Active euthanasia remains a controversial issue both among healthcare professionals and the general public. The fact that people have different opinions is one obvious reason for this; however, different surveys conducted on similar populations also give confusingly different results.¹ There are several possible explanations to this. One is that the term “euthanasia” was not defined,² or when it was defined, different conceptions were used or presupposed.^3,4 In addition, respondents in different countries or regions have different religious or cultural backgrounds influencing their moral views.^2,5–7 People may also change their view over time.⁸

There is, however, one complementary explanation. Survey results may be influenced by the way in which the questions about euthanasia and the possible answers are phrased.⁹ A classical study by Sullivan and colleagues showed that two questions with essentially the same cues could trigger different respondent inclinations and attitudes, producing sometimes remarkably different answers.¹⁰

As a matter of fact, empirical survey data often have been and will be used as a basis for normative discussions. With reference to such discourse one may question two assumptions. The first assumes that the results from surveys support normative conclusions. This has been taken for granted in two articles by Waller¹¹ and Kenis.¹² Both authors use the results from surveys on attitudes to justify the legalisation of active euthanasia. Such arguments, using empirical results as arguments for political decisions may result in these not being supported by the majority of the population, often require controversial value premises.^1,13 The second assumes that the data on attitudes to legalisation of euthanasia are trustworthy. This latter assumption on the survey instruments has, as far as we know, not been studied.

The aim of the present paper was to explore whether the phrasing of the questions and the response scales would influence the answers to questions about legalisation of euthanasia by comparing results from two different surveys.

METHODS

The target group was undergraduate students at Uppsala University. Questionnaires, in Swedish, were distributed towards the end of lectures and were voluntarily and anonymously completed immediately after distribution. We arranged distribution at times when as many students as possible were present. A similar procedure was performed when collecting the questionnaires from every group.

All students present at the times of distribution of the surveys returned the questionnaires. Of all enrolled students more than 90% responded. Internal response rate varied slightly. Demographic characteristics of all the respondents are given in table 1. Respondents were students of years 1 and 3 from four educational programmes.

The first questionnaire contained 37 questions,¹⁴ of which only one addressed euthanasia (see box 1). No explanation of the term “euthanasia” was given. Forms were distributed in the period 1998–2000. In total, 684 out of 697 individuals (98.1%) responded to the question (360 females and 324 males).

The second questionnaire contained four questions, of which one addressed euthanasia, and it was phrased differently from the first questionnaire. A text was presented in which the term “euthanasia” was defined (see box 2). This questionnaire was distributed in the spring of 2001. In total, 639 out of 711 individuals (89.9%) responded (396 females and 243 males). Respondents were in corresponding years and programmes as those subjected to the first questionnaire. The likelihood ratio χ² test was used to measure differences between groups. p Values <0.05 were considered significant. Calculations were done using the SPSS version 10.0 for Windows (SPSS Inc, Chicago, IL, USA).

RESULTS

In the first questionnaire, 43% answered “positive”, 14% answered “negative”, and 43% did not express an opinion on euthanasia. In general, the younger the students were the more likely they were to accept euthanasia (p<0.001). Students in year 1 were more likely to accept euthanasia than students in year 3 (p<0.001). Compared with the other two programmes, nursing students and medical students had the lowest approval rates of euthanasia (p<0.001) (table 2).

Comparing the first and the second questionnaire, a similar proportion of all the students were positive about euthanasia (p<0.058). The proportion of positive respondents were significantly different for two of the four programmes’ responses in the two questionnaires (engineering, p<0.881; law, p<0.001; medicine, p<0.127; nursing, p<0.012). In the second questionnaire we replaced the alternatives “negative” and “don’t know” (first questionnaire) with four types of penalty of law and with forced choice—that is, no option to answer “don’t know”. The most favoured legal consequences were “charges dropped” (32%) and “remission of sentence” (18%). A higher proportion of respondents declined to answer the second questionnaire compared with the first questionnaire (10% v 2%).

The second questionnaire showed no difference in opinion on legalisation of euthanasia according to sex, age, and year of study. A lower proportion of law students approved of the “no legal prosecution” alternative compared with students from the other programmes (p<0.001).

DISCUSSION

Forty three per cent of the respondents did not express an opinion about euthanasia in the first questionnaire. There may be at least two reasons for this. Some may not have given the issue much thought before being questioned. Others may have had an opinion but found that the response alternatives did not allow them to express their views. In the second questionnaire the students were given an explanatory text, a wider range of response alternatives, and no explicit “don’t know” alternative. The present results suggested that many of the respondents who answered “don’t know” in the first questionnaire would be replaced by respondents who could express an opinion in the second. Perhaps as a consequence of this, a higher proportion of respondents also declined to respond to the second questionnaire compared with the first. However, this group was much smaller (10%) than those who opted for “don’t know” in the first questionnaire (43%). A similar effect of the “don’t know” exclusion has been shown previously on issues not related to active euthanasia.^15,16 There were no differences in proportion of acceptance among engineering students between the two surveys (table 2). This may be because of the fact that, unlike the other three groups, they do not have any professional interest in the question. Law students who in the first questionnaire were quite positive, were in the second questionnaire given the possibility of using the legal system. Almost three out of four students in the second sample opted for such intervention. The present results further suggested that neither medical nor nursing students support intervention by the legal system and perhaps found this an issue for themselves to decide upon.

In the present study a higher proportion of the medical students accepted legalisation of euthanasia compared with a pilot study by Nilstun.¹⁷ There may be several explanations—for instance, in the pilot study the students were informed that current legislation in Sweden treats active, voluntary euthanasia as manslaughter, whereas in the present questionnaire no such information was given.

There is an extensive literature on how to construct questionnaires, and effects related to the instruments and the respondents have been discussed for many years.¹⁸ Whether or not to offer a “don’t know” option is one controversial issue.¹⁹ One problem with not offering “don’t know” is that respondents may opt for a response alternative of a question even though they do not know about the issue involved. Offering “don’t know” may on the other hand discourage some respondents from doing the cognitive work necessary to report the true opinions they do have. Further, it has been suggested that inclusion of “don’t know” options in attitude measures may not enhance data quality but instead preclude measurement of some meaningful opinions.

In the first questionnaire, the students could only answer “positive”, “negative”, or “don’t know” to the question about legalisation of active euthanasia. Almost half of the respondents answered “don’t know”. In the second questionnaire the students were, besides no legal control, given the choice between four different degrees of legal control in a split ballot fashion. The effect of excluding a “don’t know” alternative was similar to previous findings. These five options seem to have made it easier for many students to express their opinion about legalisation of active euthanasia. If this interpretation is correct, the results of the second questionnaire should be more valid than the first.

CONCLUSIONS

We trust that our study illustrates the difficulties of interpreting results from different questionnaires to support or oppose recommendations on the practices of euthanasia.

Our hypothesis was that the outcome of questionnaires might be affected by the survey instrument used. The presents study confirms this hypothesis.

These results further show the difficulties of making direct comparisons of answers to questions with different wording and response alternatives in a population with similar characteristics. Answers to questions on whether to legalise euthanasia may thus be modified by the way in which the questions and possible responses are phrased.