Abstract
Self-reported life satisfaction data are used in a project to triangulate with a longstanding analysis of Canadian regional differences in quality of life. The traditional analysis uses crime reports and vital statistics to demonstrate that indicators such as homicide, suicide, or divorce rise from east to west. The predominately westward migration in Canada is an explanation assumed, but untestable, from the vital statistics approach. The present study uses the 1985 Statistics Can ada General Social Survey to show that the satisfaction with life reported by the respondents largely follows the east-west gradient with respect to non-economic domains. Quebec, where satisfaction was very low in the '85 survey, is a marked exception. The migration interpretation of the regional pattern finds support in multivariate analysis including region, migration status, and controls.
Acknowledgements
This research was completed with support from SSHRC grant 410-92- 0821, and sabbatical leave for six months in 1992 from the University of Waterloo. I am grateful to Grace Logan for consulting on computing issues, to Tim McCutcheon f or research assistance, and to Mary Thompson for some statistical advice. Acknowledgement is owed to Statistics Canada for putting their GSS surveys into the public domain. The ideas herein were presented at the 1993 Meetings of the Canadian Sociology and Anthropology Association. I benefitted from commentary by Victor Thiessen on the paper presented, and by Robert Hiscott, Sharlene Waugh, and Lorne Tepperman on the revision
Introduction
For years it has been known that the prevalence of various social disorders such as suicide, divorce, and crime varies by region, tending to be least in Atlantic Canada and greatest in the west. Werner Cohn (1976: 90) found the pattern, which he lab elled the "Canadian syndrome of polarities," persisting as late as 1971, and inferred that the west is a more anomic society than central or Atlantic Canada. Brym et al. (1985), writing a critique of Cohn, seemed to accept the importance of regional d ifferences in social disorders even while disputing other parts of Cohn's article. Components of the polarity have been noted not only in academic journals (e.g., Pike, 1975: 127-31; Makabe, 1980; Marchak, 1980: 23; Veevers and Cousineau, 1980: 208-12; Trovato, 1986: 343-4; Gee and Veevers, 1989: 624-5; Lenton, 1989), but in the national press (e.g., Globe and Mail, 4 March 1989: A7, D2) as well as regularly in Statistics Canada publications (e.g., Beneteau, 1988: 24; Adams and Nagnur, 198 9: 26).
Such regional differences have been linked with the westward migration which originated in the frontier era and persists even today in a more complex form (Rosenbaum, 1993). It is a longstanding hypothesis, encountered, for example, in S.D. Clark's (1942: 8) pre-war analysis of social disorganization in the western frontier. More recently, Veevers and Cousineau (1980: 211) noted that "one hypothesis worthy of further investigation is that high rates of unconventional behavior [in the west] may b e related to rates of immigration per se and to the kind of persons who elect to move." Within the political realm, discussion on a recent report by the BC Attorney-General's Ministry took note of the "trend of increasing crime from east to west," and identified "migration to the province" as one cause (Globe and Mail, 7 March 1994: A4).
Migration could undermine quality of life simply because the relocation is involuntary, as is the case with the "tied" migration especially prevalent among women (Marsden and Tepperman, 1985). Even for voluntary migrants, however, enhanced economic
satisfaction may trade off for diminished non-economic satisfaction since migration within Canada is so often for economic motives (Shulman and Drass, 1979; Currie and Halli, 1989). Such trade-offs can help motivate the "return migration" described by
Richling (1985) and Hiscott (1987). But the more generic theoretical point, and the one stressed by Cohn (1976), is a view of population movement as loosening family and community bonds, thereby removing some of the most basic sources of control and r
estraint in a society (Park, 1928; Bogue, 1959: 488; Kantor, 1963; Fenelon, 1971; Kroger and Wood, 1983: 86-8). Migrants are viewed within this literature as likely to hold fewer role identities and weaker social support than non-migrants, thus being
more vulnerable to life crises (Thoits, 1983, 1986; House, 1986; Wheaton, 1990).[1] John Porter's (1965: 33-7) famous hypothesis about weak collective identities in Canada due to the "demographic railway station" use
d the same mode of reasoning about migration.
The implication that migration detracts from quality of life cannot be verified using vital statistics, which after all are counts for aggregates of people in administrative districts such as provinces or census tracts. The ecological correlations w
hich appear in Cohn (1976: 98), Makabe (1980: 175), or Fenelon (1971: 326) in the US are merely suggestive and fail to overcome the logical problems of inference to the individual as unit of analysis (Robinson, 1950). The present study is concerned wi
th solving this problem of inference regarding the implications of migration for Canadian regional differences in quality of life.
It is a topic important for our understanding of the distinctiveness of Canadian regions, for if attributable to migration, such differences in quality of life in a sense are not regional differences at all. As Pineo (Boyd et al., 1985: 510) said, r
egion then is a pseudo-variable with meaning only as the destination for migrants. To redefine regional differences in quality of life into a consequence of differential migration rates is within the "container" (Simeon, 1975: 499) interpretation of r
egion, as differentiated from a theory of indigenous regional culture such as a "frontier mentality" that might still linger in the west (Veevers and Cousineau, 1980: 211). Design
For the above objectives, a shift was needed from the yearly vital statistics to life satisfaction as self-assessed by survey respondents. Although the dependency on ecological correlations was thereby broken, a new problem arose. Self-reports of fe
elings about life quality on surveys are so different in concept from tabulations of rates for behaviours such as suicide that there is no certainty the two will agree (Schneider, 1975: 505; McCall, 1975; Wasserman and Chua, 1980; Walter- Busch, 1983).
The self-reports are susceptible to problems of reliability and validity (Connidis, 1984), but in compensation have stimulated much interest among researchers, resulting in considerable knowledge and forewarning of limitations in data quality (e.g., A
ndrews, 1974; Rodgers and Converse, 1975; Andrews and Crandall, 1976; Campbell et al., 1976; Michalos, 1980, 1983, 1985). There is, for example, a Canadian test-retest figure of r = 0.68 for general life satisfaction asked two times within one survey
interview (Atkinson, 1982: 115).
The distinction between "objective" and "subjective" indicators of life quality can actually be an advantage for present purposes. If migration as an explanation of self-reported satisfaction functioned consistently with expectations from an approac
h as distinct as ecological analysis of the vital statistics, a convincing triangulation would have occurred (Webb et al., 1966: 3,174; Denzin, 1970: 469-75). The triangulation metaphor is made viable by the commonality of some variables in the two ap
proaches. For example, divorce, a component of the vital statistics approach, is a well established source of dissatisfaction in self-reported quality of life (e.g., Tepperman, 1988: 119).
A key step toward the triangulation was incorporating the distinction already noted between satisfaction with the economic versus the non-economic domains. As a recent article in Social Indicators Research stresses, "there is a basic distin
ction between material and immaterial aspects of one's sense of well-being" (Groenland, 1990: 377). Although dissatisfaction with job, education, or finances contributes to general dissatisfaction, these economic domains are in turn affected by region
al economic disparities, the outline of which is well established in Canadian research (e.g., Matthews, 1983; Brym, 1986; Savoie, 1992). It was precisely the disjuncture between regional economic indicators (e.g., the unemployment rate or per capita in
come) and quality of life measures such as crime, divorce, or suicide, that made the vital statistics-based research compelling, and the analysis of self-reported satisfaction should follow suit. The existing evidence on Canadian regional differences
in self-assessed quality of life is limited with respect to the economic/non-economic duality, being focussed instead on global satisfaction (Atkinson, 1979: 13) or happiness (Tepperman, 1994: 201). A region such as Atlantic Canada is likely to have l
ow economic together with higher non-economic satisfaction, resulting in little summated departure from the national average. Data and Statistical Model
The present analysis used the 1985 General Social Survey (GSS), collected by Statistics Canada with telephone interviews. This edition of the GSS has province of birth, crucial for constructing a migration variable. Provincial birthplace was asked i
n the GSS for later years, but not released in the public data sets. Also, the '85 GSS contains questions on contact with relatives and friends, and these are useful for understanding the consequences of migration on life satisfaction in the friends a
nd relatives domains. Methodological details on the 1985 General Social Survey appear in a manual published by Statistics Canada (1986). The 11,200 cases for the 1985 survey were selected nationally from the population fifteen years of age and older, e
xcluding the northern territories and full-time residents of institutions, and with that many cases even small differences in perceived quality of life can be detected.[2]
The 1985 survey posed domain satisfaction questions concerning health, job, finances, housing, family relations, and friendships, along with a general satisfaction query ("How do you feel about your life as a whole?") and a similar item on happiness
. The domains deemed most clearly to address the immaterial or non-economic realm were "health," "family," and "friends." Responses were coded by Statistics Canada into a four level Likert scale with categories "very satisfied," "somewhat satisfied,"
"somewhat dissatisfied," and "very dissatisfied." Item nonresponse (coded "no opinion" or "not stated") was at most a trivial 1.3 percent, and was deleted from the list for the multivariate analysis to be described below.
An initial descriptive tabulation from the 1985 GSS appears in Table 1, giving the percentage "very satisfied" on the six life satisfaction domain questions. Regional differences in the satisfaction self-reports are not large, the table reveals, but
with the marked exception of Quebec, the non-economic items essentially follow the same east-west pattern that was seen for the vital statistics, with the lowest satisfaction figures being the ones for Alberta and BC.[3] The lack of any east-west trend for the economic domains underscores the regional satisfaction gradient for non-economic domains of the self-reported data. The modest gradient among Anglophone provinces is consistent with how the whole region
al syndrome of polarities noted by Cohn (1976) has eroded over time (Goyder, 1993), although in the vital statistics enough of the pattern has remained to warrant more than one press report as late as winter 1994 (The Globe and Mail, 11 Februa
ry 1994: A24; 7 March 1994: A4). For present objectives, it was sufficient that minimally an echo of Cohn's east-west pattern be reflected in the self-reports. With the Francophone-Anglophone difference in satisfaction so vast, Quebec was removed from
the present analysis, and the linguistic culture factor addressed in a separate analysis (Goyder and McCutcheon, 1995).[4]
As is characteristic of satisfaction questions (Stones and Kozma, 1991: 31), distributions on all the items were highly skewed, with most respondents on most domains either "very satisfied" or "satisfied." Such skews were beyond correction by transf
ormation, and due to the substantial inter-correlations between indicators even index construction by factor analysis did not eliminate the non-normal distributions. Although the main conclusions of the analysis can be represented using multiple indic
ator techniques such as Lisrel, the format adopted is logistic regression analysis of each non-economic index separately.[5] The sequence of models provides a form of internal replication. Details on independent v
ariables appear in Table 2, with means and standard deviations classified by regions. Region was scored from province of residence at the time of interview, recoded into Atlantic, Ontario, Prairie (Manitoba and Saskatchewan), and West (Alberta and Brit
ish Columbia). The collapsing of the four Atlantic provinces into a single region is in recognition of the small numbers of respondents in any one of these provinces in a survey even one the size of the GSS.
Migration was constructed from cross-classifying current region against birthplace, with attention to two types: Canadian-born migrants (termed hereafter "internal migrants"); and migration due to foreign birthplace. Ideally, migration would be take
n from a demographic history allowing separate analysis of return migrations, which account for some 24% of all interprovincial population movement (Rosenbaum, 1993: 93). That was impossible with the 1985 GSS data set, which was not primarily a survey
on geographic mobility. While immigrants might become internal migrants subsequent to landing in Canada, the data cannot discern that, which is why the operational definition of internal migration for purposes here is the native-born migrant whose reg
ion of birth differs from region of residence when interviewed for the 1985 GSS.
To begin the analysis, three logistic regression models were computed for each of the three non-economic life satisfaction domains. The first (termed the "baseline model") simply re-expressed the Table 1 patterns in the form of logistic regression.
The second (the "migration model") examined region, internal migration, and immigration, for the basic test of whether the east-west gradient in quality of life was created by historic migration flows. A third model (dubbed the "covariate model") was
then tested to determine whether the effect of migration could be translated into the main demographic characteristics of migrants, understood to comprise high education and income (Boyd et al., 1985: 501; Hiscott, 1987: 587).[6]<
a name="note6"> Education was coded within a set of dummy variable categories, described in Table 2, and the income variable referred to employment earnings. Age also was included, modelled in quadratic form, since satisfaction often is highest in ear
ly adult life, lower during the middle years, and higher again in later middle age (Tepperman, 1988: 14, 121, 153, 195-97). The migrants in the GSS sample are slightly older than non-migrants. That is not in conflict with census analyses showing high
migration among the young (e.g., Ledent, 1990: 51), due to the distinction between mobility related to province of birth and the census concept of mobility since five years previous. In the lifetime approach, probability of migration cumulates with age
. Results of the Model
Referring to Table 3, although the units for logistic regression coefficients are complex (log odds for outcomes on the dependent variable), sizes of effect can be gauged by comparing coefficients across rows, by considering which coefficients pass
significance tests, and by standardizing coefficients with division by the standard errors entered in parentheses. There is much information summarized in these tabulations, and just the data central to our main theme shall be discussed.
The east-west gradient of regional differences in satisfaction is seen by the pattern of increasingly large negative coefficients, from Atlantic to Pacific. The migration model (middle columns of figures in Table 3, under each domain heading) holds
for satisfaction with family and friendships, but not for health. That is, the coefficient for internal migrants as well as for immigrants is at least p <<˙200.05 significant for the first two domains, but not for the third. The strongest migration ef
fect is on friendship satisfaction, where the logistic regression coefficient is nearly four times the size of its standard error for internal migrants and larger still for immigrants. The effects are not, to be sure, enormously strong, but this has to
be set against the very crude measure of migration already described, making no provision for multiple migrations, motives, or recency of the geographic mobility. Expansion of the analysis into the covariate model, seen in the three longer columns of
coefficients in Table 3, does not diminish the tendency among migrants for somewhat lower satisfaction with family relations and friendships. Indeed, there is a slight suppression whereby the expanded model leaves a more powerful net effect for inter
nal migration. That is no surprise, knowing (from Table 2) that both migrants and immigrants are over-represented on such satisfaction promoting traits as post-secondary education and high income (Davis, 1984). The covariate model shows that, except fo
r this profile, the internal migrants would report even less satisfaction.
To what extent does migration account for the regional variation in life satisfaction that began the analysis? Comparison of the baseline and migration models in Table 3 shows all the coefficients for region diminishing or reversing sign in the case
of satisfaction with family and friends. The same does not hold for health satisfaction, which has already been found to be unaffected by migration. The controls in the covariate model only slightly restore the regional gradient. A summary tabulation
on the proportionate reduction in regional differences in life satisfaction appears in the bottom row of Table 3, giving percentage reduction in the chi square summed for the regional dummy variables. For satisfaction with family relations, controllin
g just for migration status reduces the regional gradient by about one-third, and a little less for the covariate model. For friendship satisfaction, the reduction of regional effect exceeds fifty percent in both the migration and covariate models.
The conclusion so far has migrants less satisfied than the native- born non-migrants with family and friendships, with no such effect for satisfaction with health. That is consistent with the initial hypothesizing (see above) about migration, for the
disruptions to social support due to migration indeed would most obviously affect satisfaction with family and friendships. Impact on health should only be indirect; it should happen, for example, as a consequence of dissatisfaction with other domain
s or with life in general. The effects under discussion do affect global satisfaction, since all the domain satisfactions are correlated with global satisfaction, but the logistic regression analysis has shown that the carry-over into satisfaction with
health amounts to no more than a trace.[7] Social Support, Migration Status, and The preceding amounts to a general sketch of the implication migration status has for regional differences in life satisfaction in some of the non-economic domains. For satisfaction with family and friends, a more complete analysis was possible usin
g the set of questions in the GSS survey concerning contact with others.[8] These questions are relevant to part of the reasoning behind migration as a determinant of life satisfaction, namely that population mo
vements loosen ties between family members and friends.
First, the implication of migration for contact with family members and friends was assessed, the data appearing in Table 4 as percentages within the three categories of migration status used already. Most (76%, from Table 4) native-born non-migrant
s are in face- to-face contact with their mothers at least once a month. In contrast, only 33% of internal migrants and 37% of immigrants see their mothers once a month or more. The pattern for face-to-face contact with the father is the same. The inte
rnal migrants and immigrants partly compensate for non-face-to-face meetings by frequent contact using mail and telephone, but even that communication is greatest among the native non-migrants. Consider, for example, the 71% of internal migrants and f
oreign-born having at least monthly letter or phone contact with a living father, compared to 77% of native non-migrants). In terms of contact with children, the foreign-born have the same profile as native non-migrants, unsurprisingly because immigran
ts would often bring younger children with them to Canada, or bear children in Canada. Even with the coarse measure of internal migration in use in these tabulations, some loss of contact with children among internal migrants is detectable (72% have a
t least monthly contact among migrants vs. 80% among native non-migrants). Again, the gap is not totally filled by letter and telephone contact. So it is, too, for contact with siblings, relatives, and even with friends. There is as much or more loss o
f physical contact among internal migrants as among the foreign-born, and the internal migrants seem only partly to substitute mail or telephone contact.
For internal migrants, contact with family and friends correlates with self-reported satisfaction in these domains, as Table 5 shows. It seems not to matter which form the contact takes, for the percentage very satisfied tends to be as strong for fr
equent telephone or postal contact as for face-to-face meetings. Contact with children is, however, far more important to family satisfaction than is contact with parents, relatives, or siblings. Some eighty percent of parents of children not living i
n the household reported themselves very satisfied with family relations so long as daily letter or telephone contact occurs, compared to a mere 26% very satisfied when there is never contact. Here is a massive strength of relationship by the standards
of life satisfaction research. In another contrast with the importance of contact with children, contact with friends relates to friendship satisfaction only in a curvilinear form.
In the equivalent figures for immigrants, entered in italics in Table 5, a difference between the two forms of migration began to emerge. Immigrants tend to be the more satisfied with family relations the less they have contact with their parents. I
t would be just one of several possibilities to speculate that links with parents complicate the adaptation of immigrants into the culture of native-born Canadians. Contact with children is important for the family satisfaction of immigrants, but not
to the same degree as for internal migrants.
The joint relationship between region, migration status, contact with family and friends, and the respective domain satisfactions was addressed in another logistic regression analysis, summarized in Table 6. Age and other controls from the "covariat
e model" in Table 3 are part of this regression model, although the respective coefficients were not entered into the Table, for simplification. The tabulation adjusts, for example, the age differences between younger respondents with living parents b
ut no children outside the home, and older people having adult children but no parents still alive. The low family satisfaction among internal migrants now is fully accounted for, since after adding the contact variables the coefficient for internal mi
grants is both non- significant and greatly reduced from the equivalent figure in Table 3. Affect is seen again (as initially sketched in Table 5) to flow mainly from senior to junior generation. For those with children, having contact with them is cru
cial to family life satisfaction, while contact with parents seems less essential. For friendships, adding contact information as intervening link between migration status and satisfaction is less illuminating than for family satisfaction, the statist
ical explanation being evident in Table 5. In terms of meaning to respondents, we are by this point deep into speculation, lacking follow- up questions in the survey as to why 100% of internal migrants who never see their friends say they are "very sati
sfied" with friendships! The adaptations newcomers make, such as involvement in leisure activities, perhaps are effective substitutes for loss of contact with friends and parents. This is consistent with Lock's research (1993: ch. 6), which showed ris
ing leisure activity rates from east to west in Canada to be a pattern largely accounted for by high activity among migrants. Conclusion
Interpreters of the vital statistics contended, with little proof, that varying rates of in-migration created the Canadian regional profile in quality of life. As discovered in the present analysis of the 1985 General Social Survey for Canada, both
native-born migrants and migrants from foreign lands tend in some key domains to have lower reported satisfaction than non-migrants, even with a crude measure of internal migration.
For reconciling self-report data with the vital statistics approach such as in Cohn (1976), Makabe (1980), or Veevers and Cousineau (1980), the form of the migration effect on reported satisfaction points towards, not away from, a successful triangu
lation. Consider that vital statistics hypothesizing derived largely from the concept of familial social support. The term "support" was expressly used by Cohn (1976: 91), and he was dealing with family issues in analyzing Jewish out- marriage. Contact
with family members (1976: 103) was an unmeasured independent variable in his conceptual scheme. Also relevant to the triangulation is that the key social support effect (ref. Tables 5 and 6) held for internal migrants more clearly than for the foreig
n-born. A form of internal migration was what Cohn (1976: 97-8) had in mind in forming his hypothesis about regional polarities, broadened slightly to include US born migrants to Canada. The majority of foreign-born were distinct from North American m
igrants and hard to assess, according to Cohn (1976: 98), due to their "varying backgrounds."
The triangulation suggests that the family disruption resulting from migration has both behavioural and attitudinal consequences. Behaviourally, as studies in the Cohn (1976) tradition established repeatedly, the high in-migration rates coincided st
atistically with the regions of Canada having the highest rates of disorder on indexes such as homicide, drug abuse, suicide, and divorce. Attitudinally, the conclusion of the analysis just described is that internal migrants articulate lower satisfac
tion with the non-economic domains most closely tied to social support. For family satisfaction, the migration can be explicitly traced back to low contact with siblings and children, but not so much with parents. Here is a reminder that the social cos
t of migration is not just a young person's issue.
It is not necessary for a triangulation between behavioural and attitudinal measures that it is the same people reporting dissatisfaction who succumb to behavioural disorders. The behaviours counted in the crime and vital statistics are most charact
eristic of the young, while the survey of self-reports approach deals with the entire age range. The level of malaise that might lead to serious behavioural consequences probably cannot be detected within a few closed-ended survey questions on life sa
tisfaction. A far more detailed psychiatric interview would be required.[9]
The hypothesizing derived from ecological indicators within the vital statistics was not making the blanket statement that the life of migrants became worse in every way as they moved westward throughout the century. As noted at the outset, our disc
ussion has no bearing on regional economic disparity, which has quite a different pattern than Cohn's (1976) syndrome of polarities. It is not possible either within the limitations of the present analysis to make more than passing mention of the mix
of west coast euphoria and angst (e.g., Seligman, 1989: 90-93). Even Garreau's (1981: 260) popular book on regional cultures, a source taken seriously by academic sociologists (e.g., Baer et al., 1993: 14), makes note of a high suicide rate in Seattle
within an approving chapter on the "ecotopia" that includes BC. One can only comment that migration like any substantial life change can be simultaneously liberating and imprisoning. As a final disclaimer, migration is no random event always having th
e unintended consequence of reduced contact with family. Decisions about migration or non-migration among older respondents would sometimes be made with proximity to adult children in mind, not, as the main analysis had to assume, centrifugally. Teppe
rman (1994: 197), in the second edition of his Choices and Chances text, was stating the same migration hypothesis we have been investigating from a different causal direction when he concluded that "strong social bonds inhibit migration." Peo
ple migrate due to problems and a migration may create problems, but in either case the mobility has helped create a longstanding difference between regions in this country.
Notes
To name just a few prominent sociological sources from a large literature which includes contributions from psychology. Much of this literature is from the US. An American analysis of east-west differences in quality of life simil
ar to the Canadian one has been made, again using official data such as the vital statistics. Straus et al. (1989: 229), for example, concluded that the western states comprise the most "stressful" region. See also a psychoanalytic approach by Seligma
n (1989: 91) which contrasts the "California" and "New England" modes of relating the individual to society. Back
Statistics Canada's sampling design has design effect due to clustering. However, from Frankel's (1971: 172-73) experiments, design effects for regression coefficients are much lower than for other statistics, and often near 1. Back
The pattern could seem anomalous in light of the tenth Annual Maclean's Poll, in which 93% of British Columbians picked their own province as the one they would like to live in the highest percentage in Canada (Maclean's 3 J
anuary 1994: 10). Liking one's province is not the same as being satisfied with domains of life such as health, family, or friends, however. The Maclean's question could be triggering feelings about geography, lifestyle, and perhaps, the liber
ating component of migration alluded to below. Back
Examination of other data sets showed that the east-west pattern for Anglo provinces tended to be reproducible at least to some reasonable degree, but figures for Quebec vs. other provinces were unstable. The quality of life survey c
onducted by the York ISR in 1977, for example, placed Ontario and Atlantic Canada as markedly more satisfied on non- economic domains than the west, but found Quebecers to be highly satisfied, in contrast to the 1985 GSS. It required a meta-analysis of
numerous Canadian national surveys, using a different theoretical grounding than the migration emphasis used here, to find some pattern to satisfaction in Quebec. Back
At the early stage of research, OLS regression was used for its easy interpretability. But since the assumption violation due to skewed distributions was so severe, I then replicated the analysis using summary indicators of satisfact
ion in the economic vs. non-economic domains. That obscured some distinctions important to the analysis, and re-directed attention back to an item-by-item focus, this time using logistic regression to handle skew. Since the satisfaction questions clea
rly were measuring degrees of satisfaction rather than satisfaction vs. dissatisfaction (typically only about 3% of replies were less than somewhat satisfied), the cutting point was placed between very and somewhat satisfied. Back
Gender and marital status were not excluded for lack of consciousness of gender issues. They are not important factors in gross Canadian interprovincial migration (George, 1970: 149-52; Hiscott, 1987: 591), and would only be useful f
or the present analysis if combined with a motivational variable to identify tied migration. Back
The impact of migration on global satisfaction was computed with a partialling out for the two most clearly economic domains, "finances" and "job." Chi square for the regional variables in the logistic regression equation predicting
global satisfaction was 29.86 (3 df), reduced by 44% in both the migration and covariate models. Back
Such questions operationalize social support in what Cohen and Syme (1985: 11) would distinguish as a structural form rather than one examining the functions in people's lives served by relationships with other people. Back
Social support, which has been used herein as a bridge between the ecological (and behavioural) and the individual (and attitudinal) levels, is itself both an "individual-level notion" and a "systems-level process" (Felton and Shinn
, 1992: 103). An environment with low social support, for example, could have consequences for all people, not just those who themselves were especially socially isolated. Back
To contact John Goyder by e-mail, click
here
Back
to the Recent
Issues Page
Back to CJS Online