1 For details on British Columbia geography see Robinson, J. Lewis (ed.), Studies in Canadian Geography: British Columbia (Toronto: University of Toronto Press, 1972)Google Scholar. On the development of the transportation and communications network, see Harold A. Innis, “Settlement and the Mining Frontier,” Part 2 in Lower, A. R. M. and Innis, Harold A., Settlement of the Forest Frontier in Eastern Canada and Settlement and the Mining Frontier (Toronto: Macmillan, 1936)Google Scholar, chaps. 5 and 6; Howay, F. W., Sage, W. N., and Angus, H. F., British Columbia and the United States: The North Pacific Slope from Fur Trade to Aviation, ed. by Angus, H. F. (Toronto: The Ryerson Press, 1942)Google Scholar, chap. 10; Roy, Patricia E., “The Rise of Vancouver as a Metropolitan Centre, 1886–1929” (B.A. essay, University of British Columbia, 1960), 67–75Google Scholar; Roy, Patricia E., “Railways, Politicians and the Development of the City of Vancouver as a Metropolitan Centre, 1886–1929” (M.A. thesis, University of Toronto, 1963)Google Scholar; Ramsey, Bruce, PGE: Railway to the North (Vancouver: Mitchell Press, 1962)Google Scholar; and this author's thesis, “The Impact of Modernization on British Columbia Electoral Patterns: Communications Development and the Uniformity of Swing, 1903–1975” (Ph.D. thesis, University of British Columbia, 1978), 228–43Google Scholar.

2 Data on telephone traffic volume were provided by Mr. Tony Farr, company historian, B.C. Telephone Company.

3 For 1921 to 1951 see Urquhart, M. C. (ed.), Historical Statistics of Canada (Toronto: Macmillan, 1965)Google Scholar, Table S 235; and for 1961 and 1971, Canada, Dominion Bureau of Statistics (Statistics Canada), The Canada Yearbook, 1963, 1973.

4 Canada, Dominion Bureau of Statistics (Statistics Canada), The Canada Yearbook, various years.

5 For 1931, Canada, Dominion Bureau of Statistics, The Census of Canada, 1931, Vol. 5Google Scholar, Table 57; and for 1941, Canada, Dominion Bureau of Statistics, The Census of Canada, 1941, Vol. 9Google Scholar, Table 18.

6 See Canada, Air Transport Board, Origin and Destination Statistics: Mainland Scheduled Traffic Suwey of Revenue Passengers, 1955–59 (Ottawa, 1961)Google Scholar; Canada, Air Transport Board (Committee); Airline Passenger Origin and Destination Statistics: Domestic Report, 1966 (1967) (Ottawa, n.d.); and Canada, Statistics Canada, A irline Passenger Origin and Destination Statistics: Domestic Report, 1975, Catalogue No. 51–204 (Ottawa, 1976)Google Scholar. The airports surveyed were Castlegar, Cranbrook, Fort Nelson, Fort St. John, Kamloops, Penticton, Prince George, Prince Rupert, Quesnel, Sandspit, Smithers, Terrace and Williams Lake.

7 See, for example, Rokkan, Stein, with Campbell, Angus, Torsvik, Per and Valen, Henry, Citizens, Elections, Parties: Approaches to the Comparative Study of the Processes of Development (New York: David McKay Co., 1970)Google Scholar; Deutsch, Karl W., “Social Mobilization and Political Development,” American Political Science Review 55 (1961), 493–514CrossRefGoogle Scholar; Deutsch, Karl W., Nationalism and Social Communication: An Inquiry into the Foundations of Nationality (Cambridge: MIT Press, 1953)Google Scholar: and essays in Jacob, Philip E. and Toscano, James V. (eds.), The Integration of Political Communities (Philadelphia: J. B. Lippincott, 1964)Google Scholar. For a good review of the literature on the relationship between geography and political integration see Taylor, R. J. and Johnston, R. J., Geography of Elections (Harmondsworth: Penguin Books, 1979)Google Scholar, Part 2.

8 Cox, Kevin R., “The Spatial Structuring of Information Flow and Partisan Attitudes,” in Dogan, Mattei and Rokkan, Stein (eds.), Social Ecology (Cambridge: MIT Press, 1969)Google Scholar.

9 Stokes, Donald E., “Parties and the Nationalization of Electoral Forces,” in Bumham, Walter Dean and Chambers, William Nisbet (eds.), The American Party Systems: Stages of Political Development (New York: Oxford University Press, 1967), 182–202Google Scholar; and Stokes, Donald E., “A Variance Components Model of Political Effects,” in Claunch, John M. (ed.), Mathematical Applications in Political Science (Dallas: Arnold Foundation, 1965), 61–85Google Scholar.

10 For a review of such theories see Abler, Ronald, Adams, John S., and Gould, Peter, Spatial Organization: The Geographer's View of the World (Englewood Cliffs: Prentice-Hall, 1971)Google Scholar, chaps. 7 and 11.

11 Our premise is that insular places are less likely to be in contact with higher order forces. An alternative conceptualization might suggest that extra-local communications do penetrate these communities. However, because of the persistence of local particularisms, these forces may have a different impact than they would in less isolated communities with similar characteristics. Thus our conceptual schema could be refined to allow us to distinguish between electoral insularity and what may be termed electoral “innoculation.” The scope of this project does not, however, allows us to ope rationalize this or other theoretical refinements.

12 Tukey, John W. and Wilk, M. B., “Data Analysis and Statistics: Techniques and Approaches,” in Tufte, Edward R. (ed.), The Quantitative Analysis of Social Problems (Reading, Mass.: Addison-Wesley, 1970)Google Scholar.

13 The constituencies considered are Albemi, Atlin, Boundary-Similkameen, Cariboo, Columbia River, Comox, Cowichan-Malahat, Fort George, Kamloops, Kootenay, Mackenzie, Nelson-Creston, North Okanagan, North Peace River, Omineca, Prince Rupert, Revelstoke-Slocan, Rossland-Trail, Shuswap, Skeena, South Okanagan, South Peace River and Yale-Lillooet.

14 Thus, for example, the part of the Mackenzie riding in the Mount Waddington- Subdivision A census area is one case, the part of Mackenzie in the Sunshine Coast-Subdivision A census area is another, and so on.

15 For the rural area cases we first aggregated the electoral results in the polls situated within the area defined by census and riding boundaries.

16 It should be noted that the opposite logic is sometimes pursued in what might be labelled the “rich get richer” argument. Basically this is the argument that a strong support base provides a context or climate which is favourable to further gains. Within certain support ranges (say perhaps the 20 per cent to 50 per cent range) the forces envisaged in the rich get richer argument might neutralize those introduced in the strata argument but our evidence suggests that, overall, the relationship between swing and E₁ support is negative. See Butler, David and Stokes, Donald, Political Change in Britain (London: Macmillan, 1969), 305–07Google Scholar. For further discussion on these questions, see Wilson, “The Impact of Modernization,” chap. 2.

17 In 1969 there were three-party (Social Credit, NDP and Liberal) races in all 23 ridings examined. Note that in this discussion we ignore the role of independent candidates and those from fifth parties like the Communists.

18 For an analysis of patterns of cross party affinity in an earlier era see Elkins, David J., “Politics Makes Strange Bedfellows: The B. C. Party System in the 1952 and 1953 Provincial Elections,” B. C. Studies, No. 30 (1976), 3–26Google Scholar. See also Ruff, Norman J., “Party Detachment and Voting Patterns in a Provincial Two-Member Constituency: Victoria, 1972,” B. C. Studies, No. 23 (1974), 3–24Google Scholar.

19 For example, the model used in regression analysis of the 1975 swings will include a dummy contest pattern variable which will take the value “1 “in ridings where there were two-party contests in 1975, and the value “0 “otherwise. On the use of dummy variables in regression analysis see Blake, Donald, “The Measurement of Regionalism in Canadian Voting Patterns,” this JOURNAL 5 (1972), 55–81Google Scholar.

20 For a discussion of electoral “safety” indices see Lovink, J. A. A., “Is Canadian Politics Too Competitive,” this JOURNAL 6 (1973), 341–79Google Scholar. An interesting attempt to use a competitiveness variable as a predictor of government outputs in the British Columbia context is Munro, John, “Highways in British Columbia: Economics and Politics,” Canadian Journal of Economics 8(1975), 192–204CrossRefGoogle Scholar. Forresearch which tests the relationship between swing and measures of competitiveness in B.C. elections from 1903 to 1975 see Wilson, “The Impact of Modernization,” chap. 4.

21 See Koenig, Daniel J. and Proverbs, Trevor B., “Class, Regional and Institutional Sources of Party Support within British Columbia,” B. C. Studies, No. 29 (1976), 19–28Google Scholar; Koenig, Daniel J., Martin, Gary R., Goudy, H. G. and Martin, Marlene, “The Year that British Columbia Went NDP: NDP Voter Support Pre- and Post-1972,” B.C. Studies, No. 24 (1974–1975), 65–86Google Scholar.

22 The indicators tested included 10 measures of ethnic-religious composition, 19 measures of occupational or industrial composition, and average family income.

23 See, for example, Black, E. R., “British Columbia: The Politics of Exploitation,” in Shearer, R. A. (ed.), Exploiting Our Economic Potential: Public Policy and the B. C. Economy (Toronto: Holt, Rinehart and Winston, 1968)Google Scholar, esp. 25; Robin, Martin, “The Politics of Class Conflict,” in Robin, Martin (ed.), Canadian Provincial Politics: The Party Systems of the Ten Provinces (Scarborough: Prentice-Hall, 1972), esp. 36–37Google Scholar; and Marchak, Patricia, “Class, Regional and Institutional Sources of Social Conflict in B.C.,” B. C. Studies, No. 27 (1975), 30–49Google Scholar.

24 The constituencies included within each region are: Peace—North Peace River and South Peace River; Northern Interior—Atlin, Fort George, Omineca, and Skeena; Island and Coastal Mainland—Alberni, Cowichan-Malahat, Comox, Mackenzie, and Prince Rupert; Central Interior—Cariboo, Kamloops, and Yale-Lillooet; Okanagan —North Okanagan, South Okanagan, Shuswap, and Boundary-Similkameen; and Kootenays—Columbia River, Kootenay, Nelson-Creston, Revelstoke-Slocan, and Rossland-Trail.

25 Only those variables which could enter with an F value greater than 3.0 were chosen. An F value slightly greater than 3.0 would be required for a significance level of .05 to be attained.

26 This averaging was designed to achieve parsimony in further stages of the analysis, and undertaken after checks indicated that there were positive and statistically significant relationships between the deviation scores calculated forthe 1972 and 1975 swings. The correlation between the deviations for the 1972 and 1975 Social Credit swings is .22 (significant at the .01 level), while the correlation for deviations from NDP swings is . 15 (significant at the .05 level). These findings may appear to indicate “rebound”-voting—that is, the tendency of places which swing strongly towards a party in one election to swing strongly against it in the next, as the party regresses (or progresses) back to its normal support level. This, however, cannot be the case because the effects of such shifts have been removed by inclusion of the base level of support variable in the equations used to produce these residuals. It must be stressed, in other words, that we are dealing here with deviation which remains after the effects of regression away from extreme support levels have been taken into account. While the correlations are not substantial, they are strong enough to suggest a general consistency in the tendency to deviate and to justify hypotheses positing relationships between enduring community characteristics and deviation. At the same time, however, it is clear that some of the electoral deviation has its roots in factors which are salient in one election and not the other. The second stage of our analysis should illustrate some of the reasons for this election-specific deviation, but separate analyses of the four sets of residuals would be needed to identify all cases where election-specific factors had a major impact.

27 On the Vancouver-Victoria metropolis see Walter G. Hardwick, “The Georgia Strait Urban Region,” in Robinson, Studies in Canadian Geography: British Columbia.

28 On regional submetropoles see J. Lewis Robinson, “Areal Patterns and Regional Character”; and K. G. Denike and Roger Leigh,’ ‘Economic Geography, 1960–70,” in Robinson, Studies in Canadian Geography: British Columbia. For research on shifts in the importance of mass media disseminating from metropolitan and submetropolitan centres and an argument about the bearing of these shifts or electoral patterns, see Wilson, “The Impact of Modernization,” 250–62.

29 The following are designated as regional centres—Nanaimo, Victoria, Kamloops, Vancouver, Prince George, Prince Rupert, Daw son Creek, Powell River, Trail or Nelson (whicheverwas closer) forthe Kootenays, and Vernon, KelownaorPenticton (whichever was closer) forthe Okanagan. The judgments concerning the quality of the linkage were made after an impressionistic consideration of maps. We took into consideration the texture of the road (whether it was paved, gravel or dirt, for example), the nature of the terrain, and special circumstances such as the necessity of ferry travel. The exact weight given to the linkage quality variable was based on an estimate of how much normal travel speed on the route would fall below the 50 mile per hour rate. Thus, for example, communities 100 miles away from a regional centre by gravel road were judged to be about as remote as those 150 miles distant by paved road, since travel time in both cases would be about three hours.

30 Cross, W. D., Goulson, C. F., and Loft, A.E. (eds.), The British Columbia Source Book (Victoria, n.p., 1966)Google Scholar.

31 The opposite arguments might be made, of course. Locally generated forces may have less impact on marginal members of society because they are more likely to get their information from the impersonal mass media which, in the cases of smaller places at least, will be beamed into the community from outside. Thus, marginal groups may be more in touch with regional and provincial influence networks than they are with local ones.

32 On the effects of education on participation and interest see, for example, Milbrath, Lester, Political Participation (Chicago: Rand McNally, 1965), 42–43Google Scholar, 53, 54, 122–24; Almond, Gabriel A. and Verba, Sidney, The Civic Culture: Political Attitudes and Democracy in Five Nations (Boston: Little, Brown and Company, 1963), 56–62CrossRefGoogle Scholar, 83–84, 248–49,258–59,315–24; and Verba, Sidney and Nie, Norman H., Participation in America: Political Democracy and Social Equality (New York: Harper and Row, 1972), 97–100Google Scholar, 180–81, 253–59. A counterargument would note that those with lower education are generally less committed in a partisan sense, and suggest that because of this, those with lower education are more likely to be swept up by trends.

33 The counterargument made in introducing the marginality-deviancy linkage is particularly applicable here. Where residential permanency is high locally unique traditions of response may be more likely to develop. And it could be argued further that those who have moved into the community from outside (especially from other parts of the province) should contribute to the erosion of local response patterns because they are more likely to have ties to extra-local influence networks. The 1971 census provides data on migration within the province which allow investigation of these possibilities. Tests with these data lend no support to the idea that insularity declines as the proportion of recent immigrants from other parts of the province increases.

34 On frontier communities see Clark, S. D., The Developing Canadian Community (Toronto: University of Toronto Press, 1962)CrossRefGoogle Scholar, chaps. 1, 5 and 7; Clark, S. D., The Social Development of Canada (Toronto: University of Toronto Press, 1942)Google Scholar, Part 4; Robinson, Ira M., New Industrial Towns on Canada's Resource Frontier (Chicago: University of Chicago, Department of Geography, 1962)Google Scholar; and Lucas, Rex A., Minetown, Milltown, Railtown: Life in Canadian Communities of a Single Industry (Toronto: University of Toronto Press, 1971)Google Scholar.

35 The coefficients were .0018 (F = 9.7, significant at .01) for the remoteness from Vancouver variable, and .0028 (F = 8.7, significant at .01) for the remoteness from regional centre variable. This equation explains 21.1 percent of the variance.

36 The coefficients were .29 (F = 11.8, significant at .001) for the per cent male variable, and .12 (F = 5.5, significant at .05) for the per cent single variable. This equation explains 16.8 per cent of the variance.

37 Across four tests (two swings in two elections), we find that the average deviation score for the 149 cases was 4.5 percentage points. The standard deviation was 2.5 points, so all of the cases selected diverged by an average of more than seven percentage points from the swings expected in the 1972 and 1975 elections.

38 Those cases remaining produced deviation scores which are more than one standard deviation greater than the scores predicted given their scores on the gender and marital status variables, and the equation: Expected deviation = - 13.7 + (.29 x % Male) + (.12 x % Single).

39 Wilson, “The Impact of Modernization.”