Attitudes and Measurement Error Revisited: A Reply to Johnston and Pattie

In a recent Note in this Journal, Johnston and PattieR. J. Johnston and Charles Pattie, ‘Inconsistent Individual Attitudes within Consistent Attitudinal Structures: Comments on an Important Issue Raised by John Bartle's Paper on Causal Modelling of Voting in Britain’, British Journal of Political Science, 30 (2000), 361–74. contend that they have discovered an ecological fallacy in the behaviour of the six-item scaleJohnston and Pattie's analysis in fact deals with two different six-item scales which were included in alternate years on the BHPS. As their conclusions applied equally to both scales, for the sake of parsimony this note refers only to the scale included in ‘odd’ years 1991, 1993, 1995, etc. developed by Heath et al. to measure the ‘left–right’ political value dimension.Anthony Heath, Geoffrey Evans and Jean Martin, ‘The Measurement of Core Beliefs and Values: The Development of Balanced Socialist/Laissez Faire and Libertarian/Authoritarian Scales’, British Journal of Political Science, 24 (1994), 115–32. Using data from the first six waves of the British Household Panel Study (BHPS), they show that, while there is remarkable over-time stability in the factor structure of these questions at the aggregate level, when the consistency of individual responses to each item is considered, a very different picture emerges; around 50 per cent of the sample fail to select the same response alternative on successive waves and a third of respondents select a response alternative on the opposite side of the agree/disagree scale from one time to the next. Correlations between the same items over time of around 0.4, they argue, bear out a picture of massive longitudinal instability at the individual level.