"Race, Party, and Representation: Health Care Attitudes in a Modern Southern State," with David A. Breaux, Doug Goodman, and Barbara Patrick. In The American Review of Politics, vol. 27, Fall 2006, pp. 231-253.

The reliability of these seven items was tested by calculating the Alpha coefficients. For the mass survey, the Alpha for all seven items was a sizable .7952. The third item pertaining to recruiting and retaining doctors was not as highly correlated with the other six items as those six items were intercorrelated with each other, but dropping this item from the Alpha scale analysis increased the Alpha coefficient only slightly to .7998. For the elite survey, the Alpha for all seven items was a significant .6957. In this case, the recruiting and retaining doctors item was essentially unrelated to the other six items, so dropping this item from the Alpha scale analysis increased the Alpha coefficient to a more impressive .7460. We are satisfied that all of these items are reliable indicators of the public’s views toward the importance of health care programs. However, these patterns suggest that the public and particularly the legislators may view health care from two or more perspectives—one focusing on the consumer (themselves), and one focusing on the providers (such as doctors and nurses). Investigating the possible multidimensionality of our health care questionnaire items will be our first substantive concern.

We also conducted a construct or criterion validity test by relating our indicators of health care priorities to a well-established indicator asking average citizens their preferred governmental spending priorities. Average Mississippians were read the following statement: “Now I'm going to ask you about some issues facing state and local government in Mississippi. As you know, most of the money government spends comes from the taxes you and others pay. For each of the following, please tell me whether you think state and local government in Mississippi should be spending more, less, or about the same as now.” Among the ten programs that average Mississippians were asked about was: “health care and hospitals.” This question was recoded so that responses ranged from a low of 1 for a desire to spend less to a high of 3 for a desire to spend more.

Our seven indicators of health care priorities exhibit considerable validity. Each of these items is significantly related to spending preferences on health care and hospitals. As average Mississippians rate a health care item as increasingly important, they are more and more likely to prefer that government spend more money on health care and hospitals (Table 1). Indeed, on six of the seven items, their responses show a steady increase in support for more government spending as they rate a health care item as increasingly important. Hence, a greater proportion of citizens rating a specific health program as Somewhat Important desire to spend more on health care and hospitals generally than those rating it as Not Important. An even greater proportion of citizens rating the health program as Important desire to spend more than those rating it as Somewhat Important, and those rating a program as Very Important desire that even more money be spent on health care in general. The only exception to this pattern of steady intervalness is on the provider dimension pertaining to recruiting and retaining doctors, where Kendall’s tau b was statistically significant though Pearson’s r was not, due presumably to insignificant differences in spending preferences between the Not Important and Somewhat Important categories, as well as between the Important and Very Important categories. The absence of a comparable government spending item in the survey of legislators precluded a validity test of that dataset, but we have no reason to believe that our seven health care priority indicators, asked with identical wording of this more informed population, would behave any differently from the mass survey in terms of validity.

To gain some insight into the possible multidimensionality of our health care priority indicators, we pooled the mass and elite responses and conducted a convergent-discriminant validity test by generating a correlation matrix (Table 2). While all seven health items were positively intercorrelated with each other, indicating that those rating one item as a very important priority were also likely to rate other items as very important, six of the items could be divided into two separate groups with items in each group more highly interrelated than were items from different groups. This suggests the existence of at least two separate (but related) dimensions of health care. A “proactive” dimension included the public education, preventive care, and minority health status items, and a “services” dimension included the items for universal care and care for children and poor adults. A principal components factor analysis with varimax rotation also produced two factors or dimensions with the same proactive and services items, though it showed the doctor recruitment item loading on the proactive factor. However, the failure of the recruit doctors item to be highly related to any of the items in either of these two dimensions, plus its unique behavior in our reliability and validity tests and forthcoming analyses, suggests that this health care concern constitutes a third dimension focusing on “providers.” The fact that a single issue (health) that is often included as merely one indicator of a social welfare/domestic economic dimension may itself be so complex as to generate multiple dimensions in Americans’ belief systems is the first noteworthy finding of our research.

Table 1. Construct Validity Test of the Seven Health Care Priorities Items

(means are for the health care/hospitals spending item)

Health Care Item	Not Important	Somewhat Important	Important	Very Important
Public education to encourage good nutrition and physical activity	2.29 (6)	2.50 (55)	2.60 (167)	2.77* (273)
Preventive health care	1.91 (9)	2.45 (42)	2.67 (188)	2.75* (258)
Improving the health status of minority groups in MS.	2.18 (35)	2.45 (66)	2.72 (236)	2.85* (157)
Providing healthcare services for children whose families cannot afford health insurance	--- (0)	2.02 (18)	2.49 (148)	2.80* (336)
Providing health care services for adults who cannot afford healthcare insurance	1.71 (10)	2.37 (56)	2.65 (191)	2.82* (238)
Universal health care coverage for Mississippians	2.06 (49)	2.58 (59)	2.74 (166)	2.86* (198)
Recruiting and retaining doctors in Mississippi	2.52 (10)	2.49 (21)	2.70 (156)	2.69+ (310)
Note: Cell entries are the means of the Health Care and Hospitals state and local spending item, with sample sizes in parentheses. This well established indicator ranges from a low of 1 for spending less to a 3 for spending more. For example, in the second to last row, the 2.06 value in the first column indicates that among the 49 Mississippians who rated universal health care as Not Important, their average preference was that government should spend about the same as it currently was spending on health care and hospitals. The 2.86 value in the last column of that same row indicates that among the 198 Mississippians who rated universal health care as Very Important, their average preference was that government should spend more than it currently was on health care and hospitals. * Pearson correlation between health care spending and health care priority item was statistically significant at .001 level. + Kendall’s tau-b was significant at .061 level.

Table 2. Dimensions of Mississippians’ Attitudes toward

Health Care Programs

(Pearson correlations for masses and elites of all parties)

	Public Education	Preventive Care	Minority Programs	Recruit Doctors	Children Programs	Poor Adults	Universal Health Care
Public Education	-
Preventive Care	.47	-
Minority Programs	.38	.43	-
Recruit Doctors	.25	.33	.21	-
Children Programs	.38	.35	.36	.22	-
Poor Adults	.31	.37	.39	.16	.59	-
Universal Health Care	.32	.27	.32	.12	.43	.59	-
Note: Cell entries are the Pearson r correlation coefficients computed for each pair of items. The sample combines both legislators and the public of all political parties. The average intra-cluster correlation among the three “proactive” questionnaire items was .43. The average intra-cluster correlation among the three “services” questionnaire items was .54. The average inter-cluster correlation between items from different clusters was .29. A principal components factor analysis with varimax rotation also produced two factors or dimensions with the same proactive and services items, though it showed the recruit doctors item loading on the proactive factor. Yet the recruit doctors item’s average correlation with items in the proactive cluster was only .26, lower than the average .29 correlation between items from different clusters, and its soon-to-be-discussed unique interrelationship with mass and elite partisanship suggests that it constitutes a separate dimension (Clausen 1973, 31, 35, 168, 213, 237).