The Body Appreciation Scale-2 (BAS-2) is the standard self-report measure of body appreciation, a core facet of positive body image, and exists in dozens of language translations. A five-country study of adults from Iran, Japan, Poland, Serbia, and the United States (N = 2944) achieved only partial metric invariance for the BAS-2 — the loading of one item had to be freed in Japan — and documented age- and gender-related differential item functioning, leaving open what causes cross-national item non-equivalence and how valid cross-cultural comparisons of body appreciation should be made. Later work transformed the landscape: a study of 65 nations and 40 languages found full scalar invariance and suggested that earlier failures may reflect differences in survey administration rather than the instrument, and a representative German study found age- and body-mass-index-related differential item functioning and published population-based norms. The surviving open problem is to identify the precise sources of BAS-2 item non-equivalence — genuine cultural differences in the construct versus translation artifacts versus respondent characteristics — to establish which respondent-level variables beyond age and gender produce item-level bias, and to develop and validate procedures (invariant anchor items, linking, culture-specific norms) for latent-mean comparison where full invariance does not hold.

Background

Positive body image research studies attitudes such as accepting, respecting, and holding favorable opinions toward one's body; its central measurable construct is body appreciation. The Body Appreciation Scale-2 (BAS-2; Tylka & Wood-Barcalow, 2015) is a 10-item questionnaire with 5-point agreement ratings that taps this construct as a single dimension, and it has been translated and validated in dozens of languages and countries. Comparing average BAS-2 scores across countries is only meaningful if the items measure the same thing, in the same units, in each group. This property is called measurement invariance, and it is tested in a standard sequence using multi-group confirmatory factor analysis (a factor model fitted simultaneously to all groups): configural invariance means the same one-factor pattern holds in each group; metric invariance means each item's loading (how strongly it reflects the latent construct) is equal across groups, which is needed to compare correlations and regression slopes; scalar invariance means each item's intercept (its baseline response level) is also equal, which is the minimum requirement for comparing latent factor means between groups. When a single item behaves differently across groups, this is called differential item functioning (DIF); one way to detect DIF for respondent characteristics is a MIMIC model (multiple indicators, multiple causes), in which the latent factor and the individual items are regressed on covariates such as age and gender, so that a significant direct effect of a covariate on an item indicates that the item functions differently at different covariate levels. Before the five-country study described below, multinational evidence was mixed: a two-country comparison of Malaysia and the United Kingdom supported only partial scalar invariance (Todd & Swami, 2020), and a cross-validation across eight countries also achieved only partial scalar invariance once several item intercepts were freed (Aimé et al., 2020). Razmus, Razmus, Tylka, Jović, Jović, and Namatame (2020) then examined the BAS-2 in adult samples from Iran, Japan, Poland, Serbia, and the United States (N = 2944, ages 18–82). They found the one-factor structure replicated in every country (configural invariance), but full metric invariance failed; freeing the loading of Item 10 in Japan yielded acceptable partial metric invariance. A MIMIC model showed significant DIF: age directly affected five items (Items 1, 7, 8, 9, and 10) and gender directly affected several items (including Items 1, 3, and 8). Their conclusion was that body appreciation may be shaped by country, language, age, and gender to different degrees, and that invariance analyses should precede any cultural group comparison on the BAS-2. Because partial metric (but not scalar) invariance was achieved, valid comparison of latent means across those five countries was not possible. These findings leave unresolved: the precise sources of the observed item non-equivalence (true cultural differences in what constitutes body appreciation, translation or linguistic artifacts, or specific cultural habits and exposures); potential predictors of noninvariance beyond age and gender (body mass index, socioeconomic status, education); the generality of the partial-invariance finding across further languages, cultures, and age ranges; and methods (invariant anchor items, item-level linking, culture-specific calibrated norms) that would permit valid latent-mean comparisons. The subsequent literature changed this picture substantially. Most importantly, Swami, Tran, Stieger, et al. (2023) analyzed the Body Image in Nature dataset (data collected 2020–2022) and reported full scalar invariance of the BAS-2 across 65 nations, 40 languages, gender identities, and age groups — the strongest possible basis for latent-mean comparison — and directly estimated large latent differences in body appreciation across nations and languages. They suggested that the discrepancy with earlier partial-invariance findings, including Razmus et al.'s, may reflect the lack of operational equivalence (identical survey format, recruitment, and administration) in earlier secondary-data studies, and reported that their results suggest translation concerns such as rendering the word 'love' in Item 6 do not unduly affect cross-language understanding. However, their samples were recruited opportunistically and cannot be considered representative of their nations, several world regions were under-represented, and the specific DIF documented by Razmus et al. (Item 10 in Japan; age and gender effects) and by Zarate, Marmara, Potoczny, Hosking, and Stavropoulos (2021), who found metric non-invariance across genders in a pooled English-speaking sample, was not explained. Meanwhile, field practice still includes cross-national comparisons of BAS-2 scores without any invariance testing (e.g., Hanson et al., 2024, six countries), which is exactly the practice Razmus et al. warned against.

Problem Statement

The four parts below are posed together as one problem: the testing designs in part 2 are the means of testing the source attribution in part 1, while parts 3 and 4 pose separate questions. No primary-data collection or new dataset analysis is required for any part.

  1. Source attribution. Identify the precise sources of the item-level non-equivalence observed for the Body Appreciation Scale-2 across languages, nations, and demographic groups — that is, determine the relative contributions of (i) genuine cultural or phenomenological differences in how body appreciation is understood, (ii) translation and linguistic artifacts, (iii) respondent characteristics and exposures such as age, gender, body mass index, socioeconomic status, and education, and (iv) operational or survey-administration non-equivalence such as non-harmonized survey format, recruitment, or sampling — and give a reasoned attribution of the relative contribution of each of these four candidate sources.
  2. Testing designs. Specify the targeted qualitative and quantitative designs (for example, cognitive interviewing across languages, controlled translation variants, within-nation cross-language comparisons, or multinational samples with harmonized administration and representative or probabilistic sampling) that would test or refine the source attribution in part 1.
  3. Predictors of noninvariance. Establish which respondent-level predictors beyond age and gender (for example, body mass index, socioeconomic status, education) produce differential item functioning on BAS-2 items in multinational samples, identifying the predictors supported by existing evidence and specifying the analyses and datasets that would establish those not yet tested.
  4. Comparison procedures. Establish defensible procedures for valid cross-cultural comparison of BAS-2 scores — identification of invariant anchor items, item-level linking strategies, or culture-specific calibrated norms — for the settings where full scalar invariance does not hold.

References

  1. Magdalena Razmus, Wiktor Razmus, Tracy L. Tylka, Marija Jović, Marko Jović, Hikari Namatame (2020). Cross-cultural measurement invariance of the Body Appreciation Scale-2 across five countries. Body Image, 34, 270–276. https://doi.org/10.1016/j.bodyim.2020.07.003
  2. Tracy L. Tylka, Nichole L. Wood-Barcalow (2015). The Body Appreciation Scale-2: Item refinement and psychometric evaluation. Body Image, 12, 53–67. https://doi.org/10.1016/j.bodyim.2014.09.006
  3. Viren Swami, Ulrich S. Tran, Stefan Stieger, et al. (2023). Body appreciation around the world: Measurement invariance of the Body Appreciation Scale-2 (BAS-2) across 65 nations, 40 languages, gender identities, and age. Body Image, 46, 449–466. https://doi.org/10.1016/j.bodyim.2023.07.010
  4. Daniel Zarate, Joshua Marmara, Camilla Potoczny, Warwick Hosking, Vasileios Stavropoulos (2021). Body Appreciation Scale (BAS-2): measurement invariance across genders and item response theory examination. BMC Psychology, 9, 114. https://doi.org/10.1186/s40359-021-00609-3
  5. P. Warschburger, N. Behrend (2023). Further evaluation of the psychometric properties of the German version of the Body Appreciation Scale-2 (BAS-2): Cross-validation, measurement invariance, and population-based norms. Body Image, 45, 105–116. https://doi.org/10.1016/j.bodyim.2023.02.004
  6. Jennifer Todd, Viren Swami (2020). Assessing the measurement invariance of two positive body image instruments in adults from Malaysia and the United Kingdom. Body Image, 34, 112–116. https://doi.org/10.1016/j.bodyim.2020.05.009
  7. Annie Aimé, Matthew Fuller-Tyszkiewicz, Jacinthe Dion, Charlotte H. Markey, Esben Strodl, Marita McCabe, David Mellor, Antonio Granero Gallegos, Giada Pietrabissa, Manuel Alcaraz-Ibáñez, Catherine Bégin, Marie-Ève Blackburn, Marie Caltabiano, Gianluca Castelnuovo, Salvatore Gullo, Naomi Hayami-Chisuwa, Qiqiang He, Gianluca Lo Coco, Gian Mauro Manzonie, Michel Probst, Rachel F. Rodgers, Alvaro Sicilia, Christophe Maïano (2020). Assessing positive body image, body satisfaction, weight bias, and appearance comparison in emerging adults: A cross-validation study across eight countries. Body Image, 35, 320–332. https://doi.org/10.1016/j.bodyim.2020.09.014
  8. Louise Nicole Hanson, Alexis Gott, Megan Tomsett, Elozino Useh, Eloise Yeadon-Caiger, Rachel Clay, Jiamin Fan, Kennice Hui, Hongdi Wang, Elizabeth Helen Evans, Dorothy Cowie, Lynda Gaynor Boothroyd (2024). Examining body appreciation in six countries: The impact of age and sociocultural pressure. PLOS ONE, 19(7), e0306913. https://doi.org/10.1371/journal.pone.0306913
  9. Minsun Lee (2023). Validation and measurement invariance of a Korean version of the Body Appreciation Scale-2 in cisgender sexual minorities and heterosexual adults. Body Image, 47, 101625. https://doi.org/10.1016/j.bodyim.2023.101625
  10. Xiaoyin Liu, Peiyi Wang, Wesley R. Barnhart, Reza N. Sahlan, Shuqi Cui, Hongjian Cao, Jason M. Nagata, Tiffany A. Brown, Jinbo He (2026). Examining differences in positive body image across sexual orientation among Chinese male and female adults: A measurement invariance study of three widely-used self-report measures. Body Image, 57, 102076. https://doi.org/10.1016/j.bodyim.2026.102076