This is an open access article distributed under the terms of the Societal threat perception—the feeling and perception that something aversive is going to happen (Brandt & Bakker, 2022)—has been at the core of research in psychology for decades (see Crawford, 2017, for a review). Previous literature has placed limited emphasis on the relation of different measures of threat perception and its adjacent constructs to one another (for exceptions, see Onraet et al., 2013; Van Hiel, 2022). As a consequence, research on societal threat perception employs a variety of different operationalizations of the construct (Jost, 2017; Onraet et al., 2013). Some studies ask about the perceived threat posed by an issue (e.g., Kahn et al., 2022), while others assess the extent to which people are worried about an issue (e.g., Brandt et al., 2021). There is also the possibility that the adjacent construct of issue salience (Leeper & Robison, 2020) captures—at least partly—people’s worries and threat perceptions (e.g., Boninger et al., 1995; Jennings & Wlezien, 2011; Krosnick, 1990). To date, no research has systematically related threat perception measures to one another or to issue salience, leaving an important gap in the literature. We address this gap, responding to calls for more rigorous assessments of psychological constructs (Flake & Fried, 2020), and answer the research question: To what extent do threat, worry, and issue salience measures produce similar empirical patterns?
When measuring threat perception, threat and worry are both commonly employed measures for threat perception. Some researchers ask participants to what extent they feel “threatened” by a societal issue (e.g., Conway et al., 2023; Hornsey & Pearson, 2024; Kahn et al., 2022; Poushter & Huang, 2020; Smith & Kim, 2026). For example, Conway et al. (2023) measure threat perception with the item: “Thinking about [a societal issue] makes me feel threatened.” Others measure threat perception by asking how “worried” participants are about an issue (e.g., Arikan, 2023; Brandt et al., 2021; Conway, 2026; Onraet et al., 2013; Woods, 2011). For example, Brandt et al. (2021) use the item: “To what degree are you worried about the following situations: [a societal issue]?”1. Yet, Van Hiel (2022) argues that threat perception consists of two distinct psychological responses: worry, a more cognitive and adaptive process, and anxiety, a more maladaptive reaction reflecting deeper, existential responses to potential harm (see also Woods, 2011). As the reviewed literature suggests, the interchangeable use of threat and worry, as well as diverging perspectives on their distinction, highlight the need for a more systematic assessment.
It is plausible that issue-specific threat and worry measures also capture the perceived importance people attach to societal issues. In public opinion research, such importance is typically studied as issue salience. Issue salience has been defined as “the degree to which a person is passionately concerned about” an issue (Krosnick, 1990, p. 60), or as “an individual’s subjective sense of the concern, caring, and significance” of an issue (Boninger et al., 1995, p. 62). These definitions suggest a motivational and evaluative orientation toward an issue, which may be emotionally charged or even threat-relevant. The survey measures of issue salience further support this interpretation. Items such as “What is the most important problem facing the country?” tend to frame issues in negatively valenced or problem-oriented terms (Jennings & Wlezien, 2011). In such cases, the salience of an issue may reflect not just its cognitive priority, but also the degree of worry or perceived threat. The conceptual and measurement-based overlaps discussed here raise the possibility that individuals may simply find the issues important that they are threatened by or worried about, and vice versa. In summary, it is unclear whether threat and worry can both be used as proxies for one another, and to what extent these measures may also reflect issue salience. For threat, worry, and issue salience to be appropriate proxies for one another, their associations with political ideology should be highly similar (e.g., Brandt et al., 2021; Kahn et al., 2022; Leeper & Robison, 2020; Onraet et al., 2013). If threat, worry, and issue salience have different associations with political ideology, treating them as equivalent proxies could result in a Type S (sign) or Type M (magnitude) error (Gelman & Carlin, 2014). A Type S error would occur when the sign of the ideology association reverses (e.g., from positive to negative), depending on whether threat, worry, or issue salience are measured. A Type M error would arise when the effect size differs across the three measures. The presence of Type S or Type M errors would affect the conclusions people draw about the threat-politics link, and add further inconsistency to the already competing theoretical perspectives in the threat-politics literature (e.g., Brandt & Bakker, 2022; Godefroidt et al., 2019; Jost, 2017). Despite their relevance, the associations of threat, worry, and issue salience with political ideology have not yet been directly compared.
To summarize, we explore to what extent threat, worry, and issue salience measures produce similar empirical patterns. We do this using a sample of the Dutch population focusing on descriptive statistics, their associations with each other, and their associations with political ideology.
Method
Openness, Transparency and Ethics
We report the results of a secondary data analysis using an existing data set. We did not preregister our expectations or analyses. The data and code to reproduce our results can be found on our study’s OSF page (see Bomm et al., 2025). This study has been approved by the Ethics Review Board of the Faculty of Social and Behavioral Sciences, University of Amsterdam, The Netherlands, under the number 2022-YME-15725. Before starting the survey, participants signed an informed consent form.
Research Design
We rely on a large and diverse sample of Dutch respondents collected by I&O Research collected between November 30, 2022 and December 11, 2022. The online survey measured threat, worry, and issue salience regarding twelve societal issues (see full survey: https://osf.io/d958h; Azrout et al., 2022). Within the survey, participants first completed the issue salience battery, then the worry battery, and finally the threat battery. The order of the blocks was fixed, while the order of the items within each block was randomized.
Sample
Our analyses included data from N = 1863 participants (51.9% male, 47.9% female, .16% other, MSample Age = 52.41, SDSample Age = 16.30; sample education levels: 20.2% low education, 39.5% medium education, 40.3% high education). The sample is, compared to the general Dutch population, comprised of slightly more men than women, older (MPopulation Age 2024 = 42.4, Centraal Bureau voor de Statistiek, 2021), and somewhat more highly educated (Population education levels: 26.0% low education, 37.0% medium education, 36.4% high education, Centraal Bureau voor de Statistiek, 2024).
Measures
Threat, worry, and issue salience were measured for the same set of twelve societal issues: climate change, Covid-19, crime, energy costs, the housing shortage, inflation, the influx of asylum seekers, the inhumane reception of asylum seekers, the labour shortage, the nitrogen problem, polarization, and the war in Ukraine. First, issue salience was measured with the item "How important do you consider the following problems the Netherlands is currently facing?"2, rated on a 7-point Likert scale ranging from 1 (Not at all important) to 7 (Very strongly important) for each issue. Worry was measured with the item "How worried are you about the (potential) consequences for you personally/your personal situation for each following problems?"3, rated on a 7-point Likert scale ranging from 1 (Not at all worried) to 7 (Very much worried) for each issue. Threat was measured with the item "To what extent do you feel threatened by the following issues?"4, rated on a 7-point Likert scale ranging from 1 (Not at all threatened) to 7 (Very strongly threatened) for each issue. Items were “forced response” and thus have no missing values.
Political ideology was measured with the item “In politics, people sometimes talk about “left” and “right”. Where would you place yourself? Display your position using a scale of 0 to 10, where 0 means “left” and 10 means “right”. Which number best describes your position?”, on a semantic differential scale from 0 (Left) to 10 (Right) (Huber, 1989). As higher values indicate more right-wing ideology, we will subsequently refer to this measure as “right-wing ideology”. Participants who answered “I don’t know” or “I don’t want to say” (8%, N = 150) were treated as missing values in analyses including right-wing ideology.
Results
Comparing the Levels of Threat, Worry, and Issue Salience
Analyses were performed using the base-R package (R Core Team, 2023). We start by examining the average levels of threat, worry, and issue salience. Figure 1 depicts the distributions of threat (green), worry (orange), and issue salience (blue) per societal issues. Dashed lines indicate mean responses for each measure per issue. To test the difference between the levels of threat, worry, and issue salience, we conducted two-sided paired-sample t-tests with Bonferroni-Holm-corrected p-values to account for repeated testing for all pairwise comparisons. Table 1 in Appendix A provides descriptive statistics for threat, worry, and issue salience for all twelve issues. Table 2 in Appendix B provides the detailed results of all t-test comparisons.
Starting with threat and worry, the self-reported levels of threat are consistently lower than the self-reported levels of worry. These differences are statistically significant (all padj < .001). In our effect size interpretation, we follow Funder and Ozer (2019) and interpret effects of d ≈ .1 as very small, d ≈ .2 as small, d ≈ .41 as medium, d ≈ .63 as large, and d > .87 as very large. The differences between threat and worry range from small to medium effects: dCovid-19 = .23 to dHousing shortage = .48. Turning to threat and issue salience, all twelve issues are statistically significantly less threatening than salient (all padj < .001). The differences between threat and issue salience are large to very large, ranging from dCovid-19 = .68 to dHousing shortage = 1.52. All threat-issue salience differences are larger than any of the threat-worry differences. Finally, for worry and issue salience, all twelve issues are statistically significantly less worrying than salient (all padj < .001). The differences between worry and issue salience are medium to very large, ranging from dClimate change = .39 to dHousing shortage = 1.08. To conclude, Dutch citizens report issues to be slightly less threatening than worrying, and substantively less threatening or worrying than salient. This is a first indication that threat and worry, but not issue salience, produce similar empirical patterns.
Figure 1
Distributions of Threat, Worry, and Issue Salience Responses per Issue
Note. Distributions of threat (green), worry (orange), and issue salience responses (blue) for each issue. Dashed lines represent mean responses. Across all issues, Bonferroni-Holm-corrected paired t-tests (two-sided) showed significant differences between threat, worry, and issue salience, across all 36 comparisons. Table 2 in Appendix B provides detailed t-test results.
Correlations Between Threat, Worry, and Issue Salience
To further assess the empirical patterns between self-reported threat, worry, and issue salience, we investigated the correlations between the constructs. Figure 2 plots the correlations between threat and worry (left), threat and issue salience (middle), and worry and issue salience (right) for the examined issues. To test whether there is a statistically significant difference in the correlation coefficients across issues, we first converted all correlation coefficients to z-scores using Fisher’s transformation and then conducted z-tests with Bonferroni-Holm-corrected p-values to account for repeated testing (see Table 3 in Appendix C for complete test results).
Figure 2
Correlation Coefficients Between Threat and Worry (Left), Threat and Issue Salience (Middle), and Worry and Issue Salience (Right)
Note. Correlation coefficients are ordered by the issues’ average correlation strength. The correlation coefficients for threat and worry, threat and issue salience, and worry and issue salience all differed statistically significantly from each other, with the exception of the housing shortage. Table 3 in Appendix C provides detailed correlation difference test results.
A visual inspection of the correlation coefficients plotted in Figure 2 shows the threat-worry associations fall within a small range from rReception of asylum seekers = .59 to rClimate change = .78, with an average correlation of rThreat-Worry = .70. The correlation coefficients for the threat-issue salience pairs are overall lower than the threat-worry correlations (average rThreat-Issue salience = .47, with a larger range from rReception of asylum seekers = .31 to rClimate change = .64). Similarly, the correlations between worry and issue salience are lower than the threat-worry correlations (average rWorry-Issue salience = .58), and have a larger range (from rHousing shortage = .39 to rClimate change = .75).
Next, we test whether, within each issue, the correlations between threat and worry, threat and issue salience, and worry and issue salience differ statistically significantly from each other. Within each issue, the correlation between threat and worry is statistically significantly stronger than the correlations between a) threat and issue salience (all padj < .001), and b) worry and issue salience (all padj < .05), see Table 4 in Appendix C. The correlations between threat and issue salience are statistically significantly weaker than the correlations between worry and issue salience (all padj <.05, except padj-Housing Shortage = .078).
To summarize, self-reported threat and worry about societal issues are strongly associated with each other, while issue salience is more weakly associated with both threat and worry. This is another indication that threat and worry, but not issue salience, produce similar empirical patterns.
Associations Between Right-Wing Ideology and Measures of Threat, Worry, and Issue Salience
Next, we assess whether we reach similar conclusions about the relation to right-wing ideology when relying interchangeably on threat, worry, or issue salience measures. Figure 3 shows the results of bivariate regression models (i.e., correlation coefficients) predicting right-wing ideology on a left-right spectrum with threat, worry, and issue salience, across twelve societal issues. Positive (resp. negative) coefficients indicate a positive (resp. negative) association between right-wing ideology and threat, worry, or issue salience for that particular issue.
Figure 3
Bivariate Correlations Between Right-Wing Ideology and Threat, Worry, and Issue Salience
Note. Correlation coefficients of threat (green), worry (orange), and issue salience (blue) with right-wing ideology on a left-right spectrum (0 = left; 10 = right). Statistically significant correlation coefficient differences from the threat baseline (square) are indicated by solid circles; non-significant differences by hollow circles. Table 5 in Appendix D provides detailed correlation difference test results.
We start by examining the correlation coefficients of threat and right-wing ideology for each of the twelve issues. In line with previous evidence in the threat-politics literature (e.g., Brandt et al., 2021), we see that the perceived threat posed by different issues is, in terms of magnitude and direction, differentially associated with right-wing ideology. The associations between threat and right-wing ideology vary considerably, from (|rWar in Ukraine| = .001) to (|rInflux of asylum seekers| = .41). Five threat ratings—crime, energy costs, inflation, the influx of asylum seekers, and the labour shortage—are positively and statistically significantly associated with right-wing ideology. At the same time, we find five negative and statistically significant associations between perceived threat—climate change, Covid-19, the nitrogen problem, polarization, and the reception of asylum seekers—and right-wing ideology.
To compare the threat-ideology correlations to the worry-ideology and issue salience-ideology correlations across the same societal issues, we conducted Fisher’s z-tests with Bonferroni-Holm-corrected p-values to account for repeated testing. Figure 3 marks statistically significant differences from the baseline threat-ideology correlation with solid (as opposed to hollow) circles. Table 5 in Appendix D provides the results of the correlation coefficient difference tests.
We do not find evidence for Type S (sign) errors when relying interchangeably on measures of threat, worry, or issue salience in separate bivariate models predicting right-wing ideology: When there is a positive association with right-wing ideology for threat posed by an issue, we see the same positive association with worry and issue salience. Out of 36 tests, only one indicates a Type S error: Perceiving the labour shortage as threatening is (statistically significantly) positively related to right-wing ideology, while perceiving it as important is negatively (but not statistically significantly) related to right-wing ideology.
We now turn to the extent to which the different measures lead to Type M (magnitude) errors in the bivariate models, i.e., errors about the magnitude of their associations with right-wing ideology conditional upon the measure used. Starting with threat and worry, as Figure 3 depicts, the correlations with right-wing ideology are highly similar for the threat and worry measures, with small differences ranging from ∆Reception of asylum seekers = −.12 to ∆Energy costs; Crime = .02. We see statistically significantly different correlations for only three of twelve issues: the influx of asylum seekers, the nitrogen problem, and the inhumane reception of asylum seekers.
Turning to issue salience, we find that issue salience is overall more strongly associated with right-wing ideology than threat, with descriptively larger statistically significant correlation differences in eight of the twelve examined issues (climate change, Covid-19, the housing shortage, the labour shortage, the nitrogen problem, polarization, the reception of asylum seekers, and the war in Ukraine). Only for the influx of asylum seekers is the threat measure more positively associated with right-wing ideology than the issue salience measure is. For the issues crime, energy costs, the housing shortage, and inflation, the associations of threat and issue salience with right-wing ideology are not statistically significantly different. Overall, issue salience is to a larger extent associated with right-wing ideology than threat is, with correlation differences ranging from ∆r Reception of asylum seekers = −.40 to ∆r Crime = .01. When comparing worry and issue salience, we see largely similar patterns: relying upon a measure of issue salience leads to larger associations with right-wing ideology compared to worry.
To summarize, using threat and worry measures interchangeably in studies on threat perception and ideology is unlikely to change conclusions about the direction of the relationship (Type S errors). However, issue salience differs: its association with ideology is substantially stronger, raising the risk of Type M errors, where the effect size is overestimated. This matters because psychology increasingly emphasizes not just whether an effect exists, but how strong it is (Lakens, 2022). For example, while all three measures show positive associations with ideology, the average correlation for issue salience (|r| = .23) is more than 1.5 times larger than for threat (|r| = .14) and worry (|r| = .16). Based on conventional benchmarks (Arceneaux et al., 2025; Funder & Ozer, 2019), this shifts the interpretation from a small to a modest effect. Such differences matter for theory and practice. Overestimating the role of threat in ideological conflict may lead scholars to exaggerate the extent of societal threat or ideological divides and misguide future studies or interventions.
Multivariate Tests of Incremental Validity
Next, we assess the incremental validity, that is, the extent to which the three measures threat, worry, and issue salience explain variance in right-wing ideology above and beyond each other (Hunsley & Meyer, 2003). To do this, we report the results of a series of multivariate regression models in which we regress right-wing ideology on 1.) threat, worry, and issue salience (Model 1; see also equation 1), and 2.) a threat-worry composite score and issue salience (Model 2, see also equation 2). Figure 4 shows the results of the multivariate regression models 1 and 2 for each of the twelve societal issues. Positive (resp. negative) regression estimates indicate a positive (resp. negative) association between right-wing ideology and threat, worry, a threat-worry composite score, or issue salience. The coefficients should be interpreted as partial associations with right-wing ideology—i.e., the association with the outcome variable controlling for the other predictors in the model. We provide the full results of the multivariate regression models in Table 6 in Appendix E.
1
2
In multivariate Model 1 (see Equation 1 and Figure 4), threat and worry show very similar partial associations across most issues; only for the influx of asylum seekers and the reception of asylum seekers are their coefficients statistically significantly different from one another (thus, for these issues, rejecting H0: βthreat = βworry). Moreover, threat and worry are, overall, less associated with right-wing ideology than issue salience. In six out of the twelve societal issues, salience has a statically significantly stronger association with right-wing ideology compared to threat, specifically for: climate change, Covid-19, the nitrogen problem, polarization, the reception of asylum seekers, and the war in Ukraine. Consistent with the bivariate correlations, the only issue on which salience is statistically significantly less strongly associated with ideology than threat is the influx of asylum seekers.
The results of Model 1 further show that, when predicting right-wing ideology and controlling for issue salience, threat and worry no longer add unique predictive value in five out of the twelve issues (climate change, energy costs, inflation, the nitrogen problem, and polarization), even though they were significant in the bivariate regressions. This suggests that when predicting ideology for these issues, salience accounts for much of the explanatory power of threat and worry, making the distinction between the measures less pronounced than when examining their descriptive scores.
To summarize, the results from Model 1 suggest that when threat, worry, and issue salience are included simultaneously, worry adds little incremental validity beyond threat (and vice versa). By contrast, issue salience shows a stronger partial association with ideology than threat or worry in half of the issues, indicating that salience is sometimes the more powerful predictor of right-wing ideology.
Second, we created latent variables combining threat and worry into a threat-worry composite for each issue to reduce potential measurement error by employing a two-item measure for each issue. We regressed ideology on issue salience and the threat-worry composite using Equation 2, see Figure 4. Issue salience is statistically significantly differently and more strongly associated with right-wing ideology than the threat-worry composite is for seven out of the twelve issues: climate change, Covid-19, the housing shortage, the nitrogen problem, polarization, the reception of asylum seekers, and the war in Ukraine. Issue salience is statistically significantly less associated with right-wing ideology than the threat-worry composite on two issues: the influx of asylum seekers and the labour shortage.
These results are consistent with those from Model 1 and the bivariate associations, and suggest that although salience and threat–worry are related, salience should not be used interchangeably with threat-worry as a proxy for perceived threat.5
Figure 4
Threat, Worry, and Salience as Predictors of Ideology in Two Standardized Multivariate Regression Models
Note. Rows 1–3 (Model 1, shaded white) and 4–5 (Model 2, shaded grey) of each plot facet depict the results of standardized multivariate regression models predicting right-wing ideology. Model 1 includes the predictors threat (green), worry (orange), and issue salience (blue). Model 2 includes a threat-worry composite (pink) and issue salience (blue) as predictors. Right-wing ideology was measured on a left-right spectrum (0 = left; 10 = right). Estimates that differ statistically significantly (p < .05) from threat (resp. the threat-worry composite in Model 2) are indicated by solid circles; non-significant differences by hollow circles. Table 6 in Appendix E provides the full output of the regression models.
Discussion
Our study demonstrates that threat and worry produce largely similar empirical patterns, warranting their use as proxies for each other in many cases—with relevant exceptions, depending on the issue under study. At the same time, citizens do not necessarily feel threatened by or worried about the issues they find most important, and vice versa. Our findings also reveal that the direction and strength of associations between political ideology on the one hand, and threat or worry about societal issues on the other, are highly similar, with the exception of asylum-related issues. Issue salience exhibits a different pattern, showing stronger associations with ideology across the board.
We see at least three opportunities to expand our research. First, by using multiple-item measures of threat, worry, and issue salience, future studies could further establish the construct validity of these concepts using factor-analytic approaches. Second, our survey used a fixed block order, raising the question of whether our results were affected by spillover effects from one measurement battery to the next. However, the fact that, overall, worry correlates more strongly with threat than with issue salience, despite being equidistant in time from both, supports our conclusion that worry is more closely related to threat than to issue salience. Ultimately, to rule out potential order effects, future studies could present issue salience, worry, and threat in a randomized order.
Third, cross-cultural research is needed to explore whether—and to what extent—our conclusions generalize to other contexts. We believe that our Dutch data can be meaningfully linked to existing research, as Dutch and English are closely related (Stefanowitsch et al., 2020). At the same time, the precise meanings attached to translations of “threat” and “worry”, as well as the relationship between these terms, may vary across languages. Cross-cultural conclusions about the relationship between threat, worry, and issue salience will therefore require additional cross-cultural validation efforts.
In sum, our findings support the practice of decades of “threat-politics research” that treated threat and worry as interchangeable indicators of a common construct (e.g., Brandt et al., 2021; Conway et al., 2023; Onraet et al., 2013), while also pointing out exceptions in which threat and worry produce different empirical patterns, and highlighting the limitations of using issue salience as a proxy. By doing so, we contribute to the growing body of work on the validation of psychological measures (Flake & Fried, 2020).