model for young-adult
women. Body Image, 7, 106-116. doi:10.1016/j.bodyim.2010.01.001
Table 1. Items of the Body Appreciation Scales-2 in English and Romanian (Italics), Descriptive Statistics for Women (n = 100) and Men (n =
100), and Item-Factor Loadings in the First Subsample.
Item Female subsample Male subsample
M (SD) Loading M (SD) Loading
1. I respect my body/Îmi respect corpul 4.24 (0.73) .64 4.10
2. I feel good about my body/Mă simt bine cu corpul meu 3.80 (0.95) .88 4.29
3. I feel that my body has at least some good qualities/Simt ca acest corp are
cel puțin câteva calități
4.13 (0.91) .75 4.24
4. I take a positive attitude towards my body/Am o atitudine pozitivă față de
3.82 (1.01) .85 4.20
5. I am attentive to my body’s needs/Sunt atent la nevoile corpului meu 3.82 (0.86) .67 3.92
6. I feel love for my body/Simt dragoste față de corpul meu 3.34 (1.16) .77 3.46
7. I appreciate the different and unique characteristics of my body/Apreciez
caracteristicile diferite și unice ale corpului meu
3.83 (1.01) .79 3.74
8. My behaviour reveals my positive attitude toward my body; for example, I
hold my head high and smile/Comportamentul meu arată atitudinea mea
3.52 (1.01) .80 3.73
pozitivă față de corpul meu
9. I am comfortable in my body/Mă simt bine în corpul meu 3.86 (1.04) .89 4.38
10. I feel like I am beautiful even if I am different from media images of
attractive people (e.g., models, actresses/actors)/Mă simt frumos/frumoasă
chiar dacă sunt diferit/ă față de imaginile din media a persoanelor
atrăgătoare (ex. modele, actrițe, actori)
3.98 (0.97) .77 4.05
Table 2. Model Fit Indices and Tests of Measurement Invariance for the One-Factor Body
Appreciation Scale-2 Model Across Participant Sex in the Second Subsample.
χ²Mdf χ²normed RMSEA
Configural Invariance 150.04
2.344 .073 (.058, .
2.192 .069 (.054, .
Intercept Invariance 198.41
2.391 .074 (.061, .
Note. χ²M = mean chi-square; χ²normed = normed chi-square; RMSEA = root mean square error
of approximation; SRMR = standardized root mean square residual; CFI = comparative fit
index; CI = confidence interval.
Table 3. Correlation Matrix with Associations for Women (n = 212) Reported in the Top
Diagonal and Men (n = 241) Reported in the Bottom Diagonal.
(1) (2) (3) (4) (5) (6)
(1) Body appreciation -.54** .71** .52** .44** -.30**
(2) Disordered eating -.45** -.45** -.31** -.13 .43**
(3) Self-esteem .52** -.32** .53** .41** -.14*
(4) Satisfaction with life .51** -.27** .58** .57** -.04
(5) Subjective happiness .41** -.12 .46** .56** -.04
(6) Body mass index -.28** .21** .15* .10 .07
Women M3.89 3.31 3.23 5.08 4.63 21.30
SD 0.72 0.59 0.52 1.16 0.80 3.65
Men M3.99 2.15 3.22 5.08 4.85 22.95
SD 0.62 0.54 0.50 1.14 0.82 3.09
Note. * p < .05, ** p < .001.
Figure 1. Path diagram and estimates for the one-dimensional model of the Body
Appreciation Scale-2. The large oval is the latent construct, with the rectangles representing
measured variables, and the small circles with numbers representing the unstandardized
residual variables (variances). Error covariances were .14 for Items #2 and #9, .21 for Items
#1 and #5, and -.16 for Items #4 and #7. The path factor loadings are unstandardised with
significance levels determined by critical ratios (all p < .001).