Attitudes are formed socialization, observation, direct experiences, cognitive evaluation or emotional reactions (Hogg & Vaughan, 2018). These influences, however, vary across social contexts. For example, children may develop negative attitudes toward diversity due to parents’ ideologies, social media hatred or peers’ behaviours observation (Priest et al., 2014). On the other hand, schools aim to foster positive attitudes toward diversity through information processing, experiential learning, or classical and instrumental conditioning (Hello et al., 2004). As a result, students may internally hold negative attitudes but adjust their explicit responses to align with school expectations. This situation when respondents provide answers that are viewed favourable by others rather than reflecting their true opinions, is referred to as social desirability bias (Holtgraves, 2004).

To address this methodological limitation, researchers employed various strategies (e.g., anonymity, confidentiality, neutral phrasing of statements, increasing cognitive load), including the development of specific measures. These measures use indirect questioning to assess tendencies to overreport socially approved behaviours (e.g., Crowne & Marlowe, 1960). The outcomes of these measures can be used to assess the extent to which responses are influenced by social desirability bias. The most prevalent methods that exist in the literature to account for this bias when social desirability is measured are:

• Method 1: Dropping respondents with high social desirability scores
  This method involves identifying and excluding respondents with high social desirability scores from the analysis (Larson, 2019). However, it is unclear about what constitutes a high score or what is the appropriate threshold. Moreover, excluding respondents from the analysis reduces the sample size.
• Method 2: Correlating key variables with social desirability scores
  This approach assesses whether social desirability bias is related with the constructs under investigation by examining their correlation (Steenkamp et al., 2010). However, there exist a debate on how strong a correlation indicates an issue. For example, some studies argue that statistically significant but small correlations do not pose a substantial issue (Nederhof, 1985).
• Method 3: Controlling for social desirability
  Incorporating social desirability as a covariate in analyses to control the effects of the bias seems to be a more straightforward approach (Podsakoff et al., 2003). Specifically, the social desirability measure can be turned into an index that could be added as independent variable in the analysis and thereby isolate the true relationships among the key variables.
• Method 4: Adjusting key variable scores after accounting for social desirability
  This method is useful when key variable scores are needed. It involves removing the systematic error of social desirability from the key measure (Fisher & Tellis, 1998; Paulhus, 1981). Specifically, the social desirability measure is regressed on the key variable. The residuals of this regression represent the corrected score, which is an estimate of the respondent's score on the variable with the bias removed. These scores are then normalized and transformed to match the original scale, in order to produce the new adjusted scores that can be directly compared to raw scores.

This study examines the extent to which social desirability bias interferes with the measurement of a self-reported construct (i.e., students’ multicultural attitudes), and how this bias may influence the relationship between multicultural attitudes and external factors. Specifically, the research questions are:

1. Does social desirability bias contaminate the factorial structure and measurement invariance of students’ multicultural attitudes? (Method 1)
2. In what extent are students’ responses on their multicultural attitudes related with a tendency to provide socially desirable responses? (Method 2)
3. Does social desirability alter potential relationships between students’ multicultural attitudes and their background characteristics? (Method 3)
4. How do factor scores of students’ multicultural attitudes change after accounting for social desirability? (Method 4)

Participants A stratified sample of 450 students from 22 classes was recruited for the study. The data was collected in the 2023–2024 school year with a paper-and-pencil format while ensuring anonymity and confidentiality. Consent forms from students’ parents and permissions from schools was obtained in advance. The 41.8% of students were in 7th grade and the rest in the 8th grade (typically represent age groups of 13-14 and 14-15 years old). Among the participants, 51.8% were male, 38.0% had an immigrant background (i.e., either the student or at least one parent being born in a country other than the local), and 6.4% reported using a language other than the local most of the time at home. Instruments The Munroe Multicultural Attitude Scale Questionnaire (MASQUE; Munroe & Pearson, 2006) was adapted to assess students’ attitudes toward diversity. The MASQUE comprises 18 items distributed across three dimensions (i.e., Knowledge, Care, and Action) in alignment with Banks’s transformative approach (2014), which integrates cognitive, affective, and behavioural domains of learning into attitudes formation. Rigorous translation procedures were followed (Fenn et al., 2020) to develop a Greek version of the instrument, including independent translation from two researchers, back-translation, and face validity procedures. The short Form C (Reynolds, 1982) of the Marlowe-Crowne Social Desirability Scale (MCSDS) was used to assess respondents’ tendency to respond "in a culturally appropriate and acceptable manner" (Crowne & Marlowe, 1960, p. 353). This 13-item short form consists of statements designed to detect the tendency for social approval and is scored on a dichotomous scale (1=True, 0=False). The translation of the MCSDS–Form C was based on the work of Lavidas and Gialamas (2019). A review of the Greek version was necessary to ensure age-appropriateness for the secondary education students who constituted the sample of this study. Data analysis Confirmatory Factor Analysis (CFA) was used to determine the factorial structure of MASQUE and MCSDS–Form C and the extent to which social desirability bias might contaminate the factorial structure and measurement invariance of students’ multicultural attitudes. Structural Equation Modelling (SEM) was then used to investigate the relationship between students’ multicultural attitudes and social desirability, and to assess whether social desirability alter the relationship between students’ multicultural attitudes and background characteristics. Last, linear regression was used to produce adjusted scores of students’ multicultural attitudes after accounting for social desirability.

First, the results (Table 1) indicated that the factorial structure and measurement invariance (across gender and immigrant background) of the MASQUE remained stable across three samples (Method 1): (a) the full sample of students (n=450), (b) a sample excluding students with social desirability scores greater than 2 standard deviations above the mean (n=430), and (c) a sample excluding students with social desirability scores greater than 1 standard deviation above the mean (n=368). To estimate the extent to which students’ multicultural attitudes were related with social desirability we modelled a correlation (Method 2) between MCSDS–Form C and each of the three MASQUE factors (Know, Care, and Act). The results revealed that social desirability was moderately related only with Care (.41) and Act (.48) but not with Know. The next step was to examine whether social desirability alter potential relationships between students’ multicultural attitudes and their background characteristics. Only gender was found to affect the three factors of MASQUE. The model was tested with and without social desirability latent variable as a covariate (Method 3). The effects of gender remained the same when MCSDS–Form C was added or not to the model (.44 on Know, .35 on Care, and .37 on Act). Last, we calculated the new adjusted scores for the Care and Act factors that were found to be related with social desirability (Method 4). The process of calculation is presented in Table 2. After accounting for social desirability, the factor means were significantly reduced (Table 3). However, the gender differences were maintained (Care: t(446)=-5.701, p>.001; Act: t(446)=-5.652, p>.001). The results shows that self-reported students’ multicultural attitudes may be influenced by a tendency to respond in a socially acceptable way. However, this bias did not seem to significantly affect neither the measurement accuracy nor the effect of the examined relationships.

Banks, J. A. (2014). An introduction to multicultural education (5th edition). Pearson Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24(4), 349–354. https://doi.org/10.1037/h0047358 Fenn, J., Tan, C., & George, S. (2020). Development, validation and translation of psychological tests. BJPsych Advances, 26(5), 306–315. https://doi.org/10.1192/bja.2020.33 Fisher, R. J., & Tellis, G. J. (1998). Removing Social Desirability Bias With Indirect Questioning: Is the Cure Worse than the Disease? Advances in Consumer Research, 25(1), 563–567. Hello, E., Scheepers, P., Vermulst, A., & Gerris, J. R. M. (2004). Association between Educational Attainment and Ethnic Distance in Young Adults: Socialization by Schools or Parents? Acta Sociologica, 47(3), 253-275. https://doi.org/10.1177/0001699304046222 Hogg, M. A., & Vaughan G. M. (2018). Social psychology (8th ed.). Pearson Holtgraves, T. (2004). Social Desirability and Self-Reports: Testing Models of Socially Desirable Responding. Personality and Social Psychology Bulletin, 30(2), 161–172. https://doi.org/10.1177/0146167203259930 Larson, R. B. (2019). Controlling social desirability bias. International Journal of Market Research, 61(5), 534-547. https://doi.org/10.1177/1470785318805305 Lavidas, K. & Gialamas, V. (2019). Adaption and psychometric properties of the short forms Marlowe-Crowne Social Desirability Scale with a sample of Greek university students. European Journal of Education Studies, 6(8). https://doi.org/10.5281/zenodo.3552531 Munroe, A., & Pearson, C. (2006). The Munroe Multicultural Attitude Scale Questionnaire: A New Instrument for Multicultural Studies. Educational and Psychological Measurement, 66(5), 819-834. https://doi.org/10.1177/0013164405285542 Nederhof, A. J. (1985). Methods of coping with social desirability bias: A review. European Journal of Social Psychology, 15(3), 263–280. https://doi.org/10.1002/ejsp.2420150303 Paulhus, D. L. (1981). Control of social desirability in personality inventories: Principal-factor deletion. Journal of Research in Personality, 15(3), 383–388. https://doi.org/10.1016/0092-6566(81)90035-0 Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879. Priest, N., Walton, J., White, F., Kowal, E., Baker, A., & Paradies, Y. (2014). Understanding the complexities of ethnic-racial socialization processes for both minority and majority groups: A 30-year systematic review. International Journal of Intercultural Relations, 43, 139–155. https://doi.org/10.1016/j.ijintrel.2014.08.003 Reynolds, W. M. (1982). Development of reliable and valid short forms of the Marlowe-Crowne Social Desirability Scale. Journal of Clinical Psychology, 38(1), 119–125. Steenkamp, J. B. E., De Jong, M. G., & Baumgartner, H. (2010). Socially desirable response tendencies in survey research. Journal of Marketing Research, 47(2), 199–214.