Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
by Derrick Thomas
What you should know about this study
What was the position of women within their households and families?
How was the fertility of women regarded?
Was the ethnic legacy of mothers acknowledged?
How was citizenship dealt with for women?
Was the work of women acknowledged?
Canadian society has changed in many ways over the past century. Gender roles and relations are among the areas that have undergone the most profound transformations. Today, legal and social equality between the sexes are explicit and virtually unquestioned societal goals. Few young people today would recognize the Canada of 1911 or 1921. There were pronounced social distinctions between men and women which prevailed a mere generation ago. The Census of Canada has been part of this history and has evolved along with society generally. Early census collections and reports largely reflected the social context of their time.
Beyond the physical differences between men and women and their different reproductive functions are separate sets of socially-determined behavioural norms and performance standards attached to each gender. For example, these norms may dictate dress or acceptable occupations. Social conventions may also set down different roles within the family or establish a hierarchy with respect to the sharing of work, resources and decision-making in the household and more broadly.
Evidence of the social construction of gender roles can be seen by examining these roles across societies or across time. The variability seen suggests that roles are not innate. They must be established and maintained via a more or less general agreement. In Canada this consensus is thought to be evolving, at least over the longer term, from a patriarchal model to a more egalitarian one.1
This article examines some of the ways in which gender roles changed over the first half of the 20th century. More specifically, it considers how the census adapted to these changes and reflected the new reality. The Census of Canada has been conducted since Confederation and provides us with a perspective on how Canadian society has changed as information on a relatively consistent set of characteristics has been captured over time. This study utilizes data collected in the decennial Censuses of Canada carried out between 1911 and 1951. For the first time computerized microdata for these census years are available.
Social constructs can be almost invisible to contemporary social actors and can form an important part of what is regarded as reality.2 Some historical perspective is usually required before they can be understood. Gender roles and expectations represent a mode of thinking and acting, and form part of an individual’s identity or relation to self and others.3 These constructs not only condition behaviour but also influence what is regarded as scientifically interesting, worthy of recording and collecting information about. The information collected and the presentation of results are indicative of the roles played by household members and what was deemed important at various junctures. This information ostensibly provided basic knowledge for the formulation of social policy and the transformation of the society that it measured.
The instructions to enumerators who collected information in these early censuses indicate who they were to question, on which information they were to focus, and which answers they were to disregard. Headquarters staff with the Dominion Bureau of Statistics (DBS)—as Statistics Canada was then called—also recoded the information taken down by enumerators and organized it for publication. Their roles are less well documented but administrative reports and the categories used in the census publications provide some insight. The instructions prepared along with the classification and recoding done by census takers reflected the attitudes and social norms of the time.
Between 2003 and 2009 the Canadian Century Research Infrastructure (CCRI) project recompiled and digitized census data from the first half of the 20th century, giving us a new perspective on the information collected from 1911 through 1951 (See “What you should know about this study”).4 A review of administrative material and census questionnaires, along with a comparison between data published by the DBS and estimated data based on the CCRI samples sheds some light on how gender roles were perceived by society.
Prior to 1971, the census questionnaires were not filled out by household members in the now-familiar way. Statistics Canada and its predecessor, the Dominion Bureau of Statistics (DBS), employed thousands of enumerators who went door to door visiting virtually every household in the country. In 1951 the enumerator was described by the Chief Statistician of Canada as the "most important man in the organization" although 40% were women that year. These enumerators questioned household members, filled out forms and submitted the data for review and editing by Census Commissioners and DBS headquarters staff. The same basic set of questions was put to each and every household.
Canadian Century Research Infrastructure
The Canadian Century Research Infrastructure (CCRI) is a five-year effort to build a comprehensive database of information on early 20th century Canada that might be used to address research questions from a wide variety of academic disciplines1. The project has been supported by the Canadian Foundation for Innovation, federal and provincial governments, a number of Canadian universities and Statistics Canada. It involves a large team of academics, researchers and specialists and is linked to a number of similar international projects.
An important part of the project has been the computerization and rejuvenation of microdata collected by the decennial Censuses of Canada from 1911 to 1951. Data from these pre-computer era censuses had never been digitized and existed only in the form of micro-filmed census returns as completed by enumerators—often in cursively written ink or pencil. Costs dictated that only a sample, about 5%, of completed census forms for each census could be captured. While this data capture cannot be regarded as a new collection it is certainly a new compilation. Data entry personnel worked from what amount to photos of original documents. The coding structure, rules and coding decisions had largely to be reinvented in the absence of corporate memory and with sketchy documentation.
CCRI operators key-entered all responses as recorded by enumerators. An effort was made to digitize the information verbatim or at the lowest level of aggregation. In cases where corrections had been made at the time of the original census by Commissioners or DBS headquarters staff, CCRI operators entered the enumerator’s original response. For some variables operators entered the DBS corrections in a second associated field. This allows a researcher to access the effects of such ‘corrections’.
CCRI teams in five universities across the country worked from the microfilmed schedules hand-written by the original enumerators. The microfilms were made available by Statistics Canada under conditions which respected the privacy of census respondents. The university centres used modern data-capture software to browse the reels of microfilm and capture computer-readable images. Sampling occurred within geographic strata corresponding to census subdistricts (enumeration areas in modern parlance). With some operator input, the software selected dwellings from a randomly determined start point at a predetermined interval. The interval or sampling fraction differed according to the size of the dwelling and, in the case of large dwellings, according to whether the dwelling contained multiple units/households or a collection of individuals in a dormitory or institutional setting. Dwellings with 30 or fewer persons were sampled at a rate of 1 in 20 for 1911, 1 in 25 for 1921, and 1 in 33 for 1931, 1941 and 1951. In 1911 all dwellings with more than 30 occupants were selected. One in 5 households (1 in 4 for 1911) in each multi-unit large dwelling was sampled and 1 in 10 individuals were selected in each large collective dwelling.2 The reciprocal of these sampling fractions in conjunction with information about the dwelling size and type were used to compute the weights.
The vast majority of women in early 20th century Canada lived in legally-constituted conjugal relationships. Common-law relationships were infrequent and not reported to census takers until 1981. According to data recompiled by the CCRI, at each census from 1911 to 1951, close to 90% of women had entered into a legal marriage by age 50. The great majority raised children and remained in these families for much of their lives as separation and divorce were extremely rare. Only 2,275 divorces were granted in Canada between 1881 and 1921.5 According to the data compiled by the CCRI, only about 1 in 1000 women were divorced or separated at each census taken between 1911 and 1941. In 1951 the rate increased to 4 in 1000. In 2006, by contrast, over 120 in 1000 women 15 years of age and over were divorced or separated.
In the early part of the last century men were apparently regarded as the persons in charge of their families. The Census reflected this view. Census takers employed the term ‘head of household’ when collecting and organizing the information gathered from each family. Other household members were defined by their relationship to the household head and household dwelling information was gathered only with reference to the household head. For example, home ownership was recorded for the household head regardless of who owned the dwelling. Information published on employment and earnings also tended to focus on the household head.
While there was no explicit order to that effect, the man or husband was clearly considered to be the ‘head’ of the household. The position of males was implicit in the instructions to enumerators which remained essentially unchanged over the 1911 to 1941 period. They stipulated that “The members of the family or household … are to be entered in the following order, namely: Head first, wife second then sons and daughters in the order of their ages, and lastly relatives, servants, boarders or other persons living in the family or household.”6 The ‘Head’/‘Wife’ distinction was also reflected in published tables.7 Only in single-parent or all-female households was a woman to be considered the ‘head’.
The data shows that the vast majority of household heads were male. However, the proportion of household heads that were female grew between 1911 and 1951 (Chart 1). This may be related to absent husbands and fathers, particularly during the period that included the Second World War. An examination of the data digitized by the CCRI indicates that only in a few thousand cases in each census year did enumerators designate a female as head of the household when a male partner was present.
By 1951 society was changing and the census instructions reflected a new awareness. For the first time, it was explicitly mentioned that the husband would be considered the head of the household. However, no hierarchy or social status was implied and census manuals noted that the designation of household head was only for the purposes of organizing the data collected. According to the Enumeration Manual for the ninth Census of Canada, “For Census purposes every household must have a head. In households consisting of husband and wife with or without children, the husband will be recorded as ‘head’.”
Husbands continued to be considered the head of their households up to the beginning of self-enumeration in 1971. In 1976 either husband or wife could be considered head of the household. With the 1981 Census the term was dropped completely.
Over the early part of the last century, one of the most prominent roles of a wife was to give birth and raise children within a male-headed family. For the 1941 Census detailed information was collected on fertility. Women who were or had been married were asked if they had ever given birth to a child. Over 83% of those 15 years old or over had birthed a child and one in five had given birth to six or more. While almost exactly the same proportion of women in relationships have children today, very few give birth to six or more (less than 3% of those surveyed in 20068).
One pronounced change has been the attitude toward the fertility of unattached women. For example, in 2006 almost 12% of single women 15 years of age or over (excluding those in common-law relationships) had given birth to at least one child. Because childbirth outside of marriage was considered rare, this question was not put to single women in 1941. Perhaps the social stigma associated with having a child outside the confines of marriage was so strong that it precluded enumerators asking the question. However, childbirth outside of wedlock did occur. Data compiled for the CCRI indicate that although unpublished, there were approximately 6,000 births outside of wedlock recorded in 1941.
While women bore and raised children, a child’s ancestry was traced through the father. A mother’s ancestry was not considered in the classification of the ancestry of their child. It was not until 1981 that both parents’ ancestry was recorded.
The 1931 Census enumerator instructions for the collection of ancestry data are typical for the period. They directed that, “A person whose father is English and whose mother is French will be recorded as of English origin, while a person whose father is French and whose mother is English will be recorded as of French origin, and similarly with other combinations…” These instructions applied to the majority of census respondents who were of European or ‘white’ origins. Aboriginals and members of visible minority groups were treated differently.9
The logic of tracing ancestry through the male line would seem to imply that each person should have a single origin. Indeed the instructions given to enumerators did not allow for multiple origins. The Dominion Bureau of Statistics did not publish any figures for people claiming more than one origin between 1911 and 1951. However, the data compiled by the CCRI clearly show that enumerators did sometimes record multiple ancestries. These data show a gradual increase in multiple ancestries from about 1 in 300 respondents in 1921 and 1931 to about 1 in 200 in 1941 and over 1 in 150 in 1951. With the 1981 Census, Canadians were permitted to write in as many ethnicities as they felt they had. They were thus able to report their maternal ancestry.
Women did not fully enjoy the rights of citizenship in early 20th century Canada. They did not, for instance, gain the right to vote in federal elections until 1920. According to Canadian law, before 1932 a woman took on the nationality of her husband when they married. A woman born or naturalized in Canada who married an ‘alien’10 lost her Canadian citizenship. Women legally gave up their right to vote and to a Canadian passport if they married someone other than a British subject or if their husband became an alien over the course of their marriage. Similarly, a woman of foreign nationality gained Canadian citizenship by virtue of her marriage to a Canadian.11 In 1921 and 1931 enumerator instructions indicated that “A married woman is to be reported as of the same citizenship as her husband.”
Regardless of gender, citizenship could be lost in a number of ways. For example, it was possible to lose citizenship via renunciation or taking on another allegiance. While it is not possible to determine from census data precisely how many Canadian-born people lost their rights, according to the 1921 to 1941 census data compiled by the CCRI, the number of Canadian- and British-born women residing in Canada without citizenship exceeded, by a considerable number, the number of men in that situation (Chart 2). Loss of citizenship status due to marriage to a non-British subject is the most likely explanation for this gender difference.
Between 1932 and 1947, a married woman’s citizenship depended on the nationality law of her husband’s country of allegiance. If she legally acquired his citizenship she lost her Canadian status. Otherwise she remained Canadian. The citizenship law of 1947 removed much of the gender discrimination with respect to citizenship and census data-collection practices quickly followed. By 1951, the number of Canadian born who were not citizens was more equal for men and women (Chart 2).
Perhaps the most interesting area in which social conventions changed over the 20th century is the participation of women in paid employment, their occupations and earnings. During the early part of the 20th century, most women did not participate in the paid labour force. In 1931, for example, just 16% of women were involved in paid employment, compared to almost 70% of men. By 2006, however, the employment rate for women was almost 60%. There continue to be differences in the employment experiences of men and women. For example, the barriers faced by women in the workforce and the gap in earnings between men and women have been well documented12 and have led to federal and provincial employment equity legislation.
While the unpaid work done in the home or for the family was important, during the first half of the 20th century governments and businesses were more interested in labour sold on markets.13 Before the 1931 Census, women working in their own home were regarded as having no occupation. For example, the enumeration manual for 1921 states that under occupation: “In the case of a woman doing housework in her own home, without salary or wages, and having no other employment, the entry should be none.” While the information was not published by the Dominion Bureau of Statistics, enumerators often recorded the answer ‘homemaker’ or ‘housewife’ in this situation. Beginning in 1931 the enumeration manual indicated that the term “homemaker” was to be used for women working in their own homes. CCRI operators recorded the answer ‘homemaker’ for about 2.45 million females and 12,600 males when they recompiled the 1941 Census information from the original returns. The figure for males had not been published by the DBS.
Little official documentation has survived about the coding structures used by those who compiled and tabulated census information in the early part of the last century. CCRI staff coded occupation anew from the information originally written down by enumerators. Needless to say, jobs and occupational categories have changed over the past half century and the occupational classification used by the CCRI project differs from the one originally employed by the DBS.
For instance, CCRI coders identified many more gainfully employed women than the original census data published by the DBS. Indeed, the 1941 Census data recompiled by the CCRI had more than three times as many unpaid family workers as did the published counts from the era (data not shown). It seems the modern coders were more likely than the original coders to describe some homemakers and women on farms as ‘unpaid family workers’. For census purposes, the term ‘unpaid family worker’ applied only to someone working in an enterprise for which someone in the family receives money. Although the modern census now gathers information on elder care, child care and other work performed in the home, homemakers are still not regarded as being a part of the labour market.
Women who worked in occupations normally deemed to be the preserve of men may not have been counted in early censuses. Enumerators were explicitly instructed to treat with skepticism any suggestion that a woman had any ‘unusual’ occupation. Further inquiry and correction was deemed necessary in such cases. For example, according to the 1931 enumeration manual, “There are many occupations such as carpenter and blacksmith which women usually do not follow. Therefore, if you are told that a woman follows an occupation which is peculiar or unusual for a woman, verify the statement.”
A similar instruction appeared in the 1921 and 1941 manuals. The occupation data published by the DBS after each census from 1911 to 1951 indicates that there were no females in these occupations. The data coded by CCRI suggests, however, that even after close questioning, hundreds of women were deemed by enumerators to have worked as carpenters and blacksmiths. These answers appear to have been recoded by headquarters staff when official tabulations were prepared.
Proportionally few women have traditionally gone into trade-related occupations and this continues today. For example, women accounted for only 3% of those registered in apprenticeship programs in the construction trades as late as 2005.14 The modern census makes no distinction between men and women when collecting occupational data and where numbers warrant these data are published.
The earnings of women have not always been published. The census reports from the beginning of the 20th century focused on the earnings of household heads. Following the 1921 Census, for example, many earnings tables were produced for male heads of households only. The data was taken down by enumerators, however, and can be tabulated based on the CCRI sample.
Where the DBS did publish earnings for women they correspond with the data obtained from the CCRI. On average for the census years under study, a working woman earned just over 50 cents for each dollar earned by a working man.15 The gap did not change dramatically between 1921 and 1941. The persistent gender wage gap has been an important focus of Statistics Canada analysis and publications over the past few decades.16
The primary role of most adult women living in the first half of the 20th century was to care for their family and home. Men more often worked outside the home for pay and assumed the role of the household head. Census collections were designed with this social reality in mind. Husbands were the official heads of their households for census purposes and other household members were defined by their relationship to the male head.
According to the norms of the time ancestry or ethnicity was traced in the census through the male line. Children were recorded as having the cultural heritage of their fathers.
Until 1947, a women’s citizenship was tied to her husband’s citizenship both legally and for census purposes.
Women working in their homes were not considered ‘gainfully employed’. Before 1931 they were considered to have no occupation. Women in some non-traditional or ‘unusual’ occupations, moreover, were not tabulated in census publications.
Census collection emphasized paid work and in some cases earnings information was published only for heads of households. According to the data recompiled for the CCRI, average weekly earnings for women and men in the labour market differed substantially for the census years examined here. On average, a woman in the paid workforce earned about half of what a man earned in 1921, 1931 and 1941. Interestingly, although the wage gap has narrowed, in 2008 the average hourly earnings for women continued to be below those of men’s—illustrating that there remain differences in labour market experiences between genders.
Censuses were planned and carried out within a social and historical context. The expectations of census takers played a role in what they looked for and in what they found. The data collected and published tended to reflect and reinforce the norms of the day. The original and rejuvenated census data, documentation and instruction manuals from the early part of the twentieth century enable researchers to examine how gender roles have evolved and changed as changes in society occurred.
Derrick Thomas is a senior analyst with Social and Aboriginal Statistics Division, Statistics Canada.