Public reasoning about voluntary assisted dying: An analysis of submissions to the Queensland Parliament, Australia

Abstract The use of voluntary assisted dying as an end‐of‐life option has stimulated concerns and debates over the past decades. Although public attitudes towards voluntary assisted dying (including euthanasia and physician‐assisted suicide) are well researched, there has been relatively little study of the different reasons, normative reasoning and rhetorical strategies that people invoke in supporting or contesting voluntary assisted dying in everyday life. Using a mix of computational textual mining techniques, keyword study and qualitative thematic coding to analyse public submissions to a parliamentary inquiry into voluntary assisted dying in Australia, this study critically examines the different reasons, normative reasoning and rhetorical strategies that people invoke in supporting or contesting voluntary assisted dying in everyday life. The analysis identified complex and potentially contradictory ethical principles being invoked on both sides of the debate. These findings deepen our understanding of the moral basis of public reasoning about end‐of‐life matters and will help to inform future discussions on policy and law reform. The findings underscore the importance of sound normative reasoning and the use of caution when interpreting opinion polls to inform policy.

conditions that cause unbearable suffering. 4 Reviews of survey and public-attitude research over the past two decades in New Zealand 5 and Australia 6 confirm that the majority of the public supports the legalization of VAD. In Australia, support for VAD has increased substantially: 85% of respondents in 2017 (up from 11% in 1996) endorsed the idea that doctors give 'a lethal dose when a patient is hopelessly ill with no chance of recovery and asks for a lethal dose '. 7 Such quantitative findings, however, should be interpreted with caution. Survey results hinge upon the precise language used to describe the practice and the context of the question asked. Magelssen and colleagues demonstrated and measured such 'framing effects' of surveys on attitudes towards VAD by distributing two versions of a questionnaire to Norwegian citizens. They found moderate to large question-wording and question-order effects on the replies received. 8 Furthermore, qualitative research on attitudes towards VAD has revealed a high level of complexity in the VAD debate, based on polarizing social expectations, moral values and concerns in facing death and dying. 9 A synthesis of the international evidence on attitudes towards VAD identified four recurring themes in the debate: concerns about poor quality of life, desire for a good death, worries about potential abuse of the practice, and the importance of personal witness of unbearable suffering. 10 Earlier research also highlighted socio-demographic variables that predict differences in the approval of VAD among individuals and between countries. These variables include the religious context, level of attachment to autonomy, and feeling of vulnerability. 11 A high level of religiosity and religious affiliation has been found to be negatively associated with permissiveness toward VAD. 12 Although the kinds of ethical arguments used in the debate about VAD and their determining factors are well known, 13 there is relatively little literature on the various reasons, normative reasoning and rhetorical strategies that people invoke in supporting or contesting VAD in everyday life. Understanding this reasoning is important if laws are to be developed that go beyond merely responding to 'polls', especially in light of research that questions such quantitative methods and reveals the complexity of socioeconomic influences on attitudes.
An opportunity to address this gap arose with the Australian State of Queensland's Parliamentary Inquiry into VAD, to which the public were invited to submit their reasons for supporting or opposing VAD.
Using a mixture of computational textual mining techniques, keyword study and qualitative thematic analysis, we analysed a large sample of these public submissions. Our analysis revealed the fundamental themes that characterize the reasoning commonly employed by the general public in debates about VAD, and the kinds of rhetorical strategies and normative argumentation to which these themes lend themselves. We discuss how some of these reasons, rhetorical strategies and normative arguments relate to each other, including possible conflicts, for example reasoning that supports VAD for competent persons versus reasoning that supports ending the life of incompetent persons on the basis of a valid advance health directive. The findings deepen our understanding of the moral basis of public reasoning about VAD and will help to inform future policy reform on end-of-life care matters.

| Data
Queensland is the second largest and the third most populous state in Australia: it covers over 22% of the total Australian continent and is home to about 20% of Australia's population. The Queensland Government launched a parliamentary inquiry into aged care, end-of-life and palliative care, and VAD in November 2018. The inquiry committee tabled the final report to Parliament on 31 March 2020. 14 As part of the inquiry, the Queensland Parliament invited the general public to provide written submissions to express their opinions on VAD. These submissions by individual Queenslanders (not institutions) are analysed in this study.

| Design and data analysis
This study combines computer-assisted keyword and text mining with more conventional qualitative thematic analysis. Data analysis consisted of the following five steps.
1. Extraction, conversion and preparation of the submissions for text mining and content analysis. All submissions in support of or in opposition to VAD were grouped together and saved as a ' Yes' or 'No' document in Microsoft Word, respectively. 2. Three lists of keywords, comprising the most frequently used and distinctive words of each of the documents (see Table 1), were identified and prepared using Voyant Tools.
3. Collocated words for each keyword were identified using Voyant Tools. Collocated words are defined as those words that occurred frequently (≧10 times) in proximity (5 words before or after) to the keywords in each list.
4. The keyword-collocate word pairs of each list were mapped using NodeXL and laid out in two broad clusters according to their position on VAD. We then examined and manually grouped the keyword-collocate pairs in each list into themes based on prior knowledge of how terms may be linked to particular reasons, reasoning or rhetorical strategies in the literature. 5. Those keywords and collocates that constituted a theme were examined in the context of the submissions using Voyant Tools to validate and refine the researchers' interpretation in step 4.

| Sampling, data extraction and preparation
The public submission process started on 14 November 2018 and ended on 15 April 2019. The submissions were made either electronically via a web form, or as emailed or handwritten letters. Most submissions were then published on the Inquiry's website, but a submission was not published if it 'uses unparliamentary language, is potentially incriminating, provides a detailed description of suicide or discloses deeply personal information '. 20 At the time of writing, the Inquiry had released a total of 4,595 (2,511 paper/email and 2,084 electronic) submissions. 21 Length ranged from one sentence to several pages (electronic submission: mean=161 words, SD=177.1; paper/email submission: mean=346 words, SD=653.4). Of these, 4,475 (97%) were submissions by individuals, and the remaining 120 submissions were by groups or organisations. Information released included the name and suburb of the submitter and his/her position on VAD (yes or no). This study focuses only on the submissions by individuals.
The first 1,400 submissions (600 paper/email and 800 electronic), or about 24% of all paper/email submissions and 38% of the electronic submissions as released by the Inquiry, were analysed.
Submissions were published largely in chronological order. A comparison of the submissions analysed with samples from later submissions found no obvious differences between the two tranches.  : die, suffering, dignity, able, terminal, want, years, day, illness, know 'No' document: euthanasia, palliative, patients, medical, care, doctors, assisted, state, patient, legislation Frequent unique keywords * 'Yes' document: choice, mother, cancer, suffer, quality, wish, having, choose, months, loved, died, allow, away, disease, days, dementia, father, living, individual, option, ones, mum, wishes, longer, allowed, let, friends, access, year, husband, left, unable, bed, body, place, painful, hope, alive 'No' document: society, suicide, vulnerable, human, killing, elderly, good, god, kill, community, government, murder, abuse, children, world, consider, families, provide, better, legal, value, love, needs, Australia, sick, treatment, cases, issues, protect, association, bill, burden, countries, committee, legalised, practitioner, disabled, doctor Distinctive keywords * 'Yes' document : slow, friend, competent, endure, advance, knowing, undignified, indignity, horrible, eat, feed, lung, bowel, saw, exit, eligible, swallow, adult, inhumane, obviously, AHD (Advance Health Directive), chair, decline, qualified, bedridden, begged, deny, endured, rational, torture, excruciating, minutes, mums, operation, tried 'No' document: association, infirm, sanctioned, devalues, AMA (Australian Medical Association), conception, peak, combating, mitigated, crossed, commandments, expendable, foster, intentionally *Frequent common keywords are those words that appear in the top 100 frequently used words list of both documents, but which have a significantly higher (p ≦ 0.001) relative frequency (per one million words) in one of the documents compared to the other document. Frequent unique keywords are those words that appear in the top 100 frequently used words in one document that do not appear in the top 100 of the other document. Distinctive keywords are those words with a high TF-IDF (term frequency-inverse document frequency) weighting and a relative frequency of use of 100/million or above in each of the documents. The TF-IDF weighting is a standard measure to evaluate how important a term is to a specific document in a collection or corpus. 19 As this study has only two documents in its corpus, the keywords identified using this method are words that occur only in one of the documents.

| Frequent common keywords-collocates
Frequent common keywords are words found in both documents but that had a significantly higher relative frequency of use in one of them. On examining these keywords and their collocates, we identified a total of seven themes (three supporting and four opposing VAD legalisation), as represented by the clusters A-G in Figure 1. 'It was entirely obvious that death was unavoidable, yet nothing was done to move it along so she could be spared the futile pain, suffering and indignity that she went through.'

| Theme B: Respect for choice and the right to die
Theme B comprises keywords like 'right', 'want', 'able', 'make' and 'decision'. Broadly, this theme appears to be concerned with personal freedom or autonomy. A closer examination of these words and collocates in context, however, suggests two related reasons that have different implications for reasoning and consequences: (a) the idea that one should respect people's self-determining choices about when and how they die (i.e., a respect for choice argument); and (b) the idea that people have a 'right' to die. As Figure 1 shows, the keyword 'right' is not associated with keywords like 'make', 'able' and 'want' that are crucial terms for the respect for choice argument. Rather, 'right' is strongly associated with 'die' and 'end', as in 'right to die' and 'right to end one's life'.

| Theme C: Dying with dignity
Theme C is clustered around the three keywords 'end', 'die' and 'dignity'. Dignity has long been a key concept in the debate. What is important about our finding is that 'dignity' language was used significantly more frequently by those supporting VAD than by those opposing it.
Analysis in context reveals a concern with the importance of the pres-

| Theme G: State-sanctioned killing
Theme G similarly insists on the moral wrongness of the act of VAD as an unjustifiable act of killing. However, here the concern is more with the consequences of legalizing such an act, effectively state- 'I believe the right to live is the only right and it is a very dangerous path to take when all of society is burdened with state sanctioned killing.' 'Please don't approve any bill to introduce state killing of those who think they are unwanted.'

| Frequent unique keywords-collocates
Frequent unique keywords are words found in the top 100 frequently used words of one of the documents but not in the top 100 list of the other document. We identified a total of nine themes (three supporting and six against VAD) (Figure 2). Seven of these were identical to, or extended or refined Themes A to G already examined above. What follows describes this new information and examines two new themes used by opponents (I and H).
Regarding themes supporting VAD, we again found a large cluster of words describing firsthand experiences of suffering (Theme A).
'Cancer' and 'dementia' are frequently mentioned as causes of the suffering or deterioration.

| Theme I: The inviolability or sanctity of life
The keywords that stand out in this theme are 'human', 'value', 'God' and 'love'. The theme invokes the ideas that human life has intrinsic value and that VAD violates this: This is the only theme that contains explicit use of religious language or claims.

| Distinctive keywords-collocates
Distinctive keywords are those words that occur in only one of the documents. Whilst not occurring as frequently as the keywords already discussed, these are included because they are characteristic only of either the 'Yes' or the 'No' group. One new theme from the 'Yes' document was identified (Figure 3).

| Theme J: Advance health directives
This theme is associated with the assertion that an advance health directive (AHD) requesting that one's life be ended should be respected, even if one is incompetent at the time that this is done: 'I believe that every adult Australian should have an Advance Health Directive which states their views

| Pathos vs ethos as rhetorical strategies
Different rhetorical strategies seem to be favoured by each side. It could be argued that pathos is also employed in the opposing view when it insists on the language of suicide, euthanasia and murder rather than VAD, or suggests slippery slopes that will lead to the abuse of the vulnerable and the demise of social mores concerning the inherent value of life. Nonetheless, the pathos strategy seems to be more effective in support of VAD because it offers actual accounts of identifiable people with whom most people can empathize. 25 Most people do not instinctively identify with vulnerable groups. In contrast, the opponents of VAD tend towards an ethical commitment to the common good or the public interest. By arguing that VAD is an act of state-sanctioned killing or suicide, they condemn the practice as a violation of the intrinsic value of life and of traditional obligations of the state to protect vulnerable members of the population. In other words, the debate on VAD legislation can be understood as a clash of individual-and community-level ethical considerations.

| Individual vs communal ethical frames
It is important to acknowledge that an opponent of VAD could also be arguing from a frame of individual self-interest, in terms of the fear of society undermining one's desire to live by placing subtle or not so subtle pressures on oneself when one is at the mercy of the 'system'.

| Normative moral reasoning: Deontology and consequentialism
Both sides employ deontological and consequentialist elements in form arguments (Table 2). That said, the individualist vs communal frame seems to be important in determining which principles or consequences are employed in public reasoning, or how these are understood.

| Pro-VAD deontological reasoning
Deontologically, the proponents of VAD employ a mix of the following principles.
1. Unnecessary suffering is evil and should be avoided (Theme A).
2. Human freedom should be respected when there is no substantial harm to others (Theme B, especially with its reference to safeguards in Figure 2).
3. Dignity (understood as self-worth or freedom from humiliation) should be protected and not violated by 'forcing' someone to 'endure' 'horrible' suffering (see Theme C, especially Figure 3).
Principle (1) is derived from the self-evident truth that good should be done and evil avoided. Because suffering is bad in itself, it should be avoided. Unnecessary suffering is a moral evil, because it could have been avoided.
Principles (2) and (3)  5. Compassion demands that we should end suffering by euthanasia, as we already do in the case of animals (Theme C in Figure 2).
The distinction between (2) respect for choice and (4) a 'right' to die is important, as the two have different philosophical and legal implications. If one recognizes a right, then it follows that a society has an obligation to ensure that the right is met. Most jurisdictions do not recognize a right to die, but rather the idea that when certain criteria are met (doctors' assessments and so on), a patient is permitted to request to have their life ended. VAD is understood as a justifiable exception to the commitment of governments to protect life. 27 The claim of a right to die thus puts additional pressure on society not merely to permit the exception but also to provide the means to die when and how one chooses, regardless of the circumstances. It is not clear from our study whether this distinction and its consequences are well understood by the public or whether there is generally a conflation of the two principles.
The appearance of (5) is interesting because it is arguably at odds with (2) and (3), as animals do not have a choice in the matter. They are euthanized because there is no instrumental or existential value in their suffering. This is, however, not the case for human beings, for whom free choice and the desire to be treated with respect and not to be humiliated are both integral to existential meaning-making. Permitting (5)

| Pro-VAD consequentialist reasoning
From a consequentialist standpoint, the pro-VAD position is rela- Palliative care will reduce physical and existential suffering and preserve dignity and freedom. VAD will lead to: • abuse of vulnerable people • undermined public interest through state-sanctioned killing • increased suicide rates • undermined core values and trust in medicine rhetoric of pathos), the principles of respect for choice and protection of dignity as self-respect and freedom from humiliation imply that the consequence of permitting VAD would be more respect for human freedom and fewer violations of dignity. These consequences are associated with reduced existential suffering and not necessarily with pain. If the premise is accepted that reducing existential suffering arising from perceived lack of control or fear of humiliation is legitimate as the sole reason for VAD, even in the absence of physical pain, then this potentially broadens the range of circumstances in which VAD would be permissible, including, for example, people experiencing psychological distress.

| Anti-VAD deontological reasoning
For those against VAD, deontological principles include the following.
a. Unnecessary suffering is evil and should be avoided.
b. The sanctity or intrinsic value of human life must be respected (Theme I).
c. Intentional killing, like suicide and murder, is morally wrong (Theme F).
d. Medicine must cure or comfort, and never intentionally kill (Theme E).
e. Government has a duty to preserve the common good (Theme G) and to protect the vulnerable (Theme H).
Like proponents of VAD, opponents believe that there is a duty to prevent unnecessary suffering, be it physical, psychological or existential (a). They are also likely to agree with the principles of respect for freedom and protection from humiliation, i.e., (2) and (3) above. For opponents of VAD, however, the latter cannot be applied in ways that violate (a)-(e).
The principles used by the public in arguing against VAD are more coherent when used together, with fewer conflicts between principles if different combinations are used as part of an argument (unlike the possible conflicts outlined above regarding (1)- (5)

| Anti-VAD consequentialist reasoning
From a consequentialist standpoint, opponents appeal to the empirical credibility of the effectiveness of high-quality palliative care (Theme D) to address the problem of suffering adequately whilst not violating (b)-(e). Against VAD, they argue that ignoring (b)-(e) will result in the abuse of vulnerable people (Theme H), in state-sanctioned killing undermining the public interest or common good (Theme G), in worsening or at least not helping the suicide problem (Theme F), and in undermining the core values of the medical profession, thereby reducing public trust in the healthcare system that is already under pressure to deliver high-quality healthcare to all Queenslanders (e.g. Figure 2, Theme E reference to the suicide problem). Whilst the deontological claims require no empirical evidence to support them, the consequentialist ones do. Opponents do make some references to examples, for instance euthanasia of children in Belgium, to support their consequentialist claims. The veracity of the evidence is beyond the scope of the present study.

| Religion less important
In contrast to previous findings regarding the importance of religion in opposing VAD, 28  But it could also be the case that theological values or arguments may have greater resonance if they are packed in secular terms. 29 We suggest that further research be conducted to clarify this important issue.

| Relationship between legalisation, trust and medicine requires further investigation
Finally, our findings have shed some light on the relationship between VAD legalisation, trust and medicine, which has long been a matter of contention. One frequently used argument against VAD is that legalizing the practice will threaten or undermine trust in the medical profession. 30 There is conflicting evidence regarding this ar-  As reported above, our findings revealed that opponents of VAD are worried that legalisation of the practice undermines the principles and authority of modern medicine, and that they tend to appeal to the credibility of professional medical bodies in their argument (Theme E). However, our analysis did not detect 'trust' as an important concept. The term 'trust*' (where '*' is a wildcard for any ending) appeared only 28 times in the corpus, and it was not identified as a keyword or a collocate in any of the submissions we analysed. To pursue further analysis of the matter of trust in the medical profession requires an in-depth qualitative content analysis of selected submissions. The scope of such an analysis is beyond the focus of the current article.

| CON CLUS ION
This study does not exhaust the possible reasons, arguments or rhetorical strategies used by the public in supporting or opposing VAD.
It does, however, help to characterize the main approaches of the

ACK N OWLED G EM ENTS
The authors want to thank Ann Heath and Kosta Simic for their assistance in processing the data, and Muhammad Akram for statistical support. They would also like to acknowledge the constructive comments of the two anonymous reviewers.