Over the past twenty years, behavioural insights (BI) and nudges have gained prominence in public policy design. Combining insights from behavioural economics, cognitive science and psychology, BI aims to improve the efficacy of policy design by guiding citizens’ behaviours toward certain policy actions (e.g. programmatic uptake or compliance) while still preserving individual choice. A ‘nudge’, more concretely, refers to a tool in the BI toolbox that consists of ‘any aspect of the choice architecture that alters people’s behavior in a predictable way without forbidding any options or significantly changing their economic incentives’ (Thaler and Sunstein, Reference Thaler and Sunstein2008, 6). Nudges may steer individuals’ behaviour in beneficial directions across a wide spectrum of policy areas, from service delivery to health to the enhancement of environmental protections. But not all nudges resonate with all segments of the population, prompting those working in the field to consider how support for nudges (the individual policy design choices) and nudging itself (the wider practice of applying nudges) vary depending on who is being nudged and how they perceive the application (Jung and Mellers, Reference Jung and Mellers2016; Loibl et al., Reference Loibl, Sunstein, Rauber and Reisch2018; Sunstein et al., Reference Sunstein, Reisch and Rauber2018; Pe’er et al., Reference Pe’er, Feldman and Gamliel2019; Almqvist and Andersson, Reference Almqvist and Andersson2024).
This literature addresses an important tension in public policy. On one hand, policies that lack broad-based public support are less likely to be successful, particularly when uptake is voluntary or when participants must opt in to certain behaviours (Hagman et al., Reference Hagman, Andersson, Västfjäll and Tinghög2015). On the other, nudges operate somewhat silently in the background of policy delivery, leading some to critique them as obscuring government motives, or manipulating the public (Haussmann and Welch, Reference Haussmann and Welch2010; Mols et al., Reference Mols, Haslam, Jetten and Niklas2015; Kuyer and Gordijn, Reference Kuyer and Gordijn2023). Since nudge policies make people ‘better off, as judged by themselves’ (Thaler and Sunstein, Reference Thaler and Sunstein2008, 5) rather than by some objective metric, the subjective component of nudge support is central to government transparency and the delivery of citizen-centred policy in a representative democracy.
In this article, we take up this line of research, asking who supports which nudges? We also explore two novel, additional questions: Why do citizens support (or not support) nudges in their own words? And, to what extent do citizens support nudges (discrete policies) or ‘nudging’ more broadly (a mode of policy design and implementation)? Our research takes up these questions in Canada—a case typically thought of as a ‘principled pro-nudge’ country (Sunstein et al., Reference Sunstein, Reisch and Rauber2018). Canada is a useful case to study public support for nudges (and nudging), having previously received comparatively little attention (though see Felsen et al., Reference Felsen, Castelo and Reiner2013). Canada was an early adopter of behavioural insights—first through establishing a branch of the Behavioural Insights Team in 2014, with the federal government later starting its own ‘Nudge Unit’ in the form of the Impact and Innovation Unit within the Privy Council Office. Two provinces, British Columbia and Ontario, currently maintain their own units in the B.C. Public Service Agency and the Ontario Treasury Board Secretariat. As of 2023, these units have implemented (either in pilot form or full rollouts) over 50 nudge policies since 2015 (Hopkins and Lawlor, Reference Hopkins and Lawlor2023). As such, Canada is not only an interesting case that contains both individualistic and collectivist social norms, but it also has ample experience with nudges.
To evaluate the Canadian public’s support for BI policies, we present data from two national surveys fielded in 2023 (N = 2020) and 2024 (N = 1991) that (1) measure public opinion toward 30 nudge policies across three policy domains (health, environment and service delivery) and (2) gauge citizens’ views of nudging as a practice employed by government. Our approach builds on previous studies examining support for nudge policies in several ways. First, we establish a comparative benchmark for support for nudges by examining 15 BI interventions from Reisch and Sunstein (Reference Reisch and Sunstein2016) (used across many subsequent studies). We then increase ecological validity by including 15 real-world BI interventions actually implemented by nudge units in Canada. Second, building on work that explores the importance of socio-demographic differences in support for nudges (Pe’er et al., Reference Pe’er, Feldman and Gamliel2019; Almqvist and Andersson, Reference Almqvist and Andersson2024), we consider how gender, ideology and identification with a visible minority group affect nudge support. Finally, we take a different lens to gauging support for nudging by offering new qualitative data analysis on citizens’ views toward nudging as a mode of policy design. We conduct thematic analysis of open-ended survey responses that allow individuals to express their views on nudging in their own words. While some of our findings are in line with the extant research on who supports nudges, we also provide new evidence about why people do or do not like nudges and the practice of nudging more broadly.
Variation in public support for BI policies across policy domains
Macro-level support
Considerable attention has been paid to the overall levels of public support for individual nudge policies. Studies that report variation in support for nudges tend to focus on the design side such as features of the policy that alter the choice architecture for citizens, while others focus on who responds positively to nudges. Factors like the perceived intent of the nudge, its potential for intrusiveness in the life of the citizen, the value system aligned with the nudge and the level of controversy that the nudge may provoke are instrumental in understanding public support. Hagman et al. (Reference Hagman, Andersson, Västfjäll and Tinghög2015), for example, emphasize the distinction between pro-social nudges, which target society-wide benefits, and pro-self nudges, which target individual welfare, finding that the former tend to be viewed as more intrusive than the latter. Sunstein and colleagues (Sunstein et al., Reference Sunstein, Reisch and Rauber2018, Reference Sunstein, Reisch and Kaiser2019) highlight that approval is nuanced by the type of nudge and its alignment with citizens’ values, the perceived legitimacy of its goals, as well as overarching levels of trust in government. Jung and Mellers (Reference Jung and Mellers2016) distinguish support for nudges along more cognitive lines, arguing that nudges that provoke ‘System 1’ thinking (Kahneman, Reference Kahneman2013) like defaults are less supported than ‘System 2’ nudges that provide opportunities for deliberation. Nudges that evoke certain types of controversy also cause variation in support, suggesting that their application should be carefully considered to avoid public backlash (Tannenbaum et al., Reference Tannenbaum, Fox and Rogers2017; Loibl et al., Reference Loibl, Sunstein, Rauber and Reisch2018).
The nudge literature also emphasizes cross-national and within-country variation in support. Different countries support the use of BI policies to different degrees, with variation existing across market economies and democratic or authoritarian structure (with some important exceptions). Sunstein et al. (Reference Sunstein, Reisch and Kaiser2019) identify three country-types by level of nudge acceptance. ‘Principled pro-nudge’ nations include industrialised liberal democracies that are largely in favour of nudges when they are perceived to support the interests and values of most citizens (Australia, Canada, France, Germany, Italy, the UK, the US). ‘Nudge enthusiasts’ (China and South Korea) are those countries that contain a strong current of support for all nudge interventions. Finally, ‘cautiously pro-nudge’ countries are those countries that express greater variation in support for nudges and where nudges may provoke controversy (Denmark, Hungary, and Japan). Alongside cross-national comparisons, studies by Almqvist and Andersson (Reference Almqvist and Andersson2024) and Kasdan and Lee (Reference Kasdan and Lee2020) contribute further results from single country studies that enrich our comparative understanding of how nudge support varies within a country’s population according to identity or attitudinal factors.
Individual level support
Embedded within the nudge literature is the recognition of personal characteristics or demographic characteristics that moderate support for nudges (Pe’er et al., Reference Pe’er, Feldman and Gamliel2019; Almqvist and Andersson, Reference Almqvist and Andersson2024). Socio-demographics, attitudinal predispositions and political anchors play a crucial role in shaping attitudes towards nudges. There is considerable evidence that women favour the use of nudges more than men (Reisch and Sunstein, Reference Reisch and Sunstein2016; Loibl et al., Reference Loibl, Sunstein, Rauber and Reisch2018; Kasdan and Lee, Reference Kasdan and Lee2020). Ideology also plays a significant role in nudge support with conservatives tending to oppose nudges more than liberals, particularly when nudges are viewed as a threat to autonomy (Jung and Mellers, Reference Jung and Mellers2016). Pe’er and colleagues (Reference Pe’er, Feldman and Gamliel2019) observe that support for nudging may be contingent on minority group status. Levels of institutional trust also enhance support for nudges (Sunstein et al., Reference Sunstein, Reisch and Kaiser2019), which is consistent with US-centred evidence that finds that perceptions of manipulation or threats to freedom by government tend to decrease support for nudges, though this is further contingent on nudge type ( Tannenbaum et al., Reference Tannenbaum, Fox and Rogers2017; Yan and Yates, Reference Yan and Yates2019, 27).
Cultural and cognitive forces also affect levels of support for nudges (Kahan and Braman, Reference Kahan and Braman2006). Those with individualistic (as opposed to communitarian) worldviews tend to resist nudges more than their empathetic counterparts (Reisch and Sunstein, Reference Reisch and Sunstein2016; Loibl et al., Reference Loibl, Sunstein, Rauber and Reisch2018; Sunstein et al., Reference Sunstein, Reisch and Kaiser2019; Almqvist and Andersson, Reference Almqvist and Andersson2024). John et al. (Reference John, Martin and Mikołajczak2023) also observe that the perceived fairness of the nudge is positively related to support, particularly where citizens are more concerned with procedural and process-oriented factors than policy efficacy. Additionally, individuals inclined towards analytical or System 2 thinking are less likely to perceive nudges as intrusive, drawing on the theory that policy responsiveness may be associated with cognitive and emotional predispositions (Amir and Lobel, Reference Amir and Lobel2008; Hagman et al., Reference Hagman, Andersson, Västfjäll and Tinghög2015).
While evidence is mounting about country-level differences in support for nudges, there remain gaps in empirical assessments for many countries that have active nudge units or policy branches that incorporate behavioural insights into policy development. We also know very little about why people like or do not like nudges. Although we have some evidence that value systems, opt-in vs opt-out structure or general perceptions of the trustworthiness of government may help identify when people view nudges as helpful versus manipulative, we also know that perceptions of nudges may be tied up in political opinions about the governments implementing them (Tannenbaum et al., Reference Tannenbaum, Fox and Rogers2017). Underscoring this is a more general question about the degree to which citizens think of nudges at all. Several of these issues are empirically taken up below.
Data, methods and hypotheses
In this paper, we focus on Canada, which presents a useful empirical case to study nudges. As stated above, it is a hybrid liberal market economy with historic economic and cultural ties to the US, but its commitment to single-payer, public health care and greater social supports marks it as more aligned with some social democratic states. To track Canadians’ perspectives toward nudges and nudging, we collected survey data from Léger Marketing’s nationally representative standing online panelFootnote 1 (N = 2020 in August 2023 and N = 1999 in June of 2024). Participants were required to live in Canada, be 18 years or older and could respond to the survey in English or French. As noted in our pre-registrations on OSF, these surveys were exploratory and draw from the same pool of participants as two other studies. Participants who did not correctly answer a pre-treatment attention check were excluded from the analysis.
Survey 1
As set out above, we are interested in gauging support for nudges and nudging. By ‘nudges’, we mean individual policies, which may be more or less popular depending on their intent and design. By ‘nudging’, we mean the practice of government using nudges or the citizen’s view of the acceptability of nudging as a policy tool more broadly. Our analysis has three components. First, we describe support for 30 nudge policies (support for nudges). In order to ground our survey in the literature, and also make it ecologically valid in the Canadian case, we ask respondents to consider 15 BI policy measures proposed in prior studies (e.g. Reisch and Sunstein, Reference Reisch and Sunstein2016; Sunstein et al., Reference Sunstein, Reisch and Rauber2018; Almqvist and Andersson, Reference Almqvist and Andersson2024), alongside 15 additional measures that were implemented by the Canadian federal government through its Impact and Innovation Unit (‘Nudge Unit’), operating out of the Privy Council Office between 2015 and present day. These measures are divided into three policy spaces: health (14); environment (7); and general service delivery (9) and include informational approaches to nudging (e.g. email campaigns that remind respondents to participate in a program) or concrete changes in policy or regulations (e.g. requiring that grocery stores move healthy foods to eye level). Respondents are asked whether they approve or disapprove of the policies (1 = yes; 0 = no). While we did not formally hypothesize the direction of differences, we suspected that the responses to nudge policies across policy areas would themselves vary, given their unique position in Canadians’ public perception of the policy spaces. Nudges may fall into more than one space (e.g. a policy that attempts to increase organ donation rates during the process of renewing a driver’s licence is both health-oriented and involves service delivery). In these cases, we use the intent of the policy (e.g. increasing organ donation) as the designated policy area. This dually-focused analysis enables us to offer broad commentary around support for hypothetical and real nudges in the Canadian context that—to this point—has not been studied systematically.
Second, we draw inferences about citizens’ characteristics and their perceptions of nudges. Owing to previous work cited above, we suspected that differences by gender, ideology, and visible minority status may exist in Canada as they do across other jurisdictions (Jung and Mellers, Reference Jung and Mellers2016; Sunstein et al., Reference Sunstein, Reisch and Kaiser2019; Pe’er et al. 2019; Almqvist and Andersson, Reference Almqvist and Andersson2024). Following these studies, we hypothesize that:
H1: Women will be more likely to support nudges than men.
H2: Those on the political left will be more likely to support nudges than those on the political right.
H3: Individuals who identify as part of a visible minority group will be more likely to support nudges than those who do not.
Survey 2
In our second survey, we employ a novel design to find out what people think of nudging as a practice using open-ended responses. Here, we provide respondents with a brief description of nudging and inform them that governments are increasingly using BI to understand how people think and make decisions in everyday life. We also tell respondents that BI policies use ideas from psychology to improve government policies and programs. Next, we ask, ‘When you think about governments in Canada using Behavioural Insights, what are the main things that come to your mind?’ and ask respondents to enter written responses in a text box.Footnote 2 Following this, we presented respondents with a close-ended question measuring their overall support for BI: ‘In general, how much do you support Behavioural Insights in government?’ Respondents indicated their level of support on a 6-point Likert scale, ranging from ‘Strongly oppose’ to ‘Strongly support’. Since our analysis is largely exploratory and inductive, we did not formally specify hypotheses for our open-ended results; however, to the extent that we are able to analyse the associated close-ended question, we expected that differences by gender, ideology and visible minority status would follow H1-H3 above.
Results
Survey 1: benchmarking support for nudges in Canada and cross-nationally
To address the question of the degree to which Canadians support nudges, we present descriptive evidence about support for all 30 BI policies in Canada. Approval was generally high across policies (see Tables 1 and 2). Only four of the 30 nudges received less than majority support. Rates of support ranged from 22% for the least supported nudge policy (a default $50 donation to the Red Cross on their tax return with an opt-out) to 89% for the most supported (promoting the convenience of online driver’s license renewals to reduce wait times at government renewal offices). Contrary to expectations, there was not a lot of variation (on average) across policy domains. The average support for administrative policies was 67%, while the average support for health and environment policies was 66% and 64%, respectively. Interestingly, we find evidence that support is stronger for the 15 BI policies that were actually implemented (mean = 71%) compared to the 15 hypothetical policies featured in prior studies (mean = 60%), though it is not clear whether this is because respondents may have had some familiarity with the implemented policies or owing to policy preferences and norms in Canada.
Table 1. Support for hypothetical BI policies (adapted from Sunstein et al., Reference Sunstein, Reisch and Rauber2018)
Table 2. Support for implemented BI policies
To contextualize our results, we benchmark support for nudges in Canada against other prominent democracies. In Figure 1, we plot the percentage of Canadians who support each of the 15 hypothetical policies, comparing it to data from Sweden (Almqvist and Andersson, Reference Almqvist and Andersson2024), Italy, the United Kingdom, France, Hungary (Reisch and Sunstein, Reference Reisch and Sunstein2016), and a pooled estimate from Belgium, Denmark, Germany, South Korea, and the US (Sunstein et al., Reference Sunstein, Reisch and Kaiser2019).Footnote 3 Support for BI is similar in Canada as in other countries, with minimal variation. Cross-nationally, support for these policies ranges from 57% (in Hungary) to 71% (in Italy) with an overall average of 64%. Overall, we find strong correspondence in the ordinal ranking of support for these policies. The most popular policies in Canada are also popular elsewhere, including public education campaigns to fight childhood obesity (proportions for Canada, 87% and cross-national, 87%) and programs to combat distracted driving (Canada, 82%, cross-national, 83%). We find a similar pattern for less popular policies, such as using tax returns to ‘default’ people into Red Cross donations (Canada, 22%, cross-national, 31%) and subliminal advertising in movie theatres to discourage unhealthy behaviours (Canada, 38%, cross-national, 45%). We compare rankings in Canada to the pooled sample from Sunstein et al. (Reference Sunstein, Reisch and Kaiser2019), finding a Kendall’s Tau of 0.75 (t-statistic = 92, P-value < 0.05), with concordant pairs in 88% of cases. In short, Canadians support the same BI policies—in roughly similar proportions—as voters in other democracies.
Figure 1. Support for BI policies in Canada and other countries.
Second, we build on evidence from other studies, testing our hypotheses about systematic variation in support for nudges across socio-demographics. Our primary dependent variable is support, defined as
\begin{equation*}Suppor{t_i} = \frac{1}{{{N_i}}}{{\Sigma }}_{j = 1}^{{N_i}}{{\text{{\textrm X}}}_{ij}}\end{equation*} where 
$Suppor{t_i}$ measures the mean (proportion) of each individual 
$i$’s support for 
$N$ nudge policies, where each policy 
$j$ takes on a value of 0 = disapproval or 1 = approval. In this framework, 
${{\text{{\textrm X}}}_{ij}}$ measures whether each individual 
$i$ supports each nudge 
$j$, and 
${N_i}$ is the total number of nudges that each survey respondent considered. In our survey, we showed each respondent a random sample of 15 out of the total 30 nudge policies. Thus, a respondent who approves of 9 out of 15 policies would receive an overall support score of 0.60. We regress this variable on several covariates that prior research has shown to be a strong predictor of support for BI (Jung and Mellers, Reference Jung and Mellers2016). Specifically, we include gender (woman/man and other gender identities), ideology (conservative/moderate/liberal), income (high—a household income over $125,000 CAD/low), education (did/did not complete postsecondary), and visible minority status (visible/non-visible minority). We use linear regression to compare support across these variables. We then split our sample and replicate this analysis, focusing on the specific set of nudges in each policy area.
Table 3 presents the coefficients and standard errors from these four regression models. Overall, we find strong support for all three of our hypotheses. Women, liberals, and visible minorities are more supportive of nudges, relative to men, conservatives/moderates, and non-visible minority individuals. Women’s support for nudges exceeds their male counterparts (full sample = 0.04, SE = 0.01, t(1731) = 3.32, p < .05, 95% CI [0.02,0.06]), supporting H1. This pattern holds across all three policy areas. Overall, we find few exceptions to this gender dynamicFootnote 4 (and none statistically significant). Cases where women’s support for BI policies were more than marginally supportive than men and statistically significant include: (1) Reminders around tax payments (8 percentage points more favourable, p < .001); (2) Health loyalty points (9 percentage points more favourable, p < .001); (3) Garbage sorting stickers (9 percentage points more favourable, p < .001); (4) Voluntary autoenrollment in green energy suppliers (11 percentage points more favourable, p < .001) and; (5) Requirements around the display of healthy foods at supermarkets (10 percentage points more favourable, p < .001); among others.
Table 3. Correlates of BI support by policy area
* p < 0.05, with robust SEs in parentheses.
While there is little to no evidence that resource level (operationalized as income or education levels) is related in a systematic way to support for nudges (full sample = − 0.01, SE = 0.01, t(1731) = − 0.85, p = .40, 95% CI [−0.04, 0.01]), there is ample evidence to suggest that ideology shapes support. Conservatives were less likely to support nudges than moderates (full sample = − 0.08, SE = 0.02, t(1731) = − 5.17, p < .001, 95% CI [−0.11, − 0.05]), while liberals were more likely to support them (full sample = 0.06, SE = 0.01, t(1731) = 5.31, p < .001, 95% CI [0.04, 0.09]). Similar to Almqvist and Andersson’s (Reference Almqvist and Andersson2024) findings, Table 3 illustrates a clear relationship between political ideology and support across different policy areas, supporting H2. Analyses in our appendix shows that this relationship is strongest in policy areas that have a progressive dimension to them (e.g. vaccination uptake, meat-free meals, personal choice measures, and almost all environmental measures). Nudges that feature both an environmental aspect and programming around energy supply have a particularly ideological distinction with respect to approval rates. Ideology remains important even in service delivery nudges, with liberals nearly 10 percentage points more likely to support such nudges than conservatives.
Finally, we find that visible minority Canadians express greater approval of nudges than non-visible minority Canadians (0.06, SE = 0.01, t(1731) = 4.15, p < .001, 95% CI [0.03, 0.08]), supporting H3. In our case, visible minority Canadians may be more supportive of these nudges because they align with prevailing attitudes about the role of government. Across all groups, these dynamics are strongest for nudges in ‘substantive’ domains such as environment and health and weakest when it comes to administration and service delivery.
Survey 2: digging deeper into support for behavioural insights in Canada
The next component of our analysis presents evidence from open-ended responses in a second survey of Canadians (N = 1999) to further examine whether Canadians support nudging as a policy tool and why. As noted above, the survey provides respondents with a brief informational text that describes nudging, acknowledging, ‘While many believe Behavioural Insights can improve public policy, others have concerns about ethics and effectiveness’. Following this, we ask respondents to consider how they think about governments in Canada using BI. We qualitatively analyze the open-ended data, thematically organizing comments using an inductive approach that sorts comments according to subject and further coding them by tone (positive, negative, neutral) (see the SIF for further detail). Second, we analyze the close-ended responses of support for BI. To bolster our findings from study 1, we compare support across the same demographic subgroups.
Qualitatively, we draw several key insights. We begin by observing that many people simply do not know what to make of BI. Even when respondents expressed an opinion, many were hesitant or qualified their response by emphasizing a lack of knowledge. Some respondents were pleased to learn government was using scientific or evidence-based approaches and felt that BI could improve decision-making and produce policy benefits for citizens in salient areas such as health and the environment. Others expressed skepticism that BI could realistically benefit Canadians. For example, many respondents mentioned Canada’s diversity and how difficult it would be for any single methodology to understand, let alone anticipate, human behaviour. Many mentioned the potential for intrusiveness in the lives of citizens, parallelling longstanding scholarly concerns over BI’s focus on individual-level behaviour (e.g., Chater & Loewenstein Reference Loewenstein and Chater2017, Reference Chater and Loewenstein2023) and failure to account for heterogeneity across populations and contexts (e.g., Bryan, Tipton & Yeager Reference Bryan, Tipton and Yeager2021).
Concerns over behavioural complexity seem to be a shared belief among BI supporters and opponents. For supporters, the fact that behaviour is hard to predict is a challenge that requires solving. These individuals often support government’s efforts to do so. For opponents, it is precisely because ‘everyone behaves differently’ that government should not be in the business of behaviour change. Those who took a more critical view tended to point to the risk of BI raising too many ethical concerns. Consistent with prior research and our findings from study 1, we find evidence of ideological disagreement here. Conservatives were more negative about BI than liberals. They were also more likely to invoke specific political leaders or objects, such as elected officials (e.g., Prime Minister Justin Trudeau) or parties (e.g., the governing Liberal Party). Such specific political mentions were almost entirely missing among those who identified as liberal. This insight is important. Given widespread uncertainty about BI, we might expect voters to rely on their political preferences as a cue. However, such cue-taking appears asymmetric: only BI skeptics associated the practice of nudging with specific political actors, which may suggest resistance to service delivery by an out-partisan government. In short, both liberals and conservatives seem to have replaced a more difficult question (‘What do you think about the government’s use of BI?’) with an easier question (‘What do you think about the government?’).
Next, we quantitatively analyze text responses to dig deeper into these insights. Two undergraduate Research Assistants manually coded each survey response by tone (positive, negative, uncertain).Footnote 5 We then used inductive thematic analysis to develop coding categories (Braun & Clarke Reference Braun and Clarke2006) and further broke these down into subthemes. Within ‘positive’ responses, we identified three sub-themes: (1a) benefits; (1b) innovation; and (1c) other. Within ‘negative’ responses, we identified (2a) ethics; (2b) trust; and, (2c) other. Our survey prompt did not mention any issue areas, however we recorded whenever a respondent mentioned one (e.g. healthcare). In Table 4, we provide examples of responses falling under each tone/subtheme (see the SIF for more information on the coding process).
Table 4. Open-ended text responses with example quotes by theme
Note: Percentages may not sum to 100% owing to rounding between categories or to responses being simultaneously coded in multiple categories
In general, most respondents expressed some opinion on BI. Just 13% selected ‘Don’t know’ in the closed response option and also expressed uncertainty in their text response. In terms of sentiment, support for BI appears weaker in the open-ended data than in the close-ended. After coding open-ended responses, we find just 20% of comments are positive while 38% are negative and 30% reflect uncertainty (another 9% were irrelevant or not applicable). This difference in sentiment could reflect acquiescence bias in the close-ended Likert question, negativity bias in the open-ended text question, or both. Building on our qualitative analysis, we find some evidence that different ideological groups may think about BI in different substantive terms. Liberals were much more likely than conservatives to express positive sentiment about BI (b = 0.14, se = 0.03, t(1724) = 5.22, p < 0.05) and to emphasize the potential for innovative policymaking (b = 0.7, se = 0.02, t(1724) = 4.60, p < 0.05). In contrast, conservatives were much more likely than moderates to express distrust about governments using BI (b = 0.05, se = 0.02, t(1724) = 3.22, p < 0.05). We also find some difference by race: White respondents were much more likely to express negative sentiment toward the use of BI than non-white respondents (b = 0.07, se = 0.02, t(1724) = 2.91, p < 0.05).
Finally, we analyze the close-ended responses of support for BI, comparing support across the same demographic subgroups from study 1. Our survey measure asks respondents to indicate their support for BI on a 6-point scale. In the full sample of respondents, about 50% indicate some type of support, 30% indicate opposition, and 20% say they ‘Don’t know’.Footnote 6 Among the 80% of respondents who provided a substantive response, a strong majority (62%) support BI. In numerical terms, the sample mean is 3.6 (SD = 1.44, 95% CI [3.53, 3.67]), suggesting moderate support of the policy tool. We also regress support for BI on gender, ideology, race, education, and income (see Table 5). Consistent with our earlier analysis, we find that women are more supportive of BI, relative to men (b = 0.16, se = 0.08, t(1416) = 2.09, p = 0.04), and that liberals are more supportive, relative to moderates (b = 0.39, se = 0.09, t(1416) = 4.34, p < 0.05) or conservatives (linear hypothesis test b = 0.52, se = 0.11, t(1416) = 4.85, p < 0.05). We also see a difference by education, with college graduates being more supportive of BI than non-graduates (b = 0.22, se = 0.08, t(1416) = 2.73, p < 0.05). While statistically significant, these differences represent small effect sizes in substantive terms.
Table 5. Support for behavioural insights
* p < 0.05, with robust SEs in parentheses.
In summary, our analysis of open- and close-ended data from survey 2 suggests that left-right ideology predicts not only support for BI but also the conceptual dimensions that come to mind when Canadians think about BI. Although we find some differences by gender and race, such as women being more supportive overall and whites being more likely to express negative sentiment, these differences are relatively small. Finally, in terms of policy topics, just 20% of respondents mentioned specific issue areas with little evidence of demographic differences. Linear regression shows that women were more likely than men to mention healthcare (b = 0.05, se = 0.01, t(1725) = 3.59, p < 0.05). However, liberals and conservatives were about as likely to mention healthcare (7% vs 8%), civil rights (2% vs 4%), or government operations (2% vs 2%).
Discussion and conclusion
Taken together, our findings show that there is broad support for nudge policies in Canada, with demographic and ideological differences that are consistent with comparative studies (Reisch and Sunstein, Reference Reisch and Sunstein2016; Pe’er et al., Reference Pe’er, Feldman and Gamliel2019; Kasdan and Lee, Reference Kasdan and Lee2020; Almqvist and Andersson, Reference Almqvist and Andersson2024). We benchmark public opinion in Canada to several other democracies and find no evidence that Canadians hold extreme opinions on this matter—if anything, Canadians are surprisingly in line with global attitudes about nudges. Broadly speaking, Canadians support individual nudge policies. Approval ranges from 60% for all 30 nudges and rises to 71% when we consider only those policies that have actually been implemented. We observe that women and liberals support the use of nudges more than their male and conservative counterparts. Similar to other studies, we find that identification as part of a visible minority group enhances support for nudges.
Our secondary, exploratory analysis combines quantitative and qualitative evaluations and provides new insight on what Canadians think of nudging as a tool for policy design and implementation. By analyzing open-ended responses, we explore what individuals think of nudging (if they think of it at all). We show that citizens are broadly more hesitant about nudging as a policy tool than they are about individual nudges, but that this effect also has an ideological component to it. We observe that nudging as a practice is associated with concerns about manipulation, and that these are felt more acutely by conservatives than liberals. In some sense this echoes academic discussions of nudge skepticism, particularly around ethics and perceptions of manipulation (Tannenbaum et al., Reference Tannenbaum, Fox and Rogers2017; Hagman, Reference Hagman2018; Lades and Delaney, Reference Lades and Delaney2022). What to make of strong support for nudges, coupled with weak support for nudging? This question requires greater probing. We believe that nudges are popular when presented as isolated policy ideas, but nudging—the act of government using nudge tools in unspecified ways—is less so. This could be a matter of well-documented ‘emphasis’ framing (Chong and Druckman, Reference Chong and Druckman2007a; Reference Chong and Druckman2007b). When surveys present individual nudge policies (e.g., text messages to reducing distracted driving), it may elicit positive considerations (e.g., savings lives) and not negative ones (e.g., invading privacy). However, when surveys present nudging as a more general policy instrument, respondents may draw upon a wider set of considerations, including more negative elements. Alternatively, this distinction may reflect citizens having varied preferences within issues, where opinions on a policy writ-large (e.g. immigration) may be negative, but opinions toward one expression of that policy (e.g. an immigrant family living next door) are more positive because the individual may more easily assess it on the merits (Jacoby, Reference Jacoby2000).Footnote 7
Our article extends the thinking on public opinion toward behavioural insights in three critical ways. First, we consider public support for nudges using real-world examples of nudges that have been applied in Canada. In this way, we not only expand our theoretical understanding of public support for nudges, but we provide concrete support for policy practitioners who aim to replicate or expand upon these styles of intervention. Second, we observe support for a variety of nudges across three areas in which nudges have become more prevalent, allowing us to explore systemic differences in support for nudges or whether support is contingent on the nudge itself (which it appears to be). Third, we dig deeper into why individuals like or dislike BI. Our findings highlight the complicated nature of public opinion toward BI, which includes optimism alongside uncertainty and skepticism. Trust in government emerges as an important factor, one that overlaps with the ideological divisions seen in prior research. Combined, these analyses advance our general understanding of public opinion toward BI policies—real and hypothetical—and go further to substantiate the connection between individual characteristics and their reasons for supporting BI.
We also observe limitations to our study. Measuring support for actual nudges that have been rolled out in the Canadian context may be affected by the that an individual has participated in one of these nudge programs. While differentiated experience with nudges may drive relative support to some degree, we observe that most public opinion research on policy does not control for direct contact with the policy area. Even where respondents may have been a recipient of a nudge through the redesign of a letter from government or through the administration of a service such as automatic licence plate renewal, we believe it is unlikely that the intervention would have been so intrusive as to radically alter opinion. Still, this prompts a relevant question that could be tested in future research.
Another limitation may be the scope of our study itself. Over the past 15 years, we have learned much about public support for BI. For example, we know that support varies by country (e.g. higher in Canada; lower in Hungary) and is sensitive to how and why a nudge is designed (e.g., voters are less approving of intrusive nudges on controversial issues). We also know that certain demographic factors (e.g., gender) are strong predictors of approval/disapproval. Yet despite this progress, the field lacks a ‘general model’ of nudge support. Future scholars might tackle this problem by considering the relationship political ideology and social identity (Pickup et al., Reference Pickup, Kimbrough and de Rooij2022; Groenendyk et al., Reference Groenendyk, Kimbrough and Pickup2023). Identification with certain ideological traditions (e.g., ‘liberal’, ‘conservative’) may serve as the basis for social group membership, which in turn, carries expectations about how members should behave. Consequently, some might approve/disapprove of BI not because of well-reasoned policy beliefs but because such an attitude follows a group norm.
Overall, our findings have several implications for the theory of behavioural public policy and its real-world application. Taking a mixed method approach to studying nudges provides a textual richness absent in many survey-based studies that gives us a much clearer understanding of why people may support or resist nudges. As public support is strongly linked to policy compliance (Cairney, Reference Cairney2009; John et al., Reference John, Martin and Mikołajczak2023), it is not enough to understand who thinks nudges are beneficial; policymakers have to know what aspects provoke discontent or lower the likelihood that the public will see nudges as legitimate policy tools. Our data suggests that policymakers do not have carte blanche to implement nudges indiscriminately, or to speak about their use in overly generic terms. Rather, policymakers need to balance the use of nudging as a policy tool with an explanation of why some of the perceived subversive effects of nudges may be questioned, even where individual nudges are well received. In sum, as is often noted, nudges are not a panacea for improved policy design; however, there is growing evidence that the public supports their careful and considered use. Policymakers should proceed with cautious optimism and clear communication.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/bpp.2025.10013.
Acknowledgements
The authors wish to thank the Journal’s anonymous reviewers and acknowledge the support of the Canadian Social Sciences and Humanities Research Council, grant 435-2021-1034.
Open access
