Experimental tests of public support for disability policy

Timmons, Shane; Carroll, Eamonn; McGinnity, Frances Research Report Experimental tests of public support for disability policy Research Series, No. 159 Provided in Cooperation with: The Economic and Social Research Institute (ESRI), Dublin Suggested Citation: Timmons, Shane; Carroll, Eamonn; McGinnity, Frances (2023) : Experimental tests of public support for disability policy, Research Series, No. 159, The Economic and Social Research Institute (ESRI), Dublin, https://doi.org/10.26504/rs159 This Version is available at: https://hdl.handle.net/10419/298323 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ EXPERIMENTAL TESTS OF PUBLIC SUPPORT FOR DISABILITY POLICY SHANE TIMMONS, EAMONN CARROLL AND FRANCES MCGINNITY RESEARCH SERIES NUMBER 159 APRIL 2023 E V I D E N C E F O R P O L I C Y EXPERIMENTAL TESTS OF PUBLIC SUPPORT FOR DISABILITY POLICY Shane Timmons Eamonn Carroll Frances McGinnity April 2023 RESEARCH SERIES NUMBER 159 Available to download from www.esri.ie https://doi.org/10.26504/rs159 The Economic and Social Research Institute Whitaker Square, Sir John Rogerson’s Quay, Dublin 2 This Open Access work is licensed under a Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. ABOUT THE ESRI The mission of the Economic and Social Research Institute (ESRI) is to advance evidence-based policymaking that supports economic sustainability and social progress in Ireland. ESRI researchers apply the highest standards of academic excellence to challenges facing policymakers, focusing on 12 areas of critical importance to 21st Century Ireland. The Institute was founded in 1960 by a group of senior civil servants led by Dr T. K. Whitaker, who identified the need for independent and in-depth research analysis to provide a robust evidence base for policymaking in Ireland. Since then, the Institute has remained committed to independent research and its work is free of any expressed ideology or political position. The Institute publishes all research reaching the appropriate academic standard, irrespective of its findings or who funds the research. The quality of its research output is guaranteed by a rigorous peer review process. ESRI researchers are experts in their fields and are committed to producing work that meets the highest academic standards and practices. The work of the Institute is disseminated widely in books, journal articles and reports. ESRI publications are available to download, free of charge, from its website. Additionally, ESRI staff communicate research findings at regular conferences and seminars. The ESRI is a company limited by guarantee, answerable to its members and governed by a Council, comprising 14 members who represent a cross-section of ESRI members from academia, civil services, state agencies, businesses and civil society. The Institute receives an annual grant-in-aid from the Department of Public Expenditure and Reform to support the scientific and public interest elements of the Institute’s activities; the grant accounted for an average of 30 per cent of the Institute’s income over the lifetime of the last Research Strategy. The remaining funding comes from research programmes supported by government departments and agencies, public bodies and competitive research programmes. Further information is available at www.esri.ie THE AUTHORS Frances McGinnity is an Associate Research Professor at the Economic and Social Research Institute (ESRI) and Adjunct Professor of Sociology at Trinity College Dublin (TCD). Shane Timmons is a Research Officer at the ESRI and Adjunct Professor of Psychology at TCD. Eamonn Carroll is a Postdoctoral Research Fellow at the ESRI. ACKNOWLEDGEMENTS This research forms part of the Joint Research Programme, which is funded by the National Disability Authority, on the experiences of persons with disabilities across key policy areas at the ESRI. We would like to thank staff at the NDA, in particular David Hallinan and Rosalyn Tamming, for helpful feedback on the study design and materials. We are also especially grateful to members of the Disability Advisory Board for their insights and input: Jacqui Browne, Fiona Ferris, Ed Harper, Sarah Moorhead, Maria Ní Fhlatharta and Aoife Price. We also thank Mat Creighton for helpful comments and advice on study design. We further appreciate helpful comments from colleagues at the ESRI at an internal seminar on the findings. This report has been accepted for publication by the Institute, which does not itself take institutional policy positions. All ESRI Research Series reports are peer reviewed prior to publication. The authors are solely responsible for the content and the views expressed. The views and opinions expressed in this publication remain those of the authors and do not represent the views or opinions of the National Disability Authority. Table of contents | v TABLE OF CONTENTS EXECUTIVE SUMMARY .............................................................................................................. IX CHAPTER 1 BACKGROUND TO THE STUDY .............................................................................. 1 1.1 Motivation for the report ................................................................................... 1 1.2 Challenges faced by disabled people in Ireland .................................................. 3 1.3 Attitudes towards disabled people ..................................................................... 5 1.4 Hidden attitudes and methods to elicit them ..................................................... 7 1.5 Inattention to policy costs and trade-offs ........................................................ 12 1.6 Study overview .................................................................................................. 14 CHAPTER 2 DATA COLLECTION AND EXPERIMENTAL DESIGN ................................................ 15 2.1 Data collection: participants ............................................................................. 15 2.2 Selection of issues to investigate ...................................................................... 16 2.3 Materials and design ......................................................................................... 17 CHAPTER 3 LIST EXPERIMENT RESULTS ................................................................................... 23 3.1 Testing design effects ....................................................................................... 23 3.2 Increased welfare payments ............................................................................. 24 3.3 Housing ............................................................................................................. 25 3.4 Parking .............................................................................................................. 26 3.5 Familiarity with disability .................................................................................. 27 CHAPTER 4 POLICY TRADE-OFFS RESULTS .............................................................................. 31 4.1 Cost of living support ........................................................................................ 31 4.2 Supports for children with disabilities .............................................................. 33 4.3 Building wheelchair accessible infrastructure .................................................. 34 4.4 Employment ...................................................................................................... 36 4.5 Familiarity with disability .................................................................................. 38 CHAPTER 5 DISCUSSION AND IMPLICATIONS ......................................................................... 41 5.1 Support for social welfare ................................................................................. 41 5.2 Accessibility and infrastructure......................................................................... 42 5.3 Other social issues ............................................................................................ 43 5.4 Familiarity with disability .................................................................................. 43 5.5 Limitations......................................................................................................... 44 5.6 Implications ....................................................................................................... 46 5.7 Conclusion ......................................................................................................... 47 REFERENCES ............................................................................................................................. 49 APPENDIX ................................................................................................................................. 57 vi | Experimental tests of public support for disability policy LIST OF TABLES Table 2.1 Socio-demographic characteristics of participants ...................................................... 16 Table 2.2 Policy statements ......................................................................................................... 21 Table 3.1 Chi-square tests of list experiment randomisation ...................................................... 24 Table 3.2a Logistic regression models predicting direct responses and list responses to increasing welfare payments ....................................................................................... 25 Table 3.2b Logistic regression models predicting direct responses and list responses to prioritising disabled people for social housing ............................................................. 26 Table 3.3 Logistic regression models predicting policy support from familiarity with disability ....................................................................................................................... 29 Table 4.1 Logistic regression models predicting support for increased cost of living supports ........................................................................................................................ 32 Table 4.2 Logistic regression models predicting support for increased child supports ............... 34 Table 4.3 Logistic regression model predicting support for building wheelchair accessible infrastructure................................................................................................................ 36 Table 4.4 Logistic regression models predicting policy support from socio-demographic characteristics ............................................................................................................... 38 Table 4.5 Logistic regression models predicting policy support from familiarity with disability ....................................................................................................................... 39 Table A.1 Joint distributions and design effect test for welfare payment list ............................. 57 Table A.2 Joint distributions and design effect test for social housing list .................................. 57 Table A.3 Joint distributions and design effect test for car parking list ....................................... 57 Table A.4 Logistic regression of cost-of-living support: education interaction ............................ 59 LIST OF FIGURES Figure 2.1a Example list experiment design (social housing) ......................................................... 18 Figure 2.1b Welfare payments list experiment ............................................................................... 19 Figure 2.1c ‘Filler’ list experiment ................................................................................................... 19 Figure 2.1d Parking list experiment ................................................................................................. 20 Figure 3.1 Percentage of participants endorsing sensitive items in the three list experiments .................................................................................................................. 27 Figure 3.2 Support for increasing welfare payments for disabled people by familiarity with disability and experimental treatment ......................................................................... 29 Figure 4.1 Support for increased cost of living associated with having a disability by version ...................................................................................................................................... 31 Figure 4.2 Support for increased supports for children with disabilities by statement version .......................................................................................................................... 33 Figure 4.3 Support for building wheelchair accessible infrastructure .......................................... 35 Figure 4.4 Beliefs that people with disabilities should work ......................................................... 37 Figure 4.5 Support for policies by familiarity with disability ......................................................... 40 Figure A.1 Distribution of list experiment responses .................................................................... 58 Abbreviations | vii ABBREVIATIONS ABC1 Chief income earner is in a higher, intermediate or junior managerial/professional role C2DE Chief income earner is in a manual or casual work role or unemployed CSO Central Statistics Office DPO Disabled Persons’ Organisations ESRI Economic and Social Research Institute IHREC Irish Human Rights and Equality Commission NDA National Disability Authority UNCRPD United Nations Convention on the Rights of Persons with Disabilities 2 | Experimental tests of public support for disability policy time, or it could instead reflect an increasing perception that positive responses to such questions are socially expected.2 Our first aim was to identify the extent of ‘social desirability’, the tendency for survey respondents to alter their responses in order to present themselves in a positive light, in reported attitudes towards disability. This report also comes at a time of substantial reform across many policy areas central to disabled people’s lives. A report on the cost of disability (Department of Social Protection, 2021) has sparked public debate about how much of this cost should be covered by the State and how much should be met privately by individuals and families. Recent research on employment among disabled people shows that the COVID-19 pandemic had a particularly detrimental impact on this group, exacerbating an existing gap in employment prospects (Emerson et al., 2021; Eurofound, 2022). In education, the School Inclusion Model is designed to make Irish schools more inclusive, while social support and medical care for children with disabilities are also being reformed through the Progressing Disabilities programme. Overall, these and other policy developments aim to bring Ireland in line with its commitments under the United Nations Convention on the Rights of Persons with Disabilities (UNCRPD), adopted in 2006 and formally ratified by Ireland in 2018. These reforms will chiefly affect disabled people, and their outcomes and experiences should be central to evaluation. However, they come with costs which will be met by the general public. Promoting a positive attitude towards these developments among the public is hence an important component of sustainable and effective change. Yet standard surveys prioritise simplicity of survey items and leave as implied that social policies entail costs. If respondents fail to consider the costs of policies, support in surveys may be artificially inflated. Hence, our second aim was to determine whether making different potential funding mechanisms of policies explicit in survey statements (e.g. a tax increase to fund the policy) influences public support for those policies. At this juncture we note the variation in conceptualisation of disability. The United Nations Convention on the Rights of Persons with Disabilities (UNCRPD) states that: persons with disabilities include those who have long-term physical, mental, intellectual or sensory impairments which in interaction with various barriers may hinder their full and effective participation in society on an equal basis with others. _______________________________________________________________________________ 2 For example Creighton et al. (2015) found that while directly-expressed support for immigration declined in the United States following the 2008 economic crash, anonymously expressed support did not change. It was thus the extent of social desirability in responding that changed. Background to the study | 3 The 2016 Census recorded whether respondents have a long-lasting difficulty or condition, but not necessarily whether any condition they have hinders their participation in society. Moreover, how disabled people are defined in law, research or policy may not be the same as how they identify themselves or how they are perceived by others. In particular, public perceptions of disability may centre on physical, visible disabilities, and exclude invisible conditions or mental health issues. The extent of this may also vary according to familiarity with disability. We return to this point in the concluding chapter. The remainder of this chapter explores the motivation for the report in greater detail. First, some of the challenges faced by disabled people in Ireland are considered, followed by an exploration of existing literature on attitudes towards disabled people with a focus on research in Ireland where possible. Approaches used to elicit ‘hidden’ attitudes (i.e. attitudes that are held but not directly expressed in surveys) are then discussed. We highlight the few international studies that have employed these methods to investigate attitudes towards disability but our focus is on the ‘list experiment’, which we exploit for this report but, as far as we know, has not yet been used in this field of research anywhere in the world. Finally, we highlight relevant research from behavioural science on how people answer survey questions. 1.2 CHALLENGES FACED BY DISABLED PEOPLE IN IRELAND According to 2016 Census data, 643,131 people or 13.5 per cent of the Irish population identify as having a long-lasting difficulty or condition or as experiencing difficulty with everyday activities (CSO, 2017). This proportion is projected to have increased by the 2022 Census data, as the total population grows and ages.3 Disabled people thus represent a sizable minority group in Irish society who face many barriers to full and equal participation in society. These challenges include but are not limited to the cost of living, employment and access to the services needed to live a full and independent life. Importantly, disabled people represent a diverse group, with needs that depend on individual circumstances. That said, even those with lower levels of additional needs are at risk of economic disadvantage, stemming from increased cost of living, lower levels of employment, and difficulty accessing appropriate housing and education. The Cost of Disability in Ireland Report (Department of Social Protection, 2021) highlights the additional spending needs faced by many people with disabilities. These include specialised care services and equipment, home adaptations, medicines and mobility costs (e.g. additional use of taxis), as well as more costly everyday expenditure, such as increased heating costs and specialised clothing. _______________________________________________________________________________ 3 The 2022 data are likely to report change in disability prevalence beyond changes in the population as the wording of the relevant question was revised. Respondents in 2022 were asked whether each long-lasting difficult or condition impacts them to a great extent, to some extent or does not impact them, rather than a binary ‘yes’ or ‘no’ response to whether they have the condition as in 2016. 4 | Experimental tests of public support for disability policy The extent of these costs varies from low to very high, depending on individual circumstances, but are estimated to be within the range of €9,482 to €11,734 per year on average. Other studies in Ireland estimate that households with members who have a disability face additional costs that equate to between 20 per cent and 37 per cent of the average household income (Cullinan et al., 2011). These costs are a direct result of having a disability. The additional cost of living faced by disabled people is exacerbated by differences in employment outcomes. Census data from 2016 show that just one-third of working age, disabled people identified employment as their main economic status compared to two-thirds of non-disabled people. This figure varies by disability type, from 15 per cent of those with an intellectual disability to 46 per cent of those with deafness or a serious hearing impairment, but employment levels for individuals with all types of disabilities remain significantly below those for persons without disabilities (Kelly and Maître, 2021; see also Watson et al., 2013). Disabled people are at greater risk of poverty than non-disabled people. This applies both when comparing disabled people in employment with non-disabled people in employment, and when comparing disabled people not in employment with nondisabled people not in employment (Kelly and Maître, 2021). The risk of economic disadvantage for disabled people also extends to housing. Analysis of discrimination (i.e. less favourable treatment because of a protected characteristic)4 in the housing market in Ireland shows that people with disabilities are twice as likely to report being discriminated against when compared to people without disabilities, while controlling for their employment status (Grotti et al., 2018). This discrimination is particularly pronounced among renters, in line with international research (e.g. Turner, 2005). Extending from the above, people with disabilities are at higher risk of experiencing discrimination in accessing housing and are over-represented among the homeless population (Grotti et al., 2018). Other analyses show that people with disabilities experience higher levels of housing deprivation and housing affordability issues5 (Russell et al., 2021). Gaps in socio-economic outcomes between disabled and non-disabled individuals appear to begin in childhood. While educational attainment of disabled people in Ireland is better than the EU average (42 per cent vs. 25 per cent hold at least a post-secondary education; Kelly and Maître, 2021) and there have been significant efforts made to improve resources and inclusivity in schools (Ring, forthcoming), _______________________________________________________________________________ 4 Gender, marital status, family status, age, disability, ‘race’ (skin colour or ethnic group), sexual orientation, religious belief, and/or membership of the Traveller Community are ‘protected characteristics’ under the legal definition of discrimination. 5 Housing deprivation is defined based on features of a home, specifically accommodation characterised by one or more of these four items: leaking roof/damp walls or /window frames or floor rot; lack of central heating; lack of double glazing, and lack of light. Housing affordability problems are measured as, for example, those experiencing income poverty after housing costs (household income lower than 60 per cent of the median equivalised income after housing costs) or being in arears on rent and mortgage repayments in the last 12 months (see Russell et al., 2021 for further details). Background to the study | 5 recent evidence highlights challenges faced by children with disabilities. Data from the Growing Up in Ireland (GUI) study show that young people with special educational needs and/or disabilities are less likely to like school, more likely to struggle academically and less likely to progress to higher education (Carroll et al., 2022). Accessing supports both inside and outside of the classroom and even enrolling in an appropriate setting can also be a significant struggle for disabled young people, with adverse effects on their educational outcomes, general wellbeing and wider family (O’Brien, 2021). The educational challenges facing disabled children and young people are compounded by the fact that they are more likely to experience economic vulnerability and attend DEIS schools (schools which receive targeted supports as they serve areas with high concentrations of socioeconomic disadvantage) than their peers (Carroll et al., 2022b). Beyond economic disadvantage and discrimination, many disabled people face everyday challenges in accessing services. Exact issues depend on individual circumstances and type of disability, but mobility and transport difficulties are widespread. For example, some use of public transport requires disabled people with particular impairments and accessibility needs to give notice 24 hours in advance, a requirement not necessary for others in society. Access to buildings and infrastructure is cited as one of the biggest issues facing physically disabled people (Irish Wheelchair Association, 2020). Three-in-four report poor access to public spaces due to steps being the only option at main entrances. While infrastructural change depends on policy, public attitudes and behaviour have further implications for accessibility. Anecdotal evidence and survey research by the Disabled Drivers Association of Ireland highlights illegal parking in disabled bays by those without permits as a major frustration in the disability community (Hutton, 2020). 1.3 ATTITUDES TOWARDS DISABLED PEOPLE The NDA have conducted regular national surveys in Ireland on attitudes to disability (NDA, 2017). These surveys were carried out face-to-face with a nationally representative sample. Between 2001 and 2017, there was a general increase in the proportion agreeing that people with physical disabilities, intellectual disabilities and autism can participate fully in life. Agreement has generally been higher among younger people, suggesting there may be generational effects alongside a general shift in attitudes (ibid). The proportion of people who agree that people with disabilities are treated fairly in Irish society decreased from 44 per cent to 36 per cent between 2011 and 2017, although explaining this decrease is not straightforward. It could reflect a perceived deterioration in how disabled people are treated, or greater awareness of the extra supports needed for ‘fair’ outcomes, or a combination of both. Either way, people in Ireland generally appear to support measures that would help disabled people. In the most recent survey (2017), the majority (almost 80 per cent) endorsed increased welfare payments for disabled people and supported prioritising them 6 | Experimental tests of public support for disability policy in social housing allocation and on hospital waiting lists.6 A similar proportion agreed that there are some circumstances where it is right to treat disabled people more favourably than others. Surveys like these allow for socio-demographic predictors of attitudes to be identified. The NDA (2017) survey shows that knowing someone with a disability predicts many positive attitudes towards disabled people, including reported comfort with having a work colleague or neighbour with a disability and belief that disabled people should have children if they wish. The importance of familiarity and contact with disabled people is replicated in multiple international studies and review papers, across multiple measures (Burke et al., 2013; Ju et al., 2013). For example, a US survey of attitudes to people with intellectual disabilities found that more frequent contact with someone who has an intellectual disability predicted greater support for the rights of people with intellectual disabilities and perceptions of their capabilities (McConkey et al., 2021). Similarly, a Swiss survey of pre-service teachers showed far more positive attitudes to disabled people among those who reported higher levels of prior contact with disabled people (Kunz et al., 2021). The focus of surveys should not solely be on the attitude of the general public; research with samples of people with disabilities is vital for understanding their lived experience. The Quarterly National Household Survey (QNHS) in 2014 featured an equality module which probed experiences of discrimination in Ireland based on protected characteristics, including disability. Analyses of this module have shown that disabled people in Ireland are significantly more likely to report experiencing discrimination in workplaces, while seeking work, and in accessing public and private services than non-disabled people (McGinnity et al., 2017). The size of the difference is large, with disabled people more than twice as likely to report discrimination while seeking employment and in accessing public services (15.5 per cent and 7.2 per cent, respectively, vs. 6.7 per cent and 2.8 per cent among people with no disability). Further analyses suggest that disabled people are more adversely affected by discrimination when they experience it. Almost one-in-five people with a disability who experienced discrimination reported that it had a ‘very serious’ effect on them, compared to one-in-ten of those without disabilities (Banks et al., 2018). More recent research in the UK found that 72 per cent of disabled people had experienced negative attitudes or behaviours in the last five years (Moss and Frounks, 2022; see also Dixon et al., 2018). These include assumptions about their disability and what they can do, impatience, dismissal of their condition or need for accommodations, accusation of faking or being lazy, exclusion and being patronised, along with many others. These attitudes and behaviours were most _______________________________________________________________________________ 6 Although the prioritisation of disabled people is not currently a policy in housing or healthcare, nor is it a goal of the NDA. Background to the study | 7 commonly experienced from the general public, but also from employers, educators, public service staff and even friends and family members. Unsurprisingly, almost 90 per cent of those who experienced these behaviours said it had a negative effect on them, with higher proportions of 18-34 year olds and women reporting a negative effect. It is difficult to gauge how likely experiences in the UK are to generalise to Ireland, as there are few similar measures recorded in both countries. Where comparisons can be made, attitudes towards disabled people in Ireland appear more positive, at least in employment settings. Whereas people in Ireland report being generally comfortable with the idea of working with a disabled person (rating a score of 8.9 out of 10 for a physically disabled person and 8.2 for a person with mental health difficulties; NDA, 2017), the British Social Attitudes survey (Cant and Bennet, 2022) reports that ‘the public support equal chances for disabled people in the workplace but would not necessarily want to work with a disabled person’ (p. 2). However, analysis of the QNHS Equality Module cited above has indicated that disabled workers are 65 per cent more likely to face workplace discrimination than nondisabled workers, with larger differences when seeking employment (McGinnity et al., 2017; Banks et al., 2018). Moreover, there are multiple international studies showing differences in reported attitudes towards employing people with disabilities and objectively measured hiring practices. Reviews show that employers tend to report positive attitudes towards the prospect of hiring someone with a disability (Bredgaard and Salado-Rasmussen, 2021; Burke et al., 2013; Ju et al., 2013), yet field experiments reveal high levels of likely discrimination. Audit experiments, in which researchers apply to job postings with fictitious CVs that vary systematically by features of interest (e.g. disability disclosure), have been run in the US, Canada and Norway. Findings show that people who disclose their disability on a job application are between 26 per cent to 50 per cent less likely to be called to interview, despite having otherwise equivalent applications (Ameri et al., 2018; Bellemare et al., 2018; Bjørnshagen, 2021). These findings point towards the importance of exploiting experimental methods to gain better insight into attitudes that may be concealed in standard surveys. The next section details some such methods for identifying concealed negative attitudes among the general public. 1.4 HIDDEN ATTITUDES AND METHODS TO ELICIT THEM Surveys often struggle to accurately capture attitudinal data about controversial or sensitive issues. This section considers evidence of potential bias in survey responses on a range of topics and methods to address this. Estimates of attitudes about sensitive topics, such as immigration, racism and abortion, vary depending on the level of anonymity respondents are provided (Creighton and Jamal, 2015; Glynn, 2013; Kulinski et al., 1997; Rosenfeld et al., 2016). Reported prevalence of sensitive behaviours, such as plagiarism, shoplifting and sexual activity, also varies (Chuang et al., 2021; Coutts et al., 2011; Krumpal, 2013). Comparisons of time-use 8 | Experimental tests of public support for disability policy diaries with survey responses show that people also sometimes over-report positive attributes, such as exercising and attending religious services (Brenner, 2011; Brenner and De Lamater, 2016). One explanation for the difference between what people report when asked directly and what they might express when offered anonymity is perceived pressure to give a response that is believed to be socially desirable and thus present themselves positively to others. A large body of research suggests that this type of misreporting of certain attitudes can lead to systematic errors. This survey bias is known as ‘social desirability bias’ (Krumpal, 2013). One of the most straightforward ways to mitigate social desirability bias is through survey mode. Surveys administered in-person or over the phone with an interviewer tend to produce lower estimates of sensitive attitudes and behaviours than surveys that are self-administered or conducted online (Krumpal, 2013; Paulhus, 2002). Hence one way to measure sensitive opinions or behaviours more accurately is to administer surveys online in a way that assures respondents of their anonymity. A number of innovative techniques have been used to further address social desirability bias, each drawing upon the experimental method. In a survey experiment, some participants see question formats in one way (e.g. a direct question) and other participants see the same question in another format (e.g. one that requires an indirect response and offers greater anonymity). Crucially, the question format that each participant sees is decided at random, meaning that any aggregate differences can be confidently attributed to the difference in question format and not to underlying differences between participants. For example, if response formats that provide greater anonymity show a higher prevalence of certain attitudes (e.g. racism), then researchers can be reasonably sure that some people conceal their true attitude when asked via standard survey questions. Experimental techniques have been widely used to measure attitudes in other fields (e.g. racism, safe sex practices), but few studies have tested social desirability effects in surveys about issues relevant to people with disabilities. This raises questions about the accuracy of responses to direct questions as reported in Section 1.3, and suggests there is a need to develop other methods to assess attitudes towards disability in Ireland. Where international studies have assessed hidden attitudes to disability issues, they indicate some evidence of social desirability bias. Specifically in relation to attitudes towards inclusive education, Lüke and Grosche (2018) used an experimental design whereby the same survey was presented to four different groups as coming from four different organisations: a university, a group that opposed inclusion as it might lower general academic standards, a group that opposed inclusion as it threatened support for children with disabilities currently attending special schools and a group which supported inclusion. They found Background to the study | 9 evidence of stronger support for inclusive education in surveys purporting to be run by the pro-inclusion group, implying that surveys of disability issues may overestimate support if respondents are aware of the purpose of the survey. Other experiments have sought to measure social desirability directly, although with potentially unreliable or potentially confusing methods. For example, the Implicit Association Test (IAT) tests how quickly and accurately people can associate positive and negative words with different groups of people (e.g. people with disabilities and people without disabilities). Differences in speed or precision in the categorisation are taken to reflect a difference in ‘implicit associations’ between the two groups. If people are slower to associate a disabled person with a positive characteristic than they are a non-disabled person, it is thought to reveal an implicit and perhaps unconscious negative attitude towards disability. One US study which used different types of IATs showed more negative implicit associations of disabled people than were reported in standard survey measures of attitudes (Thomas et al., 2014). The IAT method is, however, primarily focused on unconscious bias rather than deliberate concealing of negative attitudes and results cannot be easily compared to surveys of the general population. The IAT method itself is far from universally accepted (see for example Mitchell and Tetlock, 2017; Bartels and Schoenrade, 2022). It has poor methodological rigour, with the same people generating different scores on the same version of the test when taken again (Greenwald and Lai, 2020). Ostapczuk and Musch (2011) used two approaches to estimate socially desirable responding to a survey item about feeling ‘uneasy’ in the presence of people with disabilities. One approach asked respondents to report what they believed was ‘most people’s’ attitude rather than their own. Results showed a large discrepancy between this estimate of ‘most people’s’ attitude and the response when participants were asked about their own attitudes (55 per cent vs. 8 per cent for negative attitudes towards physical disability and 79 per cent vs. 27 per cent for mental disability). The authors acknowledge, however, that the difference likely reflects respondents’ overestimation of general negative sentiments rather than a projection of their own views onto others. The second approach they employed was the ‘Randomised Response Technique’, in which participants are instructed to answer the sensitive item either truthfully or to give a specific response based on an irrelevant rule of a known probability (e.g. to answer ‘yes’ if their mother was born in February, March or April, regardless of the question content). As researchers have no insight into the participant’s experimental condition, the approach offers a layer of anonymity. There is thus no way of knowing whether an individual is answering the question or giving a directed response, but the proportions of answers to the question at the overall sample level can be estimated. Responses recorded by Ostapczuk and Musch (2011) were not statistically different from the direct question condition 10 | Experimental tests of public support for disability policy (11 per cent reported unease around a physically disabled person and 24 per cent for mental disability). The study showed high levels of respondents not answering as instructed by the randomisation device, which the authors infer as ‘cheating’ and hence some evidence of social desirability bias. However, others have observed that the Randomised Response Technique can lead to high non-response rates due difficulty understanding the instructions (Rosenfeld et al., 2016). We could locate no other studies that have sought to measure social desirability in surveys about disability issues, yet there are other techniques that could be exploited. The ‘list experiment’ or item count technique has been frequently used to gauge social desirability in both sensitive attitudes and sensitive behaviours in other domains (see above). Like the Randomised Response Technique, list experiments provide respondents with ‘permanent’ anonymity by not asking them directly for their response (Chuang et al., 2021).7 However, it is simpler to employ and has lower non-response rates (Rosenfeld et al., 2016). As it does not yet appear to have been used in relation to attitudes or behaviours towards disability, in the following section we explain its logic and explore its use in relation to other issues. 1.4.1 List experiments In a list experiment, respondents are provided with a list of items, and they are asked how many of them they agree with. Crucially, they are not asked which of the items they agree with, simply how many. A control group, selected at random, is given a list of three items covering issues other than the one of interest, such as attitudes to education, health and the environment. The treatment group are presented with the same list, but with the addition of a focal item – in this case a potentially sensitive item related to disability. Because both samples are presented with the same control list and are randomly assigned, any difference between the average response of the two groups is due to the additional sensitive item. The technique has been most widely used in political science research on voting and research on racism and attitudes to immigration (Krumpal, 2013; Ehler et al., 2021). There have been two list experiments fielded in Ireland to date. The first investigated attitudes to immigration of Black and Muslim groups (McGinnity et al., 2020). Results showed greater support for Black than Muslim immigration when respondents were asked using standard survey techniques, but no difference when respondents were provided greater anonymity; 15 per cent of individuals were estimated to have concealed a negative attitude towards Black immigration when asked directly. The second investigated compliance with COVID-related _______________________________________________________________________________ 7 Permanent anonymity refers to the fact that researchers can never identify a specific individual’s response because of how the list experiment is designed. This is both a strength and a limitation of the method. For the respondent, it means they are not asked directly about their attitude and hence are not motivated to respond in a socially desirable way. For the researcher, this means that it is not possible to deduce, at an individual level, who was concealing undesirable attitudes; the method relies on comparisons of group differences between direct responses and anonymous responses for this. Note that, in our study, all respondents were anonymous such that no personally identifiable data were collected. Background to the study | 11 public health behaviours during the pandemic (Timmons et al., 2020). The study showed that approximately 10 per cent of participants over-reported their compliance with hand-washing and social distancing advice when asked using standard survey questions compared to when asked using a list. The extent to which participants may conceal their true attitude or behaviour varies according to the issue in question and the participant’s characteristics. McGinnity et al. (2020) observed concealed negative attitudes towards Black immigration was highest among those with a university degree (replicating research in the US; Janus, 2010) and among younger respondents. In addition, women were shown to be more likely to conceal negative attitudes towards Muslim immigration than men, but men were more likely to mask negative attitudes towards Black immigration (McGinnity et al., 2010). Timmons et al. (2020) similarly show gender differences in COVID-19 mitigation behaviours, where men over-reported hand-washing when asked directly compared to the list, but there was no such difference for women. Use of the list experiment technique to gain more accurate survey estimates has been validated against objective benchmarks. List estimates of votes by a sample of voters in an abortion referendum in the US were significantly closer to the real vote count than estimates from standard survey questions (although some underestimation remained; Rosenfeld et al., 2016). Moreover, multiple independent meta-analyses show no evidence for publication bias with list experiments, meaning that list experiments that show social desirability bias are as likely to be published as those that do not, further strengthening the evidence in favour of the method (Blair et al., 2020; Ehler et al., 2021; Li and van den Noortgate, 2022; Rosenfeld et al., 2016). It is likely to be a more reliable way to estimate social desirability than other methods used to date for disability issues (e.g. the IAT). The technique specifically targets social desirability, has a lower nonresponse rate than other methods and is less cognitively demanding for participants than the Randomised Response Technique (Rosenfeld et al., 2016). That said, the robustness of the method depends on certain design considerations. Control lists should be sufficiently long to avoid participants suspecting the focus of the experiment but short enough to limit cognitive demand (Blair et al., 2019). Lists should also be designed to preserve anonymity; any participant who responds with the minimum or maximum response has revealed their responses. As such, lists should be constructed such that at least some control items have negative correlations or that one item is expected to generate high levels of endorsement and another low levels. There are also statistical tests that are required to determine whether the presence of the sensitive items alters the pattern of responding to the control items (known simply as a ‘design effect’; Blair and Imai, 2012). Experiments should also be designed such that participants do not suspect the focus on the sensitive item (Chuang et al., 2021). For example, McGinnity et al. 18 | Experimental tests of public support for disability policy Responses varied from 0 to 4. Because both samples are presented with the same control list items, any difference between the average response to the control and treatment is due to the additional (sensitive) item. At the group level, simply subtracting the average response to the control from the average response to the treatment offers a way to ascertain support for the sensitive item among those in the treatment group (i.e. in Figure 2.1A, prioritising disabled people for social housing). An additional step, which is taken in this experiment, is to ask the control group to directly express their support for sensitive items (e.g. prioritising disabled people for social housing) via standard survey questions after the list experiment. The difference between directly expressed support on these questions and the estimate of anonymously expressed support is interpretable as a measure of the extent to which support for these issues is overor under-stated (see Figure 2.1A). The key to the success of the list experiment is that respondents in the list treatment are never asked to articulate support or report on behaviour regarding any specific item in the list, which guarantees permanent anonymity from the interviewer at the individual level.13 FIGURE 2.1A EXAMPLE LIST EXPERIMENT DESIGN (SOCIAL HOUSING) Source: Authors. Note: Because the control items are the same for both groups, any difference in average responses can be attributed to support for the sensitive item. For example, if the average response for the control group is 1.5 items and the average response for the list group is 2.3 items, 80 per cent anonymously endorsed the sensitive item in the list group (2.3 – 1.5 = 0.8). In our study, the software assigned participants at random to the treatment group (hereafter the ‘list’ group; n = 1,248) or the control group (hereafter, the ‘direct question’ group; n = 752). We pre-registered a randomisation ratio of 5:3 in favour of the list treatment, due to statistical power requirements of list experiments. Participants remained in the same treatment for each list. All participants saw four lists of items and were asked how many on each list they agreed with or applied to them. They selected a response from a drop-down menu. The first two lists contained policy statements and the last two contained behaviours (see Figures 2.1a to 2.1d). For participants in the list treatment, both policy lists and one _______________________________________________________________________________ 13 Under conditions of permanent anonymity the person’s opinion is not recorded. The survey interviewer never knows which of the items on the list the respondent supports. Data collection and experimental design | 19 of the behaviour lists contained a sensitive item about disability issues: welfare payments, social housing and disabled parking,14 respectively. The control group were asked directly about their support for prioritising disabled people for social housing and increasing welfare payments for people with a disability, as well as whether they have ever parked in a disabled space at the end of the survey (see Figure 2.1a). The other behaviour list, about frequency of exercise, was a filler list designed to prevent participants from suspecting the focus of the study was disability issues and altering their responses as a result. FIGURE 2.1B WELFARE PAYMENTS LIST EXPERIMENT Source: Authors. FIGURE 2.1C ‘FILLER’ LIST EXPERIMENT Source: Authors. Note: This list was not of analytic interest and did not have a corresponding direct question. _______________________________________________________________________________ 14 While ‘accessible parking’ is the preferred term of the research team, ‘disabled parking’ was used in the survey as it was felt that this term would be more familiar to the general population. 20 | Experimental tests of public support for disability policy FIGURE 2.1D PARKING LIST EXPERIMENT Source: Authors. Items on lists were presented in randomised order. To minimise the potential for maximum or minimum responses to the lists, which would invalidate the anonymity lists afford, the control items were constructed such that we expected one item to generate high levels of agreement, one to generate low levels of agreement and one to be supported by approximately half of participants, following best practice in the design of list experiments (Blair and Imai, 2012). 2.3.2 Policy trade-offs After completing the vignettes (reported separately in Timmons et al., forthcoming), participants saw four policy statements and were asked whether they agreed with each one (‘Yes’, ‘No’ or ‘Don’t Know’). The four issues were selected from a wider set of eight, four of which concerned disability issues (wheelchair accessible infrastructure, cost of living supports, supports for children with disabilities and employment) and four concerned other issues (refugees, parental leave, the environment and further education). Each participant saw two disability and two non-disability issues, selected at random. The focus of this report is on the disability-related items. These are presented in Table 2.2. Data collection and experimental design | 21 TABLE 2.2 POLICY STATEMENTS Version 1 Version 2 Version 3 A. Cost of Living Control: More should be done to support disabled people in meeting the extra costs of living related to having a disability. Budget: More of the Government’s budget should be allocated to helping disabled people in meeting the extra costs of living related to having a disability. Tax: A tax increase should be used to put more money towards supporting disabled people in meeting the extra costs of living related to having a disability. B. Supports for Children Control: Children with disabilities should get the supports they need. Budget: More of the Government’s budget should be allocated to making sure children with disabilities should get the supports they need. Tax: A tax increase should be used to put more money towards making sure children with disabilities should get the supports they need. C. Wheelchair accessible infrastructure Control: Local Councils should prioritise building more wheelchair accessible infrastructure. Parking: Local Councils should prioritise building more wheelchair accessible infrastructure, even if parking infrastructure needs to be removed to do so. Cycle: Local Councils should prioritise building more wheelchair accessible infrastructure instead of cycling infrastructure. D. Employment Control: Disabled people should work, in jobs which they are capable of doing. Support: Disabled people should be supported in working, in jobs which they are capable of doing. Incentive: Disabled people should be incentivised to work, in jobs which they are capable of doing. Sources: Authors. The idea of these policy questions was to elicit depth of support for progressive disability policies, by varying whether the potential funding mechanisms or tradeoffs of policies were made explicit and the nature of the policy proposed. It is wellestablished that people are often ‘rationally inattentive’, and hence are unlikely to spontaneously consider the wider implications of policies where they are implicit (e.g. Sims, 2003). For example, the public may support a policy that proposes to increase welfare payments for disabled people in principle, but support may be weakened when attention is drawn to the costs of such a policy (see also discussion in Section 1.5). For each issue, we constructed three versions and participants were assigned with balanced randomisation by the software to see one version. Table 2.2 shows the exact wording of each statement and their variants. For three of the disability issues, the versions varied by a funding mechanism or trade-off that was made explicit in the question. For the fourth (about employment), the variants presented slight variations of the nature of the statement. The nondisability issues were again designed to reduce the potential for participants suspecting the main focus of the study. Policies were presented two per page in randomised order, with the constraint that two disability issues were not presented on the same page. Participants finished the study by completing questions about background characteristics, including age, gender, educational attainment, employment status, and living area, the results of which are shown in Table 2.1, as well as whether they themselves had a disability or whether they knew anyone with a disability. List experiment results | 23 CHAPTER 3 List experiment results In this chapter we first present tests of list experiment design assumptions, including analysis of fully anonymised response rates and the presence of design effects (Blair and Imai, 2012). To compare list endorsement of sensitive items to direct question responses, we use the item-count technique with Welch’s t-tests to account for unequal variances between list and control treatments (Tsai, 2019). We use logistic regression models to test for differences between sociodemographic subgroups. We test for differences by gender, age, living location (urban or rural) and socio-economic indicators. Socio-economic indicators were educational attainment and ‘social grade’, a classification system based on the occupation of the chief income earner in the household. We compare those in households where the chief income earner is in a higher, intermediate or junior managerial/professional role (‘ABC1’) with ones where the chief income earner is in a manual or casual work role or unemployed (‘C2DE’). We use Student’s t-tests and tests of proportions for follow-up tests where models suggest significant differences. We also compare responses to direct questions to estimates from the most recent NDA survey on disability attitudes using tests of proportions (NDA, 2017). 3.1 TESTING DESIGN EFFECTS List experiments need to be designed such that the presence of the sensitive item does not alter how participants respond to other items in the list. We tested for design effects on each list using the kict package in Stata (Blair and Imai, 2012; Tsai, 2019). Results for each list showed suitable joint distributions and no indication of design effects (Tables A.1-A.3 in the Appendix). Response distributions to list experiments should also show few participants answering with the minimum or maximum response and thereby revealing their opinion. Figure A.1 in the Appendix shows no evidence for problematic ceiling or floor effects. Lastly, we tested for any differences between the groups on socio-demographic characteristics, which should be prevented by randomisation. Chi-square tests showed that there was no difference in allocation of different socio-demographic groups to the list or direction condition, as signalled by the p value being much greater than 0.05 (Table 3.1). This implies that the randomisation was effective and there were no significant differences in these groups in terms of these sociodemographic characteristics. 24 | Experimental tests of public support for disability policy TABLE 3.1 CHI-SQUARE TESTS OF LIST EXPERIMENT RANDOMISATION χ² p Gender 0.19 .660 Age 0.87 .647 Educational Attainment 0.03 .866 Working Status 0.16 .925 Socio-Economic Grade 0.24 .888 Living Area 1.11 .292 Disability Status 2.27 .132 Source: Authors’ analysis. 3.2 INCREASED WELFARE PAYMENTS When asked directly, 77 per cent of participants endorsed increased welfare payments for disabled people. This figure is the same as the level of support in response to the same question, estimated from the 2017 NDA survey on disability attitudes (N = 1,294), (77 per cent) Z = 0.00, p = .998 (see Figure 3.1). Estimates of support from the list responses were lower (66 per cent), t (1,620.69) = 2.70, p = .007. This difference implies that 14 per cent (11 percentage points) of people who endorse increasing welfare payments for disabled people may do so only to present themselves in a positive light. They do not support increasing welfare payments when provided anonymity in the list experiment (Figure 3.1). Table 3.2a presents model coefficients from logistic regressions predicting support for increasing welfare payments for different socio-demographic groups of participants. The table also shows results from a non-linear least-squares estimation of support within the list treatment (see Blair and Imai, 2012; Tsai, 2019). Table 3.2a shows that, compared to those without degrees, participants with higher educational attainment were less likely to endorse increased welfare payments for disabled people when asked in the list condition,15 but there was no evidence for a difference when both groups were asked directly.16 _______________________________________________________________________________ 15 Direct t-test: t (1237.76) = 1.99, p = .047. 16 Test of proportions: Z = 0.74, p = .458. List experiment results | 25 TABLE 3.2A LOGISTIC REGRESSION MODELS PREDICTING DIRECT RESPONSES AND LIST RESPONSES TO INCREASING WELFARE PAYMENTS Increase Welfare Payments Direct List Man (Ref: Woman) 0.11 (0.17) 0.01 (0.35) Aged 45+ years (Ref: <45 years) 0.12 (0.18) -0.36 (0.36) Degree (Ref: Less than Degree) 0.00 (0.20) -0.86** (0.40) ABC1 Social Grade (Ref: C2DEF) -0.27 (0.19) 0.38 (0.39) Urban (Ref: Rural) 0.00 (0.18) -0.36 (0.37) Constant 1.22 1.27 N 746 2,000 Sources: Authors’ analysis and NDA (2017). Note: *p < .10, ** p < .05, *** p < .01. Standard errors are in parentheses. The sample size for the Direct model varies from the control condition sample size because subgroups in which each individual responded in the same way are excluded. The sample size for the List model is the full sample because the model requires responses from both the control list and the treatment list to estimate the proportion who endorsed the target item within each subgroup. To test for educational attainment differences in social desirability bias, we repeated the item-count analysis for both groups. Those with degrees displayed a social desirability bias, with a higher percentage endorsing an increase when asked directly (75.6 per cent) compared to when asked in the list treatment (58.8 per cent), t (799.04) = 3.76, p < .001. The difference for those with lower educational attainment was in the same direction but not statistically significant (78.0 per cent vs. 71.1 per cent, respectively), t (949.51) = 1.30, p = .195. Hence, the results imply that individuals with higher educational attainment are less likely to endorse an increase in welfare payments for people with a disability, and that this difference may not be detected using standard survey methods. 3.3 HOUSING Support for prioritising disabled people for social housing showed a large decline in 2022 compared to 2017, when participants were asked directly (61.4 per cent in 2022 vs. 78.0 per cent in 2017), Z = 8.04, p < .001 (see Figure 3.1). The list condition showed higher levels of support compared to the direct condition (70.7 per cent), t (1,619.65) = 1.96, p = .050, but still marginally lower than in 2017, t (1,247.13) = 1.68, p = .094. Note that, contrary to our predictions, the direction of the difference between the direct and list treatment implies that prioritising disabled people on social housing waiting lists is judged as less socially desirable than opposing it. Table 3.2b shows that lower social grade was a predictor of direct support for housing prioritisation. Analysis of responses within these groups showed that while 26 | Experimental tests of public support for disability policy 66.5 per cent of those in the C2DE (skilled and unskilled manual workers, nonemployed) social grades supported prioritising disabled people for social housing, just 56.9 per cent of those in the ABC1 (professional, managerial, administrative) grades did.17 However, the list model in Table 3.2b shows no difference between social grades in the list treatment.18 We investigated the relationship between social grade and social desirability further by repeating the item-count analysis by subgroups. Those in the C2DE social grades did not display a list effect, with 72.0 per cent supporting in the list condition, t (768.20) = 0.78, p = .437, but those in the ABC1 grades showed higher support in the list condition (69.6 per cent), t (848.47) = 1.98, p = .047. Hence, the results imply that standard survey measures may indicate a difference in support for housing prioritisation depending on social grade that does not exist when respondents are offered greater anonymity. Importantly, social desirability may lead to lower support for disabled people among groups in higher social grades. One possibility is that, during a housing crisis where there are high levels of competition for limited social housing, some of those in higher social grades perceive prioritising any group to be socially undesirable. TABLE 3.2B LOGISTIC REGRESSION MODELS PREDICTING DIRECT RESPONSES AND LIST RESPONSES TO PRIORITISING DISABLED PEOPLE FOR SOCIAL HOUSING Housing Priority Direct List Man (Ref: Woman) 0.20 (0.15) 0.25 (0.44) Aged 45+ years (Ref: <45 years) -0.03 (0.15) 0.08 (0.44) Degree (Ref: Less than Degree) 0.21 (0.17) -0.05 (0.48) ABC1 Social Grade (Ref: C2DEF) -0.52*** (0.17) -0.09 (0.47) Urban (Ref: Rural) 0.13 (0.16) 0.18 (0.45) Constant 0.49 0.73 N 751 2,000 Sources: Authors’ analysis. Note: *p < .10, ** p < .05, *** p < .01. Standard errors are in parentheses. The sample size for the Direct model varies from the control condition sample size because subgroups in which each individual responded in the same way are excluded. The sample size for the List model is the full sample because the model requires responses from both the control list and the treatment list to estimate the proportion who endorsed the target item within each subgroup. 3.4 PARKING In the 2017 NDA survey, just 2 per cent of respondents reported that they judge parking in a disabled parking space without a permit is acceptable. As we were interested in behaviour rather than attitudes for this issue, in the 2022 survey _______________________________________________________________________________ 17 Test of proportions: Z = 2.69, p = .007. 18 T-test: t (1246) = 0.26, p = .798. List experiment results | 27 respondents were asked whether they had parked in a disabled parking space without a permit. For the analysis, we include only those who later in the survey reported that they drive (n = 1,674). When asked directly in our study, 4 per cent of drivers admit having done so in the past and this difference is statistically significant, Z = 2.57, p = .010 (see Figure 3.1). The estimate from responses to the list items (4.8 per cent) is not statistically different to the proportion who admitted parking in a disabled space without a permit to direct question, t (1162.94) = 0.22, p = .830. The number of people who reported parking in a disabled space without a permit is too low to permit analysis of differences by socio-demographic subgroups. FIGURE 3.1 PERCENTAGE OF PARTICIPANTS ENDORSING SENSITIVE ITEMS IN THE THREE LIST EXPERIMENTS Sources: Authors’ analysis and NDA (2017). Note: Error bars are the standard error of the proportion. *The parking question contains only those who reported driving (n = 1,674). The comparison with NDA (2017) is to a question about the acceptability of parking in a disabled space without a permit rather than a behaviour question. 3.5 FAMILIARITY WITH DISABILITY We were interested in the extent to which contact with disabled people or experience with disability is associated with attitudes to the policy support. We pre-registered exploratory analyses of the association between knowing someone with a disability or having a disability oneself and support for disability policy.19 We ran further logistic regression models predicting policy endorsement using an ordinal variable for familiarity. Participants were asked at the end of the study whether they or any of a list of people they knew had a disability or long lasting condition that affects their ability to carry out day-to-day activities.20 They were _______________________________________________________________________________ 19 For this analysis also, the number of people who reported parking in a disabled spot without a permit is too low for subgroup comparisons. 20 The response options were Spouse/Partner, Child, Parent, Brother/Sister/Other relative, Friend, Neighbour, Colleague/ work contact, Not sure/don’t know, None. Participants could select as many as applied to them. 77 78 2 77 61.4 4 66 70.7 4.8 0 10 20 30 40 50 60 70 80 90 Increase welfare payments for disabled people Prioritise disabled people for social housing I have parked in a disabled spot without a permit* Policy Attitudes Behaviour 2017 2022 - Direct 2022 - List 34 | Experimental tests of public support for disability policy TABLE 4.2 LOGISTIC REGRESSION MODELS PREDICTING SUPPORT FOR INCREASED CHILD SUPPORTS Table 4.2 Logistic Regression Models Predicting Support for Increased Child Supports Controla (v1) Budget (v2) Tax (v3) Male (Ref: Female) -0.11 (0.84) -0.57* (0.32) 0.11 (0.23) Age (Ref: 18-39 years) 40-59 years -0.59 (0.95) 0.85** (0.38) 0.06 (0.28) 60+ years 0.11 (1.27) 0.70* (0.41) 0.37 (0.30) Degree (Ref: No Degree) -0.99 (0.94) 0.10 (0.36) 0.09 (0.26) Social Grade (Ref: DEF) C1C2 -0.35 (1.19) -0.09 (0.40) -0.02 (0.28) AB -0.44 (1.31) -0.34 (0.48) 0.41 (0.35) Urban (Ref: Rural) - -0.01 (0.35) -0.30 (0.25) Intercept 5.12*** (1.33) 1.67*** (0.43) 0.45 (0.35) N Yes Responses 329 284 212 Total N 334 333 334 Source: Authors’ analysis. Note: *p < .10, ** p < .05, *** p < .01. Standard errors are in parentheses. a. All except for six of the 334 participants in this model supported the policy, all of whom lived in an urban area. Hence, living area is not included as a predictor in this model. 4.3 BUILDING WHEELCHAIR ACCESSIBLE INFRASTRUCTURE Support for policy C on building more wheelchair accessible infrastructure also varied across conditions, χ² (4) = 34.82, p < .001. Support was strongest when no trade-off was made explicit (Version 1) but declined when the proposal required removing parking infrastructure, (Version 2) Z = 2.36, p = .018, or for it to be built at the expense of cycling infrastructure, (Version 3) Z = 5.15, p < .001. Support was stronger if parking spaces were to be removed than if cycling infrastructure would not be built, (Version 2 vs. Version 3) Z = 2.85, p = .004. Policy trade-off results | 35 FIGURE 4.3 SUPPORT FOR BUILDING WHEELCHAIR ACCESSIBLE INFRASTRUCTURE Source: Authors’ analysis. Across all models of support (Table 4.3), men were less in favour than women (71 per cent vs. 81 per cent) and older participants were more supportive of wheelchair accessible infrastructure than younger participants (83 per cent of over 60s, 77 per cent of 40-59 year olds and 71 per cent of under 40s). Higher socioeconomic grades were less supportive than the lowest grade groups when tradeoffs were made explicit (71 per cent of the AB group and 69 per cent of the C1C2 group vs. 81 per cent of the DEF group), whereas there was no difference when the trade-off was not explicit (86 per cent of the AB group and 82 per cent of the C1C2 group and 86 per cent of the DEF group). 84.15 76.88 66.96 4.88 12.61 18.45 10.98 10.51 14.58 0 10 20 30 40 50 60 70 80 90 100 No Explicit Trade-Off (v1) By removing parking infrastructure (v2) Instead of cycling infrastructure (v3) C. Local councils should build more wheelchair infrastructure Yes No Don't Know 36 | Experimental tests of public support for disability policy TABLE 4.3 LOGISTIC REGRESSION MODEL PREDICTING SUPPORT FOR BUILDING WHEELCHAIR ACCESSIBLE INFRASTRUCTURE Wheelchair accessible infrastructure Control (v1) Parking (v2) Cycle (v3) Male (Ref: Female) -0.84*** (0.32) -0.34 (0.27) -0.51** (0.25) Age (Ref: 18-39 years) 40-59 years 0.10 (0.36) 0.47 (0.32) 0.47* (0.28) 60+ years 1.09** (0.51) 0.32 (0.35) 0.82** (0.33) Degree (Ref: No Degree) -0.40 (0.35) 0.44 (0.30) -0.44 (0.27) Social Grade (Ref: DEF) C1C2 -0.17 (0.40) -0.83** (0.38) -0.38 (0.33) AB 0.30 (0.50) -0.57 (0.44) -0.42 (0.39) Urban (Ref: Rural) 0.16 (0.33) 0.06 (0.27) 0.44* (0.26) Intercept 1.97*** (0.52) 1.54*** (0.42) 0.87** (0.38) N Yes Responses 276 256 225 Total N 327 332 336 Source: Authors’ analysis. Note: *p < .10, ** p < .05, *** p < .01. Standard errors are in parentheses. 4.4 EMPLOYMENT Unlike the previous policies, the versions for the final policy frame (policy D) did not vary by whether potential costs or funding mechanisms were made explicit. Instead, the policies, which aimed at encouraging disabled people to work in jobs they are capable of doing varied depending on how that encouragement was phrased. Version 1 assessed a general belief that people with disabilities should work (‘Disabled people should work, in jobs which they are capable of doing’), whereas Version 2 described disabled people being supported to work (‘Disabled people should be supported in working, in jobs which they are capable of doing’), and Version 3 implied that working should be motivated by financial incentives (‘Disabled people should be incentivised to work, in jobs which they are capable of doing’). Results showed an overall effect of policy version, χ² (4) = 35.78, p < .001. When framed with general support in Version 2, more people endorsed the policy than when asked about their general belief in Version 1, Z = 4.41, p < .001, or when support was implied to be purely financial in Version 3, Z = 5.75, p < .001. There was no difference between Version 1 and Version 3, Z = 1.53, p = .126. Policy trade-off results | 37 FIGURE 4.4 BELIEFS THAT PEOPLE WITH DISABILITIES SHOULD WORK Source: Authors’ analysis. Socio-demographic differences in endorsing policies aimed at encouraging disabled people to work showed similar differences by age, social grade and disability status across each version of the statement (See Table 4.4). Those aged over 60 more strongly endorsed the policies than younger respondents (96 per cent vs. 88 per cent of 40-59 year olds and 87 per cent of under 40s) and those in the AB and C1C2 groups more strongly endorsed the policies than those in the DEF group (92 per cent of both groups vs. 83 per cent, respectively). Younger respondents and those in the DEF group show response patterns most similar to those with a disability, where 82 per cent endorsed the policies compared to 92 per cent of those without a disability. 88.1 97.02 83.99 2.08 0.6 5.14 9.82 2.38 10.88 0 10 20 30 40 50 60 70 80 90 100 Control (v1) Supported (v2) Incentivised (v3) D. Should work in jobs they are capable of Yes No Don't Know 38 | Experimental tests of public support for disability policy TABLE 4.4 LOGISTIC REGRESSION MODELS PREDICTING POLICY SUPPORT FROM SOCIODEMOGRAPHIC CHARACTERISTICS Employment Control (v1) Support (v2) Incentive (v3) Male (Ref: Female) -0.62* (0.35) -0.26 (0.67) 0.03 (0.33) Age (Ref: 18-39 years) 40-59 years 0.11 (0.38) 1.14 (0.853) 0.12 (0.35) 60+ years 1.31** (0.54) 1.59 (1.10) 1.85*** (0.58) Degree (Ref: No Degree) 0.28 (0.41) 0.05 (0.71) 0.23 (0.37) Social Grade (Ref: DEF) C1C2 0.42 (0.40) 1.14 (0.71) 1.39*** (0.37) AB 0.41 (0.53) 1.58 (1.16) 0.79* (0.48) Urban (Ref: Rural) 0.50 (0.35) 0.10 (0.69) 0.08 (0.33) Intercept 1.31** (0.48) 2.11*** (0.74) 0.48 (0.37) N Yes Responses 296 326 278 Total N 333 335 330 Source: Authors’ analysis. Note: *p < .10, ** p < .05, *** p < .01. Standard errors are in parentheses. 4.5 FAMILIARITY WITH DISABILITY As before, we conducted exploratory analyses of the association between familiarity with disability and support. Table 4.5 presents logistic regression models using the same familiarity variable as before on whether the participant endorsed the policy. The models include controls for socio-demographic characteristics and the version of the policy the participant saw. Our focus on this section is on differences between those who know no one with a disability, those whose partner, child or parent has a disability and those who have a disability themselves, as the number of participants in the other groups are too low for reliable estimates but are shown in Table 4.5 for completeness (friend/neighbour/colleague ns = 60 to 67;22 brother/sister/other relative ns = 56 to 70). Results show that participants who themselves have a disability were more likely to endorse more cost-of-living supports (policy A) than those who know no one with a disability (Figure 4.5). Tests of coefficients showed there was no difference between those with a disability themselves and those whose partner, child or parent has a disability, χ² = 1.46, p = .226. The pattern is similar on policy D about whether disabled people should work. Those with a disability and those whose partner, child or parent has a disability were less supportive of statements that _______________________________________________________________________________ 22 Number of observations per policy varies because of randomisation to see two of the four policies. Policy trade-off results | 39 disabled people ‘should work’ than those who know no one with a disability, but there was no statistically significant difference between these two groups, χ² = 0.15, p = .670. This may suggest a gap between disabled people’s attitudes towards and experiences of seeking employment and non-disabled people’s attitudes towards disabled people seeking employment. TABLE 4.5 LOGISTIC REGRESSION MODELS PREDICTING POLICY SUPPORT FROM FAMILIARITY WITH DISABILITY Source: Authors’ analysis. Note: *p < .10; **p < .05; ***p < .01 On policy C about wheelchair accessible infrastructure, those related to someone with a disability were more supportive than those who know no one with a disability and there was no difference between the partner/child/parent group and those with a disability themselves, χ² = 0.33, p = .567. There were no differences on policy B about supporting children with disabilities, reflecting the high level of endorsement across all socio-demographic subgroups for this policy. Looking across the policies (Figure 4.5), there was no evidence for a difference in opinion between those with a disability themselves and those closest to someone with a disability (i.e. a partner, child or parent has a disability). A. Cost of Living B. Support for Children C. Wheelchair accessible infrastructure D. Employment Familiarity with Disability (Ref: Knows No One) Friend/Neighbour/Colleague -0.01 (0.34) 0.01 (0.36) 0.08 (0.33) 0.33 (0.63) Brother/Sister/Other Relative -0.28 (0.34) 0.17 (0.36) 0.91** (0.42) 0.39 (0.56) Spouse/Partner/Child/Parent 0.29 (0.23) 0.05 (0.29) 0.40* (0.24) -0.70** (0.32) Has a Disability 0.64*** (0.23) 0.33 (0.26) 0.24 (0.22) -0.83*** (0.26) Socio-Demographic Controls Yes Yes Yes Yes Constant 2.18 (0.31) 3.89 (0.49) 1.73 (0.29) 2.79 (0.43) N 997 1,003 997 1,001 40 | Experimental tests of public support for disability policy FIGURE 4.5 SUPPORT FOR POLICIES BY FAMILIARITY WITH DISABILITY Source: Authors’ analysis. 68.1 81.3 71.9 92.1 76.3 82.7 81.1 85.8 78.7 85.6 79.8 82.0 0 10 20 30 40 50 60 70 80 90 100 A. More support for extra cost of living B. More support for children with disabilities C. Local councils should build more wheelchair infrastructure D. Should work in jobs they are capable of % participants Knows No One Partner/Child/Parent Has a Disability Discussion and implications | 41 CHAPTER 5 Discussion and implications The majority of people in Ireland support most policies that aim to enable people with disabilities to participate fully in society, even when provided complete anonymity in answering the question or when funding mechanisms of policies are made explicit. This support is observed across cost-of-living, access to housing, and day-to-day accessibility policies, and is consistent across socio-demographic subgroups. However, the degree of support can vary depending on how surveys are implemented and the level of detail provided in survey questions. In this chapter, we summarise the main results from both stages of the study, make note of some limitations and highlight the policy implications from the findings. 5.1 SUPPORT FOR SOCIAL WELFARE The additional cost of having a disability is estimated to be €9,482 to €11,734 per year (Department of Social Protection, 2021). Our results suggest that, although they may not be aware of this figure, most people in Ireland acknowledge the financial challenges of living with a disability and are supportive of stronger government assistance. When asked using standard survey techniques, 77 per cent of people agree that welfare payments for disabled people should be increased and most (91 per cent) agree that more should be done to support disabled people to meet their extra cost of living. However, support varies as a function of survey anonymity and whether funding mechanisms are included in the survey item. When endorsement for increased welfare payments is elicited using a list technique, whereby respondents give their views under conditions of permanent anonymity, support is 14 per cent lower (11 percentage points) at 66 per cent. This difference means that one-in-seven who would have supported the policy if asked directly do not support the policy when offered greater anonymity. The difference in support for extra cost-of-living assistance is similar (15 per cent lower) when the policy is proposed to be funded through budgetary reallocation compared to when no detail is given on the funding mechanism. The alternative to budget reallocation, a tax increase, leads to a more drastic difference in support (54 per cent lower than when no funding mechanism is specified), with only 42 per cent in favour of more support in this group. This variation in support for financial assistance for disabled people depending on how questions are asked is greater among those with higher educational attainment. Results from the list experiment showed that, when provided permanent anonymity, support among those educated to degree level or above was 22 per cent lower. Support among those educated below degree level was just 9 per cent lower and was not statistically significantly different from support when asked directly. Similarly, statistical models showed that the difference in support 42 | Experimental tests of public support for disability policy for extra cost-of-living assistance through budget reallocation compared to when no mechanism was specified was more pronounced among those educated to degree level or above (20 per cent lower) than those without degrees (9 per cent). Importantly, standard survey techniques may overstate support among those with higher educational qualifications for socially desirable measures (Janus, 2010; McGinnity et al., 2020). In this study, the direct question on increased welfare payments failed to detect differences by educational attainment, whereas results from the list experiment found those with higher educational attainment to be less supportive of increased welfare payments. The control condition of the policy statement experiment, which would be a more typical formulation, suggested those with higher educational attainment are more supportive of assistance with the cost of living than their peers with lower educational attainment, whereas when presented the budget allocation trade-off the groups did not differ. These differences by educational attainment may reflect a greater tendency for those on higher incomes to oppose redistributive fiscal policies in general (Müller and Regan, 2021). Children with disabilities are especially reliant on additional support in order to achieve full inclusion (e.g. Carroll et al., 2022a), and almost everyone (98 per cent) across all socio-demographic subgroups endorses such supports when funding mechanisms are not specified. As with general cost-of-living support, however, support is lower when funding mechanisms are included in survey items, with larger differences for tax increases (35 per cent lower) than for budget reallocation (13 per cent lower). It is worth noting that the proportion in favour of a tax increase to help children with disabilities (63.5 per cent) is much larger than the level of support for tax increases for other social issues that have high levels of in-principle support, such as climate change (40-47 per cent; Leahy, 2021; Timmons and Lunn, 2022) and a United Ireland (22 per cent; Sheahan, 2021). 5.2 ACCESSIBILITY AND INFRASTRUCTURE Day-to-day accessibility issues further contribute to challenges to the rights of disabled people, including, but not limited to, access to suitable parking and insufficient infrastructure. Previous surveys show that a very small minority (2 per cent) judge parking in a disabled parking space without a permit to be acceptable (NDA, 2017). However, our findings show that a small but significantly higher proportion of drivers report having done so in the past (4 per cent). When provided permanent anonymity, the figure is 4.8 per cent (20 per cent higher), although the difference is not statistically significant. Note that, in order to detect a statistically significant difference for behaviours with very low incidence and therefore a lower absolute bias-percentage, a sample in excess of 10,000 would have been required (Blair et al., 2020). Nonetheless, a true incidence rate of even 4 per cent implies that one-in-25 drivers have parked in disabled parking spaces without a permit. If these drivers are ‘repeat offenders’, this may lead to disproportionately high encounters of such behaviour among disabled people. This estimate does not Discussion and implications | 43 include others who may not park directly in disabled parking spaces but may impede access to them in other ways. We are careful to note also, however, that there may be legitimate use of disabled parking spaces included in this estimate (e.g. taking a relative to an appointment and forgetting the permit or requiring a permit but not meeting technical requirements).23 Turning to infrastructure, the majority (84 per cent) support the proposition that local councils should build more wheelchair accessible infrastructure, with support highest among older people. As with the cost-of-living supports, the proportion who support the policy is lower when survey questions highlight a potential cost or trade-off. For this policy, the potential trade-off was one of alternative prioritisation, where wheelchair accessible infrastructure came with the removal of driving infrastructure (parking spaces) or instead of active transport infrastructure (cycle lanes). Support is 7 per cent lower in the condition where parking infrastructure would be removed and even lower (20 per cent lower) in the condition where cycling infrastructure would be de-prioritised. Note however that removing parking infrastructure and not building cycling infrastructure are not necessary trade-offs to implementing wheelchair infrastructure (e.g. some may believe the infrastructure should be installed but without cost to parking or cycling infrastructure), but are merely illustrative of the change in support when tradeoffs are specified. 5.3 OTHER SOCIAL ISSUES Offering permanent anonymity when asked about social welfare increases altered responses in line with the hypothesis that some support for disability policy is due to social desirability bias. However, the list experiment about housing priority suggested the opposite. Respondents were more supportive of prioritising disabled people for social housing than when they were asked directly about their opinion. This finding may suggest that people view the socially acceptable response to be one where no one group is ‘prioritised’ during a period of housing shortages and affordability challenges. When provided more anonymity, they reveal preferences held towards some groups (e.g. disabled people). Alternatively, or in addition, the list technique may be detecting negative sentiment towards other groups that compete for limited social housing supply (e.g. refugees). It is not possible to identify what sentiment drives the reverse-list effect observed here, but this suggestion is in line with majority support for disability policies and other evidence for concealed negative attitudes towards immigration (McGinnity et al., 2020). 5.4 FAMILIARITY WITH DISABILITY The results also reveal striking differences between those most familiar with disabled people and disability issues (i.e. people with disabilities themselves and their partners/children/parents) and those who report not knowing anyone with a _______________________________________________________________________________ 23 We thank the advisory board for the study for making us aware of both of these limitations. 50 | Experimental tests of public support for disability policy Brenner, P.S. (2011). ‘Exceptional Behaviour or Exceptional Identity: Overreporting of Church Attendance in the US’, Public Opinion Quarterly 75(1): 19-41. Brenner, P.S. and J.D. DeLamater (2016). ‘Social Desirability Bias in Self-reports of Physical Activity: Is Exercise Identity the Culprit?’, Social Indicators Research 117(2): 489-504. Burke, J., J. Bezyak, R.T. Fraser, J. Pete, N. Ditchman and F. Chan (2013). ‘Employers’ attitudes towards hiring and retaining people with disabilities: A review of the literature’, The Australian Journal of Rehabilitation Counselling, 19(1), 21-38. Cant, J. and M. Bennet (2022). ‘Disabled People at Work’, in Butt, S., E. Clery and J. Curtice (eds.) British Social Attitudes: The 39th Report. London: National Centre for Social Research. Carroll, E., S. McCoy and G. Mihut (2022). ‘Exploring Cumulative Disadvantage in Early School Leaving and Planned Post-School Pathways among Those Identified with Special Educational Needs in Irish Primary Schools’, British Educational Research Journal 00: 1-18. https://doi.org/10.1002/berj.3815. Carroll, E., K. Ye and S. McCoy (2022). ‘Educationally maintained inequality? The role of risk factors and resilience at 9, 13 and 17 in disabled young people’s post-school pathways at 20’, Irish Educational Studies, 41:3, 573-594, DOI: 10.1080/03323315.2022.2093257. Chen, C., B. MacInnis, M. Waltman and J.A. Krosnick (2021). ‘Public opinion on climate change in the USA: to what extent can it be nudged by questionnaire design features?’, Climatic Change, 167(3), 1-18. https://doi.org/10.1007/s10584-021-03194-x. Chuang, E., P. Dupas, E. Huillery and J. Seban (2021). ‘Sex, lies, and measurement: Consistency tests for indirect response survey methods’, Journal of Development Economics, 148, 102582. DOI: https://doi.org/10.1016/j.jdeveco.2020.102582. Coutts, E., B. Jann, I. Krumpal and A. Näher (2011). ‘Plagiarism in Student Papers: Prevalence Estimates Using Special Techniques for Sensitive Questions’, Jahrbücher für Nationalökonomie und Statistik, Vol. 231, no. 5-6, pp. 749760. https://doi.org/10.1515/jbnst-2011-5-612. Creighton, M.J. and A. Jamal (2015). ‘Does Islam play a role in anti-immigrant sentiment? An experimental approach’, Soc Sci Res.53, 89-103. DOI: 10.1016/j.ssresearch.2015.04.001. Creighton, M.J., A. Jamal and N. Malancu (2015). ‘Has opposition to immigration increased in the United States after the Economic Crisis? An Experimental Approach’, International Migration Review, Vol. 49 (3), 727-756. https://doi.org/10.1111/imre.12091. CSO (2017). ‘Census of Population 2016 – Profile 9 Health, Disability and Carers’, retrieved from https://www.cso.ie/en/releasesandpublications/ep/pcp9hdc/p8hdc/p9d/. Cullinan, J., B. Gannon and S. Lyons (2011). ‘Estimating the extra cost of living for people with disabilities’, Health economics, 20(5), 582-599. https://doi.org/10.1002/hec.1619. References | 51 Department of Social Protection (2021). The Cost of Disability in Ireland. Retrieved from https://www.gov.ie/en/publication/1d84e-the-cost-of-disability-inireland-research-report/. Dixon, S., C. Smith and A. Touchet (2018). The disability perception gap. SCOPE. Druckman, J.N. (2001). ‘The implications of framing effects for citizen competence’, Political behavior, 23(3), 225-256. https://doi.org/10.1023/A:1015006907312. Ehler, I., F. Wolter and J. Junkermann (2021). ‘Sensitive Questions in Surveys: A Comprehensive Meta-Analysis of Experimental Survey Studies on the Performance of the Item Count Technique’, Public Opinion Quarterly, Vol. 85, Issue 1, Spring 2021, pp 6-27, https://doi.org/10.1093/poq/nfab002. Emerson, E., R. Stancliffe, C. Hatton, G. Llewellyn, T. King, V. Totsika, Z. Aitken and A. Kavanagh (2021). ‘The impact of disability on employment and financial security following the outbreak of the 2020 COVID-19 pandemic in the UK’, Journal of Public Health, 43(3), 472-478. DOI: https://doi.org/10.1093/pubmed/fdaa270. Eurofound (2022). People with disabilities and the COVID-19 pandemic: Findings from the Living, working and COVID-19 e-survey, COVID-19 series. Luxembourg: Publications Office of the European Union. Evans, J.B.S.T. (2008). ‘Dual-processing accounts of reasoning, judgment, and social cognition’, Annual Review of Psychology, 59, 255-278. DOI: https://doi.org/10.1146/annurev.psych.59.103006.093629. Feldman, L. and P.S. Hart (2018). ‘Climate change as a polarizing cue: Framing effects on public support for low-carbon energy policies’, Global Environmental Change, 51, 54-66. DOI: https://doi.org/10.1016/j.gloenvcha.2018.05.004. Glynn, A.N. (2013). ‘What Can We Learn with Statistical Truth Serum? Design and Analysis of the List Experiment’, Public Opinion Quarterly, Vol. 77, pp. 159-172. Gollust, S.E., J. Niederdeppe and C.L. Barry (2013). ‘Framing the consequences of childhood obesity to increase public support for obesity prevention policy’, American journal of public health, 103(11), e96-e102. https://doi.org/10.2105/AJPH.2013.301271. Greenwald, A.G. and C.K. Lai (2020). ‘Implicit social cognition’, Annual Review of Psychology, 71, 419-445. DOI: https://doi.org/10.1146/annurev-psych010419-050837. Grotti, R., H. Russell, É. Fahey and B. Maître (2018). Discrimination and inequality in housing in Ireland. Irish Human Rights and Equality Commission. Hurwitz, J. and M. Peffley (2005). ‘Playing the race card in the post-Willie Horton era: The impact of racialized code words on support for punitive crime policy’, Public Opinion Quarterly, 69(1), 99-112. https://doi.org/10.1093/poq/nfi004. 52 | Experimental tests of public support for disability policy Hutton, B. (July 6, 2020). ‘Fixed-charge fines for illegally parking in disabled spaces soar’, Irish Times. https://www.irishtimes.com/news/social-affairs/fixedcharge-fines-for-illegally-parking-in-disabled-spaces-soar-1.3555499. Irish Wheelchair Association (2020). Best Practice Access Guidelines. Janus (2010). ‘The Influence of Social Desirability Pressures on Expressed Immigration Attitudes’, Social Science Quarterly, 91(4), pp. 925-946. Ju, S., E. Roberts and D. Zhang (2013). ‘Employer attitudes toward workers with disabilities: A review of research in the past decade’, Journal of Vocational Rehabilitation, 38(2), 113-123. Kahneman, D. (2011). Thinking, Fast and Slow. New York: Farrar, Strauss, Giroux. Kelly, E. and B. Maître (2021). Identification Of Skills Gaps Among Persons With Disabilities And Their Employment Prospects, ESRI Survey and Statistical Report Series 107, Dublin: ESRI, https://doi.org/10.26504/sustat107. Krosnick, J.A. (1991). ‘Response strategies for coping with the cognitive demands of attitude measures in surveys’, Applied cognitive psychology, 5(3), 213236. DOI: https://doi.org/10.1002/acp.2350050305. Krosnick, J.A. and S. Presser (2010). ‘Question and Questionnaire Design’ in Wright, J. D. and P.V. Marsden (Eds). Handbook of Survey Research. Emerald Group Publishing. Krumpal I. (2013). ‘Determinants of social desirability bias in sensitive surveys: a literature review’, Qual Quant 47, 2025-47. Kuklinski, J.H., M.D. Cobb and M. Gilens (1997). ‘Racial attitudes and the “New South”‘, Journal of Politics, Vol. 59, pp. 323-49. Kunz, A., R. Luder and W. Kassis (2021). ‘Beliefs and attitudes toward inclusion of student teachers and their contact with people with disabilities’, Frontiers in Education 6. Leahy, P. (Oct 8, 2021). ‘Poll shows high degree of public resistance to many potential climate action measures’, Irish Times. https://www.irishtimes.com/news/ireland/irish-news/poll-shows-highdegree-of-public-resistance-to-many-potential-climate-action-measures1.4694287. Li, J. and W. Van den Noortgate (2022). ‘A Meta-analysis of the relative effectiveness of the item count technique compared to direct questioning’, Sociological Methods & Research, 51(2), 760-799. Lopez de Leon, F.L. and R. Rizzi (2014). ‘A test for the rational ignorance hypothesis: Evidence from a natural experiment in Brazil’, American Economic Journal: Economic Policy, 6(4), 380-98. DOI: 10.1257/pol.6.4.380. Lüke, T. and M. Grosche (2018). ‘What do I think about inclusive education? It depends on who is asking. Experimental evidence for a social desirability bias in attitudes towards inclusion’, International Journal of Inclusive Education, 22:1, 38-53, DOI: 10.1080/13603116.2017.1348548. References | 53 Martinelli, C. (2007). ‘Rational ignorance and voting behavior’, International Journal of Game Theory, 35(3), 315-335. https://doi.org/10.1007/s00182006-0051-4. McCaffery, E.J. and J. Baron (2004). ‘Framing and taxation: Evaluation of tax policies involving household composition’, Journal of Economic Psychology, 25(6), 679-705. https://doi.org/10.1016/j.joep.2003.07.001. McConkey, R., P. Slater, A. Smith, L. Dubois and A. Shellard (2021). ‘Perceptions of the rights and capabilities of people with intellectual disability in the United States’, J Appl Res Intellect Disabil. 34: 537-545. https://doi.org/10.1111/jar.12819. McGinnity, F., D. Watson and G. Kingston (2012). Analysing the Experience of Discrimination in Ireland: Evidence from the QNHS Equality Module 2010, Dublin: The Equality Authority and the ESRI, https://www.esri.ie/publications/analysing-the-experience-ofdiscrimination-in-ireland-evidence-from-the-qnhs-equality. McGinnity, F., M. Creighton and E. Fahey (2020). Hidden versus revealed attitudes: A list experiment on support for minorities in Ireland. Irish Human Rights and Equality Commission. McGinnity, F., R. Grotti, O. Kenny and H. Russell (2017). Who experiences discrimination in Ireland? Evidence from the QNHS Equality Modules. Irish Human Rights and Equality Commission. Mitchell, G. and P.E. Tetlock (2017). ‘Popularity as a poor proxy for utility: The case of implicit prejudice’, in Psychological science under scrutiny: Recent challenges and proposed solutions, 164-195. Moss, C. and A. Frounks (2022). Attitudes and disability: The experiences of disabled people in 2022. Scope. Mu, R., Y. Li and T. Cui (2021). ‘Policy narrative, policy understanding and policy support intention: a survey experiment on energy conservation’, Policy Studies, 1-21. https://doi.org/10.1080/01442872.2021.1954609. Müller, S. and A. Regan (2021). ‘Are Irish voters moving to the left?’, Irish Political Studies, 36(4), 535-555. https://doi.org/10.1080/07907184.2021.1973737. NDA (2017). National Survey of Public Attitudes to Disability in Ireland. Retrieved from https://nda.ie/uploads/publications/2017-Main-Report-Nationalsurvey-of-public-attitudes.pdf. O’Brien, C. (2021). ‘Hundreds of special-needs children unable to access school places’. The Irish Times. https://www.irishtimes.com/news/education/hundreds-of-special-needschildren-unable-to-access-school-places-1.4609933. Ostapczuk, M. and J. Musch (2011). ‘Estimating the prevalence of negative attitudes towards people with disability: a comparison of direct questioning, projective questioning and randomised response’, Disability and Rehabilitation, 33:5, 399-411, DOI: 10.3109/09638288.2010.492067. 54 | Experimental tests of public support for disability policy Paulhus, D.L. (2002). ‘Socially Desirable Responding: The Evolution of a Construct’, in The Role of Constructs in Psychological and Educational Measurement. https://doi.org/10.4324/9781410607454-10. Porumbescu, G., N. Bellé, M. Cucciniello and G. Nasi (2017). ‘Translating policy transparency into policy understanding and policy support: Evidence from a survey experiment’, Public Administration, 95(4), 990-1008. https://doi.org/10.1111/padm.12347. Pullen, E., E.A. Ekl, E. Felix, C. Turner, B.L. Perry and B.A. Pescosolido (2022). ‘Labeling, causal attributions, and social network ties to people with mental illness’, Social Science & Medicine, 293 DOI: https://doi.org/10.1016/j.socscimed.2021.114646. Reynolds, J.P., K. Stautz, M. Pilling, S. van der Linden and T.M. Marteau (2020). ‘Communicating the effectiveness and ineffectiveness of government policies and their impact on public support: a systematic review with metaanalysis’, Royal Society Open Science, 7(1), 190522. https://doi.org/10.1098/rsos.190522. Ring, E. (forthcoming). Special Education in an Independent Ireland, 1922-2022. Rosenfeld, B., K. Imai and J.N. Shapiro (2016). ‘An empirical validation study of popular survey methodologies for sensitive questions’, American Journal of Political Science, 60(3), 783-802. DOI: https://doi.org/10.1111/ajps.12205. Russell, H., I. Privalko, F. McGinnity and S. Enright (2021). Monitoring adequate housing in Ireland. Economic and Social Research Institute. Schwarz, N. and F. Strack (1985). ‘Cognitive and affective processes in judgments of subjective well-being: A preliminary model’, in Brandstatter H. and E. Kirchler (Eds), Economic psychology (pp. 439-447). Linz, Austria: Tauner. Schwarz, N., F. Strack, H.J. Hippler and G. Bishop (1991). ‘The impact of administration mode on response effects in survey measurement’, Applied Cognitive Psychology. 5(3), 193-212. DOI: https://doi.org/10.1002/acp.2350050304. Sheahan, F. (May 1, 2021). ‘Majority favour a united Ireland, but just 22pc would pay for it’, Irish Independent. https://www.independent.ie/irishnews/centenaries/centenarypoll/majority-favour-a-united-ireland-butjust-22pc-would-pay-for-it-40375875.html. Sims, C.A. (2003). ‘Implications of rational inattention’, Journal of monetary Economics, 50(3), 665-690. https://doi.org/10.1016/S03043932(03)00029-1. Taylor, S.E. and S.T. Fiske (1978). ‘Salience, attention, and attribution: Top of the head phenomena’, in Advances in experimental social psychology (Vol. 11, pp. 249-288). Academic Press. Thomas, A., E.D. Vaughn, A. Doyle and R. Bubb (2014). ‘Implicit Association Tests of Attitudes Toward Persons With Disabilities’, The Journal of Experimental Education. 82(2), 184-204. References | 55 Timmons, S., F. McGinnity, C. Belton, M. Barjakova and P. Lunn (2020). ‘It depends on how you ask: measuring bias in population surveys of compliance with COVID-19 public health guidance’, Journal of Epidemiology and Community Health. doi.org/10.1136/jech-2020-215256. Timmons, S. and P. Lunn (2022). Public understanding of climate change and support for mitigation. Economic and Social Research Institute (ESRI) Research Series Report No. 135. https://doi.org/10.26504/rs135. Timmons, S., C.A. Belton, D. A. Robertson, M. Barjaková, C. Lavin, H. Julienne and P.D. Lunn (2022). ‘Is it riskier to meet 100 people outdoors or 14 people indoors? Comparing public and expert perceptions of COVID-19 risk’, Journal of Experimental Psychology: Applied. https://doi.org/10.1037/xap0000399. Timmons, S., E. Carroll and F. McGinnity (forthcoming). ‘Social judgements of mental and physical disabilities: Evidence from a vignette experiment in Ireland’. mimeo. Timoney, A. (2022). Long Covid: Definition, incidence and healthcare issues. Tri Handoyo R., A. Ali, K. Scior and A. Hassiotis (2021). ‘Attitudes of key professionals towards people with intellectual disabilities and their inclusion in society: A qualitative study in an Indonesian context’, Transcultural Psychiatry. 58(3), 379-391. DOI:10.1177/1363461520909601. Tsai, C.L. (2019). ‘Statistical analysis of the item-count technique using Stata’, The Stata Journal, 19(2), 390-434. Turner, M.A. (2005). Discrimination against persons with disabilities: Barriers at every step. Diane Publishing. Tversky, A. and D. Kahneman (1980). The framing of decisions and the rationality of choice. Stanford Univ Ca Dept of Psychology. United Nations Convention on the Rights of Persons with Disabilities (2006). December 13, 2006. Available at: https://www.ohchr.org/en/hrbodies/crpd/pages/conventionrightsperson swithdisabilities.aspx. Watson, D., G. Kingston and F. McGinnity (2013). Disability in the Irish Labour Market: Evidence from the QNHS Equality Module 2010. Dublin: Equality Authority/ESRI. Appendix | 57 Appendix TABLE A.1 JOINT DISTRIBUTIONS AND DESIGN EFFECT TEST FOR WELFARE PAYMENT LIST Coef. Robust SE Z p λ p Pr(R=0,S=1) 0.01 0.01 1.31 .90 Pr(R, S=0) 0.00 1.00 Pr(R=0,S=0) 0.02 0.00 5.15 1.00 Pr(R, S=1) 0.00 1.00 Pr(R=1,S=1) 0.30 0.02 13.61 1.00 Pr(R=1,S=0) 0.19 0.01 14.49 1.00 Pr(R=2,S=1) 0.26 0.02 14.52 1.00 Pr(R=2,S=0) 0.11 0.02 4.79 1.00 Pr(R=3,S=1) 0.10 0.01 11.47 1.00 Pr(R=3,S=0) 0.02 0.01 1.15 .88 Source: Authors’ analysis. TABLE A.2 JOINT DISTRIBUTIONS AND DESIGN EFFECT TEST FOR SOCIAL HOUSING LIST Coef. Robust SE Z p λ p Pr(R=0,S=1) 0.07 0.01 5.70 1.00 Pr(R, S=0) 0.00 1.00 Pr(R=0,S=0) 0.04 0.01 7.29 1.00 Pr(R, S=1) 0.00 1.00 Pr(R=1,S=1) 0.23 0.02 10.71 1.00 Pr(R=1,S=0) 0.12 0.02 6.95 1.00 Pr(R=2,S=1) 0.27 0.02 13.90 1.00 Pr(R=2,S=0) 0.12 0.02 5.22 1.00 Pr(R=3,S=1) 0.13 0.01 13.93 1.00 Pr(R=3,S=0) 0.02 0.02 1.04 .85 Source: Authors’ analysis. TABLE A.3 JOINT DISTRIBUTIONS AND DESIGN EFFECT TEST FOR CAR PARKING LIST Coef. Robust SE Z p λ p Pr(R=0,S=1) 0.07 0.01 5.70 1.00 Pr(R, S=0) 0.00 1.00 Pr(R=0,S=0) 0.04 0.01 7.29 1.00 Pr(R, S=1) 0.00 1.00 Pr(R=1,S=1) 0.23 0.02 10.71 1.00 Pr(R=1,S=0) 0.12 0.02 6.95 1.00 Pr(R=2,S=1) 0.27 0.02 13.90 1.00 Pr(R=2,S=0) 0.12 0.02 5.22 1.00 Pr(R=3,S=1) 0.13 0.01 13.93 1.00 Pr(R=3,S=0) 0.02 0.02 1.04 .85 Source: Authors’ analysis. Figure A.1 shows the distribution of responses to each of the lists. A small minority of participants in the welfare and housing lists (2.1 per cent and 4.1 per cent, respectively) revealed a socially undesirable belief (i.e. a response of ‘0’). The undesirable response (‘4’) for the parking list was 2.1 per cent. Hence the 58 | Experimental tests of public support for disability policy anonymity for socially undesirable responding was sufficiently low to avoid raising design concerns. FIGURE A.1 DISTRIBUTION OF LIST EXPERIMENT RESPONSES Source: Authors’ analysis. Note: A response of ‘4’ is not possible for the direct condition. Cost of Living by Education Interaction Table A.4 shows that support for the policy declined for both funding mechanisms relative to the control among those with no degree, but the decline in support was even larger among those with a degree, particularly for the budget reallocation policy. Tests of coefficients showed that support in the budget condition was weaker among those with degrees than without, χ² = 4.43, p = .035, but there was no difference in the tax condition, χ² = 1.49, p = .222. 2.08 4.1 2.1 0 10 20 30 40 50 60 70 80 012340123401234 Welfare Housing Parking % Response Direct List Appendix | 59 TABLE A.4 LOGISTIC REGRESSION OF COST-OF-LIVING SUPPORT: EDUCATION INTERACTION Control (v1) Male (Ref: Female) 0.35 (0.40) Age (Ref: 18-39 years) 40-59 years 0.23 (0.44) 60+ years 0.95 (0.59) Education x Statement Interaction (Ref: No DegreeControl (v1)) No Degree – Budget (v2) -0.65* (0.90) No Degree – Tax (v3) -2.52*** (0.28) Degree – Control (v1) 0.97* (0.44) Degree – Budget (v2) -1.07* (0.51) Degree – Tax (v3) -0.60 (0.49) Social Grade (Ref: DE) C1C2 -0.36 (0.55) AB -0.64 (0.65) Urban (Ref: Rural) -0.29 (0.41) Intercept 2.05*** (0.63) N Yes Responses 301 Total N 329 Source: Authors’ analysis. Note: *p < .10, ** p < .05, *** p < .01. Standard errors are in parentheses.