¹ See, for example, Carleton, David and Stohl, Michael, “The Foreign Policy of Human Rights: Rhetoric and Reality from Jimmy Carter to Ronald Reagan,” Human Rights Quarterly 7 (May 1985)CrossRefGoogle Scholar; and Mitchell, Neil J. and McCormick, James M., “Economic and Political Explanations of Human Rights Violations,” World Politics 40 (July 1988)CrossRefGoogle Scholar.

² Poe, and Tate, , “Repression of Human Rights to Personal Integrity in the 1980s: A Global Analysis,” American Political Science Review 88 (December 1994), 853CrossRefGoogle Scholar.

³ Ibid., 866–67.

⁴ See Carleton and Stohl (fn. 1); Cingranelli, David L. and Pasquarello, Thomas E., “Human Rights Practices and the Distribution of Foreign Aid to Latin American Countries,” American Journal of Political Science 29 (August 1985)CrossRefGoogle Scholar; and Henderson, Conway, “Conditions Affecting the Use of Political Repression,” Journal ofConflict Resolution 35 (March 1991)Google Scholar.

⁵ Poeand Tate (fn. 2), 853.

⁶ Vallier, , “Empirical Comparisons of Social Structure: Leads and Lags,” in Vallier, ed., Comparative Methods in Sociology (Los Angeles: University of California Press, 1971), 223Google Scholar.

⁷ Jackman, Robert, “Cross-National Statistical Research and the Study of Comparative Politics,” AmericanJournalofPolitical Science 29 (February 1985), 169Google Scholar, emphasis in original.

⁸ Verba, Sidney, Nie, Norman H., and Kim, Jaeon, Participation and Political Equality: A Seven-Nation Comparison (Cambridge: Cambridge University Press, 1978), 52Google Scholar.

⁹ Verba, Sidney, Nie, Norman H., and Kim, Jaeon, The Modes ofDemocratic Participation: A Cross-National Comparison (Beverly Hills, Calif.: Sage Publications, 1971), 8Google Scholar.

¹⁰ See Leighley, Jan E., “Field Essay: Attitudes, Opportunities and Incentives: A Field Essay on Political Participation,” Political Research Quarterly 48 (March 1995), 182CrossRefGoogle Scholar. See also Aldrich, John, “Rational Choice and Turnout,” AmericanJournal ofPolitical Science 37 (February 1993)Google Scholar.

¹¹ Poe and Tate (fn. 2), 855.

¹² In fact, the logic of the argument would extend to disaggregating torture from killing as a third dimension.

¹³ Kausikan, , “Asia's Different Standard,” Foreign Policy 92 (Fall 1993), 32Google Scholar.

¹⁴ Ibid., 39–40.

¹⁵ One analysis of the use of torture by police in India argues that because they have low status and low pay “as with many other Indian officials they feel driven to supplement their incomes. The detainees themselves, or their families, are threatened with torture if they do not bribe the police—a threat that can work only if those who do not pay, or cannot pay, are in fact tortured.” See David J. Rothman and Aryeh Neier, “India's Awful Prisons,” New York Review of Books (May 16, 1991), 54.

¹⁶ Poeand Tate (fn. 2), 867.

¹⁷ Ibid., 867–68.

¹⁸ See Mitchell and McCormick (fn. 1), 483–85.

¹⁹ Quantitative levels were employed to make this gradation. See Mitchell and McCormick (fn. 1), 485 n. 17.

²⁰ For a discussion of the logical criteria of classification, see Kalleberg, Arthur L., “The Logic of Comparison: A Methodological Note on the Comparative Study of Political Systems,” World Politics 19 (October 1966)CrossRefGoogle Scholar.

²¹ It is not simply genocidal regimes such as that in Nazi Germany or in Cambodia under Pol Pot that might aim for this combination of human rights violations. Take, for example, Amnesty International's description of human rights violations in the Philippines in its 1988 report. Descriptions focus almost entirely on extensive killings and disappearances, not on imprisonment of political dissidents.

²² There were 115 cases to analyze for 1984 and 125 cases for 1987. For each year, there were a small number of cases that Poe and Tate did not include because they lacked data on one or more of the independent variables they were analyzing.

²³ Mitchell and McCormick (fn. 1), 485 n. 17.

²⁴ Bollen, Kenneth A., “Issues in the Comparative Measurement of Democracy,” American Sociological Review 45 (June 1980)CrossRefGoogle Scholar.

²⁵ We also ran a Kolmogorov-Smirnov Z test of normality for the dependent variable, the composite scale, for 1984 and 1987. In both instances, the scale was not normally distributed and thus some caution must accompany the interpretation of our regression results. Because we are working with a population of data and because we are employing a series of other tests, we believe that it is useful to report the regression results to obtain an overall portrait of the relationship among the three human rights scales.

²⁶ As a further check on our analysis, we employed a discriminant analysis using the prisoner scale and the torture scale in separate runs as predictors of the placement of countries on the composite scale. In general, the results imply that both the prisoner and torture scales are more predictive of values at the high and low ends of the composite scale. While these results are not wholly consistent with our other analyses, they do suggest some distinctive emphases within the composite scale.

²⁷ The data are available through the following website: http://www.psci.unt.edu/ihrsc/. For ease of analysis, we truncated some of Poe and Tate's independent variables to two decimal places. Such a design decision should not affect the interpretations offered here.

²⁸ On this point, see Poe and Tate (fn. 2), 856–57.

²⁹ Poe and Tate operationalize their influential civil war variable in terms of number of deaths, the government “involved as a direct participant in the war,” and the assumption that “there must be effective resistance” on the nongovernment side. This may provide them a distinction between genocide and civil war, as they assert, but it does not permit the confident assertion that civil war as a “concept is kept distinct from our dependent variable”; Poe and Tate (fn. 2), 859. There is likely considerable circularity between the measurement of civil war and the measure of government “repression of human rights to personal integrity,” particularly at the high end (values 4 and 5 of the dependent variable). Thus, the interpretation of the relationship between the civil war variable and the dependent variable should be treated carefully.

Finally, we should note that while Table 4 reports a significant relationship for leftist government control and the unidimensional measure of human rights violations for 1984, the coefficient has the wrong sign, as it had in Poe and Tate's pooled analysis; Poe and Tate (fn. 2), 861.

³⁰ We computed t-tests for the differences between the unstandardized coefficients (assuming independence) for each model with the other two (that is, the composite model with the prison model, the composite model with the torture model, and the prison model with the torture model) for each year. For the 1984 data, three differences were significant: the coefficients for the LEFT variable for the composite and prison measures were significantly different from one another at the .05 level, the coefficients for the LPOP variable with the composite and torture measures were significantly different at the .10 level, and the coefficients for the LEFT variable with the prison and torture models were signifi-candy different at the .10 level. For the 1987 data, three were significant as well: the coefficients for the CWAR variable for the composite and prison models (.01), the coefficients for the LPOP variable for the composite and torture models (.01), and the coefficients for the LEFT variable for the prison and torture models (.10).

Following Carmines and Zeller, we also computed a series of bivariate correlations for the independent variables with each of the dependent measures as another way to test the differences among the various models. See Carmines, Edward G. and Zeller, Richard A., Reliability and Validity Assessment (Newbury Park, Calif.: Sage Publications, 1979), 17–27, 66–70, esp. 68CrossRefGoogle Scholar. To evaluate the differences for each of these bivariate relationships, we compared the unstandardized coefficients (assuming independence) from the three bivariate regression models for each variable. Comparable results obtained, with the exceptions that the VANHDEMO variable in 1984 was now significant at the .10 level for the composite model and the prison model comparison and for the composite model and torture model comparison, and that the LPOP variable comparison was now not significant in 1984. For 1987 CWAR is now significant for the composite model and prison model at .05 level (instead of .01), and CWAR is now significant for the prison and torture model comparison at the .10 level. LPOP for the composite model and torture model comparison is now significant at the .10 level (instead of .01) in 1987.

³¹ See Carmines and Zeller (fn. 30), 26. Also, as Poe and Tate (fn. 2) point out, “The development of theories to explain … such crimes … would seem to be a vital undertaking, [yet] social science scholars have only begun to use the newly developed information toward this end” (p. 853). Unfortunately, assessing the construct validity of a concept “is, by necessity theory-laden.… In a very real sense, whenever one assesses the construct validity of the measure of interest, one is also evaluating simultaneously the construct validity of measures of the other theoretical concepts”; Carmines and Zeller (fn. 30), 23–25.

³² Vallier (fn. 6), 223.