¹ Eglinton was chosen because it had elected a Liberal in the federal election of 1965 and a Conservative in the provincial election of 1963; Wellington South was chosen because it had elected a Conservative in 1965 and a Liberal in 1963. We wanted this control because our interest was in change, per se, rather than the particular pattern in aggregate shifts within the province. The surveys provided us with 236 usable returns, 100 from Eglinton and 136 from Wellington South. This was a collective return of 59 per cent, consisting in individual rates of 50 per cent for Eglinton and 68 per cent for Wellington South. The samples were gathered during the late summer of 1967, before the date of the provincial general election was announced. The summer vacation exodus from Toronto accounts for the lower return rate from Eglinton where considerable difficulty was encountered in tracking down respondents. Interviewing was done in Eglinton by Brian McFadzen and in Wellington South by Miss Peppin. Both used their samples to produce BA theses for the Department of Political Studies, Queen's University. A small grant from the Canada Council to Perlin permitted the employment of Miss Peppin for two months to integrate the two sets of data and write the programs for the collective analysis.

² We have no data which permit us to make demographic tests of representativeness, and the vote distributions within the samples contain significant error. In Wellington South for 1963 the Liberal vote was over-represented by 13 per cent, the Conservatives and New Democrats each being underrepresented by approximately 6.5 per cent; and for 1965 the Conservative vote was overrepresented by 6.8 per cent, the Liberal vote was overrepresented by 9.4 per cent, and the NDP vote was underrepresented by 16.2 per cent. An overrepresentation of close to 13 per cent in turnout for both elections is clearly one factor in this distortion. Recall error is certainly another. Allowing for these factors, our main concern becomes the underrepresentation of the NDP. When we look at vote intentions for 1967 and the outcome of the 1967 election we find almost as large an error (14.2 per cent) in NDP representation as we had in 1965. Our sample is obviously biased against the NDP.

In Eglinton for 1963 the Conservative vote is overrepresented by 5.2 per cent, the Liberal vote is underrepresented by 4.6 per cent, and the NDP vote is underrepresented by 0.6 per cent. Given about a 10 per cent overrepresentation of turnout these seem to be reasonable error margins. For 1965, how-ever, the Liberal vote is overrepresented by 16.1 per cent, the Conservative vote is underrepresented by 12.6 per cent, and the NDP vote is underrepresented by 3.5 per cent. The overrepresentation of turnout in this case is obviously the critical factor since it is of the order of 15 per cent.

Despite these discrepancies we believe our data are still useful for the kind of analysis we are pursuing here. There are two reasons. First, there is some reassurance to be found in the fact that despite the proportional deviations the relative positions of the parties in both ridings for all elections are consistent with actual outcomes. Second, to the extent they understood the questions asked them, respondents who expressed different preferences at the two levels had to believe that this was an accurate representation of what they had done or what they would have done had they voted. Thus inter-level change was a possibility for them and, for our purposes, therefore, they are still of interest.

³ John Wilson and David Hoffman point to the importance of variations in turnout in their analysis of the support bases of the Liberal party in Ontario. “The Liberal Party in Contemporary Ontario Politics,” this JOURNAL, III, no. 2 (June 1970), 181–90.Google Scholar

⁴ The high level of partisan change which we found is consistent with the findings of Laponce, J. A. in Vancouver-Burrard. La-ponce reports a shift of 49 per cent in party preferences between the federal and provincial elections of 1963. People vs Politics (Toronto, 1969), 169.Google Scholar John C. Courtney and David E. Smith report a shift of 18.6 per cent between the April 22 provincial election and June 22 federal by-election in Saskatoon City in 1964. “Voting in a Provincial General Election and a Federal By-election: A Constituency Study of Saskatoon City,” Canadian Journal of Economics and Political Science, XXXII, no. 3 (Aug. 1966), Table VII, 344.

⁵ One of the more intriguing variants is R. MacGregor Dawson's postulation of a general cyclical theory of electoral change in Canada. Dawson suggested that opposition to the federal ruling party becomes manifest first in the casting of votes for other parties in provincial elections. “… provincial governments will begin to fall away to the opposition party or parties until these are in a majority; [and then] … there is an overturn in the Dominion Parliament which brings it once more into sympathy with the provinces, whereupon the cycle begins anew.” Dawson, R. MacGregor, The Government of Canada, revised by Norman Ward (5th ed., Toronto, 1970), 486.Google Scholar Angus Taylor in another BA thesis done for Perlin in 1968 presented evidence which appears to refute the Dawson theory. From a statistical analysis of the relationships between trends in federal and provincial election results in the period from 1924 to 1963 Taylor concluded that while “there is a significant relation between trends in federal and provincial elections … The direction and strength in inter-level influence varies by party and province.” Angus Taylor, M., A Comparison of Federal and Provincial Voting Trends in Canada, unpublished BA thesis, Department of Political Studies, Queen's University, 1968.Google Scholar

⁶ “The Liberal Party in Contemporary Ontario Politics,” 198–9.

⁷ People vs Politics, 170.

⁸ Respondents were classified on the basis of their answers to the following question. “Basically what do you think the competition between politicians at election time is about? (a) basic ideas and values, like liberty or equality; (b) what particular groups of people, say farmers or trade unionists or businessman, want; (c) what position the government is to take on particular issues, for example, whether Canada should stay in NORAD or how much money the province should spend on education; (d) which group of men is to have power.” Respondents were said to have some degree of ideological orientation if they included the first statement in their response.

⁹ To measure the information level of respondents we gave them the names of ten federal and ten provincial politicians and asked them for each whether he was in federal or provincial politics, what his party was, and what position he held in the government or his party.

¹⁰ This is not really surprising since we found interest and information to be related to each other – the higher the interest the more likely the respondent was to have a higher level of information. The x² test yielded a score significant at .001.

¹¹ People vs Politics, 176.

¹² Campbell, Anguset al., The American Voter (New York, 1960Google Scholar).