We examine the effect of blame attribution and community cohesiveness (as proxied by community size) on public attitudes towards responsibility for mental health care. Data for this study were taken from the MacArthur Mental Health Module of the 1996 General Social Survey.
Moral Opprobrium, Community Membership, and Funding for Mental Health Care: A Role for Information Campaigns
Ann M. Holmes, PhD*
Sheila S. Kennedy, JD January 2005 Key words: funding of service, blame attribution, urban-rural differences Ann M. Holmes*
Associate Professor
School of Public and Environmental Affairs
Indiana University Purdue University Indianapolis
801 W. Michigan St. – 4070
Indianapolis, IN 46202-5152
Email: aholmes@iupui.edu
Phone: 317-278-1043
Fax: 317-274-7860 Sheila S. Kennedy
Associate Professor
School of Public and Environmental Affairs
Indiana University Purdue University Indianapolis
801 W. Michigan St. – 4070
Indianapolis, IN 46202-5152
Email: shekenne@iupui.edu
Phone: 317-274-2895
Fax: 317-274-7860 * Corresponding author
Moral Opprobrium, Community Membership, and Funding for Mental Health Care: A Role for Information Campaigns
Abstract This paper examines the effect of blame attribution and community size on public attitudes towards responsibility for treatment of behavioral disorders. Data for this study were taken from the MacArthur Mental Health Module of the 1996 General Social Survey. Respondents who believed behavioral illness was caused by some underlying medical problem were less likely to believe the patient’s family should be responsible for paying for care (rather than larger institutions such as government or private insurance) compared to respondents who attributed behavioral illnesses to other causes. Respondents from larger, urban communities were more likely to support government’s role in covering mental health care than people from smaller, rural communities. These results suggest that information campaigns that emphasize the underlying medical causes of mental illness may be needed to increase grassroots support for coverage of mental health care, especially in rural communities. Abstract word count: 141
Persons with mental illnesses and substance abuse problems, particularly if they reside in rural communities, face multiple barriers to care.1 One factor that contributes to unmet need is chronic under-funding of services. Between 1986 and 1996, total spending on mental health care declined as a percentage of overall health care spending, and the behavioral health care system became increasingly dependent on public funding sources during the same period.2 Even with increased public funding, resources were insufficient to meet current service needs, particularly for persons who are either underinsured or uninsured.2 As a result, effective treatments that could ameliorate the burden associated with mental illness and substance abuse are not provided.
This paper examines public attitudes regarding who should be responsible for funding behavioral health care when the afflicted individual lacks the resources to pay for services. Such attitudes depend upon an individual’s system of beliefs that, in turn, may condition perceptions of merit and entitlement on the one hand, and blame and personal culpability on the other. “Blame attribution can … condition the ways in which people respond to problems,”3, p. 46 and may explain, at least in part, the lack of parity between insurance coverage for mental health treatment and other types of health care. The primary hypothesis of this paper is that attitudes regarding fiscal responsibility for behavioral health care differ according to beliefs about the underlying causes of various behavioral disorders.
Within the United States, “moral commitment” towards others appears to depend in part on geographical proximity, suggesting that beliefs regarding familial and communal responsibility for mental health care may also depend on characteristics of the community in which an individual resides.4 Because rural populations are confronted by particularly high barriers to behavioral health care, their attitudes regarding appropriate funding channels are of particular interest. A secondary hypothesis of this paper is that the different dynamics of urban and rural life affect attitudes about responsibility for behavioral health care.
The following section of the paper reviews the relevant literatures on the effects of blame attribution and community membership on U.S. social policy. Subsequent sections describe, respectively, the data and methods, the results of the empirical analysis, and implications for behavioral health care policy. The Role of Moral Opprobrium in U.S. Social Policy
Unlike most western liberal democracies, American public policy has historically been grounded in moral judgments; notions of fault and personal responsibility have shaped everything from tort law to welfare eligibility. In the Introduction to Hellfire Nation: The Politics of Sin in American History, James Morone5 describes America as “a nation with the soul of a church,” an observation he proceeds to document at considerable length. The book details America’s various moral crusades, including those primarily concerned with temperance. The Symbolic Crusade, Joseph Gusfield’s 1963 classic examination of the American temperance movement and the politics of moral indignation remains highly relevant today6; other scholars have drawn similar comparisons between today’s emphasis upon individual responsibility and “family values” and the “moral crusades” that have characterized American social history.7
Moral judgments “blaming the victim” have not been limited to substance abuse and mental illness; ever since enactment of the 15th Century Elizabethan Poor Laws, which forbid giving alms to “sturdy” beggars, debates about poverty and welfare have revolved around moral distinctions between the “deserving” and “undeserving” poor. The belief that poverty is evidence of divine disapproval—that virtue is rewarded by material success—was held by a number of the early Protestants who settled the colonies, and that belief has continued to influence American law and culture. In the early 1900s, moral opprobrium directed at the poor found an ally in science, and poverty issues were caught up in the national debate between Social Darwinists like William Graham Summer and their critics (notably, William Jennings Bryan). In language reminiscent of earlier admonitions against rewarding “sturdy beggars,” Sumner8 wrote: “But the weak who constantly arouse the pity of humanitarians and philanthropists are the shiftless, the imprudent, the negligent, the impractical, and the inefficient, or they are the idle, the intemperate, the extravagant and the vicious. Now the troubles of these persons are constantly forced upon public attention, as if they and their interests deserved especial consideration, and a great portion of all organized and unorganized effort for the common welfare consists in attempts to relieve these classes of people….” (p. 118) Given this history and culture (presumably reinforced by the widespread and persistent concern that a too-generous government social safety net might undermine individualism, capitalism and the entrepreneurial spirit that has propelled the American economy), it would be surprising if attitudes about collective responsibility for mental health care costs did not depend significantly upon beliefs about causation and blame. The Role of Community Membership
The particularly high barriers to care facing rural populations justify a closer examination of whether urban-rural differences in attitudes regarding responsibility for funding behavioral health care exist. One reason such differences may arise is differences in the nature of social capital between rural and urban locations. Social capital is defined as “the networks, norms, values, and understandings that facilitate cooperation within or among groups.” 9 Academic attention to the concept of social capital is not new; indeed, references can be traced back as far as 1916.10 With the publication of Robert Putnam’s Bowling Alone: The Collapse and Revival of American Community,10 the concept emerged into the general public consciousness, and sparked a lively and ongoing debate about the nature of social capital, the reliability of various methods for measuring it, and its significance for public policy and democratic governance.
Putnam, Portes and others have drawn a distinction between “bridging social capital,” and “bonding social capital,” a distinction with particular relevance to the instant analysis.10-11 Bridging social capital results from ties between heterogeneous actors, and facilitates ties among people across diverse backgrounds and social cleavages. Bridging social capital is thought to generate broader identities and more generalized reciprocity. Bonding social capital, on the other hand, promotes in-group solidarity and tends to reinforce exclusive identities and homogeneity. Thus, while social capital is believed to facilitate coordination, reduce transaction costs and enhance the flow of information, effects which can enhance any number of social activities from job hunting to trade to community action,12 scholars increasingly believe that those effects will depend upon whether we are talking about bridging or bonding social capital. We suspect that bonding social capital may be more characteristic of small, homogeneous communities than their larger, more heterogeneous counterparts. If so, then residents of smaller and larger communities would have different beliefs about where responsibility for mental health care should lie—those from smaller communities might take the position that “we can take care of our own” (while more strictly limiting the definition of “our own”), while those from larger, more diverse urban settings might be more willing to see mental health care as a communal, or governmental, responsibility.
Data and Methods
Data for this analysis are from the MacArthur Mental Health Module of the 1996 General Social Survey. The survey was based on a nationwide, representative sample of adults living in non-institutionalized settings in the United States. The mental health module included a series of questions related to the recognition and knowledge of mental health problems and, of particular interest for this analysis, financial responsibility for treatment.13 The survey questions were constructed around a set of vignettes based on DSM-IV criteria for schizophrenia, major depression, alcohol dependence and drug dependence,14 as well as a control condition that did not meet any diagnostic criteria.15 While respondents randomly received only one vignette, all were asked a uniform set of questions to elicit their attitudes about mental health issues. The analysis is restricted to the set of respondents who received any of the four vignettes associated with serious mental illness or substance abuse (n=1173).
Respondents were asked, in relation to the vignette they were assigned, “In your opinion, who should be most responsible for paying the cost of [NAME]'s medical care, including mental health care and treatment?” Options included the individual described in the vignette, his/her family, government, insurance, and private charity. If the respondent initially reported the individual him- or herself, the respondent was asked a second question: “who should be next most responsible?” Responses to these two questions were used to construct the dependent variable, RESPONSIBLE, which captures who, excluding self, should pay for the costs of care. “Self” was specifically excluded from the dependent variables, since assuming responsibility for one’s own treatment would seem unreasonable in situations where one is too ill to arrange or pay for that care.
Independent variables
Two sets of explanatory variables are considered: characteristics of the health conditions described in the vignettes (particularly attribution of cause) and size of community (as a proxy for social cohesion/bonding social capital). Vignette conditions were identified as either being mental illnesses or substance abuse problems. CONDITION is a binary variable that takes on a value of 1 when the vignette condition was either depression or schizophrenia, and a value of 2 when the vignette condition was either alcohol or drug dependence.
Four variables were constructed to capture respondents’ attribution of vignette conditions. Respondents were asked how likely the condition was caused by (1) bad CHARACTER, (2) the WAY RAISED, (3) GOD’S WILL, (4) a chemical imbalance, and (5) genetics. To simplify analysis, the responses were dichotomized so that each variable was assigned a value of 1 if the respondent reported “very likely” and 0 otherwise. In addition, the last two categories were combined into a variable reflecting a MEDICAL attribution, which takes on a value of 1 if the respondent reported that the vignette condition was very likely caused by either a chemical imbalance or genetics, and 0 otherwise. Note that the attribution variables are not mutually exclusive and need not sum to one across all categories. They do not reflect respondents’ beliefs about which factor of those considered is the most likely to have caused the vignette condition, but which, if any, contributed to the problem. In fact, half of respondents exhibited ambivalence regarding the cause of the condition evaluated (identifying none of the factors as a very likely cause), and roughly a tenth identified multiple causes.
Size of community in which the respondent lived was used to differentiate between urban and rural populations. Because community size is highly skewed, community size was represented by three binary variables: RURAL takes on a value of 1 if the community had a population under 20,000 persons, 0 otherwise; URBAN takes on a value of 1 if the community had a population of at least 1,000,000, 0 otherwise; the reference category are communities between 20,000 and 1,000,000 persons. These partitions correspond roughly to categories 6-9, category 1, and categories 2-5, respectively, of the USDA rural-urban continuum codes.16 Variables are summarized in Table 1.
Analysis Plan
A regression-based approach is preferred to bivariate analysis because it estimates the effect of any one independent variable on the dependent variable while controlling for the effects of the other independent variables. In particular, by including the attribution variables in addition to the condition type variable, it is possible to determine if there were differences in attitudes to responsibility for care between mental illness and substance abuse that stemmed from other than differences in beliefs about what causes mental illness and substance abuse (e.g., personal failings). The primary variable of interest, RESPONSIBLE, is discrete rather than continuous. Linear regression techniques are inappropriate in such situations.17 Instead, relationships of interest were estimated using multinomial logistical regression using SPSS.18
Results
Table 2 presents the estimates from the multinomial logistic regression. The log likelihood statistic, the multinomial analog to the F-statistic in linear regression, indicates the regression equation is significant as a whole (χ2=80.756, p=0.000). Logistic coefficients do not reflect the marginal effects of the variables on the probabilities of selecting any one group as primarily responsible for covering the costs of care, but can be converted to relative risk ratios, which have a more intuitive interpretation. Because insurance is the reference category for the dependent variable, the relative risk ratios indicate the direction of the effect of the independent variable on the responsibility category compared to insurance. For example, a relative risk ratio of 2 (½) for the CHARACTER variable in the FAMILY column implies that people who attribute the vignette illness to bad character are twice (half) as likely to believe that the family rather than insurance should be responsible for covering the costs of care than people who do not believe bad character is the cause of such illnesses. The statistical significance of such effects can be assessed using standard t-statistics.
Beliefs about the cause of illness had a statistically significant (p<0.05) impact on who was identified as responsible for care. For instance, respondents who believed the illness was caused by the bad character of the person described in the vignette were 1.623 times more likely to believe that the family should be held responsible for care (and 1.711 times more likely to believe that government should be held responsible for care) than private insurance compared to respondents who did not believe the problem was caused by bad character. Respondents who believed the illness was caused by God’s will were 3.5 times more likely to believe private charity should be primarily responsible for care (and 2.6 times more likely to believe government should be primarily responsible) than private insurance compared to respondents who did not believe the problem was due to God’s will. In contrast, respondents who believed the illness could be attributed to medical causes were significantly less likely to believe the family should be primarily responsible for covering the costs of care relative to private insurance. Somewhat surprisingly, association of illness with the way someone was raised did not significantly affect beliefs about who should be responsible for paying for care.
After controlling for attribution, CONDITION remained a statistically significant predictor variable. Respondents who were given a vignette in which the person experienced a substance abuse problem were twice as likely to report the family should be primarily responsible for care (and were almost four times more likely to report that charity should be) than private insurance compared to respondents who received a vignette in which the person experienced a mental illness. Given that the equation has already controlled for the effect of cause of illness on responsibility for care through the inclusion of the CHARACTER, WAY RAISED, GOD’S WILL, and MEDICAL variables, these differences are probably not due to differing degrees of “blame” (either the individual with the condition or his/her family) associated with the two conditions. While other inherent aspects of the two conditions might explain the pattern of responses, we were unable to determine what these might be given the data available.
The effect of community size was also statistically significant: respondents from large communities were 2.657 times more likely to believe the government should be responsible for paying for care relative to private insurance compared to respondents from mid-sized communities. In contrast, respondents from smaller communities were only half as likely to believe that private charity should cover the costs of care relative to private insurance compared to respondents from mid-sized communities.
Conclusion
Three conclusions can be drawn from these results. First, responsibility for care varies by illness, with the responsibility for substance abuse care relatively more likely to be assigned to family or private charity, and mental illness care relatively more likely to be assigned to larger institutions, be they government or insurance. This pattern exists even after controlling for differences in attributed cause (including personal and familial blame) between the two conditions.
Second, attributed cause also affects patterns of responsibility for care: bad character (personal blame) increases the likelihood that the family or government is assigned primary responsibility relative to private insurance; God’s will (which may be a form of fatalism) increases the likelihood that government or private charity is assigned primary responsibility relative to private insurance; medical causes (arguably, the “informed” position) decreases the likelihood that the family is held accountable for the costs of care relative to private insurance. Curiously, the way someone was raised (familial blame) does not affect the assignment of primary responsibility for care (perhaps because the afflicted individual cannot be blamed for the misfortune of familial mistreatment).
Third, people from smaller communities are less likely to attribute responsibility to charities relative to insurance, and people from larger communities are more likely to attribute responsibility to government relative to insurance. Although we were not able to include a direct measure of social capital in the analysis, this result is consistent with the hypothesis that larger communities are more likely to generate bridging social capital and smaller communities are more likely to generate bonding social capital. Bridging social capital is characterized by more generalized reciprocity, which in turn is consistent with the assumption of communal, or governmental, responsibility. Bonding social capital, on the other hand, with its tendency to promote in-group solidarity and homogeneity, could be expected to accompany the belief that individuals and families should take primary responsibility for their own mental and physical health.
Implications for Behavioral Health Services
A fragmented behavioral health care system presents a number of barriers to care.1 While inadequate funding is recognized as contributing to these problems, considerable debate exists as to how funding shortages should be addressed. Some see the responsibility falling primarily on private-sector insurers (e.g., initiatives supporting coverage parity), others on the non-profit sector (e.g., the expansion of charitable choice programs), while still others on the (increasingly stressed) public sector. When these mechanisms fail, as is likely to occur when there is no general consensus as to which sector should be ultimately responsible for funding care, responsibility defaults to the afflicted individual and his/her family. When these familial resources are inadequate, treatment needs remain unmet.
The G.S.S. of 1996 surveyed attitudes regarding funding responsibility for behavioral health care. Over a quarter of respondents placed the onus on patient’s families rather than larger institutions, such as government or insurance. This multivariate analysis revealed public support for either government responsibility for behavioral health care or parity in insurance coverage is heavily influenced by cause attribution; specifically, in order to increase support for collective responsibility for mental health care, people need to be convinced that mental illness and substance abuse are caused by some underlying medical condition, and are not evidence of personal character flaws. This conclusion supports increased efforts to raise public awareness and improve education about the causes of mental illness. The results also suggest that urban populations differ from rural populations in how collective responsibility should be allocated between government payers and private insurance: urban populations are more predisposed to support government coverage of behavioral health care than rural populations. Thus, public information/education campaigns may be particularly important in smaller communities if a broad base of grass roots support for government responsibility for behavioral health care is to be achieved.
The President’s New Freedom Commission1 called for such national education initiatives to “shatter misconceptions about mental illness.” In response, SAMHSA has developed a multi-media information campaign to promote public understanding of these disorders that is currently being tested in eight states.19 The overt aim of these efforts is to reduce stigma associated with mental illness. However, as noted in the Surgeon General’s Report2, “Stigma was expected to abate with increased knowledge of mental illness, but just the opposite occurred.” (p. 8). Even if these educational campaigns fail in their primary objective to reduce stigma, however, they may inadvertently contribute to greater public support for collective responsibility for financing behavioral health care if they succeed at debunking myths regarding the origins and causes of behavioral illnesses. If so, they will succeed in reducing barriers to effective treatment of such illnesses.
References
1. New Freedom Commission on Mental Health. Achieving the Promise: Transforming Mental Health Care in America. Final Report. DHHS Pub. No. SMA-03-3832. Rockville, MD; 2003.
2. Office of the Surgeon General. Mental health: A Report of the Surgeon General. Rockville, Md: Dept. of Health and Human Services, U. S. Public Health Service; 1999.
3. Rosenthal M, Schlesinger M. Not afraid to blame: the neglected role of blame attribution in medical consumerism and some implications for health policy. The Milbank Quarterly. 2002; 80(1): 41-95.
4. Schlesinger M. Paradigms lost: the persisting search for community in US health policy. Journal of Health Politics, Policy and Law. 1997; 22(4): 937-992.
5. Marone JA. Hellfire Nation: The Politics of Sin in American History. New Haven: Yale University Press; 2003.
6. Gusfield JR. Symbolic Crusade: Status Politics and the American Temperance Movement. Chicago: University of Illinois Press; 1963.
7. Hunt A. Governing Morals: A Social History of Moral Regulation. Cambridge Studies in Law and Society, Cambridge University Press; 1999.8. Sumner WG. The Forgotten Man. In Leuchtenburg WE, Wishy B, eds. Social Darwinism: Selected Essays of William Graham Sumner. Englewood Cliffs: Prentice Hall; 1963.
9. OECD/OCDE. The Well-being of Nations: The role of human and social capital. Paris; 2001.
10. Putnam RD. Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster; 2000.
11. Portes A. Social Capital: Its Origins and Applications in Modern Sociology. Annual Review of Sociology 1998; 24:1-24.
12. Bielefeld W. Social Capital. In Burlingame DF, ed. Encyclopedia of Philanthropy in the US. San Francisco: ABC-CLIO Publishing; 2004.
13. Pescosolido BA, Monahan J, Link BG, Stueve A, Kikuzawa A. The public’s view of the competence, dangerousness, and need for legal coercion of persons with mental health problems. American Journal of Public Health 1999; 89(9): 1339-1345.
14. Diagnostic and Statistical Manual of Mental Disorders, Fourth edition. Washington, DC: American Psychiatric Association, 1994.
15. Link BG, Phelan JC, Bresnahan M, Stueve A, Pescosolido BA. Public conceptions of mental illness: labels, causes, dangerousness and social distance. American Journal of Public Health. 1999; 89(9): 1328-1333.
16. United States Department of Agriculture, Economic Research Service. Measuring Rurality: Rural-Urban Continuum Codes. Available at <http://www.ers.usda.gov/briefing/rurality/RuralUrbCon/>. Accessed July 1, 2004.
17. Maddala GS. Limited Dependent and Qualitative Variables in Econometrics. New York: Cambridge University Press; 1983.
18. SPSS for Windows, release 12.0.0. Chicago: SPSS Inc.; 2003.
19. Resource Center to Address Discrimination and Stigma. Elimination of Barriers Initiative. Available at <http://www.adscenter.org/ebi.htm>. Accessed December 29, 2004.
Table 1: Descriptive Statistics
Variable |
Percent of respondents |
RESPONSIBLE Family Government Insurance Private Charity |
28.1% 15.0% 51.8% 5.1% |
VIGNETTE CONDITION Mental illness Substance Dependence |
51.2% (49.9% depression, 50.1% schizophrenia) 48.8% (49% alcohol, 51.0% drug) |
CAUSE OF CONDITION Character Way raised God’s will Medical |
18.4% 9.8% 4.0% 29.2% (81.0% chemical imbalance; 43.9% genetics) |
COMMUNITY Rural (<20,000 population) Urban (>1,000,000) |
44.0% 7.8% |
Note: n=1173
Table 2: Coefficient Estimates for Multinomial Logistic Equations
|
|
FAMILY |
GOVERNMENT |
CHARITY |
|||
|
|
Coefficient |
Relative risk ratio |
Coefficient |
Relative risk ratio |
Coefficient |
Relative risk ratio |
Intercept |
-1.772 |
|
-1.882 |
|
-4.208 |
|
Condition |
0.706* |
2.027 |
0.319 |
1.376 |
1.355* |
3.876 |
Cause |
|
|
|
|
|
|
Character |
0.484* |
1.623 |
0.537* |
1.711 |
0.292 |
1.339 |
Way Raised |
0.201 |
1.222 |
-0.240 |
0.787 |
0.162 |
1.176 |
God’s Will |
0.634 |
1.885 |
0.940* |
2.561 |
1.262* |
3.532 |
Medical |
-0.498* |
0.608 |
-0.002 |
0.998 |
-0.124 |
0.884 |
Community |
|
|
|
|
|
|
Rural |
0.072 |
1.075 |
-0.072 |
0.931 |
-0.674* |
0.510 |
Urban |
0.300 |
1.351 |
0.977* |
2.657 |
0.235 |
1.264 |
Note: private insurance is the reference category;
n=938;
Likelihood Ratio test = 80.756 (df=21, p=0.000)
* - p<0.05