OMEGA, Vol. 51(3) 229-237, 2005
A SCALE TO ASSESS ATTITUDES TOWARD EUTHANASIA
JASON WASSERMAN, MA JEFFREY MICHAEL CLAIR, PH.D. FERRIS J. RITCHEY, PH.D. University of Alabama at Birmingham
ABSTRACT
The topic of euthanasia has been a matter of public debate for several decades. Although empirical research should inform policy, scale measurement is lacking. After analyzing shortcomings of previous work, we offer a systematically designed scale to measure attitudes toward euthanasia. We attempt to encompass previously unspecified dimensions of the phenomenon that are central to the euthanasia debate. The results of our pretest show that our attitude towards euthanasia (ATE) scale is both reliable and valid. We delineate active and passive euthanasia, no chance for recovery and severe pain, and patient’s autonomy and doctor’s authority. We argue that isolating these factors provides a more robust scale capable of better analyzing sample variance. Internal consistency is established with Cronbach’s alpha = .871. Construct external consistency is established by correlating the scale with other predictors such as race and spirituality.
INTRODUCTION AND LITERATURE REVIEW The topic of euthanasia has been a matter of public debate for several decades. As the population has aged and life-support technology has evolved, euthanasia has become of more immediate public concern. Empirical research has become increasingly important as policy reform moves to the forefront. This is exemplified by a number of high profile cases, such as the enacting of legislation 229 Ó 2005, Baywood Publishing Co., Inc.
230
/
WASSERMAN, CLAIR AND RITCHEY
permitting physician-assisted suicide (PAS) in Oregon, and in policies of other countries such as Denmark and the Netherlands. Recent events suggest that euthanasia will retain an important place in the political landscape, as evidenced by the Bush administration asking the Supreme Court for judicial review of Oregon’s “Death with Dignity Act” and the Terry Schiavo case that galvanized both supporters and critics. Since the road to policy reform should be paved with empirical research, dependable measures are needed. Our contribution to the literature here is to offer a systematically designed scale that measures attitudes toward euthanasia, specifying more distinct factors than have previously been reported. The bulk of the previous research on attitudes toward euthanasia uses data collected from various years of the NORC General Social Surveys and predominantly the 1977 survey (DeCesare, 2000; Finlay, 1985; Jorgenson & Neubecker, 1980; Rao, Staten, & Rao, 1988; Singh, 1979). In these surveys, support for euthanasia was measured using only two items. The NORC questions are: 1) When a person has a disease that cannot be cured, do you think doctors should be allowed by law to end the patient’s life by some means if the patient and his family request it? and 2) Would you approve of ending a patient’s life if a board of directors appointed by the court agreed that the patient could not be cured? A number of difficulties exist with the NORC data in that each question includes multiple objects of evaluation that can generate multiple possible points of contention for a respondent. For example, an individual could score low on the first question because they object to one or more of the following: 1) euthanasia, in general; 2) euthanasia in cases of no recovery, but where the patient is not suffering; 3) euthanasia as a practice of doctors, but perhaps not by family members or patients themselves; 4) laws permitting euthanasia, but without moral objection to the practice itself; 5) euthanasia in instances of family request, but not where requested by patient; and 6) euthanasia in instances of patient request, but where not requested by the family. NORC question one also takes as its paradigm a case of disease, but not injury. Furthermore, the phrase “by some means” convolutes the distinction between passive and active euthanasia, which is problematic since there is a traditionally recognized distinction between the two. The second NORC question does not clarify any of these issues, simply addressing concerns about the certainty of a “no recovery” diagnosis. The two NORC questions, given the number of different points at which a respondent can disapprove, may indeed underestimate support. If a respondent disagrees with any one of the standards enmeshed in the item, they may report low support for the question as a whole despite, in general, approving of euthanasia. This is the problem of a double-barrel question inflated by the inclusion of several implied contingencies. Domino, Kempton, and Cavender (1996) were sensitive to the methodological shortcomings of previous studies and developed a reliable scale for assessing attitudes toward physician-assisted suicide. They note difficulties in interpreting
ATTITUDES TOWARD EUTHANASIA SCALE /
231
results across various studies due to the use of a variety of terms (e.g., euthanasia, physician assisted suicide, etc.; Domino et al., 1996). In their scale, they chose to use “physician assisted suicide.” Physician assisted suicide represents only one type of euthanasia. The situation implied by the term is one where a doctor actively assists in the death of the patient, most typically by prescribing or administering a lethal drug. But while this is certainly a more controversial form of euthanasia, other variations of euthanasia also are debated. Passive euthanasia typically refers to the removal of life-sustaining technology. While the American Medical Association deems this an ethical practice and lawmakers in the United States have tended to concur, passive euthanasia still is contested, particularly by religious groups. Recent events surrounding the removal of life-support in the Schiavo case highlight that even passive versions of euthanasia are far from settled in the public eye. Thus, a scale which encompasses broad parameters of euthanasia would be a step forward in assessing attitudes towards the practice among the public. Moreover, terms such as euthanasia and assisted suicide have been manipulated in political rhetoric (Domino, 2003). While these terms carry a variety of connotations among various public stakeholders, it is possible to develop a scale that avoids using them. The scale produced by Domino et al. (1996) is methodologically rigorous. It focuses, however, on physician assisted suicide and does not encompass certain salient features of the broader euthanasia debate. Physician assisted suicide is defined by action on the part of the physician, but passive versions of euthanasia also are relevant. Furthermore, the items in the Domino et al. scale make no reference to the suffering of a patient, only to situations of no recovery. Circumstances of severe pain are particularly salient in the public debates and scholarly discussions about the morality of terminating the life of a patient. Finally, while Domino (2003) is sensitive to the active/passive distinction, the use of physician assisted suicide as a paradigm convolutes this key distinction. While respondents are instructed on the issue of action and inaction in Domino’s (2003) study, the questions used to delineate active and passive versions of physician assisted suicide do not adequately capture the distinction. In Domino et al. (1996) the questions are: 1) It would be OK to prescribe a lethal dose of a substance for a patient, if the patient requested it; and 2) It would be OK for a physician to actually administer a lethal dose of a substance to a patient who requested it. Contrary to what one would expect, Domino (2003) found a slightly higher proportion agreeing with the “administer” item (51%) than the “prescribe” item (43%). The absence of significantly lower levels of support for the relatively passive construction is likely the product of nesting the item in a scale whose paradigm (physician assisted suicide) is inherently active. Typically, situations of passive euthanasia are those where a patient is disconnected from life-sustaining machines and allowed to die. Active euthanasia most often concerns situations where patients are given enough medicine to end their lives. Using “prescribe” to represent situations of non-action (passivity)
232
/
WASSERMAN, CLAIR AND RITCHEY
is confusing since it is an active term. To “remove” life support and “allow to die” are passive constructions since they refer to ceasing action and non-action. While some studies have not found differences in active and passive euthanasia (e.g., Adams, Bueche, & Schvaneveldt, 1978), Weiss (1996) found significantly higher levels of support for passive euthanasia. Similarly, as we will discuss below, we found significantly higher levels of support for passive versions in our study population. This suggests that distinguishing between active and passive euthanasia is appropriate and could be informative. THEORETICAL CONSIDERATIONS We find three considerations necessary for the development of an appropriate scale of attitude toward euthanasia. First, the euthanasia debate has long been characterized by a split between active and passive euthanasia (Rachels, 1975). We believe that public attitudes have, to some degree, been shaped by the policies of the American Medical Association (AMA) and various state laws. For example, the AMA finds passive euthanasia permissible, whereas active euthanasia is not (American Medical Association, 2004). Therefore, a scale that properly delineates between active and passive euthanasia may show differences between the two despite the lack of variation found in previous work (Adams et al., 1978). There are a number of other possible standards on which someone could approve or disapprove of euthanasia. The two most commonly recognized reasons for the termination of the life of a patient are situations of severe pain and those of no possible recovery. No possible recovery is referenced in the first NORC question, but severe pain is not. Pain is at the center of current debate and should presumably register among the public as an important consideration. Finally, the issue of decision-making has been of paramount importance in public and academic debates. Discourse on this issue is dichotomized on two factors: patient’s autonomy and doctor’s authority. The first of the two NORC questions addresses patient request, but doctor authority is not included. Although question 2 ambiguously refers to a board of directors making a medical diagnosis, this addresses concerns about the accuracy of the diagnosis rather than proper jurisdiction over decision-making. ATTITUDES TOWARD EUTHANASIA (ATE) SCALE Table 1 presents the items of our ATE scale and indicates conceptual dimensions tapped by each item. No dimension discussed above can be completely isolated from the others. For example, there is not a circumstance where euthanasia is performed by patient request, but where there is no standard for the decision (most commonly severe pain or no recovery) or where the method is not either passive or active. Therefore, the questions in the ATE scale represent
ATTITUDES TOWARD EUTHANASIA SCALE /
233
Table 1. Atttiudes toward Euthanasia (ATE) Scale— Items and Dimensions Dimensiona
Item 1. If a patient in severe pain requests it, a doctor should remove life support and allow that patient to die.
SP / PR / PASSIVE
2. It is okay for a doctor to administer enough medicine to end a patient’s life if the doctor does not believe that they will recover.
NR / DA / ACTIVE
3. If a patient in severe pain requests it, a doctor should prescribe that patient enough medicine to end their life.
SP / PR / ACTIVE
4. It is okay for a doctor to remove life-support and let a patient die if the doctor does not believe the patient will recover.
NR / DA / PASSIVE
5. It is okay for a doctor to administer enough medicine to a suffering patient to end that patient’s life if the doctor thinks that the patient’s pain is too severe.
SP / DA / ACTIVE
6. Even if a doctor does not think that a patient will recover, it would be wrong for the doctor to end the life of a patient.b
NR
7. It is okay for a doctor to remove a patient’s life-support and let them die if the doctor thinks that the patient’s pain is too severe.
SP / DA / PASSIVE
8. If a dying patient requests it, a doctor should prescribe enough medicine to end their life.
NR / PR / ACTIVE
9. Even if a doctor knows that a patient is in severe, uncontrollable pain, it would be wrong for the doctor to end the life of that patient.b
SP
10. If a dying patient requests it, a doctor should remove their life support and allow them to die.
NR / PR / PASSIVE
aSP = severe pain, NR = no recovery, PR = patient requests, DA = doctor’s authority, ACTIVE = active euthanasia, PASSIVE = passive euthanasia. bIndicates items that need to be reverse coded.
234
/
WASSERMAN, CLAIR AND RITCHEY
the variety of possible combinations of these dimensions. Questions six and nine are phrased negatively and require reverse coding. For these questions, the method (active or passive) is not specified. The purpose of these questions is to provide a check on response set bias, the situation where a respondent simply checks responses without reading questions. If a respondent scores high on all other items, they should score low on the reversed items. Questions 1 and 3 deal with circumstances of severe pain where the patient has requested to die, but in question 1 the method is passive whereas in question 3 it is active. Similarly, questions 8 and 11 deal with circumstances of no recovery where the patient has requested to die. For these questions, the method is active in 8 and passive in 11. Questions 5 and 7 deal with circumstances where the doctor thinks the patient’s pain is too severe, but make no reference to the patient’s desires. Here too, question 5 is active and question 7 is passive. Similarly, questions 2 and 4 deal with circumstances where a doctor believes there is no chance of recovery, but make no reference to the patient’s desires. The method is active in question 2 and passive in question 4. Our scale was pretested in several introductory sociology classes at a large, urban university in the southeastern United States. Using the 10-item scale, we conducted both a pretest (n = 47) and then tested the same scale in a larger sample (n = 176) drawn from the same population but containing no respondents from the first sample. For both, we used the Likert scale response categories of: 1) strongly disagree, 2) disagree, 3) undecided, 4) agree, and 5) strongly agree. Demographic characteristics for the pretest (n = 47) showed the sample was disproportionately female and largely comprised of first- and second-year students. Roughly half of the students were African American, while the other half were white. This sample had a mean age of 20.7, slightly younger than the mean age for students at the university. Demographics from the larger sample (n = 176) reflected a similar composition, but with a mean age of 21.7, which is still slightly younger than the mean age of students at the university. While particular sample characteristics might affect the mean level of support for the sample relative to other populations, we do not expect they will affect assessment of the internal reliability of the scale or the correlation of attitudes toward euthanasia and other variables such as spirituality measured within the samples. Admittedly, additional pretesting in random samples would buttress these conclusions. In a preliminary pretest (n = 47) the scale exhibited a Cronbach’s alpha of .914 with item to scale correlations ranging from .578 to .821. Research into survey methodology suggests that pretest samples range from 25 to 75 participants and those participants be slightly more educated than the general public (Converse & Presser, 1986). Our population of university students fits this standard. In a larger sample from the same population (n = 176) the scale had a Cronbach’s alpha .871 with item-to-scale correlations ranging from .481 to.670. No item could have been deleted to improve the internal reliability of the scale in either trial.
ATTITUDES TOWARD EUTHANASIA SCALE /
235
A confirmatory factor analysis extracted two components in both trials. The first component was clearly dominant. In the prestest (n = 47), component one had an initial eigenvalue of 5.73 and explained close to 60% of the variance. Proportions for scale items on component one ranged from .634 to .810. In the follow up study (n = 176) component one had an initial eigenvalue of 4.81 and explained close to 50% of the variance. Proportions for scale items on component one in this trial ranged from .516 to .793. Essentially, the factor analysis confirms that the items all tend to correlate with a single underlying concept. While no statistic can ever confirm substantive interpretation, together with face validity, we feel the scale does, in fact, measure attitudes toward euthanasia. External reliability and validity are less concrete issues. However, there is evidence that our scale is actually measuring what it is intended to measure. In a study on racial differences on attitudes toward euthanasia, the ATE scale correlated with several variables in the expected direction (Wasserman, Clair, & Ritchey, 2005). Consistent with previous research, African Americans were less supportive of euthanasia than whites (r = –.155; p < .05; Adams et al., 1978; Caralis, Davis, Wright, & Marcial 1993; DeCesare, 2000; Jorgenson & Neubecker, 1980; Litchenstein, Alcser, Corning, Bachman, & Doukas, 1997; Rao, et al., 1988; Singh, 1979; Wade & Anglin, 1987). Also, whereas Domino (2003) found no correlation between “degree of religious involvement” and PAS scores, our research does find a correlation between ATE and “spiritual meaning of health and illness.” Respondents scoring high on measures of spirituality tended to score low on support for euthanasia (r = –.399; p < .001). This, too, is consistent with previous research on the relationship of support for euthanasia and spirituality/religiosity (DeCesare, 2000; Finlay, 1985; Jorgenson & Neubecker, 1980; Lichenstein et al., 1997; Rao et al., 1988; Singh, 1979; Wade & Anglin, 1987; Weiss, 1996). Further, while other research has failed to find empirical evidence for the distinction between active and passive euthanasia, our scale finds higher levels of support for passive euthanasia, which is consistent with expectation (Adams et al., 1978; Domino, 2003). Within the scale, four items specifically reference passive euthanasia and four parallel questions reference active euthanasia. For the active subscale, the mean score is 9.18, which is significantly lower than the mean for the passive subscale (p < .001), which is 10.44. We feel that the ability of our scale to distinguish active and passive dimensions represents a significant contribution to the literature, although one that needs to be replicated in further research. LIMITATIONS AND CONCLUSIONS Our scale does not capture every possible variation in circumstance on which a respondent might approve or disapprove of euthanasia. For example, no question makes specific reference to family members performing euthanasia. Some
236
/
WASSERMAN, CLAIR AND RITCHEY
questions make no reference to doctors at all, leaving room for those who support euthanasia, but not as an act of medicine. But our scale may not adequately capture this variation, particularly if those questions are listed in close proximity with other items that deal with the medical arena. While our scale is certainly not all-inclusive, we attempt to systematically incorporate the most prominent features of discussion on euthanasia. In this regard, we feel that it is an incremental improvement to previous work. Further validation of the measure can be accomplished by its incorporation into empirical research. Those studying euthanasia might correlate this measure with variables such as spirituality/religiosity measures, gender, age, health status, or any number of attitudinal variables. While some of these variables have been tested in previous research, this new measure could potentially yield different results. Finally, additional pretesting is needed for other populations to establish its broader applications. Studying attitudes toward euthanasia is increasingly relevant as the issue becomes of more immediate concern for public policy. Previous research has largely been limited by insufficiently operationalized variables. As a systematically designed measurement instrument that is both statistically reliable and substantively valid, we hope our 10-item ATE scale rectifies some operational problems. Future scale development on attitudes toward euthanasia might benefit from comparing our ATE scale to other measures, particularly the scale by Domino et al. (1996). Empirical research might benefit from using the ATE scale to identify correlates of support for euthanasia and thereby focus policy efforts. ACKNOWLEDGMENTS The authors would especially like to thank Kenneth J. Doka, Michael Flannery, and two anonymous reviewers for their helpful comments on earlier versions of this article. REFERENCES American Medical Association. (2004). AMA policy on end-of-life care. [Electronic version]. Retrieved February 25, 2005 from http://www.ama-assn.org. Adams, G., Bueche, N., & Schvaneveldt, J. (1978). Contemporary views of euthanasia: A regional assessment. Social Biology, 25, 548-560. Caralis, P. V., Davis, B., Wright, K., & Marcial, E., (1993). The influence of ethnicity and race on attitudes toward advance directives, life-prolonging treatments, and euthanasia. The Journal of Clinical Ethics, 4, 155-165. Converse, J. M., & Presser, S. (1986). Survey questions: Handcrafting the standardized questionnaire. Newbury Park: Sage Publications. DeCesare, M. A. (2000). Public attitudes toward euthanasia and suicide for terminally ill persons: 1977 and 1996. Social Biology, 47, 264-276.
ATTITUDES TOWARD EUTHANASIA SCALE /
237
Domino, G. (2003). Community attitudes towards physician assisted suicide. Omega, 46, 199-210. Domino, G., Kempton, S., & Cavender, J. (1996). Physician-assisted suicide: A scale and some empirical findings. Omega, 34, 247-257. Finlay, B., (1985). Right to life vs. the right to die: Some correlates of euthanasia attitudes. Sociology and Social Research, 69, 548-560. Jorgenson, D. E., & Neubecker R. C. (1980). Euthanasia: A national survey of attitudes toward voluntary termination of life. Omega, 11, 281-291. Lichenstein, R. L., Alcser, K. H, Corning, A. D., Bachman, J. G., & Doukas, D. J. (1997). Black/white differences in attitudes toward physician-assisted suicide. Journal of the National Medical Association, 89, 125-133. Rachels, J. (1975). Active and passive euthanasia. The New England Journal of Medicine, 292, 78-80. Rao, V. V., Staten, F., & Rao, V. (1988). Racial differences in attitudes toward euthanasia. The Euthanasia Review, 2, 260-277. Singh, B. K. (1979). Correlates of attitudes toward euthanasia. Social Biology, 26, 247-254. Wade, C. H., & Anglin, M. D. (1987). Factors influencing decisions to terminate life. Social Biology, 34, 37-47. Wasserman, J. W., Clair, J. M., & Ritchey, F. J. (2005). Racial differences in attitudes toward euthanasia. Unpublished manuscript, presented at the American Sociological Association Annual Meeting, Philadelphia, PA. Weiss, G. L. (1996). Attitudes of college students about physician assisted suicide: The influence of life experiences, religiosity, and belief in autonomy. Death Studies, 20, 587-599.
Direct reprint requests to: Jason Wasserman, MA Department of Sociology University of Alabama at Birmingham 1212 University Boulevard 237 Ullman Building Birmingham, AL 35294-3350 e-mail:
[email protected]
