Participants were recruited using convenience sampling. At the start, 558 individuals began participating in this study, but 140 (25.1%) of them did not finish the survey. Additionally, 70 (16.7%) of the remaining participants were excluded because they did not correctly respond to the attention check item. Therefore, the final sample consisted of 348 university students (72.4% female) from various departments in state universities, aged between 18 and 61 years (M = 22.27; SD = 4.80). Of these, 19.5% were in Year 1, 23.9% in Year 2, 18.7% in Year 3 and 37.9% in Year 4. In terms of socioeconomic status, 17.8% evaluated themselves as having low socioeconomic status (i.e., lower-middle income, poor or very poor), with 52.9% in the middle and 29.3% having high socioeconomic status (i.e., good income, rich or very rich).

The Climate Self-Protection Scale was developed by Wullenkord and Reese (2021). For use in this study, for the first time, the scale was translated into Turkish by the authors and the translation was checked by three experts in social psychology. The original scale had 26 items and five factors: (1) rationalisation of own involvement; (2) avoidance; (3) denial of personal outcome severity; (4) denial of global outcome severity; and (5) denial of guilt. The scale ranged from 1 (Strongly disagree) to 6 (Strongly agree), with a higher value indicating a stronger construct.

Exploratory factor analysis was required to determine the factorial structure of this version. The Kaiser–Meyer–Olkin coefficient was 0.90, and Bartlett’s test of sphericity was statistically significant (χ²(325) = 5664.25, p < .001), indicating that the scale was appropriate for factor analysis (see Supplementary Table 1 in Kıral Uçar et al., 2025 for the results of the exploratory factor analysis). Therefore, a principle component analysis was performed, followed by a varimax rotation. The principle component analysis identified five factors with eigenvalues > 1.0, and the scree-plot confirmed the five-factor solution. Supporting H1a, apart from Item 25 (I don’t need to make climate change a matter of conscience), all the items were loaded as in the original scale. Item 25 was loaded strongly on both the rationalisation of own involvement (λ = .50) and denial of guilt (λ = .47) factors. However, in terms of interpretability, it was decided to leave it on the denial of guilt factor, as in the original scale. The rationalisation of own involvement factor included seven items (e.g., My personal influence on climate change is negligible) and accounted for 32.89% of the total variance, with an eigenvalue of 8.55. The avoidance factor consisted of eight items (e.g., When I get worried about climate change, I try to think of something else) and explained 15.17% of the total variance, with an eigenvalue of 3.94. The denial of personal outcome severity factor, which had four items (e.g., Nothing will happen to me as a consequence of climate change because Turkey is a safe country) explained 9.33% of the total variance, with an eigenvalue of 2.43. The denial of global outcome factor severity had three items (e.g., I believe that climate change won’t be as severe as expected in the future) and accounted for 6.03% of the total variance, with an eigenvalue of 1.57. Last, the denial of guilt factor included four items (e.g., I have a guilty conscience because I know that I should behave more sustainably) and explained 4.09% of the total variance, with an eigenvalue of 1.06. The factor loadings of the items ranged from .67 to .86 for rationalisation of own involvement, from .54 to .86 for avoidance, from .70 to .81 for denial of personal outcome severity, from .66 to .82 for denial of global outcome severity and from .47 to .84 for denial of guilt. Cronbach’s alpha was computed as .91, .90, .84, .86 and .75 for rationalisation of own involvement, avoidance, denial of personal outcome severity, denial of global outcome severity and denial of guilt, respectively, which indicate higher levels of scale reliability.

Literal denial was assessed using a five-item scale developed by McCright and Dunlap (2011) (sample item—There is no scientific consensus that climate change is occurring). The scale was used based on the unifactorial Turkish version (Nartova-Bochaver et al., 2022) as a six-point Likert scale ranging from 1 (Strongly disagree) to 6 (Strongly agree). Cronbach’s alpha was satisfactory for this study (α = .83).

The Ecological Dominance Orientation Scale is a one-item iconographic scale developed by Uenal, Sidanius, Maertens et al. (2022). The participants were asked to indicate their preference for human–environment relations using a slider ranging from 1 (Less hierarchical) to 7 (More hierarchical), with a higher value indicating a stronger construct. The two ends of the scale were illustrated with an appropriate image. The scale was adapted to Turkish by the authors of this study. It was found that EDO was significantly correlated to SDO (see Table 1), indicating criterion validity.

The Social Dominance Orientation Scale SDO₇ is a 16-item scale (Ho et al., 2015) ranging from 1 (Strongly disagree) to 6 (Strongly agree), with a higher value indicating a stronger construct. The scale was adapted into Turkish by Kaynak et al. (2021). Cronbach’s alpha was satisfactory for this study (α = .91).

Past Pro-Environmental Behaviours within the last three months were assessed using seven items (including major activities, such as using public transportation, energy saving, using nature-friendly products) (Kıral Uçar, 2020). The scale ranged from 1 (Not suitable at all) to 6 (Totally suitable). Cronbach’s alpha was acceptable (α = .66).

In terms of demographics, gender, age, department and perceived economic status were obtained. Anonymized raw data and syntax for all the analyses conducted in this study can be found at Kaynak Malatyalı et al. (2023).

The Ethics Committee of the corresponding author`s university authorised the research. The participants were then recruited online, and the questionnaires were distributed via email lists, social media and in classes. The participants were briefly informed about the purpose of the study and completed an informed consent form before responding.¹1

We also applied pro-environmental behavioural intentions and social desirability scales but did not use these variables in this study.

The correlation analysis (see Table 1) revealed that, in line with H1b, all types of climate self-protection strategies were positively related to literal denial. Additionally, all types of climate self-protection strategies and literal denial were negatively related to past pro-environmental behaviour, whereas SDO was negatively related to past pro-environmental behaviour and EDO was not related to it, SDO was positively related to all types of climate self-protection strategies, and EDO had a similar correlational pattern, apart from for avoidance, which was unrelated.

First, to test the predictive role of gender, age, SDO and EDO on climate change denial, a series of separate hierarchical multiple regression analyses were run for each climate self-protection strategy dimension and literal denial. In these analyses, age and gender were added in the first step. The results indicated that gender predicted literal denial, rationalisation of own involvement, denial of personal outcome severity, denial of global outcome severity and denial of guilt, but not avoidance (see Table 2), mostly supporting H3. That is, male participants had higher scores on all forms of climate change denial, except for avoidance. Age only predicted the avoidance dimension of climate self-protection strategies in which older participants were less likely to avoid climate change. In the second step, SDO and EDO were added to the model. The results supported H2a and partially supported H2b, in which SDO predicted literal denial and all forms of climate self-protection strategies, but EDO only predicted denial of guilt after controlling for age and gender. That is, while SDO was associated with an increase in literal denial, rationalisation of own involvement, avoidance, denial of personal severity, denial of global outcome severity and denial of guilt, EDO was only linked to an increase in denial of guilt after controlling for age and gender. Additionally, gender no longer predicted literal denial and denial of personal severity after controlling for EDO and SDO. The final model explained 16% of the variance in literal denial, 15% of the variance in rationalisation of own involvement, 6% of the variance in avoidance, 13% of the variance in denial of personal outcome severity, 8% of the variance in denial of global outcome severity and 26% of the variance in denial of guilt.

Finally, to test the associations between gender, age, SDO, EDO, literal denial, climate self-protection strategies and past pro-environmental behaviour, a hierarchical multiple regression analysis was run.²2

Similar results for pro-environmental behavioural intentions were obtained and presented in Supplement Table 2 of Kıral Uçar et al. (2025).