¹ The literature is reviewed by Miller, William L., ‘Studying How the Economy Affects Attitudes and Behavior: Problems and Prospects’, in Clarke, Harold D., Stewart, Marianne C. and Zuk, Gary, eds, Economic Decline and Political Change (Pittsburgh: University of Pittsburgh Press, 1989), pp. 143–72.Google Scholar See also Clarke, Harold D. et al. , Controversies in Political Economy: Canada, Great Britain, the United States (Boulder, Colo.: Westview Press, 1992)Google Scholar; Norpoth, Helmut, Lewis-Beck, Michael and Lafay, Jean-Dominique, Economics and Politics: The Calculus of Support (Ann Arbor: University of Michigan Press, 1991).Google Scholar

² See, inter alia, Beck, Nathaniel, ‘The Economy and Presidential Approval: An Information Theoretic Perspective’Google Scholar in Norpoth, , Lewis-Beck, and Lafay, , eds, Economics and Politics, pp. 85–102Google Scholar; Beck, Nathaniel, ‘Comparing Dynamic Specifications: The Case of Presidential Approval’, in Stimson, James A., ed., Political Analysis, vol. 3 (Ann Arbor: University of Michigan Press, 1992), pp. 51–88Google Scholar; Chappell, Henry W. and Keech, William, ‘A New View of Political Accountability for Economic Performance’, American Political Science Review, 79 (1985), 10–17CrossRefGoogle Scholar; MacKuen, Michael, ‘Political Drama, Economic Conditions, and the Dynamics of Presidential Popularity’, American Journal of Political Science, 27 (1983), 165–91CrossRefGoogle Scholar; Mueller, John E., ‘Presidential Popularity from Truman to Johnson’, American Political Science Review, 64 (1970), 18–34CrossRefGoogle Scholar; Mueller, John E., Wars, Presidents and Public Opinion (New York: John Wiley & Sons, 1973)Google Scholar; Norpoth, Helmut, ‘Economics, Politics, and the Cycle of Presidential Popularity’, Political behavior, 6 (1984), 253–73CrossRefGoogle Scholar; Norpoth, Helmut and Yantek, Thorn, ‘Macro-Economic Conditions and Fluctuations in Presidential Popularity: The Question of Lagged Effects’, American Journal of Political Science, 27 (1983), 785–807CrossRefGoogle Scholar; Ostrom, Charles and Simon, Dennis, ‘Promise and Performance: A Dynamic Model of Presidential Popularity’, American Political Science Review, 79 (1985), 334–58CrossRefGoogle Scholar; Whiteley, Paul, ‘Inflation, Unemployment and Government Popularity: Dynamic Models for the United States, Britain and West Germany’, Electoral Studies, 6 (1984), 395–410Google Scholar; Yantek, Thom, ‘Polity and Economy Under Extreme Economic Conditions: A Comparative Study of the Reagan and Thatcher Experiences’, American Journal of Political Science, 32 (1988), 196–216.CrossRefGoogle Scholar

³ MacKuen, Michael B., Erikson, Robert S. and Stimson, James A., ‘Macropartisanship’, American Political Science Review, 83 (1989), 1125–42.CrossRefGoogle Scholar

⁴ Weisberg, Herbert F. and Smith, Charles E. Jr., ‘The Influence of the Economy on Party Identification in the Reagan Years’, Journal of Politics, 53 (1991), 1077–92.CrossRefGoogle Scholar

⁵ Many of the major theoretical, conceptual and methodological issues in the debate are summarized in Asher, Herbert B., ‘Voting Behavior Research in the 1980s: An Examination of Some Old and New Problem Areas’, in Finifter, Ada W., ed., Political Science: The State of the Discipline (Washington, DC: American Political Science Association, 1983), pp. 339–88.Google Scholar See also Abramson, Paul R., ‘Generational Change and the Decline of Party Identification in America: 1952–1974’, American Political Science Review, 70 (1976), 469–78CrossRefGoogle Scholar; Abramson, Paul R. and Ostrom, Charles W. Jr, ‘Macropartisanship: An Empirical Reassessment’, American Political Science Review, 85 (1991), 181–92CrossRefGoogle Scholar; Abramson, Paul R. and Ostrom, Charles W. Jr, ‘Response’, American Political Science Review, 86 (1992), 481–5Google Scholar; Achen, Christopher H., ‘Social Psychology, Demographic Variables, and Linear Regression: Breaking the Iron Triangle in Voting Research’, Political behavior, 14 (1992), 195–212CrossRefGoogle Scholar; Budge, Ian, Crewe, Ivor and Farlie, Dennis, eds, Party Identification and Beyond (New York: John Wiley & Sons, 1976)Google Scholar; Dennis, Jack, ‘Political Independence in America, Part I: On Being an Independent Partisan Supporter’, British Journal of Political Science, 18 (1988), 77–110CrossRefGoogle Scholar; Dennis, Jack, ‘Political Independence in America, Part II: Toward a Theory’, British Journal of Political Science, 18 (1988), 197–220CrossRefGoogle Scholar; Fiorina, Morris, Retrospective Voting in American National Elections (New Haven, Conn.: Yale University Press, 1981)Google Scholar; Franklin, Charles H., ‘Measurement and the Dynamics of Party Identification’, Political behavior, 14 (1992), 297–310CrossRefGoogle Scholar; Green, Donald P. and Palmquist, Bradley, ‘Of Artifacts and Partisan Instability’, American Journal of Political Science, 34 (1990), 872–902CrossRefGoogle Scholar; MacKuen, Michael B., Erikson, Robert S. and Stimson, James A., ‘Macropartisanship’Google Scholar; MacKuen, Michael B., Erikson, Robert S. and Stimson, James A., ‘Question Wording and Macropartisanship’, American Political Science Review, 86 (1992), 475–81CrossRefGoogle Scholar; Miller, Warren E., ‘Party Identification: Back to Basics’, American Political Science Review, 85 (1991), 557–70CrossRefGoogle Scholar; Shively, W. Phillips, ‘The Nature of Party Identification: A Review of Recent Developments’, in Pierce, John C. and Sullivan, John L., eds, The Electorate Reconsidered (Beverly Hills, Calif.: Sage Publications, 1980), pp. 219–36Google Scholar; Weisberg, Herbert, ‘A Multidimensional Conceptualization of Party Identification’, Political behavior, 2 (1980), 33–60.CrossRefGoogle Scholar

⁶ Asher, , ‘Voting Behavior Research in the 1980s’Google Scholar; Beck, Paul Allen, ‘The Dealignment Era in America’, in Dalton, Russell J., Flanagan, Scott C. and Beck, Paul Allen, eds, Electoral Change in Advanced Industrial Democracies (Princeton, NJ: Princeton University Press, 1984), pp. 240–66Google Scholar; Dennis, , ‘Political Independence in America, Part I’Google Scholar; Dennis, , ‘Political Independence in America, Part II’Google Scholar; Keith, Bruce E. et al. , The Myth of the Independent Voter (Berkeley: University of California Press, 1992)Google Scholar; Petrocik, John R., ‘An Analysis of Intransitivities in the Index of Party Identification’, Political Methodology, 1 (1974), 31–48Google Scholar; Shively, , ‘The Nature of Party Identification’Google Scholar; Weisberg, , ‘A Multidimensional Conceptualization’.Google Scholar

⁷ Budge, , Crewe, and Farlie, , eds, Party Identification and BeyondGoogle Scholar; Butler, David and Stokes, Donald E., Political Change in Britain, 2nd college edn (New York: St Martin's Press, 1976)Google Scholar; LeDuc, Lawrence et al. , ‘Partisan Instability in Canada: Evidence From a New Panel Study’, American Political Science Review, 78 (1984), 470–84.CrossRefGoogle Scholar

⁸ Franklin recently has demonstrated the validity of the traditional 3- and 7-point party identification scales and, hence, the distinction between weak identifiers and leaning independents. See Franklin, , ‘Measurement and the Dynamics of Party Identification’, p. 306.Google Scholar Note also the inconsistency in the treatment of the face validity of party identification questions by analysts who use the observation that some of the attitudes and behaviour of ‘leaning’ independents resemble those of weak party identifiers to claim that the former are really identifiers. This observation might be employed with equal force to argue that the weak identifiers should be classified as independents. Presumably, the reason such arguments have not been made is because weak identifiers say that they identify with a party.

⁹ See, for example, Abramson, , ‘Generational Change’, p. 473Google Scholar; Beck, , ‘The Dealignment Era in America’, p. 250Google Scholar; Norpoth, Helmut and Rusk, Jerrold, ‘Partisan Dealignment in the American Electorate: Itemizing the Deductions Since 1964’, American Political Science Review, 76 (1982), 522–38, p. 536.CrossRefGoogle Scholar

¹⁰ See, for example, Converse, Philip E., The Dynamics of Party Support: Cohort-Analyzing Party Identification (Beverly Hills, Calif.: Sage Publications, 1976), pp. 91–7.Google Scholar

¹¹ Beck, , ‘The Dealignment Era in America’, p. 261.Google Scholar

¹² Niemi, Richard, Katz, Richard and Newman, David, ‘Reconstructing Past Partisanship: On the Failure of the Party Identification Recall Question’, American Journal of Political Science, 24 (1980), 633–51.CrossRefGoogle Scholar

¹³ See, for example, Franklin, , ‘Measurement and the Dynamics of Party Identification’, pp. 306–7.Google Scholar

¹⁴ The Gallup question is: ‘In politics, as of today, do you consider yourself a Republican, a Democrat, or an Independent?’

¹⁵ Abramson, and Ostrom, , ‘Macropartisanship: An Empirical Reassessment’.Google Scholar In a rebuttal, MacKuen et al. contend that CBS-New York Times party identification data gathered using a virtually identical question wording to the NES show that misgivings about wording differences are unfounded. Abramson and Ostrom have responded by reiterating their earlier claim on this point, as well as that concerning the lack of significant relationships between NES-type party identification variables and the economic measure used by MacKuen et al. See Abramson, and Ostrom, , ‘Response’Google Scholar; MacKuen, , Erikson, and Stimson, , ‘Question Wording and Macropartisanship’.Google Scholar

¹⁶ The relationship between the Gallup independent series and the NES pure independent series is also very strong (r = + 0.84) but, as Figure 1 shows, the Gallup numbers are consistently much larger, the mean difference being 16.6 per cent.

¹⁷ The Durbin-Watson d = 2.15 indicated that (first order) serial correlation does not pose a threat to inference. Also, all residuals are quite small (maximum = -3.8 per cent in 1962).

¹⁸ The inflation and unemployment data are obtained from the Citibase data tape. Inflation is measured as the annualized percentage increase in the consumer price index, and unemployment is the percentage wholly unemployed.

¹⁹ The Durbin-Watson d's are 1.52 (NES analysis) and 1.52 (Gallup analysis) respectively. Since the first of these test statistics is only slightly below, and the second, exceeds, the upper bound of the critical values for the test (d₁ = 1.046, d_u = 1.535), we conclude that the analyses are not confounded by (first-order) serial correlation.

²⁰ See, for example, Brody, Richard A., Assessing the President: The Media, Elite Opinion and Public Support (Palo Alto, Calif.: Stanford University Press, 1991)Google Scholar; Iyengar, Shanto and Kinder, Donald R., News That Matters: Television and American Opinion (Chicago: University of Chicago Press, 1987)Google Scholar; Miller, William L., Media and Voters (Oxford: Clarendon Press, 1991)Google Scholar; Mosley, Paul, ‘“Popularity Functions” and the Role of the Media: A Pilot Study of the Popular Press’, British Journal of Political Science, 14 (1984), 117–29CrossRefGoogle Scholar; Sanders, David, Marsh, David and Ward, Hugh, ‘The Electoral Impact of Press Coverage of the British Economy, 1979–87’, British Journal of Political Science, 23 (1993), 175–210.CrossRefGoogle Scholar

²¹ Although we model a number of political interventions, we do not include presidential approval because Granger causality tests suggest that the flow of causality is from independence to presidential approval rather than vice versa. Using four lags on the quarterly independence and presidential approval series, the test statistic for lagged effects of approval on independence is F_4,129 = 2.07, p = 0.092. The comparable test statistic for lagged effects of independence on approval is F_4,129 = 3.68, p = 0.007. See Freeman, John R., ‘Granger Causality and the Time Series Analysis of Political Relationships’, American Journal of Political Science, 27 (1983), 327–58.CrossRefGoogle Scholar

²² The variable is analysed in (natural) log form.

²³ The scandals include Sherman Adams (1959Q1), Walter Jenkins (1964Q4), Watergate (1973Q2–1975Q4), Bert Lance (1977Q3–1977Q4), EPA (1983Q1) and Iran-Contra (1986Q4–1988Q2).

²⁴ Mueller, , ‘Presidential Popularity from Truman to Johnson’Google Scholar; Mueller, , War, Presidents and Public Opinion.Google Scholar

²⁵ Aggregate analyses do not enable us to distinguish the alternative of no net effect from the simple null hypothesis of no effect whatsoever.

²⁶ There is no ‘canonical list’ of rally events. MacKuen, Erikson and Stimson, for example, include fourteen events. See ‘Macropartisanship’, p. 1140, n. 9.Google Scholar Here, we use their list plus fourteen others (Vietnam is considered separately). The events are: Korean War ceasefire (1953Q3), Eisenhower's heart attack (1953Q4), Hungary-Suez crisis (1956Q4), sending of Marines to Lebanon (1958Q3), Khrushchev's visit to the United States (1959Q3), Paris Summit/U-2 incident (1960Q2), Bay of Pigs invasion (1961Q2), Berlin crisis (1961Q3), Cuban missile crisis (1962Q4), civil rights march (1963Q3), Kennedy assassination (1963Q4–1964Q1), Gulf of Tonkin incident (1964Q3), sending of Marines to Dominican Republic (1965Q2), Glassboro summit (1967Q2), Detroit-Newark riots (1967Q3), announcement of Nixon's wage and price control program (1971Q3), Nixon's China visit (1972Q1), Nixon's Soviet Union visit (1972Q3), Vietnam War ceasefire (1973Q1), Mayaguez incident (1975Q2), Ford's China visit (1976Q1), Camp David Accord (1978Q3), Iranian hostage crisis (1979Q4), Reagan assassination attempt (1981Q2), KAL 007 incident (1983Q3), Beirut bombing-Grenada invasion (1983Q4), TWA hijacking (1985Q2), Achille Lauro terrorist capture/Reagan-Gorbachev summit (1985Q4), Challenger explosion (1986Q1), attack on Libya (1986Q2).

²⁷ On rational choice theories of party identification, see, for example, Achen, , ‘Social Psychology’Google Scholar; Fiorina, , Retrospective Voting, pp. 65–83.Google Scholar The classic statement of the social psychological approach is Campbell, Angus, Converse, Philip E., Miller, Warren E. and Stokes, Donald E., The American Voter (New York: John Wiley & Sons, 1960), pp. 120–67.Google Scholar

²⁸ On the theoretical basis of the partial adjustment model, see Gujarati, Damodar N., Basic Econometrics (New York: McGraw-Hill, 1988), pp. 520–2.Google Scholar The model enables us to analyse how short-term economic and political forces affect the percentage of independents in the electorate net of any longer-term forces that may be operating. The partial adjustment model resembles an infinite distributed lag model which, via the Koyck transformation, may be rendered in auto-regressive form. However, unlike the latter, the partial adjustment model does not necessarily have a first-order moving-average error term. Rather, if e_t satisfies conventional linear regression assumptions, φe_t will do so as well, and least square estimates will be consistent. See Gujarati, , Basic Econometrics, p. 523.Google Scholar

²⁹ Breusch, T. S., ‘Testing for Autocorrelation in Dynamic Linear Models’, Australian Economics Papers, 17 (1978), 334–55CrossRefGoogle Scholar; Godfrey, L. G., ‘Testing Against General Autoregressive and Moving Average Error Models When the Regressors Include Lagged Dependent Variables’, Econometrica, 46 (1978), 1293–302.CrossRefGoogle Scholar

³⁰ Goldfeld, Stephen M. and Quandt, Richard E., ‘Techniques for Estimating Switching Regressions’, in Goldfeld, S. M. and Quandt, R. E., eds, Studies in Nonlinear Estimation (Cambridge, Mass.: Ballinger Publishing Company, 1976), pp. 3–35.Google Scholar

³¹ The log-like function is: where n is the number of observations, and and are the variance of the residuals in the old and new regimes, respectively.

³² The value of D(t) increases from 0.173 in 1967Q1 to 0.309 in 1967Q2 to 0.999 in 1967Q3. Since the point where D(t) = 0.5 is where the probit curve has its inflection, these figures indicate that 1967Q2 is the most likely switching date.

³³ The log-likelihood ratio test statistic = are the variance of the residuals in where L* is the log-likelihood function of the regression without accounting for regime change, and L is the log-likelihood function of the switching regression. The statistic is distributed as X² with degrees of freedom equal to the 2 plus the sum of the number of regressors and the number of regimes. Here, L* = -267.09 and L = -245.82, and = 42.56 (p < 0.001), thus rejecting the equality restriction between the two regimes.

³⁴ The rapid evolution of D(t) (see fn. 32 above) indicates that the transition to the new independence regime was largely competed after four quarters.

³⁵ As a check on the switching regression estimates, we performed separate OLS regression for the two time periods. The results are very similar to those discussed here.

³⁶ The effects cumulate over time in the model through the presence of the lagged endogenous variable (INDEP_t-1) with the size of its coefficient (l-φ) = β_y governing the magnitude of the lagged effects. If β_x is the coefficient for the independent variable of interest, the effect after four quarters is See, for example, Stewart, Jon, Econometrics (Deddington, Oxon: Philip Allan, 1991), p. 178.Google Scholar

³⁷ Recall that the Vietnam variable is logged but the dependent variable is not. Thus, the immediate effect is β_x * relative change in Vietnam = absolute change in independence. See Gujarati, , Basic Econometrics, p. 148.Google Scholar Effects in subsequent periods are calculated taking account of the lagged endogenous variable (see fn. 36 above).

³⁸ Beck, , ‘The Dealignment Era in America’.Google Scholar

³⁹ These findings differ from those for Britain where recent analyses indicate long-term (1951Q1–1989Q4) stability in the magnitude of the impact of economic conditions on party support (measured as vote intention). See Price, Simon and Sanders, David, ‘Modeling Government Popularity in Postwar Britain: A Methodological Example’, American Journal of Political Science, 37 (1993), 317–34, at pp. 331–2.CrossRefGoogle Scholar

⁴⁰ For the first period, the range is 18.3 per cent and 28.3 per cent and σ² is 3.88; for the second, the range is 25.3 per cent to 35.3 per cent and σ² is 4.24.