The methodology chapter is divided into four parts. Information on the sample is presented first. This is followed in Section 2 by a description of how key constructs, such as perceived health and work-life conflict, were operationalized in this study. A brief discussion of the statistical techniques used in the analysis is found in Section 3. The last section outlines the reporting protocols followed throughout the report.
The sample for the "National Study on Balancing Work, Family and Lifestyle" was drawn from 100 Canadian companies with 500+ employees. Forty of these organizations operated in the private sector, 22 were from the public sector and 38 were from the not-for-profit (NFP) sector. Private sector companies from the following sectors were included in the sample: telecommunications, high technology, retail, transportation, pharmaceutical, financial services, entertainment, natural resources and manufacturing. The public sector sample included 7 municipal governments, 7 provincial government departments, and 8 federal government departments/ agencies. The NFP sector sample consisted of 15 hospitals/district health councils, 10 school boards, 8 universities and colleges, and 5 "other" organizations that could best be classified as NFP/greater public sector (e.g. social service, charity, protective services).
A total of 31,571 people responded to the survey. The sample is distributed as follows:
A full description of the sample can be found in Higgins and Duxbury (2002). Details on work-life conflict and other outcome data can be found in Duxbury and Higgins (2003). A summary of key findings of relevance to the readers of this report is given below.
The 2001 study sample is well distributed with respect to age, region, community size, job type, education, personal income, family income, and family's financial well-being. In many ways, the demographic characteristics of the sample correspond to national data, suggesting that the results from this research can be generalized beyond this study. Approximately half of the respondents to the survey can be considered to be highly educated male and female knowledge workers. The majority of respondents are part of a dual-income family and indicate that they are able to "live comfortably" (but not luxuriously) on two full-time incomes. Respondents who belong to a traditional, male breadwinner family are in the minority (5% of total sample, 11% of the sample of men) and outnumbered by respondents who are single parents. The fact that the traditional families tended to be headed by highly paid male managers and professionals suggests that this family arrangement is restricted to those with higher incomes.
The mean age of the respondents to this survey is 42.8 years which puts them in the mid-career/fast-track stage of the career cycle, the "full-nest" stage of the life cycle and the 40's transition stage of adult development. Each of these stages is associated with increased stress and greater work and family demands. Three quarters of the respondents to this survey are presently married or living with a significant other and 69% are part of a dual-income family. Eleven percent of the respondents are single parents. Twelve percent of the sample live in rural areas. One in three is a clerical or administrative employee with a lower level of formal education (i.e. reduced job mobility) and lower personal and family incomes. One quarter of the respondents indicated that money is tight in their family; 29% of respondents earn less than $40,000 per year and just over one quarter live in families with total family incomes that are less than the Canadian average. One in three of the respondents has a high school education or less.
In this report, we focus on four different types of work-life conflict: role overload, work to family interference, family to work interference and caregiver strain.
Role overload is having too much to do in a given amount of time. This form of work-life conflict occurs when the total demands on time and energy associated with the prescribed activities of multiple roles are too great to perform the roles adequately or comfortably. The majority of employees in our sample (58%) are currently experiencing high levels of role overload. Another 30% report moderate levels of role overload. Only 12% of the respondents in this sample report low levels of overload.
Work to family interference occurs when work demands and responsibilities make it more difficult for an employee to fulfil family role responsibilities. One in four of the Canadians in this sample reports that his or her work responsibilities interfere with the ability to fulfil responsibilities at home. Almost 40% of the respondents report moderate levels of interference. The proportion of the Canadian workforce with high levels of work to family interference has not changed substantially over the past decade.
Family to work interference occurs when family demands and responsibilities make it more difficult for an employee to fulfil work role responsibilities. Only 10% of the Canadians in this sample reported high levels of family to work interference. Another third reported moderate levels of family to work interference. The percentage of working Canadians who give priority to family rather than work has doubled over the past decade.
Caregiver strain is defined as physical, financial or mental stress associated with providing care or assistance to a disabled or elderly dependent. Approximately one in four of the individuals in this sample experiences what can be considered to be high levels of caregiver strain.16 The rest of the respondents to this survey (74%) rarely experience caregiver strain.
Who, in this sample, has more problems balancing work and family responsibilities? The evidence is quite clear-employed Canadians with dependent care responsibilities (i.e. child care, elder care or both). Employees who have child and/or elder care responsibilities report higher levels of work-life conflict than those without such responsibilities regardless of how work-life conflict was assessed. The finding that employees without dependent care responsibilities are more able to balance work with life can be attributed to two factors: fewer demands outside of work and more degrees of freedom to deal with work issues (e.g. more control over their time).
Job type is associated with all but one of the measures of work-life conflict examined in this study. Employees with higher demands at work (i.e. managers and professionals) are more likely than those in "other" jobs to experience high levels of overload and work to family interference. Those in "other" jobs, on the other hand, are more likely to report higher levels of caregiver strain due to the financial stresses associated with elder care.
Women are more likely than men to report high levels of role overload and high caregiver strain. This is consistent with the fact that the women in this sample devote more hours per week than men to non-work activities such as child care and elder care and are more likely to have primary responsibility for non-work tasks.
A 12-page survey produced in a mark-sensitive format with a unique bar code given to each organization participating in the study was used to collect the data. This survey was divided into nine sections: your job; your manager; time management; work, family and personal life; work arrangements; work environment; family; physical and mental health; and "information about you." Virtually all of the scales used in the questionnaire are psychometrically sound measures that have been well validated in other studies.
One major objective of this research was to attempt to put some kind of dollar value on work-life conflict. To do this we developed, for the purposes of this study, a number of questions focusing on the respondents' use of Canada's health care system, and their use of prescription medicine. These measures were modelled on the questions developed by Moos et al. (1988) for use in the Health and Daily Living Form. Information on the constructs used in this analysis is given below and in Boxes 3 (definition of work-life conflict), 4 (measurement of work-life conflict) and 5 (health and health care outcomes).
Box 3 Defining Work-Life Conflict Work-life conflict is conceptualized broadly in this study to include role overload, work to family interference, family to work interference and caregiver strain. The working definition of each of these constructs is given below. Role overload is having too much to do in a given amount of time. This form of work-life conflict occurs when the total demands on time and energy associated with the prescribed activities of multiple roles are too great to perform the roles adequately or comfortably. Role interference occurs when incompatible demands make it difficult, if not impossible, for an employee to perform all roles well. Role interference is conceptualized as having two distinct facets:
Caregiver strain: Caregiver strain is an outcome which may arise due to responsibility for the care of an elderly or disabled dependent. Caregiver strain is a multidimensional construct which is defined in terms of "burdens" or changes in the caregivers' day-to-day lives which can be attributed to the need to provide care (Robinson, 1983). Four types of caregiver strains resulting from stress have been identified: emotional strain (i.e. depression, anxiety, emotional exhaustion), physical strain, financial strain, and family strain. It should be noted that research on caregiver strain has typically focused on strains associated with the provision of elder care or care for a disabled dependent rather than those linked to child care. |
Box 4 Measurement of Work-Life Conflict Role overload was assessed in this study using five items from a scale developed by Bohen and Viveros-Long (1981). Role overload was calculated as the summed average of these five items. High scores indicate greater role overload. In this study, Cronbach's alpha for this scale was 0.88. Work to family interference was measured by means of a 5-item Likert scale developed by Gutek, Searle and Kelpa (1991). Work to family interference was calculated as the summed average of these five items. High scores indicate higher levels of perceived interference. In this study, Cronbach's alpha for this scale was 0.92. Family to work interference was assessed by means of a 5-item Likert scale developed by Gutek, Searle and Kelpa (1991). Family to work interference was calculated as the summed average of these five items. High scores indicate higher levels of perceived interference. In this study, Cronbach's alpha for this scale was 0.87. Caregiver strain was quantified using a modified three-item version of Robinson's (1983) Caregiver Strain Index (CSI) (family strain, a key family outcome, was assessed separately). This index measures objective (rather than subjective) burden in four areas. Respondents were asked to indicate (using a 5-point Likert scale) how often they had difficulty in caring for an elderly or disabled relative or dependent because of physical strains, financial strains or because it left them feeling completely overwhelmed. Options given included never, monthly, weekly, several days per week or daily. Total caregiver strain was calculated as the summed average of these three items. Higher scores indicate greater strain. This measure has been used in a number of studies with good results (Robinson reports a Cronbach alpha of 0.91). In this study, the Cronbach alpha was 0.78. |
Box 5
Measurement of Health Outcomes Perceived health was measured by asking respondents the following: "Compared to other people your age, how would you describe your usual state of health?" A 5-point Likert scale with the following anchors were used for the responses: 1 = poor, 2 = fair, 3 = good, 4 = very good and 5 = excellent. This measure of health status has been found by Statistics Canada (1999) to be a good predictor of the presence of more "objectively" measured health problems as well as health care utilization and longevity. Use of Canada's health care system by Canadian employees was estimated by asking respondents to indicate (yes or no) whether or not in the past six months they had:
Respondents who indicated that they had used any of the above services were then asked to record either the number of visits they had made (or, in the case of hospital use, the number of days they had spent in hospital) in the six-month period. These measures are used two ways in this study: (1) as an indicator of demands on various facets of the health care system, and (2) as indicators (albeit crude) of an employee's physical and mental health. The first question in this series (visits to the physician) was developed by Moos, Cronkite, Billings and Finney (1988) for use in the Health and Daily Living Form. Previous studies have determined that this measure is a good proxy for actual health status and accurately reflects actual physician visits. Accordingly, the other measures of use were designed using the same format. Use of prescription medicine: The perceived use of prescription medicine was also collected in this study. Respondents were asked if in the past six months they had purchased prescription medicine for their own personal use (yes/no response). Those who answered yes were asked to indicate approximately how much they had spent on prescription medicine in the last six months. They were given the following choices: $0, under $50, $50 to $99, $100 to $149, $150 to $199, $200 to $299, $300 to $399 and $400 or more. They were also asked to indicate who paid for the prescription medicine that they had purchased for their own personal use. Responses ranged from respondent pays 100% to company pays 100%. This measure was developed for this study. |
The following health and health care outcomes were included in this study: perceived health, use of Canada's health care system, and perceived use of prescription medicine. Details on each of these measures are given below and in Box 5.
Measures of health status may be subjective (e.g. self-rated health status) or objective such as instrumented measures of blood pressure. They may describe health directly (e.g. incidence of cancer) or indirectly (e.g. health care utilization is a proxy for the existence of a health problem) (Statistics Canada, 1999). The measure of self-rated health status used in this study (see Box 5) was developed by Statistics Canada and summarizes physical and mental health as experienced by the individual according to his or her values (Statistics Canada, 1999, p. 217). This measure of health status has been found by Statistics Canada (1999) to be a good predictor of the presence of more "objectively" measured health problems, as well as health care utilization and longevity.
Use of Canada's health care system: This study examines the extent to which Canadian employees use six facets of Canada's health care system: physicians, other health care professionals, mental health professionals, hospitals (both emergency wards and overnight stays) and outpatient clinics. While the use of health services is a less satisfactory indicator of health status than physical health measures, such as blood pressure or cholesterol, it is a useful measure of health (Statistics Canada, 1999). Furthermore, information on the use of physicians, other health care professionals and hospital services can indicate emerging trends that may have an impact on health care budgets (Statistics Canada, 1999) and allows us to estimate how much high work-life conflict costs Canadian society in terms of demands on the health care system.
Prescription drug use: Prescription drug use and its associated costs are escalating in Canada (Statistics Canada, 1999). It has been estimated, for example, that Canadians spent over $15.5 billion on drugs in 2001-an increase of 8.6% over the previous year. Why is spending on drugs rising? Reasons cited in the literature include a decline in the health of Canadians, a trend toward drug therapy (rather than surgery), the emergence of new diseases for which drugs are the treatment of choice, and the development of new drugs to treat old diseases (CIHI, 2002). In Canada, both the public and private sectors pay part of the drug bill and many companies are concerned with the increase they are experiencing in their benefits costs as employees purchase more prescription medicine. This measure will be used in two ways in this study: as a surrogate indicator of health status (higher prescription drug expenditures are assumed to reflect poorer health) and to further assess the impact of high work-life conflict on the bottom line.
The following types of analysis were undertaken to meet the research objectives outlined above:
Perceived health: These data are reported in two ways in this report: mean level of perceived health and the frequency with which:
Use of health services: One of the key objectives of this study was to use our sample to estimate the extent to which Canadian employees were using various facets of the health care system. Two aspects of use will be examined in this study: likelihood of use (i.e. the percent of the sample who did and did not use the service) and the mean amount the service was used. This second statistic is reported in two different ways:
Use of prescription medicine: After examination of the distribution of the responses, this variable was collapsed into three groups as follows: those who had spent nothing, those who had spent $1 to $150 and those who had spent more than $150 on prescription medicine in the six months prior to the survey being conducted. Frequencies were then calculated for the collapsed variable.
The impact of work-life conflict: A second objective of this research was to look at the impact of high work-life conflict on perceived health, use of Canada's health care system and prescription drug use. The procedures used to examine the impact of work-life conflict on each of these health outcomes are shown in Box 6.
Examination of the impact of gender, job type, dependent care status and sector of employment on health outcomes: Research done in this area suggests that gender, job type, dependent care status and sector of employment might all influence the outcomes (i.e. work-life conflict, perceived health, use of the health care system) included in this study.17 The procedure used to examine between-group differences in health outcomes can be found in Box 7.
Relative risk is a way to measure strength of association between two constructs. The higher the relative risk, the stronger the association (CIHI, 2003). For example, researchers have determined that the relative risk of developing lung cancer in smokers versus non-smokers is approximately 3.0 (CIHI, 2003), meaning that smokers are 3.0 times more likely to develop lung cancer than non-smokers. Relative risk is calculated in this report by dividing the percent of the sample with high work-life conflict who report a particular outcome by the percent of the sample with low work-life conflict who report this outcome (or vice versa depending on the item).
Box 6 Methodology Used to Examine the Impact of Work-Life Conflict This report looks at the impact of high work-life conflict on perceived health, prescription drug use and use of Canada's health care system (the measures used to quantify each of these outcomes are shown in Box 5). The procedure for this analysis can be summarized as follows:
|
Box 7 Methodology Used to Examine the Impact of Gender, Job Type, Dependent Care Status and Sector of Employment This paper examines how key contextual factors affect work-life conflict. It also explores the association between these contextual factors and the attitudes and outcomes under study. The contextual factors were operationalized as follows:
This research series takes a fairly unique approach to the analysis of gender impacts on work-life conflict by examining gender differences within job type, dependent care status and sector of employment. Such an analysis recognizes that Canadian men and women have different realities and that it may be these realities, rather than gender itself, that have an impact on the attitudes and outcomes being examined in this analysis. This type of analysis should be invaluable to policy makers who need to know if supports and interventions should be targeted to a particular group (e.g. women, parents) or an environmental condition (e.g. low-control jobs). Two different statistical tests were used to determine if the differences were statistically significant. Crosstab procedures were used when the data were reported as a frequency (i.e. % using a service) to determine if the effect of job type, employment sector and dependent care status on the outcome of interest was the same for men and women. In those cases where we report a mean (i.e. number of visits per six-month period), an F test was used to test for between-group differences. Where more than three groups were involved, Scheffe's follow-up tests were used to determine which groups were different from each other. |
Finally, it should be noted that the cross-sectional nature of the data collected for this study means that the direction of causality between the various measures of work-life conflict examined in this research and the use of the health care system cannot be determined. Theoretically, most of the research done in this area supports our hypothesis that higher levels of work-life conflict contribute to poorer health and greater use of medical services. The authors do, however, acknowledge that it is possible that in some cases the direction of causality may be reversed (i.e. people who are sick and make greater use of the health care system are more likely to experience work-life conflict as a result of their health problem).
All of the differences discussed in the report meet two criteria: they are statistically significant and substantive (i.e. the differences matter in a practical sense). This second requirement was necessary as the large sample sizes meant that differences as small as 0.5% were often statistically significant. In interpreting the data, the reader should use the following rule of thumb: the greater the difference, the more important the finding.
The reader should also be aware that most numbers reported in the text have been rounded off to the nearest decimal or presented as whole numbers and thus may not add up to 100%.
In the survey, we asked respondents to use the six months prior to the study as their frame of reference when answering questions on use of the health care system, prescription drug use, etc. This time frame was selected as research in this area indicates that recall data become less reliable when people are asked to consider longer time frames. The data reported are therefore presented for the six-month period; however, the data used in the conclusion section and in the executive summary are presented as visits or use per year. These estimates were obtained by multiplying the findings for a six-month period by two.
Finally, it should be noted that to make the text easier to follow, most of the source data have been put in Appendices at the end of the report. The data are grouped as follows:
[Previous] [Table of Contents] [Next]