Abstracts

Applications of Demographic Analysis in the Evaluation of 1996 Census Results
Don Kerr, Demography Division, Statistics Canada and D. Morrisette, Demography Division, Statistics Canada

Statistics Canada has produced population estimates, fully adjusted for coverage error, from 1971 to the present. Several procedures, developed by both demographers and statisticians have historically played an important role in evaluating census coverage in Canada. Most methods can be grouped as involving either (i) demographic analysis (macro-level approaches), or (ii) case by case matching procedures and record linkage techniques (micro-level approaches). The present paper reviews selected procedures, and outlines the planned application of demographic analysis in the evaluation of the 1996 Census. It is argued that there is substantial potential in increasing the emphasis placed on demographic analysis in the estimation of census coverage, particularly in improving estimates for specific age and sex groups.

Assimilation and the Altar: Canada 1991
Madeline A. Kalbach, Ethnic Studies Chair, Sociology, University of Calgary

This paper examines patterns of ethnic intermarriage for Canada's ethnic populations at the time of the 1991 Census. Many of Canada's older immigrant populations reveal relatively high levels of ethnic intermarriage, especially for the native born. Newer immigrant groups generally exhibit significantly lower levels of ethnic intermarriage than their older immigrant population counterparts. However, some exceptions were evident.

Canadian Children in the 1990s: Selected Findings of the National Longitudinal Survey of Children and Youth (NLSCY)
Alice Peters, Canadian Social Trends, Statistics Canada

Data for the NLSCY show that on the whole, Canadian children are physically, emotionally, and socially healthy. Some children, however, are experiencing difficulties that my lead to inferior school performance, unsatisfactory social relationships, and may ultimately, hinder their transition to a healthy adulthood.

In 1994, 79% of the population under the age of 12, lived in two-parent families (with both biological parents). Sixteen per cent lived with a lone parent, the vast majority with their mother, with the remainder living in some form of blended family. Even though most women avoided risk behaviours during their pregnancies in 1994, almost one in five newborns needed special medical care at birth.

A number of key components of development are identified as being important to later success in school and work and to be influenced by the nature of parents' interaction with their children. Analysis of NLSCY data indicates that parenting practices were most closely associated with children's social relationships. As a result of changing family structures which reflect the long-term trends in marriage and divorce, most stepfamilies today are the result of divorce followed by remarriage. It is not surprising that parent-child relationships in stepfamilies often seem more problematic than those of other families.

Changing Sex Compostion in the Canadian Physician Work Force: Implications for Health Care
Raymond W. Pong, Laurentian University

Abstract unavailable

Chronic Social Stress and Emotional Well-being: An Analysis of Mental Health of Immigrants in Alberta
M. Acharya, Sociology, University of Alberta

This study examines the association between various life events and emotional well-being of immigrants aged 20 to 64 years living in Alberta. For the purpose of this analysis, secondary data from the 1994-95 National Population Health Survey were used. A multivariate analysis using stepwise regression was utilized to highlight the significant predictor variables of emotional well-being of the immigrants. The paper proposes a explanation using the socio-demographic stressor model.

Results showed that the main health problem causing restricted activities had the strongest association with emotional well-being of these immigrants. These immigrants have excellent health status. Furthermore, the paper also suggest that general chronic social stress, main activity/labour force participation and socioeconomic classification of occupation variables are also important predictor of emotional well-being. The results support the social stressor theory. Further research is needed to expand knowledge with regard to the mental health issues of immigrants in Alberta, Canada.

A Cluster Analysis of Ethnic Attachment among Immigrants
Derrick Thomas, Citizenship and Immigration Canada/Statistics Canada

The intensity and complexity of their attachment to ethnic groups is thought to exert an influence on the adjustment of immigrants. It is not a simple matter, however, to objectively measure the depth of ethnic feeling or to identify overlapping attachments. The proposed study will apply the principles of numerical taxonomy to the delineation of ethnic groups and the measurement of ethnic attachment. Distance or pattern similarity measures will be employed to define clusters around ethnic centroids or prototypical patterns.

Among the variables used in the measurement of distances will be: languages spoken, languages spoken at home, religion, ethnic ancestry, and perhaps citizenship or status in Canada. The computed distances will also utilize the patterns exhibited by spouses and children on the same indicators. It should be possible to get an impression of the density, cohesion, and distinctiveness of the delineated groups. The influence of time in Canada, human capital, and unchosen characteristics such as membership in a visible minority will also be examined. The study will utilize data on families from the 1991 census, but the method should be generalizable to almost any dataset with the requisite indicators.

Comparative Socio-economic Analysis of the Metis Settlements of Alberta
Melinda Mills, Population Research Center, University of Groeningen

This paper is a comparative socio-economic analysis of selected demographic characteristics of the Metis Settlements in the province of Alberta, Canada. Differences in population distribution, education, labour force and income are examined to compare the socio-economic conditions of the Settlements with their provincial and national counterparts. A central conclusion of this work is that not only is the Metis Settlement population composition substantially different than other Canadians, but that persons living on the Settlements have a significantly lower standard of living than persons who reside in other parts of Alberta and Canada. To unify the main ideas of this analysis, a final theoretical synthesis is presented to further understanding and research in this field.

Demographic Estimation Program: Achievements and Challenges
Réjean Lachapelle, Demography Division, Statistics Canada & Ronald Raby, Demography Division, Statistics Canada

The demographic estimation program has undergone numerous changes since its origin. The most significant changes occurred after the 1991 Census. Since then, the target population has included non-permanent residents, and the Census not only takes into account the number of persons enumerated but also an estimate of net undercoverage. Today, the postcensal estimates are based on components not considered in the past (non-permanent residents and Canadians returning to Canada). The challenges posed by the correction of the postcensal data and the estimation of the various components will be presented. The results of the work undertaken to identify these challenges will be summarized and we will indicate the avenues of research which seem most promising.

Determinants of Re-Employment Probability in Canada: Competing Risks Analyses between Full-Time and Part-Time Employment.
Junjie Zhang, Sociology, University of Western Ontario & Rod Beaujot, Sociology, University of Western Ontario

Longitudinal Data from the Survey of Labour and Income Dynamics 1992-1993 are used to test the hypotheses that marital status and presence of young children are important factors in determining the re-employment probability of the jobless. The full-time employment and part-time employment are treated as competing risks. Double-Decrement life table and simultaneous competing-risks model results indicate that, married men have higher probability of re-employment, while the divorced women have much lower probability of re-employment. The women with young children are found less likely to enter full-time employment but more likely to work part-time. Additionally, immigrant women, older men and minority men are identified as disadvantaged groups in the labour market.

Determinants of Return Migration
Eric Fong, Sociology, University of Toronto; Kumiko Shibuya, University of Chicago, Ming-Long Lam, SPSS, Inc. & Clement So, Chinese University of Hong Kong)

This paper examines the determinants of return migration among Chinese immigrants in Toronto, Canada. Based on a telephone survey conducted in October 1994 among Chinese in Toronto area, we identified various economic, social and demographic factors that influence the decision about return migration. Truncated logistic regression model is used to account for truncated nature of the data. Results show that location specific capital is significant for the decision of return migration among business migrants and both social networks and age are important in the decision of return migration.

Determinants of the Risk and Timing of Alcohol and Illicit Drug Use Onset Among Natives and Non-Natives: Similarities and Differences
Margaret DeWit, University of Toronto at Mississauga; Bryan Embree, University of Western Ontario; & David J. DeWit, Addiction Research Foundation of Ontario

Using data from the Ontario Health Survey Supplement (1990/91) and a sample of native Ontario reserve residents (Embree, 1993), proportional hazards modelling was employed in order to examine factors associated with the risk and timing of onset of substance use (alcohol and illicit drugs) among both Natives and Non-Natives. In addition to considering demographic characteristics like age and sex, special emphasis is placed on the role of family background characteristics (e.g., parental substance abuse, paternal history of depression, quality of parental relations, and sexual abuse during childhood) as a precursor to early alcohol and drug-use onset. The multivariate results reveal some important similarities and differences between Natives and Non-Natives in their patterns of alcohol and drug use timing.

For both Natives and Non-natives alike, age cohort predominates as a risk factor, with younger groups at greatest risk. The trends in both drinking and drug use appear to be ones of steadily increasing risk over time for young people. As expected, males also tend to exhibit consistently higher risks of both alcohol and other substance use. For the model of drug use timing, age of alcohol use onset is the second best predictor for both Natives and Non-Natives, although the effect is much more powerful in the case of Natives. This result lends further support to previous findings that alcohol use is a powerful predisposing factor to the use of illicit substances, but the evident disparity between cultural groups in the predictive power of this measure also suggests that natives may lag behind the general population with respect to recently observed shifts in the pattern of substance use progression.

Despite certain inconsistencies, the findings point overall to the salience of a number of family background characteristics in affecting early onset drinking and drug use, behaviours well recognized to have potentially adverse mental and physical health consequences, as well as negative social outcomes.

Economic Constraint among Senion Citizens
Karen Kampen, Centre on Aging, University of Manitoba & David Cheal, Sociology, University of Winnipeg

It has been suggested that the concept of an economic life course be re-designed. The post-war notion of becoming educated and working steadily throughout one's life until retirement at age 65 is becoming irrelevant. Further, changes in the nature of work, the composition of the work force, as well as the family will mean changes to the process of leaving the workforce. Statistics Canada's Survey of Labour and Income Dynamics Public Use Microdata, 1993, permits analysis of some of these issues.

The first wave of data in the survey sampled 29,934 individuals. From this data we have created two derived measures, the Low Income Measure (LIM) as well as a measure of financial dependence. Each can be viewed as a different dimension of economic constraint. Using these measures, analysis focuses upon Senior Citizens aged 54-63, who have elevated rates of poverty and dependence compared with individuals at other stages of the life course. Factors associated with poverty and dependence among Senior Citizens will be explored, with special emphasis placed upon policy relevant questions of economic security and insecurity among older women.

Entropy of Disability-free Life Expectancy in Canada
Gerry Hill, 333, Helene Street, Rockland, ON K4K 1G4, Jean-Marie Berthelot, & Wayne Millar, Ottawa

The simplest index of health which incorporates morbidity as well as mortality is Sullivan's index of Disability-Free Life Expectancy (DFLE). This index requires only estimates of the age-specific prevalence of disability, however defined, and the current life table population.

In this paper we show how the sensitivity of the DFLE to changes in the incidence of disability (entropy) can be estimated by a minor extension to the calculations. We illustrate this method using Canadian disability data from the 1994 Health Survey and the 1990-92 Canadian life table.

Estimating a Population's Age Structure from Past Fertility and Mortality Patterns
Jacques Ledent, Institut Nationale Recherche Scientifique (INRS)-Urbanisation

Recently, Claude Dionne has set forth an intergenerational population model which emphasizes a two-way linkage of the women in successive generations by means of two subsets of equations: one subset of descent equations reflecting a forward relationship from mother to daughter cohorts through period births and one subset of ascent equations reflecting a backward relationship from daughter to mother cohorts through surviving ascendants to daughters in mother cohorts. Such a model, it turns out, could be used for recovering the age structure of a population from past data on period births (and deaths) and thus possibly for adjusting a population's age structure obtained from a census.

This paper reconsiders Dionne's model by bringing two major improvements: 1) a more correct specification of the ascent equations and 2) a revision of the overall structure of the model by way of a reintroduction of some valuable information on past fertility and mortality patterns discarded by Dionne.

These improvements pave the way for an enhanced formulation of the model in a matrix format which eventually leads to a characteristic equation, the dominant root of which necessarily takes on the expected value of 1. In addition, a dual version of the model is suggested. Reversing the roles of mothers and daughters between the two subsets of equations, it can be used for estimating the age structure of a population from the age distributions of 1) period births and of 2) daughters in mother cohorts. Such a dual version could prove useful for gaining insights into the population age structure of countries with no or limited census data.

Estimating Changes in Canadian Diversity
John Kralt, Heritage Canada

There is the perception of a drastic increase in diversity since the liberalization of the immigration laws in the late sixties. These changes in diversity have generally been measured through the use of the census data on ethnic origins and it is often assumed that the changes in ethnic origins as reported mark the actual changes in the origins of the Canadian population, often very large changes.

The extent to which these changes in the origins reflect real changes and the extent to which some of these changes are a reflection of changes in reporting patterns by Canadians is a question of some debate. In this paper, I present a more accurate picture of the actual increase in the numbers of persons reporting specific origins and visible minority origins as assigned than is available from a simple comparison of the 1986 and 1991 Census data by examining the various components of the increase such as births, immigrants, and changing population definitions, i.e.. who is included in the Census count and who is excluded from the Census count. The results show that although there are large increases in the reporting of ethnic origins or the assignment of visible minority status, without making a number of adjustments, these changes are exaggerated.

Fertility Trends and Projections for Census Metropolitan Areas, 1986-2016
Ravi Verma, Demography Division, Statistics Canada, Shirley Loh, Demography Division, Statistics Canada; & M.V. George, Demography Division, Statistics Canada

The objective of this paper is to study the trend of total fertility rates (TFRs) for Census Metropolitan Areas (CMAs), 1986 to 1994 and discuss the projections of these rates until 2016. On the whole, TFRs for CMAs were found to be lower than the provincial values. The projected fertility rates for the CMAs were derived using an index method based on the observed CMA/provincial ratios. The seven methods that were used to derive the relative indices for the base year (1994) and the projected period (1995-2016) will be presented in the paper.

Immigrant Participation into Politics: The Role of Ethnic Organizations
Tina Chui, Housing, Family & Social Statistics Division, Statistics Canada

While research in the immigrant population has been an important focus in Canadian sociology, the literature on immigrant and ethnic political behaviour has been fairly thin. The social and political participation of immigrants and minority ethnics is a key to the understanding of ethnic relations. Participation into the polity is an indicator of integration into the broader society. People participate when they acquire certain knowledge of and become familiar with the political process. But politics is also about power. The ability to involve reflects the responsiveness of the political system to its participants. Equity of participation is a key to stratification because participation implies the opportunity to influence political decisions and the distribution of resources.

Immigrants often do not fare as well as the Canadian-born in political participation. This paper examines the extent to which immigrants get involved in politics and assesses the role of ethnic community groups in the immigrants' political behaviour. Different measures of political behaviour are used in the analysis of data from a Toronto immigrant survey. The findings indicate that ethnic organizations do not only provide political socialization, but also bear the potential to mobilize their members into collective action. Implications of the findings to our understanding of immigrant political behaviour are discussed.

The Impacts of Demographics and Legislation on the Future Size, Structure and Composition of the Aboriginal Population in Canada
Mary Jane Norris, Demography Division, Statistics Canada

This paper examines the interplay between demographic components and the legislation of the revised Indian Act in terms of population implications. It demonstrates how Aboriginal group differentials in fertility and mortality, combined with legislative impacts, will determine the size, age-sex structure and group composition of the Aboriginal population in Canada over the next twenty years. The revised Indian Act affects both the registered and non-status Indian populations. The Act implements provisions for Indian status through: reinstatements of those individuals eligible for status and, inheritance rules affecting descendants of the current registered Indian population through intermarriage. The population implications of the demographic and legislative components of growth are demonstrated through the use of various projection scenarios.

Importance of Cohorts in Assessing Implications of Population Aging
Leroy O. Stone, Analytical Studies Branch, Statistics Canada

Books and articles concerning the actual or potential effects of population aging have rarely laid out systematic theory designed to explain the processes or mechanisms by which population aging exerts the causal forces that are often attributed to it. Among the classic writings on this subject we find that some recent works reject a number of key claims made in the earlier literature concerning the 'more direct' effects of population aging. It is now timely to raise the following question: by what strategies will we arrive at systematic statements about the processes by which population aging becomes a causal agent in exerting forces for change in societies? The paper's thesis is that the development of understanding of population aging as a potentially causal force is greatly assisted by paying more attention to cohort analysis and, specifically, the aging of real cohorts in a given society. This may seem like a strange idea because population aging involves the whole population; studying it causes us to look at the structure of the entire population. However, the implications of population aging for a society are not readily discernible until we take time to study the potentially contrasting patterns by which different cohorts have gone or are going through certain major life course transitions. The paper illustrates this idea by making some generalizations about what's been happening to the pattern of marital status transitions in Canada when they are viewed from a cohort perspective.

Language Maintenance and Language Shift among Selected Immigrant Groups in Canada, 1971-1991
T.R. Balakrishnan, Population Studies Centre, University of Western Ontario & J. Rao, Population Studies Centre, University of Western Ontario

Language maintenance and language shift among some recent immigrant groups are examined using data from the Public Use Sample Tapes of the 1971, 1981, and 1991 Canadian censuses. Those whose home language is the same as their mother tongue is found to be related to age, age at immigration, education, labour force participation, and occupation. Language maintenance seems also to be related to place of birth, and city and ethnic group size. Implications for assimilation and policy are also examined.

Living Arrangements of Elderly Immigrants in Canada: Selected Aspects
K.G. Basavarajappa, Analytical Studies Branch, Statistics Canada

Elderly immigrants aged 55 years and over from devleoping regions have increased markedly since 1971 (52%). Such increases have implications for their well-being and integration into the host society. As they do not become eligible for government transfer payments for considerable length of time (e.g., 10 years for OAS/GIS), they have to depend on their sponsors for their maintenance. Consequently, many of them choose to live with their relatives in three or more generation households. The propensity for immigrant males to live in such arrangements is almost three times and females, almost four times that of their Canadian born counterparts. Immigrant elderly coming from developing regions do so to a much greater extent than their counterparts coming from developed regions. The South Asian group shows the highest proportions: 36% of males and 46% of females aged 55 years and over live in such arrangements, and these are nearly 15 and 18 times those of their Canadian born counterparts. Average income, percent receiving OAS/GIS, percent widowed, and duration of residence in Canada are significantly associated with living in three or more generation households, the proportion widowed being directly related and the other three factors, inversely related. These factors, together explain about 84% of variation among the 15 birthplace groups for males and about 81% for females. The groups from the developing regions also show much lower propensity for living with non-relatives or for living alone when compared with groups from the developed regions.

The Mature Community College Student: Needs and Expectations
Jim Jackson, Humber College

This report examines the change in the age structure of the student population at Humber Community College of Applied Arts and Technology, for 1978 and 1994. The registration data, for the age of the students, illustrates that in 1994, proportionally more older students enrolled in a career program, than in 1978. From graphs and the tables, the trend of the aging of the student population is clearly observed.

In 1996, a pilot survey was carried out to investigate the needs and expectations of the mature group of students upon entry into the College. Information on the mature students' program registration, early classroom experiences and the use of college services were highlighted in this survey. Other demographic variables, such as educational status and sex, were also examined in this study. This information could aid the College in offering the services and developing the programs which are needed by signficant numbers of students.

Measures of Identity and Acculturation among Ethnic Groups in Canada
Gustave Goldmann, Statistics Canada/Israeli Central Bureau of Statistics

Researchers and analysts using Canadian census data to study the socio-economic and demographic conditions of specific sub-groups of the population face an embarrassment of riches. Canadian decennial censuses provide data on the religious identification, ethnic origin and a host of socio-economic and demographic characteristics of the population.

Although the data refer to ancestry, I propose that it is possible to derive measures of identity and acculturation by analyzing the character of the resposes to the question on ethnic origin. Over 30% of respondents declared more than one ethnic origin in the 1991 Census. Furthermore, approximately 4% of respondents declared at least one of their origins to be Canadian. If, for example, we examine only the Jewish population, over 30% declared multiple origins and almost 5% declared Canadian as one of their origins. In this paper, I analyze the characteristics of the subset of the population of Canada who declared at least one of their origins as Canadian in the census with a view to determining whether these data may be used to derive proxmy measures of identity and acculturation.

Medicare file and Canada Post change of Address File as a Source of Data for the Population Estimation Program: the Current Situation
Sylvain Rémillard, Demography Division, Statistics Canada

This research conducted by the Estimation Section of Statistics Canada, is divided into two parts, the objective of which is to identify alternate sources of data for estimating interprovincial migration. Part 1 examines the potential for using data from provincial and territorial health insurance plans beginning with a general description of the health insurance plans. It then reviews provincial responses to the questionnaire which furnish a description of the health insurance plan and the amount of detail in available and potential data. We also compare migration data from the different health insurance files. Part 2 examines the potential for using data from Canada Post Corporation's change-of-address file. The general framework of the project is presented, after which there is a brief description of the file and the methodology used. The research findings are then presented.

Natives' and Non-natives' Relative Risk of Early Parental Loses: The Roles of Family Background Characteristics and Parental Substance Abusers
Margaret DeWit, University of Toronto at Mississauga; M. Sahota, M. Vermani, Bryan Embree, University of Western Ontario; & David J. DeWit, Addiction Research Foundation of Ontario

Abstract unavailable.

Patterns of Disability in Canada
Parameswara Krishnan, Sociology, University of Alberta

The 1991 census of Canada has collected data on disability in Canada. This is further supplemented with sample data from disability surveys. These data are analyzed here to set up various disability rate estimates by province, age, gender, etc. Since the age breakdown is very broad, interpolation is done to find age-specific rates. Using the population of Canada as the standard, decomposition of the incidence/prevalence of disability is attempted. Disability-free life expectancy is also shown for the country as a whole and for each province. Trends in disability are presented.

Population Size and Age Structure of Nunavut, 1986-2016
Cameron W. Stout, Demography Division, Statistics Canada & Jeanine Perreault, Demography Division, Statistics Canada

Nunavut will officially become Canada's third northern territory in 1999. This paper will focus on Nunavut's changing age structure and population growth over three decades, 1986-2016. Recent and projected trends in the territory's components of population growth will be analyzed. Three scenarios of Nunavut's potential population growth over the next two decades will be provided.

Receipt of Assistance: Chinese Community-Dwelling Elders
Ellen M. Gee, Sociology and Anthropology, Simon Fraser University & Neena L. Chappell, Centre on Aging, University of Victoria

While we have substantial information about care receipt among mainstream Canadian elders, we know much less about informal (and formal) assistance among sub-groups who are outside the normal research ken, due to language and other cultural barriers. One such sub-group is the Chinese Canadian elderly (a sub-group that is growing substantially due to past and present immigration).

A 1995-96 survey of 830 randomly sampled Chinese persons aged 65 and over residing in Vancouver and Victoria (and interviewed in the language of their choice -- which involved Cantonese, Mandarin, English, and five Chinese dialects) provides us with data to examine patterns of health status and care receipt, including social support, among this minority group in comparison with Canadian averages (as reported in the 1990 General Social Survey). Of particular interest is the more than one-half (437 or 52.7%) of Chinese elders who reported having 3 or more functional disabilities. We examine the receipt of assistance among this more disabled sub-sample of the survey, focussing on variations by age, gender, marital status, living arrangements (living with child(ren) or not), type of functional disability, and measures of acculturation such as age at immigration and ability to speak English. We also focus on the first three (ranked) providers of assistance (in terms of relationship and ability to speak Chinese), the perceived need for more assistance, and the use of formal care services, particularly home care.

Our overall goal is to assess the patterns of receipt of informal elder care, the degree of unmet need, and the factors involved in quality of life among Chinese Canadian elders. We discuss our findings in terms of their implications for aging policy in a society in which the elderly are becoming increasingly diverse.

Socio-Economic Correlates of Life Expectancy among Health Regions of Alberta
George Parakulam, Health Surveillance, Health Strategies & Research Division, Alberta Health & Yan Jin, Health Surveillance, Health Strategies & Research Division, Alberta Health

Life expectancy is a widely used health outcome indicator to measure health status of a population. Life tables for the 17 health regions in Alberta were constructed. The data were available from Alberta Vital Statistics from 1989 to 1993 and from Census 1991. Issues of unstable mortality rate for some small regions in terms of population size was addressed.

Comparison of life expectancy at birth across regions showed that the difference between the highest and lowest life expectancy at birth was large. Correlations with many socio-economic indicators demonstrated that life expectancy is sensitive to social and economic factors which are known to influence health. More specifically, indicators of economic status and development are positively correlated to life expectancy. Indicators of social problems on the other hand show moderate to strong negative correlation with life expectancy among the health regions. Further research examining in healthy life expectancy and multiple decrement life stables is required to better understand the dynamics of population health and well-being in Alberta.

Third Birth Intentions in Canada
Hui Wang, Sociology, University of Victoria & Zheng Wu, Sociology, University of Victoria

Third Birth Intentions and Uncertainty in Canada The survival of a society depends on its ability to replace its members. In modern societies, this, to a large extent, depends on whether a significant number of women would bear three or more children, for the simple reason that third birth represents the dividing line between above and below the replacement-level fertility. Using data from the 1990 Canadian General Social Survey (GSS), this paper studies the intention of women to bear a third child. We examine the individual-level factors that influence this intention with a sample of married or cohabiting women who were in their childbearing ages and who had two children of their own. We found that desirability of a third birth declines with a woman's age, her age at union, being employed outside home and regular use of childcare services, but increases with the frequency of her church attendance and having had two boys.