Psychological Distance to Science: Psychometric Evaluation of the Swedish PSYDISC-Scale and as a Predictor of Science Skepticism
Funding: The authors received no specific funding for this work.
ABSTRACT
The main purpose of the present study was to test if PSYDISC demonstrates incremental validity in predicting science skepticism across domains, beyond science attitude measures, in a Swedish context. Swedish-speaking participants (N = 314) completed an online survey containing the PSYDISC scale and two science attitude scales (Faith in Science and Negative Perceptions of Science), as well as measures of science skepticism regarding climate change, vaccines, and genetically modified organisms (GMO). Results from confirmatory factor analysis found support for the previously found four-factor model (hypothetical distance, social distance, spatial distance, and temporal distance). Additionally, measurement invariance was supported for age and gender, and local psychometric properties indicated good internal consistency with good discriminant validity but somewhat poor convergent validity. Regression analyses showed that PSYDISC demonstrated incremental predictive validity for climate change and vaccine skepticism but not for GMO skepticism. The present study thus supports PSYDISC as an incremental predictor of science skepticism.
Summary
- The study tested if PSYDISC predicts science skepticism beyond traditional science attitude measures in Sweden.
- 314 Swedish-speaking individuals completed an online survey including PSYDISC and science attitude scales.
- Confirmatory factor analysis supported a four-factor model
- PSYDISC predicted skepticism
Even though there is a general trust in science, a considerable level of skepticism within certain scientific domains exists, with divergent views being observed between public opinion and scientific consensus in the United States, particularly on topics such as the safety of genetically modified foods and the influence of human activity on climate change (Funk and Rainie 2015). Despite the widespread agreement on the safety of vaccines worldwide, there are notable levels of skepticism regarding vaccines in high-income countries (Gallup 2019). Rutjens et al. (2022) conducted a comprehensive study across 24 countries, including Sweden, exploring skepticism toward climate change, vaccination, genetically modified organisms (GMOs), and evolution. In the Swedish context, the results indicated a moderate level of skepticism toward climate change and GMOs, with skepticism toward evolution and vaccination being relatively low.
Većkalov et al. (2024, 18) defined science skepticism as “the systematic and unwarranted rejection of empirical evidence or well-established scientific findings” (see also Rutjens and Hornsey 2025, for a more nuanced definition of science skepticism). While a critical approach is a cornerstone of science, this poorly grounded skepticism can have harmful consequences for human and environmental health. Denial or ignorance of scientific consensus can undermine public support for taking action against climate change (Gifford 2011; van der Linden et al. 2015), which is widely recognized as necessary to mitigate harm to ecosystems and societies (Steffen et al. 2018). Additionally, distrust in scientists and the acceptance of misinformation are associated with a reduced willingness to comply with public health guidance, including vaccination (Roozenbeek et al. 2020).
Due to the prevalence and consequences of science skepticism, studies concerning this phenomenon and its possible psychological underpinnings have increased in recent years. Earlier research on science skepticism can be described as consisting of two main perspectives (Većkalov et al. 2024; Zarzeczna et al. 2021), with the first being centered around science understanding and knowledge as predictors of science attitudes, and the second being focused on the underlying motivations for rejecting science, such as ideologies and worldviews. Većkalov et al. (2024) argued that both perspectives, although having il supplement the knowledge of science skepticism, have been focusing on predictors that are both difficult to change and only specific to scientific domains.
Within the perspective focused on understanding and knowledge of science as predictors of science attitudes, the “information deficit model” of science communication has had great influence (Cortassa 2016). The model suggests that science skepticism is due to a lack of knowledge and understanding about science among the public. The deficit account has been criticized for being too simplistic and paternalistic (e.g., Simis et al. 2016), and while a meta-analysis (Allum et al. 2008) suggests only weak correlations between science knowledge and science attitudes, newer studies suggest that this is only the case within specific science domains (Rutjens et al. 2018, 2022). For example, informing about the scientific consensus about genetically modified foods is effective in correcting misperceptions, but not for climate change (van Stekelenburg et al. 2021). Recent research (e.g., van Stekelenburg et al. 2021; Većkalov et al. 2024) also shows the presence of effects for climate change, but the effects tend to be very small.
Within the other perspective of research on science attitudes, motivational correlates of science rejection have been emphasized, such as ideologies and worldviews, identity needs, fears, and conspiratorial ideation (Hornsey and Fielding 2017). However, motivational predictors also seem to be largely domain-specific. Political ideology is a consistent predictor of climate change skepticism, with higher levels of political conservatism being associated with higher levels of climate change skepticism (Hornsey et al. 2016; Rutjens et al. 2021, 2022). Political conservatism has also been shown to positively predict skepticism regarding vaccines and evolution, but not GMO (Rutjens et al. 2022). Religious beliefs are positively associated with evolution skepticism (Rutjens and van der Lee 2020; Rutjens et al. 2022), while spirituality consistently positively predicts vaccine skepticism (Rutjens and van der Lee 2020; Rutjens et al. 2021, 2022). However, neither religiosity nor spirituality is consistently associated with GMO skepticism (Rutjens and van der Lee 2020; Rutjens et al. 2022). In addition to being domain-specific, ideological predictors are relatively stable beliefs (Peterson et al. 2020; Većkalov et al. 2024); attempts to reduce science skepticism can be challenging and ethically questionable, particularly when targeting factors that are difficult to influence. As argued by Većkalov et al. (2024), previous research from both the knowledge-focused and ideology-focused perspectives has not been able to identify psychological predictors of science skepticism that are applicable across different domains and can be effectively modified.
In light of earlier research being concentrated on relatively stable and domain-specific predictors, Većkalov et al. (2024) developed and tested a new construct, “psychological distance to science,” as a general predictor of science skepticism. PSYDISC addresses the individual's perspective of science, and to what degree the individual finds science relevant and tangible (Većkalov et al. 2024). According to the theoretical framework of PSYDISC, attitudes to science can be predicted by how the individual perceives the relation between the self and science. If science is perceived to be within psychological proximity to oneself, it will also be seen as more relevant in terms of its tangibility and consequences (Zarzeczna et al. 2022). In response to previous research focusing on stable and domain-specific predictors, Većkalov et al. (2024) introduced a novel construct called “psychological distance to science” (PSYDISC) as a broad predictor of science skepticism, which examines individuals' perspectives on science and their perception of its relevance and tangibility. The theoretical framework of PSYDISC suggests that attitudes toward science can be predicted by how individuals perceive the relationship between themselves and science. When science is perceived to be psychologically close to oneself, it is also viewed as more relevant in terms of its tangibility and consequences (Zarzeczna et al. 2022).
In the context of PSYDISC, the experience of psychological proximity to science involves perceiving science as relevant for the present time (temporal) and for the local community (spatial). It also involves perceiving science as performed by individuals who are easily accessible and similar to oneself (social) and that science is perceived as tangible, with practical and noticeable effects on the world (hypothetical; Većkalov et al. 2024).
Većkalov et al. (2024) argue that people who perceive science in their psychological proximity will be more sensitive to developing attitudes and behaviors that are consistent with the scientific facts available to the public.
In a series of studies conducted in the UK and the US, Većkalov et al. (2024) developed and tested a scale measuring psychological distance to science (PSYDISC) and assessed its validity in predicting science skepticism. Factor analyses were used to develop and validate the four-factor model with the distance dimensions, which offered a good model fit to the data. When Većkalov et al. (2024) tested if PSYDISC could predict science skepticism across science domains, they found that PSYDISC predicted skepticism toward climate change, vaccination, evolution, and GMO after controlling for demographics, religiosity, science knowledge, and ideology. Moreover, in Study 4, they further tested its incremental predictive validity by specifically exploring its contribution beyond established measures of general science attitudes, which have been shown to predict science skepticism in earlier studies (Farias et al. 2013; Morgan et al. 2018; Rutjens et al. 2018). The results showed that PSYDISC predicted skepticism beyond established predictors across different science domains. Interestingly, recent experimental research indicates that PSYDISC can be influenced to decrease science skepticism (Zarzeczna et al. 2022). While the malleability of PSYDISC is highly relevant, further investigation is warranted regarding the existing knowledge on the fundamental connections between PSYDISC and science skepticism, particularly since previous studies have been confined to English-speaking settings (the UK and the US). Additionally, the incremental predictive validity of PSYDISC was evaluated in only one of Većkalov et al. (2024) studies, showing rather modest effect sizes of the measure's unique predictive power (ranging from 2% to 5% of explained variance).
The present study had two objectives, of which the first was to assess the factor structure and psychometric properties of a Swedish translation of the PSYDISCs using confirmatory factor analysis (CFA). We expected to find support for the previously found four-factor model (hypothetical distance, social distance, spatial distance, and temporal distance). The second objective was to replicate PSYDISCs' incremental predictive validity in Study 4 Većkalov et al. (2024), but with some minor adjustments for the present context. Specifically, we examined the contribution of PSYDISCs on science skepticism regarding climate change, vaccines, and GMOs, beyond the effect of two science attitude scales (Faith in Science and Negative Perceptions of Science).
1 Method
1.1 Participants and Procedure
Swedish-speaking participants (N = 317, ≥ 18 years) were recruited via the authors' accounts on social media (Facebook, Instagram, LinkedIn, Reddit and Discord) as well as physical posters in public places (e.g., the university). The social media posts contained information about the study and an anonymous survey link, while the physical posters contained the same information and a QR code leading to the survey. The survey was hosted by the online survey platform Qualtrics (https://www.qualtrics.com). Before being published, the survey was piloted for estimations of time consumption and for improvements in user-friendliness. Distribution data on Qualtrics showed that the great majority of respondents found the survey through social media. Respondents did not receive compensation for their participation. The participants could choose not to participate or to terminate their participation by closing down the survey. Two out of 317 who opened the survey terminated their participation.
After having given their informed consent, participants first responded to the PSYDISC scale, second the science attitude scales in randomized order, and third the science skepticism scales in randomized order. All items within respective scales were randomized. Finally, participants answered questions regarding demographics, scientific work experience, political orientation, and religiosity. The survey included a Captcha to exclude potential bots and two test questions acting as attention checks. Three participants were excluded from the study due to failing at least one of the attention checks, leaving us with a final sample of N = 314 participants (n = 196 female, one nonbinary; Mage = 39.70, SDage = 14.0). The average formal education level was 16.6 (SD = 2.9) years. The study received ethical approval from the local committee as the subjects were anonymous, and similar surveys (with similar subjects) had been previously approved, making it unnecessary to seek approval from the regional ethics committee.
Our power analysis, conducted with the semTools R package (Version 0.5–6; Jorgensen et al. 2022), determined that, given a df (98) for the four-factor model, a sample size of 143 participants was required to closely fit the model and detect model misspecifications, while 161 participants were needed for a less close fit. Sample size calculations were based on 95% power, with an acceptable fit defined by a cut-off value of RMSEA ≤ 0.06. With respect to the multivariable regressions, an expected R squared change of 0.05 (equivalent to observed in Većkalov et al. 2024) for the inclusion of one predictor to the baseline model (consisting of 6 predictors) required 249 participants based on 95% power. Data are available from https://osf.io/ze4c7.
1.2 Design and Instruments
Variables used in the present study were the same as the ones used in Study 4 by Većkalov et al. (2024) with a few exceptions, as described below. In addition to the PSYDISC scale, two measures of general science attitudes were used as predictor variables: the Negative Perceptions of Science Scale (NPSS; Morgan et al. 2018) and Faith in Science (Farias et al. 2013; Rutjens et al. 2018). Većkalov et al. (2024) used a third attitude measure, the Credibility of Science Scale (CoSS; Hartman et al. 2017). However, due to CoSS being strongly correlated with NPSS, which captures a wider variety of science attitudes, it was excluded from their data analysis; therefore, it was excluded in the present study. As outcome variables, in line with Većkalov et al. (2024), measures of vaccine skepticism, climate change skepticism, and GMO skepticism were used. However, a fourth outcome variable used in Većkalov et al. (2024), evolution skepticism, was not included in the present study due to evolution skepticism being very rare in Sweden (Jerkert and Hansson 2005; Rutjens et al. 2022).
The PSYDISC, NPSS, and Faith in Science scales, as well as the three science skepticism scales, were translated into Swedish. The translation procedure of all items was composed of five steps: (1) Machine translation using the online translator DeepL (https://www.deepl.com/translator). (2) Evaluation of the machine translation. (3) Proposed adjustments (if considered needed). (4) Rationale for adjustments. (5) References for adjustments (i.e., alternative machine translation of specific words). Rationales for adjustments concerned adherence to the wording of the original items, sensitivity to the Swedish context, reader accessibility, and grammar. The translation procedure involved the authors of the current study and was completed when consensus was reached regarding all items.
1.2.1 PSYDISC
The PSYDISC scale is a validated measure (Većkalov et al. 2024) consisting of 16 items assessing psychological distance to science, of which four items pertaining to each of the four dimensions: hypothetical distance, social distance, spatial distance, and temporal distance. The items consist of statements that respondents were asked to indicate to what degree they agreed with, on a scale ranging from 1 (strongly disagree) to 7 (strongly agree). Items were summarized into a total score for each factor (and factor scores were summarized into a total score for the scale), with higher scores indicating a larger distance to science. All PSYDISC items and their respective translations are listed in Table 2. Cronbach's α for the Swedish version of the total scale was α = 0.86; hypothetical distance, α = 0.70; social distance, α = 0.75; spatial distance, α = 0.69; and temporal distance, α = 0.73.
1.2.2 Faith in Science
This 5-item scale is a shortened version of the validated Belief in Science Scale (Farias et al. 2013), used in earlier studies (Rutjens et al. 2018; Većkalov et al. 2024), which measures evaluations of the epistemic value of science. The translated version demonstrated good internal consistency (α = 0.83). Items include: “The only real kind of knowledge we can have is scientific knowledge” and “Science is the most efficient means of attaining truth.” Respondents were asked to indicate the degree to which they agreed with the statements on a scale from 1 (strongly disagree) to 7 (strongly agree). Items were summarized into a total score, with higher scores indicating higher faith in science.
1.2.3 Negative Perceptions of Science Scale (NPSS)
The NPSS was developed and validated by Morgan et al. (2018) and consists of 20 items gauging negative perceptions of science, including science being corrupt (e.g., “Scientists are often dishonest about their research findings”) and complex (e.g., “Scientific jargon is too complex to understand”). Items also tap into perceptions of science as heretical (e.g., “God is the ultimate way of knowing, not science”) and limited (e.g., “Science cannot explain everything”). Participants indicated to what degree they agreed with the statements on a 5-point scale (1 = strongly disagree to 5 = strongly agree). Items were summarized into a total score, with higher scores indicating a more negative perception of science. Cronbach's α for the Swedish version was α = 0.86.
1.2.4 Science Skepticism
Three 5-item scales used in earlier research on science skepticism (Lewandowsky et al. 2013; Većkalov et al. 2024) were translated and used in the present study. Items on climate change skepticism include “Humans are too insignificant to have an appreciable impact on global temperature” and “Human CO2 emissions cause climate change” (reverse coded). The vaccine skepticism scale includes the items “I believe that vaccines have negative side effects that outweigh the benefits of vaccination for children” and “Vaccinations are one of the most significant contributions to public health” (reverse coded). Finally, items on GMO skepticism include “I believe genetically engineered foods have already damaged the environment” and “GM of foods is a safe and reliable technology” (reverse coded). Respondents indicated to what degree they agreed with each statement, from 1 (strongly disagree) to 7 (strongly agree). For each form of science skepticism, items were summarized into a total score, with higher scores indicating higher science skepticism. Cronbach's α for the Swedish versions were; climate change skepticism, α = 0.85, vaccine skepticism, α = 0.84, and GMO skepticism, α = 0.86.
1.2.5 Background Variables
Data on demographics, scientific work experience, political orientation, and religiosity were collected to assess sample characteristics and to enable analysis of associations between these background variables, predictor variables, and outcome variables. The same background variables collected in Study 4 by Većkalov et al. (2024) were included, except for a few exclusions that were made to restrict survey length and to adjust to the Swedish context. The gender variable was binary (Female = 1, Male = 2) to allow meaningful use in statistical analyses. Political orientation was measured with two variables: (1) political ideology (left vs. right) and (2) political conservatism (progressive vs. conservatism). These variables were measured through questions that participants answered on a 9-point scale (1 = Left, 9 = Right; or 1 = Progressive, 9 = Conservative). The question regarding political ideology read as follows: “People often refer to their political orientation in the context of the left- and right-wing spectrum. What is your political orientation on the left/right-wing dimension?” The question regarding political conservatism was: “To what extent would you describe yourself as progressive or conservative?”. Religiosity was measured with the question: “How religious are you?” Participants responded on a 7-point scale (1 = Not religious at all, 7 = Very religious).
1.3 Data Analytic Procedures
1.3.1 Validation of the PSYDISC
Using confirmatory factor analysis (CFA), we initially tested if the structure could be represented by four correlated first-order factors. Due to some minor departure from the normality of the data, robust maximum likelihood estimation was used. Model fit was evaluated using chi-square statistics (χ2), Comparative Fit Index (CFI), Tucker-Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). With respect to the CFI and TLI, values above 0.95 indicate a good fit, and values above 0.90 suggest an adequate fit. SRMR values around 0.08 or lower indicate a good fit for the data. For the RMSEA, values below 0.06 are considered a good fit, while values below 0.08 suggest an adequate fit (Hu and Bentler 1999). In addition, the Akaike's Information Criterion (AIC) index (the smallest AIC indicating the best-fitting model) was used to identify the best-fitting solution.
For the model with the best fit, measurement invariance tests were conducted across age groups based on a median split (younger/equal to 37 years; older than 37 years) and gender. Ensuring measurement invariance (MI) is crucial for our analysis, as it confirms that the PSYDISC factors are consistently interpreted across different age and gender groups. Without this step, observed differences or similarities in responses might be misinterpreted due to variations in measure operation within these groups. We aimed to test MI for age and gender, as these demographic factors significantly influence psychological processes.
A sequential strategy was used to test the invariance at different levels. In order to establish equivalence in factor structure across the two groups (configural) model, all parameters were freely estimated across groups. Second, a metric model was fitted, in which the factor loadings were constrained to be equal, and the fit of this model was compared to the configural model. Third, a scalar model was fitted, in which factor loadings and item intercepts were constrained to be equal, which was compared to the second (metric) model. We report the Yuan-Bentler scaled difference chi-square test statistic in comparing competing nested models. Even though a scaled chi-square difference test for nested models can be used to index invariance between models, it suffers from the same dependency on sample size as the minimum fit function statistic. Thus, changes in model fit according to CFI, RMSEA, and SRMR were used. According to the criteria suggested by Chen (2007), a decrease in CFI of ≥ −0.01 in addition to an increase in RMSEA of ≥ 0.015 and SRMR of ≥ 0.030 corresponds to an adequate criterion indicating a decrement in fit between models for sample sizes > 300.
Finally, for the model with the best fit, composite reliability (CR) was calculated as a measure of internal consistency of the factors (values > 0.70 are indicative of good reliability). CR is computed from the squared sum of factor loadings and the sum of the error variance terms for the latent variable. Two measures of validity were calculated: (1) Discriminant validity is achieved when average variance extracted (AVE) is greater than maximum shared squared variance (MSV), and (2) convergent validity, AVE should be equal to or greater than 0.50 and lower than CR (i.e., variance explained by the construct should be greater than measurement error and greater than cross-loadings). The AVE is calculated as the mean-variance extracted for item loadings on the latent variable and represents the mean percentage of variation explained among the items of a construct (Hair et al. 2014).
The CFAs were carried out using the R (R Core Team 2021) package lavaan (Rosseel 2012).
1.3.2 Regression Models Predicting Skepticism
Since vaccine skepticism and climate change skepticism were non-normally distributed, robust multivariable regression models were used to investigate the association of robust regression with R (version 3.5.3), using the package robustbase 0.93–5 (Maechler et al. 2017). As GMO skepticism was normally distributed, an ordinary hierarchical linear regression with GMO skepticism as the outcome variable was conducted.
To test the incremental predictive validity of PSYDISC, we entered NPSS and Faith in Science in the first step for all three outcome variables, followed by PSYDISC in the second step in the regression models. The other covariates entered in the first step were determined by the zero-order correlations between each outcome variable, with significant correlations meeting the criteria for inclusion in the model.
All other statistical analyses were performed using JASP version 0.18.3 (JASP Team 2024).
2 Results
2.1 Confirmatory Factor Analyses
The CFA showed that the four-factor model was good according to all of the fit indices. The modification indices suggested that the inclusion of correlated error terms (Item 3 and Item 4 within the spatial distance scale) in the model would improve the model fit substantially. With this modification, all fit indices were further improved (see Table 1). Factor loadings of the four-factor model are shown in Table 2.
| Model | χ2 (df) | χ2/df | CFI | TLI | SRMR | RMSEA (95% CI) | AIC |
|---|---|---|---|---|---|---|---|
| Four-factor | 201.49 (98) | 2.06 | 0.93 | 0.92 | 0.059 | 0.059 (0.047–0.071) | 16,519.221 |
| Four-factor, modifieda | 129.32 (97) | 1.33 | 0.98 | 0.97 | 0.058 | 0.033 (0.016–0.046) | 16,450.922 |
- Note: Square error of approximation.
- Abbreviations: AIC = akaike information criterion, FI = Comparative Fit Index, SRMR = standardized root mean square residual, RMSEA = root mean.
- a The proposed 4-factor structure with error terms for Item 3 and Item 4 for the Spatial distance scale is correlated.
| Items (In Swedish) | HYP | SOC | SPAT | TEMP | |
|---|---|---|---|---|---|
| 1 |
Scientific knowledge is a reliable way to solve important issues (R) (Vetenskaplig kunskap är ett pålitligt sätt att lösa viktiga problem) |
0.631 | |||
| 2 |
Science provides accurate information about the world we live in (R) (Vetenskap ger korrekt information om den värld vi lever i) |
0.753 | |||
| 3 |
I can see the effects of science, whether positive or negative, on the world (R) (Jag kan se vetenskapens effekter på världen, vare sig de är positiva eller negativa) |
0.472 | |||
| 4 |
We can rely on science to deliver results that can be implemented in real life (R) (Vi kan lita på att vetenskapen ger resultat som kan tillämpas i verkligheten) |
0.605 | |||
| 1 |
The prospect of working as a scientist seems beyond my reach (Möjligheten att arbeta som forskare verkar vara utom räckhåll för mig) |
0.623 | |||
|
Scientists are very different from me (Forskare är väldigt olika mig) |
0.617 | ||||
| 3 |
I rarely interact with scientists in real life (Jag har sällan kontakt med forskare i det verkliga livet) |
0.751 | |||
| 4 |
It would be difficult for me to meet with a scientist (Det skulle vara svårt för mig att träffa en forskare) |
0.669 | |||
| 1 |
People from my local area don't become scientists (Människor från mitt närområde blir inte forskare) |
0.639 | |||
| 2 |
Very few scientists live or work in my town (Väldigt få forskare bor eller arbetar på min ort) |
0.673 | |||
| 3 |
Science and scientific research play a big role in my local area (R) (Vetenskap och vetenskaplig forskning spelar en stor roll i mitt närområde) |
0.463 | |||
| 4 |
Scientific research really contributes to my local area (R) (Vetenskaplig forskning bidrar verkligen till mitt närområde) |
0.518 | |||
| 1 |
Most of today's science is concerned with solving problems of the distant future (Merparten av. dagens vetenskap handlar om att lösa problem i en avlägsen framtid) |
0.831 | |||
| 2 |
We will see the impact of science more in the distant future than we do in the present (Vi kommer att se vetenskapens effekter mer i en avlägsen framtid än vad vi gör i dag) |
0.502 | |||
| 3 |
Science is mainly focused on the distant future (Vetenskapen är huvudsakligen inriktad på en avlägsen framtid) |
0.791 | |||
| 4 |
Scientists spend most of their time working on issues of the distant future (Forskare ägnar större delen av. sin tid åt frågor som rör en avlägsen framtid) |
0.872 |
- Note: R = reversed item.
- Abbreviations: HYP = hypothetical distance, SOC = social distance, SPAT = spatial distance, TEMP = temporal distance.
Measurement invariance tests regarding age showed support for configural invariance (suggesting a similar factor structure across the two age groups). There was no substantial decrease in model fit in the metric model, indicating that full metric invariance was achieved (i.e., the strength of the relationship between the items and constructs is equivalent across groups). However, with respect to the scalar invariance model (i.e., the item intercepts are equivalent across groups), the invariance exceeded the criteria for invariance based on the change in CFI, but not the change in RMSEA and SRMR. Inspection of the residuals indicated that releasing the constraints for the intercept that had the highest variability (Item 1 within the social distance scale) across groups was sufficient to achieve partial invariance in the CFI.
With respect to invariance for the genders, the results showed that the configural, metric, and scalar models also passed the criteria for invariance (see Table 3 for estimates).
| Model | Δχ2(df) | CFI | ΔCFI | RMSEA | ΔRMSEA | SRMR | Δ SRMR |
|---|---|---|---|---|---|---|---|
| Age invariance | |||||||
| Configural | — | 0.966 | — | 0.043 | — | 0.065 | |
| Weak/Metric | 10.17 (12) | 0.966 | 0.000 | 0.041 | −0.002 | 0.070 | 0.005 |
| Strong/Scalar | 61.89 (12) | 0.935 | −0.031 | 0.055 | 0.014 | 0.076 | 0.007 |
| Partial/Scalar | 42.94 (1) | 0.958 | −0.008 | 0.043 | 0.002 | 0.071 | 0.001 |
| Gender invariance | |||||||
| Configural | — | 0.965 | — | 0.043 | — | 0.069 | |
| Weak/Metric | 00.84 (12) | 0.972 | 0.007 | 0.037 | −0.006 | 0.068 | −0.001 |
| Strong/Scalar | 19.27 (12) | 0.967 | −0.005 | 0.039 | 0.002 | 0.069 | 0.001 |
- Note: Δχ2-values are with respect to the previous level of measurement invariance. With respect to partial invariance, the Δ-values are changes with respect to the metric model.
Regarding the reliability and the factorial validity (i.e., convergent and discriminant) of the final model, Table 4 shows that CR indices indicated good reliability for all the factors (all above 0.70) apart from Spatial distance. Apart from the Temporal distance factor (AVE = 0.582), all indices indicated insufficient convergent validity (AVE < 0.50). However, indices of discriminant validity still point to some validity concerns for the Social and Spatial distance factors, as estimates of AVE are lower than MSV, which indicates nonindependence between these two factors in the model (see Table 4).
| Type of factors | |||||||
|---|---|---|---|---|---|---|---|
| Estimates | CR | AVE | MSV | 1 | 2 | 3 | 4 |
|
0.712 | 0.388 | 0.178 | — | |||
|
0.761 | 0.445 | 0.659 | 0.405 | — | ||
|
0.664 | 0.336 | 0.659 | 0.422 | 0.812 | — | |
|
0.843 | 0.582 | 0.327 | 0.329 | 0.572 | 0.528 | — |
- Abbreviations: ASV = average shared squared variance, AVE = average variance extracted, CR = composite reliability, HYP = hypothetical distance, SOC = social distance, MSV = maximum shared squared variance, SPAT = spatial distance, TEMP = temporal distance.
2.2 Predicting Science Skepticism
Descriptive statistics of the assessed variables and associations among variables are summarized in Table 5. PSYDISC (and all subscales) correlated as expected with the science attitude scales. To assess the additional predictive power of PSYDISC beyond the two science skepticism measures (Faith in Science and Negative Perceptions of Science), we performed three hierarchical multivariable regression analyses. In these analyses, we used climate change skepticism, vaccination skepticism, and genetically modified foods skepticism as the outcome variables.
| Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | M | SD |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Gender | — | — | — | ||||||||||||||
| 2. Age | −0.131* | — | 39.78 | 13.98 | |||||||||||||
| 3. Years of education | −0.122* | 0.044 | — | 16.55 | 2.85 | ||||||||||||
| 4. Religiosity | −0.148** | 0.213*** | 0.083 | — | 2.02 | 1.47 | |||||||||||
| 5. Left–Right | 0.168** | 0.221*** | −0.145* | 0.044 | — | 3.91 | 2.04 | ||||||||||
| 6. Progressive-conservative | −0.024 | 0.228*** | −0.112* | 0.245*** | 0.487*** | — | 3.43 | 1.68 | |||||||||
| 7. FIS | 0.149** | −0.041 | 0.065 | −0.207*** | −0.089 | −0.238*** | — | 21.62 | 6.36 | ||||||||
| 8. NPSS | 0.076 | −0.115* | −0.081 | 0.232*** | 0.027 | 0.197*** | −0.517*** | — | 45.59 | 9.34 | |||||||
| 9. HYP | −0.070 | 0.115* | −0.116* | 0.152** | 0.188*** | 0.269*** | −0.461*** | 0.495*** | — | 9.14 | 3.17 | ||||||
| 10. TEMP | −0.024 | 0.065 | −0.234*** | 0.021 | 0.106 | 0.150** | −0.148** | 0.325*** | 0.277*** | — | 11.79 | 4.97 | |||||
| 11. SOC | 0.046 | 0.249*** | −0.371*** | 0.086 | 0.228*** | 0.254*** | −0.204*** | 0.367*** | 0.275*** | 0.469*** | — | 11.84 | 5.44 | ||||
| 12. SPAT | 0.048 | 0.041 | −0.259*** | 0.069 | 0.166** | 0.243*** | −0.280*** | 0.389*** | 0.351*** | 0.433*** | 0.588*** | — | 11.21 | 4.63 | |||
| 13. PSYDISC | 0.006 | 0.168** | −0.351*** | 0.080 | 0.224*** | 0.297*** | −0.331*** | 0.490*** | 0.535*** | 0.759*** | 0.822*** | 0.800*** | — | 43.99 | 13.75 | ||
| 14. CC | −0.014 | 0.236*** | −0.186*** | 0.075 | 0.384*** | 0.322*** | −0.231*** | 0.295*** | 0.397*** | 0.369*** | 0.346*** | 0.379*** | 0.474*** | — | 9.38 | 4.52 | |
| 15. VAC | 0.018 | 0.073 | −0.163** | 0.127* | 0.209*** | 0.293*** | −0.329*** | 0.437*** | 0.437*** | 0.306*** | 0.376*** | 0.411*** | 0.484*** | 0.479*** | — | 8.95 | 4.15 |
| 16. GMO | −0.273*** | 0.160** | −0.032 | 0.195*** | 0.001 | 0.193*** | −0.338*** | 0.210*** | 0.194*** | 0.042 | 0.137* | 0.146** | 0.158** | 0.064 | 0.263*** | 21.17 | 5.58 |
- Note: Gender = Female =1, Male = 2; Religiosity = Higher scores, more religiosity; Left–Right = Higher score, more right-wing political ideology; Progressive-Conservative = Higher score, more political conservatism; PSYDISC = Total score.
- Abbreviations: CC = climate change skepticism, FIS = faith in science, GMO = genetically modified organisms skepticism, HYP = hypothetical distance, NPSS = negative perceptions of science, SOC = social distance, SPAT = spatial distance, TEMP = temporal distance, VAC = vaccination skepticism.
- * p < 0.05.
- ** p < 0.01.
- *** p < 0.001.
When climate change skepticism was used as a criterion, apart from the contribution of Faith in Science (β = −0.064, p = 0.27) and Negative Perceptions of Science (β = 0.19, p < 0.001) there was a significant increment in explained variance from step 1 to step 2 (Δ R2 = 0.16), due to the inclusion of PSYDISC (β = 0.40, p < 0.001).
When vaccination skepticism was used as a criterion, apart from the contribution of Faith in Science (β = −0.07, p = 0.10) and Negative Perceptions of Science (β = 0.25, p < 0.001) there was a significant increment in explained variance from step 1 to step 2 (Δ R2 = 0.06), due to the inclusion of PSYDISC (β = 0.21, p < 0.001).
When genetically modified foods skepticism was used as a criterion, apart from the contribution of Faith in Science (β = −0.28, p < 0.001) and Negative Perceptions of Science (β = 0.09, p = 0.14), there was no significant increment in explained variance from step 1 to step 2 (Δ R2 = 0.00), due to the inclusion of PSYDISC (β = 0.05, p = 0.42).
After confirming the predictive ability of the PSYDISC scale in relation to science skepticism across two domains, our next objective was to investigate whether the predictive validity of PSYDISC extends beyond self-reported science skepticism and adjusting for demographics and ideological beliefs. Specifically, apart from controlling for self-reported science skepticism, we also included demographic and ideological belief variables that were significantly associated with climate change skepticism and/or vaccination skepticism when examining the predictive validity of PSYDISC and its subscales.
When Climate Change Skepticism was used as a criterion, apart from higher Negative Perceptions of Science (p = 0.43) and stronger right-wing political ideology (p < 0.001) being associated with stronger skepticism, there was a significant increment in explained variance from step 1 to step 2a due to higher PSYDISC being associated with stronger skepticism (p < 0.001). Step 2b indicated that while hypothetical and temporal distance were most strongly associated with skepticism, social distance was not significantly associated with skepticism (see Table 6).
| Step 1 | Step 2a | Step 2b | |||||||
|---|---|---|---|---|---|---|---|---|---|
| β | t | p | β | t | p | β | t | p | |
| Age | 0.19 | 3.34 | < 0.001 | 0.13 | 2.46 | 0.015 | 0.17 | 2.90 | 0.004 |
| Years of education | −0.06 | −1.07 | 0.285 | 0.02 | 0.35 | 0.728 | −0.02 | −0.37 | 0.736 |
| Left–right | 0.27 | 4.62 | < 0.001 | 0.26 | 4.41 | < 0.001 | 0.25 | 4.19 | < 0.001 |
| Progressive-Conservative | 0.04 | 0.81 | 0.420 | 0.01 | 0.22 | 0.827 | 0.01 | 0.27 | 0.784 |
| FIS | −0.02 | −0.34 | 0.731 | −0.01 | −0.23 | 0.815 | 0.02 | 0.54 | 0.589 |
| NPSS | 0.23 | 4.52 | < 0.001 | 0.10 | 2.04 | 0.043 | 0.09 | 1.72 | 0.086 |
| PSYDISC | — | — | — | 0.28 | 4.86 | < 0.001 | — | — | |
| Hypothetical distance | — | — | — | — | — | — | 0.18 | 2.84 | 0.005 |
| Temporal distance | — | — | — | — | — | — | 0.14 | 2.78 | 0.006 |
| Social distance | — | — | — | — | — | — | −0.06 | −0.79 | 0.432 |
| Spatial distance | — | — | — | — | — | — | 0.16 | 2.17 | 0.031 |
| R 2 | 0.28 | 0.35 | 0.38 | ||||||
| ΔR2 | 0.07 (p < 0.001) | 0.03 (p < 0.01) | |||||||
- Note: Gender = Female = 1, Male = 2; Left–Right = Higher score, more right-wing political ideology; Progressive-Conservative = Higher score, more political conservatism; FIS = Faith in Science; NPSS = Negative Perceptions of Science; PSYDISC = Total score.
When vaccination skepticism was used as a criterion, apart from higher Negative Perceptions of Science (p < 0.001) and stronger right-wing political ideology (p = 0.43) being associated with stronger skepticism, there was a significant increment in explained variance from step 1 to step 2a due to higher PSYDISC being associated with stronger skepticism (p = 0.002). Step 2b indicated that hypothetical distance was the only factor significantly (p = 0.002) associated with skepticism (see Table 7).
| Step 1 | Step 2a | Step 2b | |||||||
|---|---|---|---|---|---|---|---|---|---|
| β | t | p | β | t | p | β | t | p | |
| Years of education | −0.06 | −1.60 | 0.111 | −0.02 | −0.53 | 0.595 | −0.04 | −0.99 | 0.339 |
| Religiosity | −0.03 | −0.78 | 0.437 | −0.02 | −0.56 | < 0.579 | −0.01 | −0.43 | 0.669 |
| Left–right | 0.09 | 2.17 | 0.031 | 0.08 | 2.06 | 0.040 | 0.08 | 1.99 | 0.048 |
| Progressive-Conservative | 0.09 | 2.11 | 0.035 | 0.06 | 1.39 | 0.166 | 0.06 | 1.42 | 0.155 |
| FIS | −0.05 | −1.17 | 0.241 | −0.05 | −1.12 | 0.269 | −0.02 | −0.43 | 0.667 |
| NPSS | 0.25 | 5.76 | < 0.001 | 0.18 | 3.65 | < 0.001 | 0.15 | 3.20 | 0.001 |
| PSYDISC | — | — | — | 0.17 | 3.11 | 0.002 | — | — | |
| Hypothetical distance | — | — | — | — | — | — | 0.17 | 3.08 | 0.002 |
| Temporal distance | — | — | — | — | — | — | 0.04 | 0.97 | 0.332 |
| Social distance | — | — | — | — | — | — | 0.04 | 0.77 | 0.442 |
| Spatial distance | — | — | — | — | — | — | 0.04 | 0.74 | 0.463 |
| R 2 | 0.23 | 0.27 | 0.31 | ||||||
| ΔR2 | 0.04 (p < 0.001) | 0.04 (p < 0.01) | |||||||
- Note: Gender = Female =1, Male = 2; Religiosity = Higher scores, more religiosity; Left–Right = Higher score, more right-wing political ideology; Progressive-Conservative = Higher score, more political conservatism; FIS = Faith in Science; NPSS = Negative Perceptions of Science.
When genetically modified food skepticism was used as a criterion, apart from the contributions of Age and Gender and Negative Perceptions of Science (all p < 0.001), there was no significant increment in explained variance from step 1 to step 2 (Δ R2 = 0.00), due to the inclusion of the four PSYDISC factors (see Table 8).
| Step 1 | Step 2a | Step 2b | |||||||
|---|---|---|---|---|---|---|---|---|---|
| β | t | p | β | t | p | β | t | p | |
| Gender | −0.19 | −3.53 | < 0.001 | −0.19 | −3.53 | < 0.001 | −0.20 | −3.60 | < 0.001 |
| Age | 0.09 | 1.72 | 0.087 | 0.09 | 1.54 | 0.126 | 0.101 | 1.64 | 0.102 |
| Religiosity | 0.05 | 0.90 | 0.371 | 0.05 | 0.94 | 0.349 | 0.052 | 0.91 | 0.363 |
| Progressive-conservative | 0.08 | 1.38 | 0.170 | 0.07 | 1.28 | 0.203 | 0.068 | 1.21 | 0.229 |
| FIS | −0.25 | −3.97 | < 0.001 | −0.24 | −3.94 | < 0.001 | −0.24 | −3.70 | < 0.001 |
| NPSS | 0.08 | 1.22 | 0.223 | 0.07 | 0.93 | 0.355 | 0.08 | 1.04 | 0.301 |
| PSYDISC | — | — | — | 0.02 | 0.31 | 0.756 | — | — | |
| Hypothetical distance | — | — | — | — | — | — | −0.02 | −0.25 | 0.801 |
| Temporal distance | — | — | — | — | — | — | −0.04 | −0.58 | 0.564 |
| Social distance | — | — | — | — | — | — | −0.01 | −0.18 | 0.854 |
| Spatial distance | — | — | — | — | — | — | 0.08 | 1.17 | 0.245 |
| R 2 | 0.19 | 0.19 | 0.19 | ||||||
| ΔR2 | 0.00 (p = 0.756) | 0.00 (p = 0.814) | |||||||
- Note: Gender = Female =1, Male = 2; Religiosity = Higher scores, more religiosity; Progressive-Conservative = Higher score, more political conservatism; FIS = Faith in Science; NPSS = Negative Perceptions of Science.
The inclusion of two science skepticism measures (Faith in Science and Negative Perceptions of Science) as additional predictors may provide a conservative test. Thus, we first examined the contribution of Faith in Science and Negative Perceptions of Science (comparing it with the baseline model with demographics and ideological beliefs, see Tables 6 and 7 for covariates) and then examined the PSYDISC as a stand-alone measure of science skepticism, by examining the contribution of PSYDISC (comparing it with the baseline model). When Climate Change Skepticism was used as the criterion, PSYDISC explained an additional 13.4% of the variance (compared to 7.3% for the model with two science skepticism measures). When Vaccination Skepticism was used as the criterion, PSYDISC explained an additional 13.2% of the variance (compared to 13.5% for the model with two science skepticism measures).
3 Discussion
The present study is the first to translate the PSYDISC-scale developed by Većkalov et al. (2024) and to assess its incremental predictive validity for science skepticism in a Swedish context. After the inclusion of correlated error terms (Item 3 and Item 4 within the spatial distance scale), the four-factor structure of the translated scale provided a good fit for the data, indicating that the scale structure generalizes to the Swedish sample. With respect to invariance for the genders and age groups, the results showed that the configural, metric, and scalar models passed the criteria for invariance. When looking at the local psychometric properties, the results indicated good reliability for all the factors. Moreover, apart from the Temporal distance factor, all indices indicated insufficient convergent validity (AVE < 0.50), meaning that the latent construct explains less than 50% of the indicator variance. Indices of discriminant validity were good except for some validity concerns for the Social and Spatial distance factors, which indicate nonindependence between these two factors in the model.
Importantly, the PSYDISC scale contributed a unique predictive value beyond measures of science attitudes for both climate change skepticism and vaccine skepticism. Thus, the scale was found to have incremental validity in predicting skepticism in these science domains, with effect sizes that were marginally stronger than those observed by Većkalov et al. (2024). However, in predicting GMO skepticism, PSYDISC did not contribute additional explained variance beyond the measures of attitudes to science, which is in contrast with results from Većkalov et al. (2024). Additionally, while both Faith in Science and negative attitudes to science (along with PSYDISC) were significant predictors across all skepticism outcomes in Većkalov et al. (2024), Faith in Science was the only significant predictor of GMO skepticism in our study and failed to reach significance for the other skepticism domains.
While the main objective of the study is to test the predictive validity of the overall PSYDISC scale, replicating Study 4 by Većkalov et al. (2024), the present study did address the predictive validity of the respective dimensions of the scale, which have been found to differ in earlier research (Većkalov et al. 2024). Our findings reveal that while negative perceptions of science are only weakly related to climate change skepticism, stronger right-wing political ideology and psychological distance, especially in hypothetical and temporal dimensions, are significant contributors to skepticism. Social distance, however, showed no significant influence. These trends were in line with Većkalov et al. (2024), even though the effects of temporal and spatial distance were stronger in our study. With respect to vaccination skepticism, our results indicate that it is strongly linked to negative perceptions of science, while right-wing political ideology is only weakly related. Psychological distance also played an important role, with hypothetical distance emerging as the only significant factor influencing skepticism, which in large part is in line with Većkalov et al. (2024).
Through demonstrating that PSYDISC predicts climate change and vaccine skepticism beyond science attitude measures, the present study does provide some support for the generalizability of the PSYDISC construct as an incremental predictor of science skepticism across domains. The study also indicates weaknesses of PSYDISC as a domain-general predictor of science skepticism, due to this measure being found not to contribute unique explained variance in GMO skepticism. However, it is important to note that when analyses were repeated without the inclusion of two science skepticism measures (Faith in Science and Negative Perceptions of Science), PSYDISC on its own explained equal or more (13% vs. 7% for climate skepticism, 13% vs. 13% for vaccine skepticism) of the variance as when the two science skepticism measures were included in the model. Therefore, one can argue for the use of PSYDISC as a stand-alone measure of science skepticism rather than relying on several measures in future research. Consequently, the result from the present study encourages further investigation and development of PSYDISC, highlighting both potential and ambiguity regarding this new construct.
Većkalov et al. (2024) hypothesized that PSYDISC would demonstrate incremental predictive validity across science domains, since it addresses how the individual positions science in relation to herself, in contrast to measures that address evaluations of the epistemic value of science (i.e., Faith in Science) or negative attitudes to science (i.e., NPSS). The results from the present study suggest that this reasoning needs to be developed, since PSYDISC lacked incremental predictive validity for GMO skepticism, while Faith in Science was a stronger predictor of skepticism within this domain. One notable fact is that, while climate change and vaccination skepticism were quite strongly correlated, they were only weakly correlated with GMO skepticism. In addition, GMO skepticism levels were much higher than for the other two forms of skepticism, indicating that GMO skepticism is, at least in Sweden, quite different from the other two forms of skepticism. While climate change and vaccination benefit from a strong scientific consensus, GM technology faces division within the scientific community due to ongoing debates about its long-term impacts, leading to larger uncertainty and skepticism among the public. Therefore, one can speculate that GMO skepticism should be primarily associated with low science literacy or trust in scientific processes rather than resulting from certain ideologies/worldviews (see Rutjens et al. 2022), and that is why Faith in Science is related to GM skepticism, as it is related to in that trust in scientific processes and authorities. Hence, the present study points to the need to develop an understanding of the distinguishing features of psychological distance to science, not least regarding the reasons why PSYDISC would predict science skepticism beyond other predictors.
The procedures of the present study have limitations that need addressing. Regarding the sampling method, we relied on convenience sampling, due to having limited resources while aiming to recruit a large sample. Our distribution method might have generated an unrepresentative sample, since it was largely based on our social media networks, with sample characteristics likely not corresponding to the characteristics of the entire Swedish population. Another aspect of the present biased sample was the fact that our respondents did not receive compensation for their participation, which could have led to an overrepresentation of participants with a previous interest in science as a topic.
Some demographic characteristics of our sample need mentioning with regard to the generalizability of our results, as well as the comparability of our results with related research. Our sample has a slightly asymmetrical gender distribution, which is not surprising considering women in general are more likely to self-select into online surveys (Smith 2008). We also had an overrepresentation of people living in the north of Sweden, with the population of this region only constituting 11.8% of Sweden's total population (Statistics Sweden, 2022). Moreover, highly educated people were overrepresented in our sample, since only 30% of the Swedish population between 25 and 64 years have studied for 15 years or more (Statistiska centralbyrån, 2022). A great majority of our participants also reported research experience from university studies.
The overrepresentation of highly educated people with some science experience in our sample might explain relatively low scores on both predictor variables (PSYDISC and NPSS) and outcome variables (climate change skepticism, vaccine skepticism, and GMO skepticism). Yet, the data from our sample contain variations within both the predictor variables and the outcome variables, acting as the basis for our analyses of the associations between these variables. Hence, the non-representativity of our sample does not necessarily weaken the internal validity of the present study, although it does limit the external validity.
Although the direct consequences of the current study are theoretical rather than practical, it contributes to laying the foundation for possible practical implications. Većkalov et al. (2024) developed the PSYDISC scale with hopes of providing not only a domain general but a malleable predictor of science skepticism, in contrast to previously studied predictors. Portraying science as closer to the individual, they argued, could prove to be a more feasible strategy to decrease science skepticism than addressing knowledge, ideologies, or attitudes.
In the first research with this objective, conducted in the UK, Zarzeczna et al. (2022) showed that psychological distance to science could be influenced through linguistic manipulation of science information (i.e., advertisements of gene editing and nanotechnology), and that framing information as psychologically close reduced skepticism toward the science topics. As the present study found that PSYDISC predicts science skepticism in at least two important science domains (climate change and vaccination), it provides reasons for conducting experimental research concerning the possible practical implications of PSYDISC also in Sweden.
With the impending risk of science skepticism causing harm to societies and our planet, there is a need for increased understanding of psychological predictors of this unfounded rejection of science. The present study adds support to the generalizability of the new PSYDISC-scale as an incremental predictor of science skepticism across at least two important science domains. At the same time, by demonstrating a lack of incremental predictive value in the GMO domain, the study indicates a need to further examine both the distinguishing features and the domain-generality of PSYDISC as a predictor of science skepticism. Hopefully, the present study contributes to developing PSYDISC as a tool for predicting science attitudes, laying the foundation for possible approaches to decrease unfounded science skepticism.
Author Contributions
The author takes full responsibility for this article.
Open Research
Data Availability Statement
The data that support the findings of this study are openly available in OSF at https://osf.io/ze4c7.
