Introduction
Debates over social programs are often about perceptions of deservingness: Who is worthy of public resources and why? When assessing deservingness, voters often rely on heuristics—especially race, gender, and citizenship (Watkins-Hayes and Kovalsky, Reference Watkins-Hayes, Kovalsky, Brady and Burton2017). Recently, evidence from the United States points to another heuristic: administrative burden (Keiser and Miller, Reference Keiser and Miller2020). This refers to the learning, psychological, and compliance costs that citizens experience when they access a public service or program (Herd and Moynihan, Reference Herd and Moynihan2019). The logic is straightforward: People who overcome significant burdens to access social spending are seen as more deserving than those who access equivalent benefits more easily. This insight sets up a basic tension for the design of social programs, such as Canada's new Early Learning and Child Care Plan. On the one hand, governments looking to expand access to child care might simplify the application process for existing low-income supplements, such as Ontario's child-care fee subsidy or British Columbia's Affordable Child Care Benefit. On the other, governments looking to shore up support for affordable child care spending may be inclined to further restrict access through complex eligibility and reporting requirements.
In this research note, we present results from a pre-registered survey experiment (N=821) embedded in Canada's 2022 Democracy Checkup (Harell et al., Reference Harell, Stephenson, Rubenson and Loewen2022a). We present a hypothetical scenario to measure support for a low-income supplementary cash benefit paid directly to child care providers. Using a vignette experiment, we manipulate administrative burden (high vs low vs control) and measure the impact on public support for the supplement. We also assess voters’ overall willingness to support and pay for such a program.
There are two main findings. First, support for a low-income child care supplement is strong, but there is little enthusiasm to pay for it through new income taxes. Second, we find weak evidence that administrative burdens affect public support for an affordable child care supplement. People in the high-burden condition were slightly less supportive of the benefit, relative to the control group. People in the low-burden condition were also slightly less supportive, but the effect goes away after adjusting for pre-treatment covariates. We find no evidence of any treatment effects on willingness to fund the supplement through income tax. In sum, we find no evidence that burdensome child care benefits are more popular than easily accessible benefits. If anything, burdensome benefits reduce support. These insights nuance previous findings from the United States (for example, Keiser and Miller, Reference Keiser and Miller2020), which suggest that administrative burdens can increase support for social spending. Building off these insights, we make timely recommendations for governments in Canada, arguing for easily accessible child care benefit programs.
Deservingness, Administrative Burden, and Support for Social Programs
Decades of research suggest voters, including Canadians, rely on heuristics when thinking about who deserves public resources (Watkins-Hayes and Kovalsky, Reference Watkins-Hayes, Kovalsky, Brady and Burton2017). For example, Harell et al. (Reference Harell, Soroka and Ladner2014) embedded a survey experiment in the 2011 Canadian Election Study to show that voters’ support for social welfare programs is lower when recipients are Indigenous compared to white. Doberstein and Smith (Reference Doberstein and Smith2019) conducted a survey experiment on support for homelessness spending in Canada. They found systematic variations in perceived deservingness: Voters see homeless individuals who had previously experienced victimization as more deserving of public assistance than those who did not. Recently, Harell et al. (Reference Harell, Banting, Kymlicka and Wallace2022b) drew on social psychology to show the power of in/out group dynamics—voters see beneficiaries who are more “committed” to Canada as more deserving of redistribution, and vice versa. These findings dovetail with a large body of research: Voters are less supportive of social welfare spending when beneficiaries are depicted as either lazy or unable to meet stereotypical ideals (Watkins-Hayes and Kovalsky, Reference Watkins-Hayes, Kovalsky, Brady and Burton2017). In short, the attributes of beneficiaries may serve as a heuristic for assessing whether beneficiaries deserve support.
Recent evidence from the United States suggests voters may also use the design of social programs as a heuristic of deservingness—specifically, administrative burden (for example, Burden et al., Reference Burden, Canon, Mayer and Moynihan2012; Herd and Moynihan, Reference Herd and Moynihan2019; Moynihan et al., Reference Moynihan, Herd and Harvey2015). Common burdens include complex eligibility rules, heavy paperwork, and stigmatizing benefits. In two recent survey experiments, Keiser and Miller (Reference Keiser and Miller2020) and Nicholson-Crotty et al. (Reference Nicholson-Crotty, Miller and Keiser2021) experimentally manipulated the burdens of common social programs like Temporary Assistance for Needy Families (TANF) and the Supplemental Nutrition Assistance Program (SNAP). They show that higher levels of administrative burden can increase support for these programs and their beneficiaries. These results are primarily driven by perceptions of deservingness and by partisan identity, with Republicans being more likely to support burdensome social programs. These findings align with work by Ellis and Faricy (Reference Ellis and Faricy2020) and Faricy and Ellis (Reference Faricy and Ellis2014, Reference Faricy and Ellis2021). These authors used survey experiments to show that Americans who access social welfare spending through indirect transfers (for example, refundable tax credits) are seen as more deserving than people who access equivalent benefits through direct cash transfers (for example, monthly cheques).
Building off this literature, we identify at least two gaps in the literature. First, we do not know how generalizable this “burden heuristic” is—that is, whether it works outside the United States and whether it works similarly in other social programs with a different composition of beneficiaries and perceived program integrity. Keiser and Miller (Reference Keiser and Miller2020) focused on TANF, which is well-known in the United States, both for its welfare-to-work requirements and the racial composition of many of its beneficiaries. From this perspective, burdens could increase deservingness (and public support) not solely due to eligibility and screening, but also due to the imposition of burdens that disproportionately affect Black and Hispanic applicants. Keiser and Miller (Reference Keiser and Miller2020) mention this in their discussion section (p. 145), prompting questions about when burdens might help build public support for social programs and when they might undermine equitable redistribution.
Second, and relatedly, we do not fully understand how burdens shape support for social spending. As Keiser and Miller (Reference Keiser and Miller2020) show, burdens may increase deservingness perceptions and support for welfare, as people view the necessary effort as a cue for peoples’ need. Yet, this finding could be specific to “classic” liberal welfare regimes like the United States, where universal social programs are rare (Béland and Wadden, Reference Béland and Waddan2017; Esping-Anderson, Reference Esping-Andersen1990). In such settings, means testing and selective eligibility rules may lead voters to think of administrative burden as a way to differentiate deserving from undeserving individuals. Yet in other liberal welfare regimes, such as Canada, universal programs exist alongside targeted or means-tested programs (Béland et al., Reference Béland, Marchildon and Prince2020; Rice and Prince, Reference Rice and Prince2023). In such settings, people may come to see universality as a normal way to design public services. We thus hypothesize that voters might expect government would only impose burdens on those who are undeserving—in this sense, administrative burden could actually decrease deservingness perceptions of recipients and reduce support for social programs.
Vignette Experiment
Case Study
Our study considers administrative burden and support for a hypothetical low-income supplement to Canada's new national child care program. In considering our case study, we briefly discuss perceptions of the “deserving” poor. We then consider child care subsidies in relation to other social welfare policies.
Deservingness is socially constructed and contextual. In general, low-income individuals become part of the “deserving poor” when they are in need and unable to improve their financial situation. Jensen and Petersen (Reference Jensen and Petersen2017) show that public support for healthcare is higher than support for unemployment programs, largely because voters see sick people as more deserving than those who are jobless. Ellis and Faricy (Reference Ellis and Faricy2020) argue that programs like the Earned Income Tax Credit are popular precisely because they benefit working families who file taxes. These families are seen as more deserving than families who do not work or do not file taxes. Sociodemographic identities matter too. Depending on their circumstance, nonwhite individuals are often seen as less deserving, as are able-bodied men (Watkins-Hayes and Kovalsky, Reference Watkins-Hayes, Kovalsky, Brady and Burton2017). From an intersectional perspective, identities may overlap to mutually reinforce deservingness perceptions—such as women of a minority ethnic group who are seen as even less deserving than men from the same group, relative to members of an ethnic majority (Assouline and Gilad, Reference Assouline and Gilad2022). In Canada, Harell et al. (Reference Harell, Banting, Kymlicka and Wallace2022b) argue that national identity exerts a strong force on deservingness perceptions, over and above the effect of poverty. In a large survey, they show that Francophones, Indigenous people, and newcomers are often seen as less deserving than members of the English-language majority.
These patterns can help us understand how potential beneficiaries of child care subsidies may be perceived, and how these perceptions compare to other social welfare policies. For example, income assistance programs like TANF and SNAP in the United States provide basic necessities like food and shelter for low-income families with young children. Healthcare programs like Medicare and the Children's Health Insurance Program help minors experiencing illness or injury. Similarly, child care subsidies help vulnerable children by increasing their access to caregiving supports. From the perspective of deservingness, we might therefore expect voters to think of the families who benefit from these subsidies as highly deserving.
However, subsidized child care is also a labour market program. It is designed, and often communicated, as a way to improve parents’ economic opportunities—especially for mothers. This gendered economic dimension makes child care unlike other social welfare policies in ways that could matter for deservingness. Hansen (Reference Hansen2019) found deservingness is highly conditional on prior beliefs. People with individualistic attitudes may be skeptical of welfare programs, especially when beneficiaries have some measure of control over their circumstance. This skepticism may be especially strong for child care, which has not typically been thought of as a government responsibility (Mettler and SoRelle, Reference Mettler, SoRelle, Weible and Workman2018: 108). In the Canadian context, Pasolli (Reference Pasolli2015) highlights the gendered nature of this belief: People have historically been uncomfortable with the idea of working mothers. Thus, voters may see the decision to enroll one's child in subsidized child care as a personal choice—one that constitutes a deviation from a woman's traditional gender role—and thus undeserving of public support.
Child care is also worth studying given ongoing global efforts to enhance access to caregiving supports. In Australia, the federal government instituted universal free early education during the early months of the COVID-19 pandemic. In the United Kingdom, the pandemic prompted a large reduction in property taxes for public child care providers. In Canada, the federal government has gone even further, announcing a new national program to bring down child care costs—on average to just $10 a day. But even with Canada's new affordable plan, child care will be too expensive for many low-income families. Fortunately, several provinces offer targeted subsidies for low-income families (Pasolli, Reference Pasolli2015). Unfortunately, these subsidies can be burdensome and difficult to access. As Canada overhauls its approach to child care, there is an urgent need to understand the popularity of these benefits and the role of administrative burden in shaping public opinion.
Hypotheses
We use a pre-registered vignette experiment to explore the relationship between administrative burden and support for such a supplement. In contrast to Keiser and Miller (Reference Keiser and Miller2020), we explore whether administrative burdens might convey a negative signal of deservingness and reduce support and willingness to pay (WTP) for a low-income child care supplement. This is our primary hypothesis: people in the high-burden condition will exhibit weaker support and smaller WTP, relative to people in the control and low-burden conditions. In the subsequent analyses, we also test for treatment effect heterogeneity by political party, as well as the impact of administrative burdens on deservingness perceptions. See the Online Appendix for our pre-registration materials, including our other two hypotheses and analysis plan.
Data
We collected data through an online national survey, administered as part of the C-Dem 2022 Democracy Checkup (Harell et al., Reference Harell, Stephenson, Rubenson and Loewen2022a).
Our total sample size was 821.Footnote 1 We designed a three-arm vignette experiment. Participants were randomly assigned to one of the three conditions (see online Appendices 1 and 2 for the summary statistics of sample characteristics across treatment conditions and the results of a balance test). In the control condition (N=276), we presented participants with information about a hypothetical child care supplement. In the “low-burden” condition, we presented the same information and described a user-friendly enrollment procedure, consisting of automatic enrollment (N=277). In the “high-burden” condition, we again presented the same information and described a complicated enrollment procedure loosely based on British Columbia's Affordable Child care Supplement (N=268). See Table 1 for the wording of our interventions.
Treatments
Outcome Measures
Our primary outcomes are support for the hypothetical program (Likert scale, where 1 = “Strongly oppose” and 5 = “Strongly support”) and a two-part WTP question asking to fund the program through a potential tax increase (whether people are willing to pay a tax increase to support this program, and if yes, what the maximum annual amount is from the range of $0 to $200 or more). We include both of these measures to capture what has elsewhere been conceptualized as unconstrained and constrained measures of policy preferences (see, for example, Bremer and Bürgisser, Reference Bremer and Bürgisser2023). Our variable captures unconstrained preferences for child care subsidies—that is, the extent to which voters prioritize such a policy notwithstanding the implications for taxation or other demands for public attention. By contrast, our WTP variable captures constrained preferences—that is, the extent to which voters prioritize the policy in light of potential tax increases. Our empirical models include socioeconomic and demographic pre-treatment variables from the 2022 Democracy Checkup. We also explore the effect of burdens on five deservingness questions based on the well-known CARIN framework (Van Oorschot, Reference Van Oorschot2000, Reference Van Oorschot2006).
Methods
We use linear regression to estimate the treatment effect of administrative burden on attitudes toward the hypothetical child care supplement.Footnote 2 Our empirical models are:
where Support i ∈ (1, 5) measures support for each individual i on a 1-5 scale, WTP i ∈ (0, 200) measures each individual's WTP a tax increase in dollars on a $0–$200 or more sliding scale, T i is a categorical variable measuring each individual's treatment assignment (high-burden vs. low-burden vs. control), and ∈i is the error term. Unless otherwise indicated, we use HC1 (“Stata”) robust standard errors.
We also provide an Online Appendix with supplementary information and analyses, including summary statistics, a balance test, full regression results, a robustness check using an ordered logit model, and results from a manipulation check.
Results
Administrative Burden and Public Support for Child Care Supplement
In Figure 1 (top), we consider support for the program (the full regression results and tables are available in Online Appendix 3, Table A4). Overall, support for the supplement is high. In the control group, 57 percent of respondents “strongly” or “somewhat” support the supplement, while just 20 percent of respondents are in opposition (“strongly” or “somewhat” oppose). This represents a mean score of 3.53 on a 1–5 scale. The figure plots mean support among people in the control. Consistent with our pre-specified hypothesis, linear regression shows that people who received the high-burden intervention showed a lower level of support to the program than those in the “control” condition. The magnitude of the effect is approximately -0.21 (se = 0.10, P-value = 0.04), or a 6 percent reduction in support compared to the control group. We find a similar effect for the low-burden condition, which is contrary to our prior expectations. The magnitude of the effect is approximately -0.23 (se = 0.10, P-value = 0.02), or a 7 percent reduction in support compared to the control group. However, and contrary to our primary hypothesis, we find no difference in support between the high- and low-burden conditions (b= 0.02, se = 0.10, P-value = 0.84).
Treatment Effects of Administrative Burden on Support and WTP for Hypothetical Low-Income Child Care Supplement.
Next, we consider WTP for a tax increase to fund the program. Overall, there is little enthusiasm for tax increases to fund the child care supplement. This variable is strongly right-skewed, with 65 percent of people in the control group saying they would pay $0 in taxes. In the bottom panel of Figure 1, we show this represents a mean WTP of $23.94 (se = 2.63). Contrary to our expectations, linear regression shows no evidence of treatment effects. In the high-burden condition, WTP is slightly lower than control (b= -0.98, se = 3.92) but this difference is not statistically significant (P-value = 0.80). In the low-burden condition, WTP is slightly lower than control (b= -0.44, se = 3.80) but again this difference is not statistically significant (P-value = 0.91). We find no difference in WTP between the high- and low-burden conditions (b= -0.55, se = 4.00, P-value = 0.89).
Descriptive statistics suggest some minor imbalance on variables that theory suggests could shape perceptions on child care, such as gender and employment status (see Tables A1 and A2 in the Online Appendix). Thus, we extend both models using pre-treatment covariates: party identification (that is, the political party each individual i voted for in the 2021 federal election); ideology (Left/Right); household income; marriage status; age; gender identity; employment status, and region. Re-estimation results with pre-treatment covariates are available in Online Appendix 3 (Table A4). Looking at support, we find the low-burden treatment effect is no longer significant (b = -0.16, se = 0.10, P-value = 0.08) while the high-burden intervention is still significant (b = -0.20, se = 0.10, P-value = 0.04).Footnote 3 Looking at WTP, we find no substantive change in our findings. Note, however, that this robustness check is sensitive to the selection of covariates, and was not specified in our pre-registration plan.
In short, we find little to no evidence of our main hypothesis. If administrative burdens influence support for a low-income child care supplement, the effect is minor—about a 6 percent decrease relative to control. This effect is concentrated among people in the high-burden condition, as we find no evidence of a treatment effect among people in the low-burden group when we include pre-treatment covariates. This result, taken together with the lack of evidence of any effect on WTP, leads us to conclude that there is no substantial effect of administrative burden on public attitudes toward a low-income child care supplement.
Partisanship Differences
Next, we consider the impact of partisanship on public support and WTP for this hypothetical low-income child care supplement. Rather than pool all three experimental groups, we focus on people in the control group. This eliminates any potential confounding effects from the experiment, however, it comes at the cost of significantly reducing the sample size, to just 276 respondents. We add a word of caution when interpreting these exploratory findings: The sample sizes are at times very small, resulting in underpowered analysis. In Figure 2, we plot mean support for the hypothetical affordable child care supplement by vote choice in the 2021 federal election. Child care was a key election issue. The Liberal Party and the New Democratic Party each proposed national programs with universal elements, including cost reductions to $10-a-day. In contrast, the Conservative Party proposed a refundable tax credit. Consistent with these competing proposals, we find a clear partisan divide. Support for the hypothetical supplement was higher among Liberal (mean=3.69, SE=0.13) and NDP voters (mean=3.71, SE=0.16), relative to people who voted for the Conservative Party in the 2021 election (mean=3.02, SE=0.17). We estimate pairwise comparisons using Tukey's Honest Significant Differences test. We find statistically significant differences between Conservatives and Liberals (difference in means = -0.67, 95% CI = [-1.27, -0.07], P-value=0.02) and Conservatives and NDP voters (difference in means = -0.69, 95% CI = [-1.36, -0.03], P-value=0.03).
Support for a hypothetical child care supplement by 2021 vote choice (control group).
In Figure 3, we plot mean WTP by vote choice. Again, we find a clear partisan divide. WTP for the hypothetical supplement was higher among Liberal voters (mean=29.4, SE=5.34) and NDP voters (mean=31.3, SE=5.75), relative to people who voted for the Conservative Party in the 2021 election (mean=13.2, SE=4.36). However, using Tukey's Honest Significant Differences test, we find none of these differences are statistically significant.
WTP for a hypothetical child care supplement by 2021 vote choice (control group).
In our Online Appendix, we also explore the heterogeneity of treatment effect by political party and gender. We also compare perceptions of deservingness across our treatment groups.
Conclusion
Child care supports early childhood development and helps families, especially women, participate in the economy. Recent evidence suggests the design of social programs can influence citizens’ beliefs about social spending. In this research note, we leverage these insights to shed light on an urgent question in Canada: How should governments design child care benefits? We conducted an experimental survey to inform development of a low-income supplement to Canada's new national child care program. In terms of theory, we complement the large literature on deservingness and the burgeoning literature on administrative burden. Our results indicate that the “burden heuristic” identified in the United States may differ across different contexts of social programs, such as differing beneficiaries and public confidence in program integrity. Future research is warranted to unpack the relationship between administrative burden and public support, especially the potential for deservingness perceptions to mediate this relationship.
In terms of policy, we offer three evidence-informed recommendations for governments looking to expand access to child care. First, our survey results suggest that governments can afford to be even bolder when designing child care programs. Most Canadians (56 percent) would support greater spending on affordable child care supplements, while a smaller minority (20 percent) would oppose it. Yet despite broad popularity, the average voter is only willing to pay roughly $25 more a year in taxes to fund child care subsidies. We find a stark difference between unconstrained support for child care benefits and support when tax increases are introduced. Second, we find that voters could potentially support initiatives that help low-income families more easily access child care benefits. Some provinces already offer some type of supplement, including British Columbia, Ontario, and Quebec. But the administrative burden is high. In British Columbia, for example, families must file as many as six forms, including consent forms, income declaration forms, work search forms, and forms to apply for special needs supplements. Our study finds no evidence these burdens build public support for child care benefits—if anything, they reduce support. Third, our results show strong differences in support for child care by partisanship. Liberal and NDP voters are more supportive and more willing to pay for child care programs than Conservative voters. This may make sense given partisan disagreement over social policy more generally (Merkley, Reference Merkley2022), as well as the prominent debate over child care spending during the 2021 federal election. We note that a similar divide also exists in the United States, where child care has not featured prominently in national debates. While three-quarters of the US public supports child care subsidies for low-income families, such policies are much more popular among Democrats (86 percent in support) compared to Republicans (68 percent in support) (YouGov, 2023). Given the partisan difference in support for child care subsidies, the well-established influence of partisan cues, and Canada's increasingly polarized political climate (Armstrong et al., Reference Armstrong, Lucas and Taylor2022; Johnston, Reference Johnston2023; Kevins and Soroka, Reference Kevins and Soroka2018), we highlight the risk of governments and parties negatively framing beneficiaries. The main concern is stigma: Ensuring low-income parents feel no shame accessing benefits to help themselves and their children.
Supplementary Material
The supplementary material for this article can be found at https://doi.org/10.1017/S0008423924000350
Acknowledgement
The authors greatly acknowledge the funding support provided by the Houston Family Trust Research Grant through the Johnson Shoyama Graduate School of Public Policy, University of Saskatchewan. We extend our appreciation to the C-Dem for their assistance in survey programming and data collection as part of its 2022 Democracy Checkup. We also thank two anonymous CJPS reviewers for their valuable feedback.
Competing interests
The authors declare none.
Open access
