María Soledad Torregrosa Díez, María Isabel Gómez Núñez, Ricardo Sanmartín López, José Manuel García Fernández,

Annette M. La Greca, Xinyue Zhou, Jesús Redondo Pacheco, and Candido J. Ingles Saura

ISSN 0214 - 9915 CODEN PSOTEG

Psicothema 2022, Vol. 34, No. 1, 126-133
Copyright © 2022 Psicothema
doi: 10.7334/psicothema2021.42 www.psicothema.com

Article

Measurement Invariance and Latent Mean Differences Between

American, Spanish and Chinese Adolescents Using the Social Anxiety
Scale for Adolescents (SAS-A)
María Soledad Torregrosa Díez1, María Isabel Gómez Núñez1, Ricardo Sanmartín López2,
José Manuel García Fernández2, Annette M. La Greca3, Xinyue Zhou4, Jesús Redondo Pacheco5,
and Candido J. Ingles Saura6
1
International University of La Rioja (Spain), 2 University of Alicante (Spain), 3 University of Miami (USA), 4 Zhejiang University (China),
5
Pontificia Bolivariana University (Colombia), and 6 Miguel Hernández University (Spain)

Abstract Resumen
Background: Social anxiety is one of the most prevalent disorders among Análisis de Invarianza y Diferencias de Medias Latentes Entre
adolescents (Stein et al., 2017). The main aim of this study was to analyze Adolescentes Americanos, Españoles y Chinos Usando la Escala de
the equivalence of scores on the Social Anxiety Scale for Adolescents Ansiedad Social para Adolescentes (SAS-A). Antecedentes: la ansiedad
(SAS-A) using structural equation modeling and identify differences in social es uno de los trastornos con mayor prevalencia en adolescentes
latent means of social anxiety in China, Spain, and the USA. Method: (Stein et al., 2017). Así, el propósito principal de este estudio fue analizar
Random sampling was used to recruit participants, which included 536 la invarianza de la Escala de Ansiedad Social para Adolescentes (SAS-A)
Chinese (46% girls), 1,178 Spanish (55.3% girls) and 866 North American mediante un modelo de ecuaciones estructurales y examinar las diferencias
(55.1% girls) adolescents. The participants’ ages ranged between 14 and de medias latentes en ansiedad social en adolescentes de China, España y
17 years old. Results: The SAS-A three-factor correlated model of social EE.UU. Método: los participantes se seleccionaron a través de muestreo
anxiety remained invariant between the Spanish and North American aleatorio: 534 chinos (46% chicas), 1.178 españoles (55,3% chicas) y
adolescents, but results could not be replicated in the Chinese adolescents 866 norteamericanos (55,1% chicas), con edades comprendidas entre los
[M2 = ΔS-Bχ2 (Δdf, p) = 4732.56 (36, < .01)]. Analyses of latent differences 14 y 17 años. Resultados: las puntuaciones del modelo de tres factores
between Spain and the USA showed that Spanish adolescents had higher correlacionados de ansiedad social de la SAS-A resultaron invariantes
scores than North Americans for Fear of Negative Evaluation (TS = -9.630; entre adolescentes españoles y norteamericanos, pero estos resultados no
d = .44) and for Social Avoidance and General Anxiety towards people fueron replicados en adolescentes chinos [M2 = ΔS-Bχ2 (Δdf, p) = 4732.56
(TS = -2.717; d = .12). Conclusions: The results are interpreted according (36, < .01)]. El análisis de medias latentes entre España y EE.UU. mostró
to the cultural traits of individualism-collectivism and self-construal, and que los adolescentes españoles manifestaban niveles más altos de Miedo
practical implications are discussed. ante las evaluaciones negativas (TS = -9.630; d = .44) y Evitación social y
Keywords: Anxiety Scale for Adolescents, SAS-A, measurement invariance, ansiedad general hacia las personas (TS = -2.717; d = .12). Conclusiones:
social anxiety, cross-cultural. estos hallazgos fueron interpretados atendiendo al de individualismo-
colectivismo y las concepciones culturales de la propia persona, analizando
sus implicaciones prácticas.
Palabras clave: Ansiedad Social para Adolescentes, SAS-A, invarianza,
ansiedad social, transcultural.

Adolescence is the developmental stage in which great Anxiety Disorder (SAD) (Bandelow & Michaelis, 2015; Stein
biological, cognitive, and social changes take place, which in turn et al., 2017). According to the American Psychiatric Association
may increase anxiety (Arnett, 2011; Crawley et al., 2018). One of (APA, 2014), the prevalence of SAD in the USA is 7%, while in
the most prevalent anxiety disorders among adolescents is Social Europe it is around 2.3%, with an age of onset during childhood
and adolescence (8-15 years). The World Health Organization
(WHO, 2019) notes that SAD is characterized by a disproportionate
Received: February 5, 2021 • Accepted: October 5, 2021 expression of intense fear or anxiety when a person faces social
Corresponding author: María Isabel Gómez Núñez situations such as social interactions, acting in the presence
Faculty of Social Sciences and Humanities of others or feeling observed while performing an action. The
International University of La Rioja
26006 Logroño (Spain) individual is concerned because the behavior or anxiety symptoms
e-mail: [email protected] they exhibit are negatively evaluated by others (fear of offending

126
 Measurement Invariance and Latent Mean Differences Between American, Spanish and Chinese Adolescents Using the Social Anxiety Scale for Adolescents (SAS-A)

other people). Therefore, individuals tend to avoid these social the sources of evidence of the internal structure of an instrument is
situations or to face them with high levels of fear or anxiety. SAD essential to assess items differential functioning. If a model is not
symptoms are persistent for several months, causing a clinically invariant or equivalent across cultural groups, differences could
significant deterioration in social, family, educational, work and be due to biases in measurements (Dimitrov, 2010) or to different
personal functioning. individual characteristics or item interpretations in each group
In the case of adolescents, SAD symptomatology can negatively (Byrne, 2008; Byrne & van de Vijver, 2017; Muñiz & Fonseca-
influence their quality of life (Vidal-Arenas et al., 2021) and their Pedrer, 2019). Thus, not taking into account the invariance of
functioning in different areas, such as school performance, self- the questionnaires among populations of different cultural origin
esteem, and friendship and intimate relationships (Chiu et al., can lead to interpreting differences in results that are not really
2021; La Greca & Danzi, 2016; Vilaplana-Pérez et al., 2021). comparable (Krieg et al., 2018).
Negative effects have also been reported for subclinical levels One of the main difficulties identified in cross-cultural studies
of SAD (Delgado et al., 2014). To prevent those adverse effects is having a valid measuring instrument in the populations of
Girio-Herrera et al. (2019) suggested implementation of evidence- interest. Despite the numerous studies that analyze social anxiety
based interventions in schools, as they will have an impact on the in adolescents, to our knowledge, the SAS-A (La Greca & López,
mental health of adolescents and, therefore, on public health. 1999) is the only instrument that has been validated in the three
To understand a psychological or social phenomena, such as countries considered in this research: China (Zhou et al., 2008),
social anxiety, it is essential to examine the context in which it Spain (García-López et al., 2001; Inglés et al., 2010; Olivares et
occurs (Hofmann et al., 2010; Jefferies & Ungar, 2020). In that al., 2005) and the USA (Inderbitzen et al., 2004; Inderbitzen &
sense, cross-cultural studies have highlighted the relevance of Walters, 2000; La Greca et al., 2015; Myers et al., 2002; Storch
attending to individualism-collectivism (Liddell & Williams, et al., 2004).
2019; Schreier et al., 2010). Based on prior research, the SAS-A appears to be an appropriate
Individualism refers to societies in which individuals do not instrument to evaluate and compare social anxiety among adolescents
have close bonds, and where each individual must take care of from China, Spain, and the USA. For this reason, the main aim of
themselves and their close relatives (Errasti-Pérez et al., 2018). this study was to analyze the equivalence of scores on the Social
Collectivism, in contrast, refers to societies in which individuals Anxiety Scale for Adolescents (SAS-A) using structural equation
belong to strongly cohesive groups, who take care of the group modeling and identifying latent means differences of social anxiety
members in return for their loyalty. Hofstede et al. (2010) found in China, Spain, and the USA. Based on the differences that previous
that individuals in the USA showed high scores in individualism literature has revealed regarding the functioning of these cultural
whereas those in China showed low scores in individualism, with groups, it is expected that: (a) the SAS-A three-factor correlated
Spanish individuals in between. Findings from Oyserman et al. model of social anxiety will be invariant in youth across the three
(2002) also show, when comparing the Spain and China with the countries, as validation studies in these countries have confirmed its
USA, that those in both Spain and China had a moderate magnitude three-factorial structure, and; (b) there will be statistically significant
in collectivism (whereas individuals in the USA had lower scores). differences in latent means among adolescents from the three
These differences among cultural groups could influence prevalence countries studied, with higher levels of social anxiety occurring in
and expression of social anxiety. In fact, previous research has adolescents from collectivistic cultures.
shown that social anxiety levels are significantly higher in Asian
than in Western cultures (Essau et al., 2011; Krieg & Xu, 2018; Method
Schreier et al., 2010; Woody et al., 2015).
In order to understand social anxiety variation levels among Participants
the adolescent populations included in this study (China, Spain,
and the USA), it is necessary to consider how adolescents The initial sample consisted of 2634 adolescents from 9th to
conceive social interaction in these countries. Thus, in Chinese 12th grade from public and private schools in China, Spain and
adolescents, shyness, behavioral inhibition, respect, and obedience the USA, all from community samples. A total of 54 participants
are considered maturity signs (Arnett, 2011; Chen, 2012) and were eliminated due to omissions or errors in the answers and/
contribute to adequate group functioning; assertive behaviors or or for not obtaining parental informed consent (China: 2.6%
standing out from others is discouraged (Xu et al., 2008). However, = 14 adolescents; Spain: 2.1% = 25 adolescents; USA: 1.7% =
North American society values assertiveness, expressiveness, 15 adolescents). For the 54 eliminated participants, using the
and competitiveness in social contexts (Chen, 2012). Spain is Mahalanobis distance (Field, 2018), 4 atypical multivariate values
characterized by features of both cultures (Hofstede et al., 2010; were detected and excluded; other incomplete cases were removed
Oyserman et al., 2002). because of their small percentage of the final sample and because
Despite a broad interest in the study of social anxiety in more than 60% of the SAS-A items were incomplete. The final
different countries and populations (e.g., Caballo et al., 2016; sample consisted of 2580 adolescents (M = 14.32, SD = 4.38): 536
Caballo et al., 2019; Krieg et al., 2018), previous research that Chinese (45.9% female), 1178 Spanish (55.3% female) and 866
examines the differences among Chinese, Spanish and North North American adolescents (55.1% female), aged between 14 and
American youth are scarce. To our knowledge, only Zhou et al. 17 years (China: M = 14.35, SD = 1.21; Spain: M = 15.45, SD =
(2008) addressed this issue, finding that the Chinese adolescents 1.28; the USA: M = 16.57, SD = 1.02). Chi-square tests confirmed
showed significantly higher scores on all SAS-A scales than did homogeneity of the samples across the countries based on sex and
Spanish and North American adolescents. However, the model’s age (China: χ² = .37, Spain: χ² = .97, the USA: χ² = .08).
measurement invariance was not analyzed for all three samples. For data collection, a random cluster sampling of the schools
According to Muñiz and Fonseca-Pedrero (2019) the analysis of in the participating cities was used, identifying rural and urban

127
 María Soledad Torregrosa Díez, María Isabel Gómez Núñez, Ricardo Sanmartín López, José Manuel García Fernández, Annette M. La Greca, Xinyue Zhou, Jesús Redondo Pacheco, and Candido J. Ingles Saura

areas and diverse socioeconomic levels. In the schools, a random was 116.15, higher than the 5 points proposed by Bentler (2005).
sample of classrooms was conducted depending on the number of Statistical properties of the confirmatory analysis were assessed
classrooms per academic year. Adolescents were included in the using the following goodness of fit indexes: χ²(S-Bχ²), R-RMSEA
study, if they: (a) were born in the country of analysis; (b) their (< .08 acceptable fit), SRMR (near .08 acceptable fit), R-CFI (near to
parents were born in that same country; and (c) they lived in that .90 acceptable fit) and TLI (>.90 good fit) (Brown, 2006). Cronbach’s
country for at least the last 8 years. alpha coefficient and Omega coefficients (McDonald, 1999) were
used to analyze reliability of each factor, with values higher than .70
Instruments considered as acceptable (Cho & Kim, 2015).
Subsequently, invariance analyses among China, Spain and the
Social Anxiety Scale for Adolescents (SAS-A; La Greca & USA groups were performed through multi-group confirmatory factor
López, 1999). This questionnaire assesses social anxiety in analyses following the hierarchic method. The Robust Maximum
adolescents. The SAS-A contains 22 items rated on a 5-point Likelihood procedure was also used. To analyze the invariance of
Likert scale (ranging from 1 = not at all to 5 = all the time). The the model, firstly, the psychometric properties of the model were
questionnaire includes three subscales: Fear of Negative Evaluation analyzed and, after that, restrictions to the preceding model were
(FNE; 8 items) that assesses concerns and fears related to negative included. Thus, consistent with Dimitrov (2010), the first model (M0)
peer evaluation (e.g.: “I worry about what others say about me”); analyzed configural invariance and did not include any restriction.
Social Avoidance and Distress in New Situations (SAD – New; Model 1 imposed equal factor loadings to M0. Model 2 imposed the
6 items) that assesses adolescents’ degree of discomfort and equivalence of intercepts to Model 1. Model 3 imposed the equality of
avoidance in new situations or with unfamiliar peers (e.g.: “I get error variances and covariances to Model 2. Finally, Model 4 imposed
nervous when I talk to peers I don’t know very well”); and Social equal factor variances and covariances to Model 2. In order to consider
Avoidance and Distress – General (SAD – General; 4 items) that that nested models are invariant the following criteria need to be met:
assesses discomfort and avoidance in any social situation (e.g.: (a) values of goodness of fit indexes are acceptable; (b) ΔS-Bχ² is non-
“I’m quiet when I’m with a group of people”). The remaining four significant (p > .05); and (c) ΔR-CFI is higher than -.01.
items are fillers. Items from each subscale are summed to obtain If invariances are confirmed, latent means analyses would be
a subscale score and those subscales are added to obtain a total performed comparing China, Spain, and the USA. In order to
score for social anxiety. Higher scores reflect greater social anxiety. perform those analyses, one of the groups is set as the reference
The Spanish and Chinese adaptations of the SAS-A confirmed that group and is fixed to 0 (here the reference group is Spain). The
psychometric properties of the scores of the scale were adequate in Critical Ratio (CR) is used to assess differences compared to the
both populations and the three-dimensional structure was replicated reference group, considering scores higher than 1.96 or lower
(García-López et al., 2001; Olivares et al., 2005; Zhou et al., 2008). than -1.96 as statistically significant (Tsaousis & Kazi, 2013).
In addition, sources of convergent validity of the scores of the The magnitude of the differences was analyzed according to the
SAS-A have been confirmed in several studies (García-López et al., typified mean difference, criteria stated by Cohen (1988).
2001; Inderbitzen et al., 2004; Inderbitzen & Walters, 2000; Storch
et al., 2004; Zhou et al., 2008). For the current study, the American Results
(La Greca & López, 1999), Chinese (Zhou et al., 2008) and Spanish
(Inglés et al., 2010) versions of the scale were employed. Descriptive Statistics and Reliability of Scores

Procedure For all the samples, Cronbach’s alpha coefficients were: .87
(FNE), .80 (SAD-New) and .73 (SAD-General), and Omega
The SAS-A questionnaire was answered collectively and coefficients [95% CI] were: .87 [.86, .88] (FNE), .80 [.79, .81]
voluntarily in the classroom. Informed written consent was (SAD-New) and .74 [.72, .75] (SAD-General). For the Chinese
requested from families or legal guardians for the participation of sample, Cronbach’s alpha coefficients were: .80 (FNE), .76 (SAD-
minors. In all the countries, a member of the research team was New) and .59 (SAD-General), and Omega coefficients [95% CI]
always present during the SAS-A administration process to ensure were: .79 [.75, .82] (FNE), .76 [.73, .80] (SAD-New) and .60 [.52,
the adolescents answered the questionnaire independently and to .65] (SAD-General). For the Spanish sample, Cronbach’s alpha
solve any problems that could arise. The research was approved coefficients were: .86 (FNE), .79 (SAD-New) and .73 (SAD-
by the ethical committees of the involved institutions and was in General), and Omega coefficients [95% CI] were: .86 [.84, .87]
accordance with the Helsinki ethical standards. The questionnaire (FNE), .79 [.77, .81] (SAD-New) and .74 [.71, .77] (SAD-General).
was administered in the native language of the participating For the North-American sample, Cronbach’s alpha coefficients
country (Chinese, Spanish, and English, respectively), following were: .89 (FNE), .82 (SAD-New) and .70 (SAD-General), and
the International Testing Commission (ITC) guidelines for test Omega coefficients [95% CI] were: .89 [.88, .91] (FNE), .83 [.80,
adaptation (Hernández et al., 2020; Muñiz et al., 2013), in order to .84] (SAD-New) and .70 [.67, .74] [(SAD-General). The descriptive
avoid possible biases due to questionnaire language. statistics for the items of the SAS-A, their factor loadings, and their
discrimination indexes for the total sample and each country can
Data Analysis be seen in Tables 1 and 2. A summary of this information across
countries is as follows: China (M = 2.03, 3.44; SD = 1.02, 1.19;
Confirmatory factor analyses were performed to analyze the Factor loadings = .323, .791; Discrimination indexes = .276, .660),
internal structure of the three-factor correlated model on the scores of Spain (M = 1.37, 2.81; SD = .73, 1.21; Factor loadings = .390,
the SAS-A items. The Robust Maximum Likelihood procedure was .789; Discrimination indexes = .388, .735) and North America
used because the kurtosis multivariant coefficient for the total sample (M = 1.48, 2.86; SD = .82, 1.24; Factor loadings = .405, .881;

128
 Measurement Invariance and Latent Mean Differences Between American, Spanish and Chinese Adolescents Using the Social Anxiety Scale for Adolescents (SAS-A)

Discrimination indexes = .394, .798). Regarding the correlations and SAD-G = .605, SAD-N and SAD-G = .723), Spanish Sample
among factors, the following data were obtained: Total Sample (FNE and SAD-N = .654, FNE and SAD-G = .623, SAD-N and
(FNE and SAD-N = .666, FNE and SAD-G = .692, SAD-N and SAD-G = .705) and North American Sample (FNE and SAD-N =
SAD-G = .747), Chinese Sample (FNE and SAD-N = .694, FNE .692, FNE and SAD-G = .694, SAD-N and SAD-G = .768).

Confirmatory Factor Analyses in China, Spain, and the USA

Table 1
Descriptive Statistics for Each Item of the SAS-A of the Three Countries and the The three-factor correlated model achieved adequate fit indexes
Total Sample
values (R-RMSEA [90% CI] = .054 [.051, .057]; SRMR = .053;
Items Mean SD RCFI = .935; TLI = .923) for the total sample, composed by
Chinese, Spanish, and North-American participants.
China Spain USA General China Spain USA General

1 2.66 2.80 2.59 2.70 1.12 1.05 1.04 1.07 Factorial Invariance Among Adolescents From China, Spain and
3 3.10 2.81 2.31 2.71 1.12 1.15 1.14 1.18
the USA
4 3.06 2.75 2.86 2.85 1.13 1.15 1.18 1.16
Table 3 includes data for the model in each country as well
5 2.85 2.23 2.57 2.48 1.19 1.10 1.24 1.19
as configural, measurement, and structural invariance values. As
6 2.58 2.30 2.32 2.37 1.05 1.06 1.11 1.08 shown, the model did not achieve an adequate fit in the Chinese
8 3.44 2.80 2.49 2.84 1.06 1.21 1.19 1.22 population, as TLI and R-CFI are lower than .90. Furthermore,
9 3.08 2.38 2.14 2.45 1.17 1.05 1.13 1.16 invariance analyses showed that the inclusion of restrictions
10 2.81 2.32 2.30 2.42 1.19 1.07 1.10 1.13 (M1 and M2) made ΔS-Bχ² significant, despite the adequate fit
12 2.85 2.74 2.23 2.60 1.11 1.20 1.11 1.18 of configural model (M0). Together, these data show the lack of
invariance of the model among the three groups of adolescents
13 2.70 2.34 2.38 2.43 1.06 1.13 1.10 1.11
(from China, Spain and the USA). However, because the model
14 2.96 2.58 2.02 2.48 1.10 1.20 1.04 1.18
fit indexes were acceptable for Spain and the USA, invariance
15 2.98 2.30 2.32 2.46 1.08 1.05 1.13 1.12 between those populations was checked.
17 2.47 1.87 1.77 1.97 1.02 .92 .90 .97
18 2.71 2.16 1.85 2.18 1.11 1.11 1.00 1.12 Confirmatory Factor Analysis in Spain and the USA
19 2.81 1.90 1.62 2.01 1.08 1.03 .93 1.10
20 2.94 2.31 2.47 2.50 1.05 .96 1.03 1.04 Fit values for the total sample (composed by participants from Spain
21 2.03 1.37 1.48 1.55 1.08 .73 .82 .89
and the USA) were adequate (R-RMSEA [90% CI] = .054 [.050, .057];
SRMR = .055; RCFI = .934; TLI = .923). Cronbach’s alpha coefficients
22 2.45 1.85 1.83 1.98 1.14 .96 1.05 1.06
were: .87 (FNE), .80 (SAD-New) and .71 (SAD-General).

Table 2
Standardized factor loadings and discrimination indexes of the Chinese, Spanish and USA samples

Factor loadings [Discrimination index]

Items China Spain USA General China Spain USA General China Spain USA General

1 .323 [.276] .550 [.452] .522 [.450] .458 [.388]

3 .592 [.513] .661 [.605] .658 [628] .669 [.622]
4 .716 [.609] .683 [.610] .688 [.652] .683 [.628]
5 .539 [.464] .436 [.388] .493 [.460] .496 [.443]
6 .332 [.350] .390 [.404] .405 [.415] .389 [.400]
8 .527 [.517] .712 [.684] .760 [.720] .726 [.695]
9 .642 [.562] .722 [.658] .866 [.790] .778 [.708]
10 .612 [.523] .742 [.665] .785 [.683] .743 [.647]
12 .593 [.498] .765 [.735] .808 [.773] .740 [.705]
13 .740 [.642] .692 [.603] .801 [.709] .743 [.655]
14 .791 [.660] .784 [.700] .881 [.798] .825 [.744]
15 .496 [.312] .533 [.427] .499 [.394] .562 [.442]
17 .438 [.436] .504 [.507] .586 [.599] .560 [.555]
18 .586 [.511] .607 [.531] .616 [.567] .643 [.576]
19 .574 [.332] .789 [.616] .721 [.547] .748 [.576]
20 .626 [.493] .630 [.513] .688 [.585] .671 [.549]
21 .472 [.389] .500 [.428] .610 [.515] .574 [.493]
22 .488 [.466] .757 [.624] .658 [.524] .675 [.584]

129
María Soledad Torregrosa Díez, María Isabel Gómez Núñez, Ricardo Sanmartín López, José Manuel García Fernández, Annette M. La Greca, Xinyue Zhou, Jesús Redondo Pacheco, and Candido J. Ingles Saura

Factorial Invariance Between Spain and the USA established three-factor correlated model in the three samples,
consistent with previous studies with Chinese (Zhou et al., 2008),
Table 4 shows configural, measurement, and structural
Spanish (Inglés et al., 2010) and North American (La Greca et al.,
invariance between Spain and the USA. The model showed
2015) adolescents. However, the first hypothesis was not supported
adequate fit indexes despite subsequent constraints applied to the
because analyses failed to confirm measurement invariance across
initial model (M0).
the three groups. These results potentially suggest a different
interpretation of the construct of social anxiety in Chinese
Latent Mean Differences Between Spain and the USA
adolescents, compared to those in Spain and the USA. In addition,
In the analysis of latent means, adolescents from Spain was alpha and omega coefficients reached acceptable values for all
considered as the reference group (fixed to 0) and the USA data SAS-A scales in the general population. However, scores on the
was freely estimated. Goodness of fit indexes of the model were SAD-General scale in Chinese adolescents were lower than .70,
appropriate (S-Bχ² = 1410.56; df = 290; R-CFI = .929; TLI =.916; indicating a low internal consistency for this scale (in line with

Table 3
Goodness-of-fit Indexes for SAS-A Models in China, Spain and the USA

R-RMSEA
χ2 S-Bχ² df TLI R-CFI SRMR ΔS-Bχ² (Δdf, p) ΔR-CFI
[90% CI]

China 685.409 495.440 130 .811 .840 .070 [.064, .077] .076
Spain 835.267 618.699 130 .907 .921 .057 [.052, .061] .060
USA 523.503 433.123 130 .935 .945 .052 [.047, .057] .050
M0 2044.228 1555.646 390 .902 .917 .034 [.032, .036] .063
M1 2192.345 1694.992 420 .901 .909 .034 [.032, .036] .072 144.60 (30, < .01) -.008
M2 3403.881 2809.739 456 .904 .919 .044 [.043, .046] .131 4732.56 (36, < .01) .010

2 2
Note: Model 0 = free model; Model 1 = Model 0 with factor loadings; Model 2 = Model 1 with intercepts; S-Bχ = Satorra-Bentler χ escalated; df = degrees of freedom; TLI = the Tucker-Lewis
Index; R-CFI = robust comparative fit index; R-RMSEA = robust root mean square error of approximation; SRMR = standarized root mean square residual; ΔR-CFI = robust comparative fit index
difference test. ΔS-Bχ² = χ² difference model comparison test

Table 4
Goodness-of-fit Indexes for SAS-A Models in Spain and the USA

R-RMSEA
χ2 S-Bχ² df TLI R-CFI SRMR ΔS-Bχ² (Δdf, p) ΔR-CFI
[90% CI]

Spain 835.267 618.699 130 .907 .921 .057 [.052, .061] .060
USA 523.503 433.123 130 .935 .945 .052 [.047, .057] .050
M0 1358.779 1062.081 260 .919 .932 .039 [.036, .041] .055
M1 1382.117 1088.546 275 .922 .930 .038 [.036, .041] .057 21.19 (15, .131) -.002
M2 1400.338 1109.927 293 .920 .930 .038 [.036, .040] .057 16.01 (18, .592) .000
M3 1441.638 1128.049 313 .921 .930 .037 [.035, .040] .058 27.20 (20, .130) .000
M4 1411.145 1119.676 299 .921 .930 .038 [.036, .040] .060 9.03 (6, .172) .000

Note: Model 0 = free model; Model 1 = Model 0 with factor loadings; Model 2 = Model 1 with intercepts; Model 3 = Model 2 with error variances; Model 4 = Model 2 with factor variances and
covariances; S-Bχ2 = Satorra-Bentler χ2 escalated; df = degrees of freedom; TLI = the Tucker-Lewis Index; R-CFI = robust comparative fit index; R-RMSEA = robust root mean square error of
approximation; SRMR = standarized root mean square residual; ΔR-CFI = robust comparative fit index difference test. ΔS-Bχ² = χ² difference model comparison test

R-RMSEA [90% CI] =.044 [.041, .046]; SRMR = .061). Table 5 Table 5
shows that statistically significant differences were found between Latent Mean Scores Between Spain and the USA in SAS-A
both groups in FNE and SAD-General. Adolescents from the
USA reported significantly lower social anxiety scores than did FNE SAD-New SAD-General

adolescents from Spain, with small and insignificant effect sizes Spain (reference)
for FNE and SAD-General, respectively, according to Cohen’s USA
criteria. ME -.367 .040 -.079
SE .038 .028 .029
Discussion CR -9.630* 1.470 -2.717*
d .44 - .12

The main aim of the study was to analyze the equivalence of Note: ME = Mean estimate; SE = Standard error; CR = Critical Ratio; d = effect size;
scores on the SAS-A using structural equation modeling and to FNE = Fear of Negative Evaluation; SAD-New = Social Avoidance and Distress in New
identify latent means differences of adolescent social anxiety in Situations; SAD-General = Social Avoidance and Distress–General; * = Statistically
significant difference (>1.96 or <-1.96)
China, Spain, and the USA. Data showed an acceptable fit to the

130
 Measurement Invariance and Latent Mean Differences Between American, Spanish and Chinese Adolescents Using the Social Anxiety Scale for Adolescents (SAS-A)

the adaptated version; Zhou et al., 2008). A possible explanation North-American adolescents. However, the study was performed
of this finding is that western definitions of social anxiety include using community samples, without considering clinical pathologies
symptoms (e.g., shyness) that are considered as acceptable or in participants. Attending to clinical information and including
normative in eastern cultures. It may be the case that some items adolescents with SAD diagnoses would help to increase construct
of the SAS-A may not represent social anxiety symptomatology validity of SAS-A. Another limitation is that cultural characteristics
in China in the same ways they do in western countries, thereby of each country were not directly measured. However, previous
contributing to the lack of model invariance among Chinese studies have shown that the country of analysis is a crucial factor
adolescents. Further examination of the social anxiety construct to assess the individualism-collectivism dimension, and thus,
among Chinese adolescents would be important and desirable. cultural differences among groups (Hofmann et al., 2010; Woody
Furthermore, there are differences among disorders included et al., 2015). The use of specific self-reported measures to assess
in the Chinese Classification and Diagnostic Criteria of Mental individualism-collectivism has been questioned because of being
Disorders (3rd ed.; CCMD-3; Chinese Society of Psychiatry, 2001) influenced by culture (Chen et al., 2015) or by self-construal
and those of western manuals, such as the Diagnostic and Statistical (Cross et al., 2011), as culture gives structure to behaviors without
Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric individual’s consciousness of it (Morris et al., 2008). For that
Association, 2014) or the International Classification of Diseases reason, Morris et al. proposed the analysis of interpersonal nets
(11th ed.; ICD- 11; WHO, 2019). The lack of an independent to understand social relations in social groups. Attending to such
categorization of social anxiety disorder in the Chinese classification a perspective in future research and considering different cultural
system makes difficult the consensus about symptoms or reactions dimensions (i.e., power distance, uncertainty avoidance, and
that could integrate this complex emotional response in Chinese masculinity) will help to provide a more in-depth analysis of cross-
population. The present study points out the relevance of attending cultural invariance among countries.
to the Chinese cultural context to improve the comprehension of One of the main contributions of validation studies is to support
social anxiety as a prerequisite for group comparisons. the use of assessment tools in different population groups, not only
Although the findings for Chinese adolescents differed, for use in research studies but also to identify symptomatology
the invariance analyses between Spanish and North-American and facilitate early intervention. In the research field, the present
adolescent groups showed that the three-factor correlated model study has shown that scores on the SAS-A are valid for the
was equivalent in both populations. Thus, the three dimensions of assessment and comparison of the social anxiety construct in
FNE, SAD-New and SAD-General, identified in previous validation Spanish and North-American adolescents. In the applied area, the
research in these countries, were also confirmed (Inderbitzen et al., present results show that the SAS-A could be used as a screening
2004; Inderbitzen & Walters, 2000; Myers et al., 2002; Storch et questionnaire in educational contexts to identify those adolescents
al., 2004; La Greca & López, 1999; García-López et al., 2001; who need individual attention. Adolescents suffering from social
Inglés et al., 2010; Olivares et al., 2005). These data suggest that anxiety rarely seek professional help, so it is difficult to identify
social anxiety, as measured by the SAS-A, is understood similarly them in clinical contexts. In that sense, La Greca et al. (2016) have
across Spanish and North-American adolescents, which might adapted preventive programs (i.e., UTalk) for use with adolescents
reflect how individualism influences self-interpretation and the in educational settings, in order to identify those with social
value of social interactions in western societies. anxiety symptomatology and help them to manage it. Cultural
Regarding the second hypothesis, latent mean differences group differences would help to identify the most representative
were analyzed only for Spanish and North-American samples and features of social anxiety for a specific cultural group when
revealed significant differences between groups in FNE and SAD- planning interventions (Jankowska, 2019). Thus, it would be
General factors of social anxiety. Spanish adolescents received useful to focus on specific contents in cognitive restructuring
higher scores on these factors than North-American youth. (e.g., to work more in depth on the relevance of significant-others’
Perhaps the greater emphasis on collectivism in Spanish culture opinions) or exhibition to fear situations (e.g., to include situations
could explain this finding, in that family attachment and sense where adolescents have to show opinions contrary to the reference
of belonging are more salient. Thus, Spanish adolescents might group).
be more fearful of not meeting the expectations of significant Our results indicate that the social anxiety construct is not
others or feel more pressured by the peer group. Those situations conceived similarly in China as in western countries. Thus, the
could contribute to high levels of fear of negative evaluations manifestation of social anxiety symptomatology and its intensity
or discomfort in general social situations when not fitting the may be influenced by cultural values and social norms. That
established standard (Ohannessian et al., 2017; Zhou et al., 2008). finding highlights the use of sound validated measures, like the
The present study provides valuable information regarding SAS-A, to improve efficacy of social anxiety programs in clinical
differences in the social anxiety construct in Chinese, Spanish and and educational areas attending to cultural differences.

References

American Psychiatric Association (2014). Diagnostic and statistical Bandelow, B., & Michaelis, S. (2015). Epidemiology of anxiety disorders
manual of mental disorders (5th ed.). Author. http://dx.doi.org/10.1176/ in the 21st century. Dialogues in Clinical Neuroscience, 17(3), 327-
appi.books.9780890425596 335. https://doi.org/10.31887/DCNS.2015.17.3/bbandelow
Arnett, J. J. (2011). Adolescent psychology around the world. Psychology Bentler, P. M. (2005). EQS 6.1: Structural Equations Program Manual
Press. [computer software].

131
 María Soledad Torregrosa Díez, María Isabel Gómez Núñez, Ricardo Sanmartín López, José Manuel García Fernández, Annette M. La Greca, Xinyue Zhou, Jesús Redondo Pacheco, and Candido J. Ingles Saura

Brown, T. (2006). Confirmatory factor analysis for applied research. Girio-Herrera, E., Ehrlich, C. J., Danzi, B. A., & La Greca, A. M.
Guildford Press. (2019). Lessons learned about barriers to implementing school-based
Byrne, B. M. (2008). Testing for multigroup equivalence of a measuring interventions for adolescents: Ideas for enhancing future research and
instrument: A walk through the process. Psicothema, 20(4), 872-882. clinical projects. Cognitive and Behavioral Practice, 26(3), 466-477.
http://www.psicothema.com/pdf/3569.pdf https://doi.org/10.1016/j.cbpra.2018.11.004
Byrne, B. M., & van de Vijver, F. J. R. (2017). The maximum likelihood Hernández, A., Hidalgo, M. D., Hambleton, R. K., & Gómez-Benito, J.
alignment approach to testing for approximate measurement invariance: (2020). International test commission guidelines for test adaptation: A
A paradigmatic cross-cultural application. Psicothema, 29, 539-551. criterion checklist. Psicothema, 32(3), 390-398. https://doi.org/10.7334/
https://doi.org/10.7334/psicothema2017.178 psicothema2019.306
Caballo, V. E., Salazar, I. C., Arias, B., Calderero, M., Irurtia, M. J., Hofmann, S. G., Asnaani, M. A. A., & Hinton, D. E. (2010). Cultural
Ollendick, T. H., & CISO-A Research Team. (2016). The Social Anxiety aspects in social anxiety and social anxiety disorder. Depression and
Questionnaire for children: Cross-cultural assessment with a new self- Anxiety, 27(12), 117-1127. https://doi.org/10.1002/da.20759
report measure. Journal of Psychopathology and Behavioral Assessment, Hofstede, G., Hofstede, G. J., & Minkov, M. (2010). Cultures and
38(4), 695-709. https://doi.org/10.1007/s10862-016-9562-x organizations: Software of the mind. McGraw-Hill.
Caballo, V. E., Salazar, I. C., Arias, V., Hofmann, S. G., & Curtiss, J. (2019). Inderbitzen, H. M., & Walters, K. S. (2000). Social Anxiety Scale for
Psychometric properties of the Liebowitz Social Anxiety Scale in a Adolescents: normative data and further evidence of construct validity.
large cross-cultural Spanish and Portuguese speaking sample. Brazilian Journal of Clinical Child Psychology, 29, 360-371. https://doi.
Journal of Psychiatry, 41(2), 122-130. https://doi.org/10.1590/1516- org/10.1207/S15374424JCCP2903_7
4446-2018-0006. Inderbitzen, H. M., Davies, C. A., & McKeon, N. D. (2004). Investigating
Chen, X. (2012). Culture, peer interaction and socioemotional development. the construct validity of the SPAI-C: Comparing the sensitivity and
Child Development Perspectives, 6(1), 27-34. https://doi.org/10.1111/ specificity of the SPAI-C and the SAS-A. Journal of Anxiety Disorders,
j.1750-8606.2011.00187.x 18, 547-560. https://doi.org/10.1016/S0887-6185(03)00042-2
Chen, X., Gong, J., Yu, B., Li, S., Striley, C., Yang, N., & Li, F. (2015). Inglés, C. J., La Greca, A. M., Marzo, J. C., García-López, L. J., &
Constructs, concept mapping, and psychometric assessment of the García-Fernández, J. M. (2010). Social Anxiety Scale for Adolescents:
Concise Scale of Individualism-Collectivism. Social Behavior and Factorial invariance and latent mean differences across gender and age
Personality: An International Journal, 43(4), 667-683. https://doi. in Spanish adolescents. Journal of Anxiety Disorders, 24(8), 847-855.
org/10.2224/sbp.2015.43.4.667 https://doi.org/10.1016/j.janxdis.2010.06.007
Chinese Society of Psychiatry (2001). The Chinese Classification and Jankowska, M. (2019). Cultural modifications of cognitive behavioural
Diagnostic Criteria of Mental Disorders (CCMD-3). Chinese Society treatment of social anxiety among culturally diverse clients: A
of Psychiatry. systematic literature review. The Cognitive Behaviour Therapist, 12,
Chiu, K., Clark, D. M., & Leigh, E. (2021). Prospective associations Article e7. https://doi.org/10.1017/S1754470X18000211
between peer functioning and social anxiety in adolescents: A Jefferies, P., & Ungar, M. (2020). Social anxiety in young people: A
systematic review and meta-analysis. Journal of Affective Disorders, prevalence study in seven countries. Plos One, 15(9), e0239133. https://
279, 650-661. https://doi.org/10.1016/j.jad.2020.10.055 doi.org/10.1371/journal.pone.0239133
Cho, E., & Kim, S. (2015). Cronbach’s coefficient alpha: Well-known but Krieg, A., & Xu, Y. (2018). From self-construal to threat appraisal:
poorly understood. Organizational Research Methods, 18, 207-230. Understanding cultural differences in social anxiety between Asian
http://dx.doi.org/10.1177/1094428114555994 Americans and European Americans. Cultural Diversity and Ethnic
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Minority Psychology, 24(4), 477-488. https://doi.org/10.1037/
Erlbaum. cdp0000194
Crawley, S.A., Settipani, C.A., & Kendall, P.C. (2018). Anxiety Disorders. Krieg, A., Xu, Y., & Cicero, D. C. (2018). Comparing social anxiety
In R. J. R. Levesque (Eds.), Encyclopedia of Adolescence (pp. 236-243). between Asian Americans and European Americans: An Examination
Springer, Cham. https://doi.org/10.1007/978-3-319-33228-4_180 of measurement invariance. Assessment, 25(5), 564-577. https://doi.
Delgado, B., Inglés, C. J., & García-Fernández, J. M. (2014). Subclinical org/10.1177/1073191116656438
social anxiety and academic performance in adolescence: Analysis of La Greca, A. M., & Danzi, B. A. (2016). Addressing issues of anxiety
theoretical and practical implications. Estudios de Psicología, 35(1), specificity and comorbidity in understanding developmental outcomes
58-79. https://doi.org/10.1080/02109395.2014.893649 of youth anxiety. Clinical Psychology: Science and Practice, 23(4),
Delvecchio, E., Mabila, D., Di Riso, D., Miconi, D., & Li, J. B. (2015). 436-440. https://doi.org/10.1111/cpsp.12170
A comparison of anxiety symptoms in community-based Chinese and La Greca, A. M., Ehrenreich-May, J., Mufson, L., & Chan, S. (2016).
Italian adolescents. Journal of Child and Family Studies, 24(8), 2418- Preventing adolescent social anxiety and depression and reducing
2431. https://doi.org/10.1007/s10826-014-0045-y peer victimization: Intervention development and open trial. Paper
Dimitrov, D. M. (2010). Testing for factorial invariance in the presented at the Child & Youth Care Forum, 45(6), 905-926. https://
context of construct validation. Measurement and Evaluation doi.org/10.1007/s10566-016-9363-0
in Counseling and Development, 43(2), 121-149. https://doi. La Greca, A. M. (1999). Manual for the Social Anxiety Scale for Adolescents.
org/10.1177/0748175610373459 Revised version. University of Miami
Errasti-Pérez, J. M., Amigo-Vázquez, I., Villadangos-Fernández, J. M., La Greca, A. M., Inglés, C. J., Lai, B. S., & Marzo, J. C. (2015). Social
& Morís-Fernández, J. (2018). Differences between individualist and Anxiety Scale for Adolescents: Factorial invariance across gender and
collectivist cultures in emotional facebook usage: Relationship with age in Hispanic American adolescents. Assessment, 22(2), 224-232.
empathy, self-esteem, and narcissism. Psicothema, 30, 376-381. https:// https://doi.org/10.1177/1073191114540749
doi.org/10.7334/psicothema2018.101 Liddell, B. J., & Williams, E. N. (2019). Cultural differences in interpersonal
Essau, C. A., Ishikawa, S. I., & Sasagawa, S. (2011). Early learning emotion regulation. Frontiers in Psychology, 10, 1-15. https://doi.
experience and adolescent anxiety: A cross-cultural comparison between org/10.3389/fpsyg.2019.00999
Japan and England. Journal of Child and Family Studies, 20(2), 196- McDonald, R. P. (1999). Test Theory: a Unified Treatment. Erlbaum.
204. https://doi.org/10.1007/s10826-010-9404-5 Morris, M. W., Podolny, J., & Sullivan, B. N. (2008). Culture and coworker
Field, A. (2018). Discovering statistics using IBM SPSS Statistics (5th relations: Interpersonal patterns in American, Chinese, German, and
Edition). Sage. Spanish divisions of a global retail bank. Organization Science, 19(4),
García-López, L. J., Olivares, J., Hidalgo, M. D., Beidel, D. C., & Turner, 517-532. https://doi.org/10.1287/orsc.1070.0333
S. (2001). Psychometric properties of the Social Phobia and Anxiety Muñiz, J., Elosua, P., & Hambleton, R. K. (2013). International Test
Inventory, the Social Anxiety Scale for Adolescents, the Fear of Commission Guidelines for test translation and adaptation: Second
Negative Evaluation Scale and the Social Avoidance and Distress Scale edition. Psicothema, 25(2), 151-157. https://doi.org/10.7334/
in an adolescent Spanish-speaking sample. Journal of Psychopathology psicothema2013.24
and Behavioral Assessment, 23, 51-59. https://link.springer.com/ Muñiz, J., & Fonseca-Pedrero, E. (2019). Ten steps for test development.
article/10.1023/A:1011043607878 Psicothema, 31(1), 7-16. https://doi.org/10.7334/psicothema2018.291

132
 Measurement Invariance and Latent Mean Differences Between American, Spanish and Chinese Adolescents Using the Social Anxiety Scale for Adolescents (SAS-A)

Myers, M. G., Stein, M. B., & Aarons, G. A. (2002). Cross validation of the Adolescents and the Social Phobia and Anxiety Inventory for Children:
Social Anxiety Scale for Adolescents in a high school sample. Journal Construct validity and normative data. Journal of Anxiety Disorders,
of Anxiety Disorders, 16, 221-232. https://doi.org/10.1016/S0887- 18, 665-679. https://doi.org/10.1016/j.janxdis.2003.09.002
6185(02)00098-1 Tsaousis, I., & Kazi, S. (2013). Factorial invariance and latent mean
Ohannessian, C. M., Cavanaugh, A., & Cheeseman, K. (2017). Gender differences of scores on trait emotional intelligence across gender and
and racial/ethnic differences in anxiety disorders during adolescence. age. Personality and Individual Differences, 54(2), 169-173. https://
Adolescent Psychiatry, 7(1), 13-24. https://doi.org/10.2174/22106766 doi.org/10.1016/j.paid.2012.08.016
07666170908160057 Vidal-Arenas, V., Ortet-Walker, J., Ibáñez, M. I., Ortet, G., & Mezquita, L.
Olivares, J., Ruiz, J., Hidalgo, M. D., García-López, L. J., Rosa, A. I., & (2021). Self-reported DSM-5 anxiety severity measures: Evidence of
Piqueras, J. A. (2005). Social Anxiety Scale for Adolescents (SAS-A): validity and reliability in Spanish youths. Psicothema, 33(2), 312-319.
Psychometric properties in a Spanish-speaking population. International https://doi.org/10.7334/psicothema2020.398
Journal of Clinical and Health Psychology, 5(1), 85-95. https://www. Vilaplana-Pérez, A., Pérez-Vigil, A., Sidorchuk, A., Brander, G., Isomura,
redalyc.org/articulo.oa?id=33701005 K., Hesselmark, E., Kuja-Halkola, R., Larsson, H., Mataix-Cols, D.,
Oyserman, D., Coon, H. M., & Kemmelmeier, M. (2002). Rethinking & Fernández-de la Cruz, L. (2020). Much more than just shyness: The
individualism and collectivism: Evaluation of theoretical assumptions impact of social anxiety disorder on educational performance across the
and meta-analyses. Psychological Bulletin, 128(1), 3-72. https://doi. lifespan. Psychological Medicine, 53, 861-869. https://doi.org/10.1017/
org/10.1037//0033-2909.128.1.3 S0033291719003908
Schreier, S. S., Heinrichs, N., Alden, L., Rapee, R. M., Hofmann, S. G., Woody, S. R., Miao, S., & Kellman-McFarlane, K. (2015). Cultural
Chen, J., Oh, K. J., & Bögels, S. (2010). Social anxiety and social differences in social anxiety: A meta-analysis of Asian and European
norms in individualistic and collectivistic countries. Depression and heritage samples. Asian American Journal of Psychology, 6(1), 47-55.
Anxiety, 27(12), 1128-1134. https://doi.org/10.1002/da.20746 https://doi.org/10.1037/a0036548
Stein, D. J., Lim, C. C., Roest, A. M., De Jonge, P., Aguilar-Gaxiola, S., World Health Organization (2019). ICD 11. International Classification
Al-Hamzawi, A., Alonso, J., Benjet, C., Bromet, E. J., Bruffaerts, R., of Diseases for Mortality and Morbidity. World Health Organization.
De Girolamo, G., Florescu, S., Gureje, O., Haro, J. M., Harris, M. G., https://icd.who.int/en
He, Y., Hinkov, H, Horiguchi, I., Hu, C. (…) and WHO World Mental Xu, Y., Farver, J. M., Yang, Y., & Zeng, Q. (2008). Chinese children’s
Health Survey Collaborators (2017). The cross-national epidemiology conceptions of shyness: A prototype approach. Merrill-Palmer
of social anxiety disorder: Data from the World Mental Health Survey Quarterly, 54, 515-544. https://www.jstor.org/stable/23096270
initiative. BMC Medicine, 15(1), 1-21. https://doi.org/10.1186/s12916- Zhou, X., Xu, Q., Inglés, C. J., Hidalgo, M. D., & La Greca, A. M. (2008).
017-0889-2 Reliability and validity of the Chinese Version of the Social Anxiety
Storch, E. A., Masia-Warner, C., Dent, H. C., Roberti, J. W., & Fisher, P. Scale for Adolescents. Child Psychiatry and Human Development,
H. (2004). Psychometric evaluation of the Social Anxiety Scale for 39(2), 185-200. https://doi.org/10.1007/s10578-007-0079-0

133
© 2022. Notwithstanding the ProQuest Terms and conditions, you may use
this content in accordance with the associated terms available at
https://www.psicothema.com/PublicationNorms2022.pdf