Validity evidence of the Brief Fear of Negative Evaluation Scale for children and adolescents adapted for physical activity and sports settings

Aim: To investigate the validity evidence of the adapted Brief Fear of Negative Evaluation Scale in physical activity and sports settings BFNE-PAS. Methods: Participants included 721 children and adolescents (376 girls and 345 boys), 10 to -14-year-olds (total sample Mage = 12.1, SD = 1.3) from Brazil. Results: Exploratory factorial analysis confirmed the unidimensionality of the scale. Confirmatory factorial analysis showed adequate indexes (RMSEA = 0.05, [90% C.I. = 0.04 to 0.06], CFI = 0.98 TLI = 0.98). Network analyses indicate the highest influence of the item: “afraid to expose failures when practicing physical exercises” and/or participating in games and play in physical education classes” upon all others. The polychoric alpha showed adequate internal consistency (total scale: α = 0.86; items: α values > 0.84). The intraclass coefficient correlation showed strong reliability test-retest (ICC > 0.90). Conclusion: BFNE-PAS showed adequate validity evidence in Brazilian children.


Introduction
The fear of negative evaluation, or social phobia, is a crucial demand related to social anxiety 1 . Individuals with high fear of negative evaluation show exaggerated apprehension when face exposure in social circumstances, avoidance of achievement settings, and frequent feelings of expectation that others will evaluate them negatively 2 . As an aspect of social anxiety, this fear of evaluation from others is related to a range of disruptive behaviors, such as social isolation, difficulty in making friends 3 , and communication in public 4 . Unfavorable health conditions, for example, depressive symptoms 5 and quality of life 6 also related to fear of negative evaluation. Besides, specifically, during childhood and adolescence, the fear of negative evaluation affects stronger and undesirably the academic performance and the acquisition of social skills 3,7 .
This fear may occur in various social settings evaluate situations 8 . These require specific performance, such as talking in public and interacting with others 9 . Notably, in physical education and sport contexts, the situations naturally expose the practitioners to social evaluation. Usually, sports practices take place in groups, mediated by a teacher or coach, in settings that allow the evaluation of others (i.e., gymnasium, sports courts, playgrounds, pool). Eventual failure to complete the task satisfactorily is more visible and exposes the practitioners to a social evaluation of their capabilities. Other aspects are related to the use of appropriate dress for physical activity practices (e.g., sports shorts and t-shirts, swimsuits, lightweight t-shirt, or sweatshirts), which may cause distress since it exposes shape or parts of the body. Children and adolescents may be more sensitive to those factors 10,11 . Higher fear of negative social evaluation may lead them to demonstrate high levels of tension, apprehension, and shame of failure during physical activity practices 7,12 . Besides, athletic performance, playing with other children, and physical education classes were reported by children with a social phobia as potentially socially fearful situations 7 . Given this particularity, assessing the fear of negative evaluation and physical education class and sports contexts is essential.
Various self-report scales had been proposed to assess the fear of negative social evaluation in children and adolescents. The Social Phobia and Anxiety Inventory for adolescents 13 ; the Social Phobia and Anxiety Inventory for Children 14,15 ; the Social Phobia Inventory 16 ; Fear of Negative Evaluation scale 2 are examples of scales that assess this construct. Mainly, the Fear of Negative Evaluation scale had been widely utilized in the literature. The original version was developed from an American adult population and contained 30 dichotomous items in a truthfalse format. Posteriorly, Leary 17 proposed a Brief Fear of Negative Evaluation -BFNE, arguing about the excessive quantity of the item in the original scale. This brief version was composed of 12 items organized in a 5-point scale format rating ranging from 1 (not at all characteristics of me) to5 (extremely characteristic of me. The BFNE showed adequate psychometric properties and has been validated for different adult 18-22 and adolescent 23-25 populations.
Although considerable research reports the importance of understanding social anxiety for mental, emotional, and social health in childhood and adolescence, little is known how the fear of negative evaluation interferes with the engagement and permanence of children and adolescences in physical education lessons and motor and sports programs. Nevertheless, this factor is related to avoidance in physical activities, and peers and teammate interactions in sports practices have been understudied. This limitation, in part, is due to the lack of valid and reliable instruments that assess how this construct is manifested in physical education and sports contexts. To the authors' knowledge, no instrument assesses children's and adolescents' fear of negative social evaluation in physical activity and sports settings. Thus, this study aimed to adapt and examine the validity of the Brief Fear of Negative Evaluation Scale for children and adolescents in physical activity and sports settings.

Participants
Participants in the present study included a final sample size of 741 children and adolescents (376 girls; 365 boys), 10-to -14-year-olds (total sample Mage = 12.1, SD = 1.3; boys: Mage = 12.1, SD = 1.3; girls: Mage = 12.2, SD = 1.3) from eight public school in Ceará, Brazil. Children with cerebral paralysis, motor disorder, learning disabilities, and anxiety disorder, reported by parents, teachers, or caregivers, were excluded from the study. Consent was obtained from the custodial caregiver (s), and each child verbally agrees to participate in the study. Three Ph.D. professors, experts in development or sports psychology, and five professionals in motor behavior, sport, and development psychology with five years' experience participated in the face validity. To data analysis the children were randomly allocated in two sample groups as following: 321 children (sample 1: M = 12.1, SD = 1.2; 149 girls, 46.5%), 420 children (sample 2: M = 12.2, SD = 1.1; 220; girls, 47.6%). the scores from sample 1 were used to examine the initial structure of BFNE-PE. The scores from sample 2 were used in the cross-validation factorial structure. From the total sample, 150 children (age M = 12.2 SD = 1.0; 75 girls; 50%) were randomly selected to be assessed twice; the scores were used in the test-retest analysis.

Instruments
The Brief Fear of Negative Evaluation Scale -BFNES 17 was translated and adapted to the present study. The original BFNES is used to determine the degree of apprehension that the individuals experience in a situation in which they are exposed to social evaluation. In doing so, they prospect an unfavorable evaluation from others. The original BFNE is composed of 12 items, being eight specific items and four reverse items, about the fear concerning the lack of approval from others, worries about what kind of impression the individual makes on others, and the worry about saying or doing wrong things. The BFNE is a Likert-type scale from 1 (not at all characteristics of me) to 5 (extremely characteristic of me). The items sum composes a total score, ranging from 12 to 60. The validity (r = 0.90), reliability (r = 0.75) and internal consistency (r = 0.96) have demonstrated initially for undergraduate students (17) 14 . Psychometric studies of BFNES have been demonstrating adequate evidence of validity and reliability of this scale for children and adolescents 19,23 . In this study, a 5-points Likert scale ("unclear/irrelevant" = 1; "more or less clear/relevant" = 2; "clear/relevant" = 3; "very clear/relevant" = 4; "optimal clear/relevant" = 5) was used to assess the expert's judgment about the clarity and relevance of each item of the BFNE.

Procedures
The ethics committee from proponent University approved the study (CAAE number: 10270118.6.0000.5589). First, four translators participated in the retro translation of BFNE-PAS. Independently, two translators (T1 and T2) with fluency in the original language of the instrument (English) and the target language (Portuguese) translate the scale to Portuguese. An initial Portuguese version was obtained according to the synthesis of these two initial translations. The other two translators (T3 and T4) independently conducted the retro translation of the initial version in Portuguese back to the English language. A translator meeting was held, and a final version in Portuguese was obtained by confronting the final Portuguese version with the original version in English.
Since this study proposes an adaptation of the original Brief Fear of Negative Evaluate scale for children and adolescents in physical activity and sport, we consider increasing the construct precision by inserting new items. Although all items from the original version were adapted, we considered that it did not reach the latent construct's totality. Thus, three new items were formulated to increase the construct precision for physical activity and sports practice. One item included the worry concerning wearing appropriate clothes (e.g., gym pants, T-shirt, shorts, shorts) in physical activities. One item included the worries about embarrassment or shame caused by a failure in physical activities or sports practices. Lastly, the one new item intended to reinforce the assessment about the fear of exposing failures when practicing physical activity and/or participating in sports matches. The original scale, and the adapted scale, proposed in the present study with three new items, are presented in Table 1.
We contacted the board of education and school administrators from two cities of Ceará, Brazil. Eight public schools responded positively to participation in the study. In a meeting with the school administrator and staff, the researchers explained the study's aims and procedures, and parents were informed about the research. We randomly selected children within the school.
Parents or legal guardians were contacted and received all information about the research procedures; parents' informed consent was obtained from each child participate in this study. The children were assessed individually at schools by trained professionals. The time of the test application was approximately 15 minutes for each participant. The sample size for CFA was estimated, considering a model with 54 degrees of freedom (Number of distinct sample moments -Number of distinct parameters to be estimated: 78 -24 = 54), a significance level at 0.05, and the desired power at 0.98. Considering a sample loss of 10%, our calculated final sample size for CFA was n = 420 participants. Table 1 -Original scale, and the adapted scale for physical activity and sports.

Items Original Brief Scale (Lary, 1983):
1 I worry about what other people will think of me even when I know it doesn't make any difference. Regarding EFA, we follow the recommended literature of minimum sample size in absolute numbers n = 300 26 . Considering a sample loss of 10%, our calculated sample size for EFA was 330 participants. Considering that nine participants discontinued this study, our final sample size for EFA was 321 children and adolescents.

Statistical Analysis
Descriptive analysis was provided using mean, standard deviation, and percentages. The Content Validity Coefficient (CVC) was used to estimate each item's language clarity and relevance and total items 27 . The Gwet's Agreement Coefficients (AC1) weighted and unweighted were used to estimate experts' agreement; values above 0.80 were considered a high agreement 28 . Mahalanobis squared distance (D 2 ) was used to examine multivariate outliers and the Omnibus test based on Small's (chisquare) multivariate nonnormality of the data. We used Exploratory Factorial Analysis (EFA) and Baye's modal estimation polychoric correlation matrices to examine if the new items affected the unidimensional structure 29 . To estimate the adequacy of the correlation matrix, the Kaiser-Meyer-Olkin (KMO) with asymptotic Covariance/Variance matrix, using bootstrap sampling (95% confidence interval (CI95%), and Bartlett's Sphericity Test were conducted. Values higher than 0.80 for KMO and significant statistical differences for Bartlett's test (p < 0.05) were considered as adequately. Principal Components extraction method and Promax rotation for EFA analysis and the optimal implementation of Parallel Analysis (PA), to estimate the number of scale dimensions was used 30 .
A confirmatory factorial analysis was conducted to verify the FNE-PE structural model, adopting the weighted least square mean and variance adjusted (WLSMV) estimator's method. The model's overall fit was tested using multiple fit indexes since different measures present strength and weakness 31 . The Tukey Lewis Index (TLI) 31 and the Comparative Fit Index (CFI); were used to verify model adjustment -values greater than or equal to 0.95 and 0.90 as considered as appropriate 32 . The Root Mean Square Error of Approximation (RMSEA) with 90% confidence interval (CI90%), with recognized cut-off 32 ; good: values lowest 0.05; acceptable: values between 0.06 and 0.08) was also used.
The invariance of the model for sex and age (10 to 11-year-old & 12 to 14-year-old) was examined using the multigroup CFA invariance factorial loads. The CFI, GFI, and RMSEA (90% CI) were adopted as fit indexes for the configurational invariance. The metric invariance was used to verify if loadings varied across sex and age by groups 33 . Test invariance was used to examine if the intercept for each variable and construct did not vary. The comparisons of the models were conducted using means differences between constrained and unconstrained models, the delta of the RMSEA (Δ RMSEA), adopting the recommended cut-off (< 0.015) to support the invariance assumption 33 .
Besides, a network analysis was conducted to investigate the structure of partial relationships and the relative influence of items on each other. The estimated Gaussian Graphic Model, with a partial correlation matrix, was used for the network analysis. Thus, the edges represent the conditional association between variables by model, the pairwise Markov field, and it was estimated using the L1regularized neighborhood regression. The regularization was obtained using the graphical Least Absolute Shrinkage and Selection Operator -gLASSO 34 , controlling the model sparsity. Using the Extended Bayesian Information Criteria (EBIC), the model was selected, levels of penalty parameter between 0.25 and 0.50 were adopted. The function "cor_auto" in R-program was used to estimate the correlations matrix. Three centrality measurements were adopted: betweenness, closeness, and strength.
The internal consistency and reliability were conducted to investigate the construct validity using alpha for ordinal data based on polychoric correlations; values ≥ 0.70 were considered acceptable 35 . Alternatively, the composite reliability (CR) was conducted 30 ; CR values equal or superior to 0.70 were considered adequate 36 . In this study, the average variance extracted (AVE) was utilized as a measure of precision. Values superior to 0.50 were considered adequate 37 . The Intraclass Correlation Coefficient (ICC) was utilized to investigate the test-retest reliability. A two-way mixed effect model, based on the mean of multiple measures, was adopted to measure consistency 38,39 .
The analyses were conducted using Agree.15 software, Factor analysis software version 10.9.01, Mplus version 7.4, Psych, and qgraph packages from R-free software. The significance level was set at α less than or equal to 0.05.

Translation and adaptation
After the translation and retro translation process, the version of BFNE was adapted to be used in physical activity and sports settings. All items were rewritten (i.e., when I participate in the physical activity and sports, I am frequently afraid of other people noticing my shortcomings; when I am talking to someone from my sport and physical activity team or group, I worry about what they may be thinking about me). Furthermore, three items to the scale: I worry about not being ashamed when I participate in physical education or sports; I get worried when I have to wear fitness clothes (gym pants, t-shirt, shorts, shorts); I am afraid of exposing my flaws by doing physical exercises and/or participating in sports matches.

Content validity evidence
The results showed high CVC values concerning total items among experts for linguistic clarity (94.7% to 99.8%) and relevance (96.4% to 97.5%). The high CVC results for each item were found (values ranged from 92% to 100% for linguistic clarity and 97% to 100% for relevance among expert responses). The Gwet's AC1 results ranged from 0.75 to 0.78 for clarity and from 0.84 to 0.93 for relevance, showed high concordance among experts. Table 2 shows the content validity coefficient (CVC) and Gwet's AC1 for language clarity and item relevance.

Construct validity evidence
The initial EFA analysis enrolled 321 children (sample 1) showed adequate values for KMO (0.83, CI95% = 0.81 to 0.85) and Bartlett sphericity test (χ 2 (105) = 1727.3; p < 0.001). However, were observed inadequate values for load factor (λ < 0.50) and communalities    The network and centrality plot analyses for the BFNE-PAS items were conducted in the total sample (741 children). The highest partial correlations were observed between items 3 and 6 (r = 0.26), items 9 and 11 (r = 0.024), items 1 and 11 (r = 0.20) and items 13 Table 4 presents the alpha results for ordinal data based on polychoric correlations of the items, the item-rest correlation, and the descriptive statistics (median, minimum, and maximum values) of the scores BFNE-PAS, according to sex and age groups. The results showed an appropriate reliability for all items (values > 0.84) and total scale (α = 0.86). In addition, the composite reliability analysis showed adequate value (0.87). Yet, individual item reliability showed appropriate values (values between 0.25 to 0.53) except for item 12 (value = 0.19). In general, these results reinforcing the internal consistency evidence of the instrument. The AVE value was near to expected (0.40).

Discussion
This study aimed to examine the validity and reliability of the BFNES-PAS for children and adolescents. After translating and retro translating, a pre-final version was sent to experts. The sentence "I worry about not being ashamed" was replaced by "I worry about embarrassing myself" (item 13). Also, in item 10, the sentence "it has little effect on me" was replaced by "I don't care", following experts' suggestions. This pre-version received strong agreement among experts concerning high or total clarity and relevance 27,28 . Professionals also reported clarity and pertinence of BFNRS-PAS items. The BFNRS-PAS showed adequate content validity evidence.
The initial EFA analysis showed adequate values for KMO and Bartlett sphericity test, but inadequate load factor (λ < 0.50) and communalities (h < 0.40) values for the items 2, 4, 7, and 10. The items were excluded, and the model showed adequate KMO, Bartlett sphericity, and adjustment indexes 30 . Besides, the version with specific items showed adequate load factor and communalities. The optimal results, in several tests, confirmed the unidimensionality of the scale, namely the implementation of parallel analysis 30 , the congruence, the common variance, and the mean of item residual absolute loadings analyses 41 .
Items 2, 4, 7, and 10 correspond to reverse-worded items from the original short version of the scale. Similarly, other studies reported difficulty in the assessment of the fear of negative evaluation with reverse items. So, compared to the original scale with the brief form that contains only direct items, the original version presented inferior validity indexes 18, 19 . For example, in studying Chinese students, the results showed stronger parsimony and excellent reliability and validity only for the BFNE with specific items 19 . Although a two-dimensional model, with one dimension composed only by straightforward items and others with reverse-worded items, has been provided in the literature 20,21,42 our results provided evidence for unidimensionality of the scale.
The CFA results showed an adequate fit of the model 31 . Most of the load factors showed values considered adequate 33 . Items 12 and 14 showed loads near to adequate, and when eliminated, it produced a small modification in the model. So, for providing additional information to the model, these items were maintained. The multigroup analysis showed configurational, metric, and scalar invariance for sexes and age groups. Therefore, the struc- ture, including configuration, load factor, and error, did not vary in these groups 32 . Results of an adequate internal structure of BFNE-S were also reported previously 21,42 , similar to our results. Our results for alpha and based on polychoric correlations of the items 43 and item-total correlations 44 showed adequate values. The individual item reliability showed appropriate values for most of the items. The composite reliability and AVE 36,44 showed fair values and provided additional support for evidence of measure precision 37 of the BFNES-PE. Also, the intraclass correlation coefficient presents adequate test-retest reliability 38,39 . These results support the high internal consistency.
The network analysis showed adequate good-of-fitindex for relational structure measured. Item 11 showed the highest closeness, betweenness, and strength. Besides, items 15 and 6 showed high closeness values. Closeness, a measure that reveals the shortest path between a given node and the remaining nodes, suggested that these items given a central influence upon others in this network 45 . Betweenness, quantification of the number of times a node acts as a bridge along the shortest path between two other nodes 46 indicates that, in our data, item 11 seems to mediate the relation between various other nodes in this network. Strength is a measure that indicates a more robust relationship among particular nodes through the sum of the absolute value of its connections with other nodes in a network 46 suggested that, in our data, the item maintains relationships more robust with other nodes. These results showed that these items have the optimal ability to discriminate between fear of negative evaluation levels in physical education contexts.
These items, especially item 11, are related to what other people think about him/her. It seems that when the judgment from others is more "out of our control" or is less dependent on "own performance", the fear of negative evaluation may be more expressive. So, this is a critical point in the assessment of the fear of negative evaluation. Contrarily, in this network, item 12 showed the lowest closeness, betweenness, and strength. It refers to worry about saying or doing the wrong things, factors that are more within one's control, or depend more on one's attitudes, and therefore were more easily avoidable. Individuals who are high in fear of negative evaluation behave in ways to avoid, when possible, evaluative settings 47,48 . Further, youths with social phobia frequently display a wide range of avoidant strategies when encountering socially distressing situations 7 . These strategies included noncompliance and refusal, such as pretended not to hear the person talking to them, hid their eyes so they would not be called on, and refuse to do as they were asked 7 . So, more avoidable situations may cause less fear of negative evaluation from others.
Explicitly, in physical activity and sports practices, to say or do wrong things depends partly on the individual level, quality of practice, and level of engagement in motor activities. For example, if the children are not required to participate effectively in practices or do practice or do not offer adequate experiences (i.e., challenging tasks), they may choose to avoid the experiences that can fail and consequently receive negative judgment from others. Thus, the feelings of individual control about participation in these situations may affect the fear of negative evaluation from others less strongly. The argument may explain why item 12 showed the lowest influence upon others.

Limitations and strength of the study
The study's Strength was using robust psychometric procedures in a large and representative Brazilian sample to provide the first validity version of a scale that measures the fear of negative social evaluation in physical activity and sport settings. A limitation of this study was not investigating the criterion validity evidence of the scale. Futures studies can address this unassessed point.

Conclusion
Based on translation and adaptation procedures and statistical model support, the BFNES -PAS presented substantial evidence of validity to assess the appropriateness of the fear of negative evaluation of Brazilian children in the physical activity and sports settings. This scale could be used to investigate further the complex relationships among other essential aspects, such as engagement in physical activities, self-perceptions of competence, and physical appearance among children and adolescents. Besides, BFNES -PAS's information may help the teachers, coaches, psychologists, and administrators make decisions about the interventional procedures related to social anxiety in sports and physical activity programs.

Declaration of interest statement
The authors declare no conflicts of interest.