1.1 Political polarization of Americans’ support for climate policies

Unfortunately, in the United States, people’s attitudes about climate policies have been polarized along political lines [12–14]. Historically, political liberals in the United States have been more concerned about climate change and expressed greater demand for policy solutions, compared to conservatives [15]. Even in the last five years, liberals have been much more likely to agree that there is anthropogenic climate change [16], to self-report engaging in pro-environmental behaviors [17], and to say the national government should prioritize “dealing with climate change and global warming” [18]. Researchers have concluded that such polarization constitutes a barrier towards the adoption of climate policies in the United States, at both national and state levels [19–21].

One explanation for Americans’ polarization on climate policies is provided by cultural cognition theory, which argues that opposing beliefs on the values that underpin an ideal society lead to political differences in climate change concerns and policy preferences [22]. Under this account, political liberals are motivated by communitarian worldviews and political conservatives by individualistic worldviews [22, 23]. As a result, liberals are more likely to support government interventions on climate because they perceive climate change as a significant threat to collective welfare; conversely, conservatives are more likely to question the merits of climate policies that might infringe on personal autonomy [23]. Americans also tend to trust media sources that are “culturally congruent”, or in alignment, with their political ideology, leading to enhanced polarization [24]. Cultural cognition theory has also been proposed as an explanation for Americans’ politically polarized attitudes about other threats to public wellbeing, such as nuclear weapons [25].

Political polarization may also persist because of motivated reasoning, or the tendency to selectively interpret evidence that confirms preexisting beliefs [26]. For example, one nationally representative survey of Americans in 2009 found that, for liberals, greater knowledge about climate change was associated with heightened support for climate mitigation policies, whereas the opposite association was found in conservatives [27]. The evidence base for scientific literacy leading to motivated reasoning about climate policy is mixed, however [26], including two nationally representative U.S. studies reporting null effects for motivated reasoning as a driver of climate policy polarization [28, 29].

1.2 Generational differences in climate policy attitudes and decision-making over time

Irrespective of the roots of politically polarized views on climate change, the strong association between Americans’ political ideology and their support for climate policies might not hold among young Americans today [30]. Nationally representative surveys from 2020 and 2021 found that younger Republicans (Millennials and Generation Z) were more supportive of climate policies and are more concerned about climate change compared to older Republicans (Baby Boomers and older). In contrast, Democrats of all ages were highly supportive of climate policies [31, 32].

However, it is unclear whether younger adults in the United States have always been less politically polarized in their support for climate policies. Existing longitudinal studies on the evolution of Americans’ climate policy opinions over time do not directly address this question. For example, existing analyses are limited by a reliance on regional survey data rather than national samples [33], or by collapsing across years of public opinion data without analyzing time-series trends [13]. Additionally, such studies do not compare how patterns of support for climate policies compare to public support for other policy domains [13, 33]. Both studies do show that younger Republicans have been more supportive of climate policies than older Republicans since 2016.

Why are younger Republicans today more supportive of climate policies than older Republicans? One potential explanation is due to cohort differences between younger versus older conservatives. Younger conservatives are part of a generation that is more aware of and more concerned about climate change [34, 35]. Additionally, today’s news media prominently discusses climate change policies and affords high social capital to many climate activists, especially on social media [36, 37]. Younger conservatives may therefore be more influenced by social norms to address climate change [38]. Moreover, younger conservatives may be more concerned about climate change simply because they are more likely to be alive when more serious consequences of climate change manifest [39].

Additionally, older adults might be more likely than younger adults to rely on their political ideology when responding to climate policy survey questions. Two findings from the literature on aging and decision-making offer mechanisms for this hypothesis. First, socioemotional selectivity theory posits that older people are highly motivated to maintain positive emotional wellbeing, compared to younger adults [40]. Accordingly, if negative emotions like fear come to mind when being asked to think about climate change policies, older adults might be more motivated to quickly resolve upsetting thoughts (known as the “positivity effect”) [41, 42]. A second finding is that fluid intelligence, or the ability to reason flexibly and solve novel problems, declines as people age [43]. On complex tasks, such as forming opinions on the nuanced topic of climate science, this could result in older people preferentially relying on their political party’s views rather than engaging in more deliberate decision-making [44, 45]. Taken together, younger conservatives might be more likely to deeply reflect on climate policy questions and offer answers that conflict with their political ideology.

1.3 The current study

The primary goal of this study is to better understand the precise interplay between political ideology and age in predicting support for climate policies. Such knowledge is valuable, as it would aid social scientists in designing psychological interventions and communication strategies that overcome partisan divides on climate policy [46]. Additionally, this manuscript helps contextualize previous research on climate policy polarization in three ways. First, we analyze attitudes towards a diverse set of climate policies. In doing so, we answer recent calls to study public opinion of a wide range of environmental solutions [47]. Second, in analyzing four decades of time series data, we identify when in history current patterns of support for climate policies might have emerged [48]. Third, we compare attitudes about climate change policy to attitudes about other polarized policy domains. Thus, this analysis can speak to the issue of whether differential political polarization based on age is unique to climate change, or whether it instead applies to Americans’ policy attitudes more broadly.

In a secondary analysis of nationally representative surveys conducted from 1982–2020 by American National Election Studies (ANES; 16 waves, ranging from n = 822 to n = 6,531, combined sample N = 29,467), this paper sought to answer the following research questions:

Research Question 1: Are younger Americans today less politically polarized than older Americans when it comes to support for climate policies?

Research Question 2: If younger Americans are less politically polarized than older Americans in support for climate policies, when in time did this pattern first emerge?

Research Question 3: Are younger Americans less politically polarized than older Americans when it comes to policy preferences in non-climate policy domains?

2.1 American National Election Studies: Overview

Since 1948, the American National Election Studies (ANES) has conducted public opinion surveys. Each ANES survey recruits a fresh cross-sectional panel of Americans, selected for national representativeness using an area probability sampling frame. All ANES data are freely available for download: https://electionstudies.org/data-center/.

2.2 American National Election Studies: Participants

ANES interviewed Americans aged 18 and older, with a lower age limit (15 or 16) in earlier years. All participants provided informed consent. Respondents could select an English- or Spanish-language interview. In 2004 and in all prior waves, ANES surveys were administered face-to-face or over telephone. Since 2008, ANES has also allowed for self-administered online surveys. In 2020, ANES used online video interviews to safeguard against COVID-19. ANES compensated participants for their time. Before 2004, ANES conducted this survey every two years, but since 2004, only every four years. ANES conducts interviews before national elections (usually two months before election day in early November) and attempts a post-election follow-up.

In this paper we analyze all ANES surveys from 1982–2020 (16 waves, combined N = 29,467). We start with 1982 because this is the first year that ANES began consistently polling Americans on their environmental policy attitudes. Response rates and survey lengths varied year-to-year. A yearly breakdown of ANES sample sizes across political ideology and age groups is available in Supporting Information, see S1 Table.

2.3 American National Election Studies: Survey questions

The primary measures used in the study are listed in Table 1. Table 1 shows the survey questions as phrased in ANES 2020 and lists the transformed response scales we used for modeling (e.g. after dichotomizing and reverse-scoring questions as necessary). Question wording sometimes varied year-to-year. Full details on year-to year question design, control variables, and other dependent measures are available in S1 Appendix.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Primary American National Election Studies survey questions.

https://doi.org/10.1371/journal.pone.0302434.t001

In addition to the repeatedly asked climate policy questions presented in Table 1, ANES also asked survey questions about specific policy preferences in just one or two survey waves over 1982–2020. Before 2008, ANES survey questions tended to focus on topics like pollution and cleaning up nature, as well as one question on global warming in 1996. From 2008, ANES questions focused on climate change-specific policies, such as emissions standards and rising temperatures, reflecting greater public awareness of climate change as a specific threat to the environment. We analyzed all available climate and environmental policy questions to answer Research Question 2. Full details on these questions’ designs are available in S1 Appendix and S2 Table.

2.4 Dependent variable coding

For dependent variables with three response options (e.g. increase, decrease or keep the same), we dichotomized responses to allow for easier comparisons across time and question types. Response options to the federal environmental spending question varied year-to-year, so we also dichotomized this question to allow for direct comparisons across survey waves. In dichotomizing, responses were coded such that 1 represented favoring greater environmental intervention, whereas 0 represented opposing environmental regulation or preference for the status quo.

For variables with response scales (mostly 1–7), we kept the original scale, but reverse-scored many questions to allow for easier comparisons to the binary-transformed variables. We coded those such that the high scale point represented favoring greater environmental intervention and the low scale point represented opposing those intervention or preference for the status quo. The defense spending and private vs public health insurance questions were also reverse-scored to align liberal vs conservative scale direction with the welfare spending question (for which the transformed high scale point indicates the more typical position of political liberals in the United States).

To answer Research Question 2, we created indices of climate policy questions asked in each ANES survey wave, resulting in eight “climate policy support” indices for 1990, 1992, 1996, 2000, 2008, 2012, 2016 and 2020. Climate policy support indices were calculated by averaging z-standardized responses to the constituent questions.

2.5 Analytical strategy

Participants’ responses to survey questions were modeled using either logistic regression (with logit link function, for binary-coded questions) or multiple linear regression (for Likert-scale questions and the policy indices), using political ideology, age, and their interaction as predictor variables. We used the same set of predictors for all models and analyzed standardized regression coefficients across question types. Models in which the political ideology × age interaction p-value was less than α = 0.05 were deemed to have statistically significant differential age-based political polarization.

To answer Research Question 1, we examined whether political ideology interacted with age for three climate policy questions asked in ANES 2020: 1) whether federal government spending on protecting the environment should be increased or decreased; 2) whether to prioritize environmental regulations versus business interests; and 3) whether to increase regulations on businesses that emit greenhouse gases.

To answer Research Question 2, we tested for a significant political ideology × age interaction in two analyses of historical ANES environment and climate policy questions. First, in eight ANES survey waves that asked multiple climate policy questions, we averaged normalized responses to all climate policy items available (see S2 Table) to form a within-year “climate policy support index”, then regressed this index onto the same model used in the preceding analysis. Second, ANES repeatedly asked two questions over 1982–2020; 1) whether federal government spending on protecting the environment should be increased, and 2) whether to prioritize environmental regulations versus business interests. We analyzed the time series for these items, again testing for a political ideology × age interaction at the 0.05-signifance level. ANES used three different wordings of the environmental regulations versus business interest question across survey waves, but we decided to treat different versions of this question as a unified time-series because they assess the same underlying policy preference (year-to-year question wording is discussed in S1 Appendix).

To answer Research Question 3, we analyzed time-series questions that asked about non-climate policies: 1) defense spending, 2) public versus private health insurance, and 3) welfare spending. We selected these three survey questions because ANES has mostly complete time-series over 1982–2020 and because public opinion data reveals significant political polarization on all three topics [49]. Copying the analytical strategy used in Research Question 2, we tested for a significant political ideology × age interaction in predicting responses to a z-standardized average index of all non-climate questions within each survey wave, as well as testing for the same interaction in all questions analyzed individually.

All models controlled for participants’ highest level of education completed (dichotomized in our analyses as College degree or No college degree), the interaction between education and political ideology, gender and household income in the year prior to the survey date, based on association between these demographic variables and climate policy preferences in other studies [50]. We included the interaction between education and political orientation because it significantly improved model fit in 2020 data (which was analyzed first), and because other authors have reported effects of this interaction on support for climate policies [51]. The three-way interaction between political ideology, age and education did not significantly improve model fit, and was thus excluded from our models. Fig 1 provides a visualization of our modeling approach.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Visualization of our conceptual model.

Note: Political ideology, age and their interaction are the main independent variables of interest in predicting Americans support for climate policies and other policy topics.

https://doi.org/10.1371/journal.pone.0302434.g001

We did not use survey weights in our analyses, for three reasons. First, survey weights might be inaccurate due to high proportions of missing data in every survey wave (see S1 Table) [52]. Second, ANES did not calculate survey weights before 1992, so unweighted data analysis offers better consistency over time. Third, there are concerns that using survey weights for regression analysis may lead to less reliable results when analyzing relationships between individual-level variables [53].

Data were analyzed in R version 4.2.2, primarily using functions from the “stats” package to model survey responses [54] and functions from the “effectsize” package to calculate standardized regression coefficients [55].

3.1 Research finding 1: Younger Americans were less politically polarized than older Americans in 2020

In survey data from the 2020 ANES (n = 6,531), younger Americans were significantly less politically polarized than older Americans in support for three different climate policies. As can be seen in Fig 2, liberals were highly supportive of these climate policies regardless of age, whereas conservatives’ support tended to decline with age, with political moderates’ support falling between these groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Younger Americans were less politically polarized than older Americans on support for three climate policies in 2020.

Note: Response variables are z-standardized and displayed with 95%-confidence interval smoothing. Asterisks indicate significant political ideology × age interaction at p < 0.05. Full regression outputs are available in S3–S5 Tables. Participants expressed their support for three climate policies, as well as their political ideology (1 = extremely liberal to 7 = extremely conservative) and age. Models treated political ideology as a continuous variable, but for visualization purposes, people who responded 1 or 2 are coded as “liberal”, 3 or 4 or 5 are coded as “moderate”, and people who responded 6 or 7 coded as “conservative”. For the federal spending on the environment question, participants were asked, “What about protecting the environment–should federal spending on protecting the environment be increased, decreased, or kept the same?”, with response coding 1 = Increase, 0 = all other responses. For the environmental regulations versus business interests question, participants were asked, “Where would you place yourself on this scale, or haven’t you thought much about this? 1 = Tougher regulations on business needed to protect environment to 7 = Regulations to protect environment already too much a burden on business”, though responses were reverse-coded. For the greenhouse gas regulations question, participants were asked, “Do you favor, oppose, or neither favor nor oppose increased government regulation on businesses that produce a great deal of greenhouse emissions linked to climate change?”, with responses ranging from 1 = Oppose regulations a great deal to 7 = Favor regulations a great deal.

https://doi.org/10.1371/journal.pone.0302434.g002

For all three climate policy questions, differential political polarization by age is reflected in the statistically significant standardized regression coefficient and consistent direction for the interaction between political ideology and age (all ps < 0.001). For the federal spending on the environment question, the standardized coefficient was β = –0.13; for the environmental regulations versus business interests question, β = –0.05; for the question about increasing regulations on greenhouse gas emitters, β = –0.05. For all three climate change questions, younger Americans’ responses were less politically polarized than older Americans. Comprehensive modeling results are available in S3–S5 Tables.

3.2 Research finding 2: Younger Americans became less politically polarized than older Americans on climate policies during 2008–2012

Fig 3 shows that, since 2012 but usually not before, younger Americans have been less politically polarized than older Americans in terms of their support for climate policies (full question list in S2 Table). We did not detect a significant political ideology × age interaction in 1992, 1996, 2000, and 2008 (ps > 0.10), whereas younger Americans were significantly less politically polarized than older Americans in climate policy support in 2012, 2016, and 2020 (ps < 0.001). Surprisingly, 1990 also exhibited a significant political ideology × age interaction (p = 0.04).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Younger Americans have been less politically polarized than older Americans on climate policy support since 2012, but not before (Except 1990).

Note: Response variables are displayed with 95%-confidence interval smoothing. Years with an asterisk (e.g. 2012*) indicate significant political ideology × age interaction at p < 0.05, whereas non-asterisked survey years do not exhibit significant age-based political polarization. In all surveys, participants expressed their support for climate policies, as well as their political ideology (1 = extremely liberal to 7 = extremely conservative) and age. Models treated political ideology as a continuous variable, but for visualization purposes, people who responded 1 or 2 are coded as “liberal”, and people who responded 6 or 7 coded as “conservative”. A version of this chart with political moderates included is available in S1 Fig, and regression models for each year are available in S6–S13 Tables.

https://doi.org/10.1371/journal.pone.0302434.g003

For the federal environmental spending question, in ten survey waves before 2012 (combined N = 11,567), we did not find a significant interaction between political ideology and age (all ps > 0.21). The only exception is in 2000 (n = 1,434), when the political ideology-age interaction was significant (p < 0.02). In 2012, 2016 and 2020 (combined N = 14,685), younger Americans were significantly less politically polarized than older Americans (all ps < 0.01). Annual regression outputs are available in S14 Table.

For the environmental regulations versus business question, in five survey waves before 2012 (combined N = 5,845), younger Americans were not differentially polarized compared to older Americans (all ps > 0.11). Since 2012 (combined N = 14,685), younger Americans have been significantly less politically polarized than older Americans in support for environmental regulations at the expense of business interests (all ps < 0.001). Annual regression outputs are available in S15 Table.

Analyzing other survey questions individually yielded convergent results. ANES asked additional specific climate policy questions 20 times over 1982–2020, such as support for a clean air and water tax (ANES 1990 and 1992) and offshore drilling (ANES 2012; full list in S2 Table). 18 of 20 ANES questions analyzed reinforced the general finding that younger Americans have been less politically polarized than older Americans from 2012 on, but not before. The two exceptions were for: 1) an ANES 1990 question about whether to strictly enforce pollution standards (significant political ideology × age interaction, p < 0.05); and 2) an ANES 2012 question about nuclear energy (no significant political ideology × age interaction, p = 0.08). Otherwise, 14 climate policy questions asked before 2012 did not exhibit a significant political ideology × age interaction (all ps > 0.10), while for four climate policy questions asked after 2012, younger Americans were significantly less polarized than older Americans (all ps < 0.05). Annual regression outputs are available in S5 and S16–S34 Tables.

3.3 Research finding 3: Younger versus older Americans are not differentially politically polarized in three non-climate-related domains

Fig 4 shows that in modeling three ANES time-series on public support for policies not relating to climate change (e.g. defense spending, public vs private health insurance, and welfare spending), we found that younger people have not been consistently more politically polarized than older people since 2012. In nine survey waves that asked all three questions, only 2012 and 2020 featured statistically significant political ideology × age interactions (p < 0.01 in 2012 and 2020; for other years, ps > 0.05). Additionally, the direction of the interaction differs between 2012 and 2020; in 2012, older people are less politically polarized than younger people, but in 2020, older people are more polarized than younger people.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Americans are not consistently differentially politically polarized based on age in support for defense spending, public health insurance, and welfare spending.

Note: Response variables are displayed with 95%-confidence interval smoothing. Years with an asterisk (e.g. 2012*) indicate significant political ideology × age interaction at p < 0.05, whereas non-asterisked survey years do not exhibit significant age-based political polarization. In all surveys, participants expressed their support for three non-climate policies, as well as their political ideology (1 = extremely liberal to 7 = extremely conservative) and age. Models treated political ideology as a continuous variable, but for visualization purposes, people who responded 1 or 2 are coded as “liberal”, and people who responded 6 or 7 coded as “conservative”. A version of this chart with 1992 non-climate data (excluded here for space considerations) and political moderates included is available in S2 Fig. For the federal spending on defense question, participants were asked, “Some people believe that we should spend much less money for defense. Suppose these people are at one end of a scale, at point 1. Others feel that defense spending should be greatly increased. Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5 or 6. Where would you place yourself on this scale, or haven’t you thought much about this?” Responses were reverse-scored such that values ranged from options ranged from 1 = greatly increase defense spending to 7 = greatly decrease defense spending. For the public vs private health insurance question, participants were asked, “There is much concern about the rapid rise in medical and hospital costs. Some people feel there should be a government insurance plan which would cover all medical and hospital expenses for everyone. Suppose these people are at one end of a scale, at point 1. Others feel that all medical expenses should be paid by individuals through private insurance plans like Blue Cross or other company paid plans. Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5, or 6. Where would you place yourself on this scale, or haven’t you thought much about this?” Responses were reversed-scored such that they ranged from 1 = private insurance plan to 7 = government insurance plan. For the welfare spending question, participants were asked, “Should federal spending on welfare programs be increased, decreased, or kept the same?”, with responses coded such that 1 indicated a preference to increase welfare spending, while 0 reflected keep welfare spending the same or decrease welfare spending.

https://doi.org/10.1371/journal.pone.0302434.g004

Analyzing time-series for the individual policy questions also yielded patterns of age-based political polarization that differed from the climate policy items. For defense spending, the political ideology × age interaction was significant only in 1994, 2008, and 2020 (ps < 0.05). For private versus public health insurance, the political ideology × age interaction was significant only in 2008 and 2012 (ps < 0.05). For welfare spending, the political ideology × age interaction was significant only in 1982 and 2012 (ps < 0.05). These time-series were distinct from the climate policy time series, which showed that younger Americans have been consistently less politically polarized since 2012 but not before. Within-wave estimates of the political ideology × age interaction effect size for the non-climate survey items are available in S35–S37 Tables.