ALTERNATE ROUTES

A Critical Review

Editorial Board : Marybelle Myers, Erika van Meurs

Distribution and Promotion : Brian McKee

Alternate Routes is a multi-disciplinary journal published

annually by graduate students in the Department of Sociology

and Anthropology at Carleton University in Ottawa. It is not,

however , restricted to work from students at Carleton. We are

interested in receiving papers written by graduate students

(or co-authored with faculty) in any university . Editorial

emphasis is on provocative analysis of contemporary issues and

we particularly welcome papers dealing with debates within

Canadian society. The theme of volume 7, to be published in

1984 is Class and Ethnicity

.

Ve also welcome critical responses to articles published and

we occasionally publish review articles of important works.

Two copies of your typewritten manuscript should be sent to:

Alternate Routes,

Department of Sociology and Anthropology

,

Carleton University

,

Ottawa, Ontario, KlS SB6

The editors gratefully acknowledge the financial assistance of

the Carleton University Community : the Department of Sociology

and Anthropology and its chairperson, Gordon Irving; the Dean of

Social Sciences , Dennis Forcese; the Graduate Students ' Association

and the Sociology and Anthropology Graduate Students ' Caucus. We

also wish to thank Eva Hegmann, Jennifer Davidson and Kim Mitchell

for their assistance

.

Subscriptions : Individuals and Institutions : $4.50.

Make cheques payable to A Iternate Routes .

copyright 1983 Alternate Routes
Printed in Canada



EDITORIAL STATEMENT

The papers in this special issue reflect both the

versity of perspectives which have been applied to

eminist issues and the wide range of substantive problems

eing addressed.

Armstrong and Armstrong examine quantitative data

ollection by Statistics Canada. They argue that the

onceptualizations and operationalizations employed reflect

male bias. The authors suggest changes to remedy this

ias, and in addition to this, emphasize the need for a

reater use of qualitative techniques

.

In a second paper, written from a psychoanalytic

erspective , Hamilton contributes to the ongoing feminist

ebate on the oppression of women by capitalism and

atriarchy

.

Saunders focuses on the roles of women during

'.iffevent stages of post revolutionary China. She argues

hat since revolution is an engineered process , the status

f women is to some extent, determined by those who have the

ower to direct this process

.

Morris examines the history , structure and operation

f the National Action Committee on the Status of Women (NAC)

lather than simply endorsing the popular view that the

organization is elitist and ineffective , the author stresses

he structural constraints which prevent it from realising it<

pal cf representing aZZ women.

In the last contribution , Bourgeault deals v'.z'-. the

~ub$ugation of native (Indian and Half-breed) women in

lorthern and western Canada since the development of the

K-r trade. Be emphasizes that an analysis of the rac' . ,

colonial and class divisions which came into being is crucial
por our understanding of the process of oppression cf native

oomen

.



BIOGRAPHICAL SKETCHES

Hugh Armstrong obtained his B.A. (political science and

history) and his M.A. (sociology ) from Carleton University

.

His Ph.D. thesis was recently 'deposited' at the Universite

de Montreal . He now teaches sociology and humanities at

Vanier College , Montreal. Most of his research is conducted

with Pat Armstrong on women and work. Their book, A Working

Majority : What Women 'lust Do for Pay , was published earlier

this year and they are at present revising The Double Ghetto

Canadian Women and Their Segregated Work , first published

in 1978. He is a manuscript editor for Studies in Political

Economy .

Pat Armstrong obtained her B.A. (sociology) from the

University of Toronto and her M.A. (Canadian studies) from

Carleton University . Her M.A. thesis formed the basis of

The Double Ghetto . She now teaches sociology and women's

studies at Vanier College, Montreal and is a Ph.D. candidate

in sociology at Carleton University . Her thesis is on women

and the economic crisis. She has been on the executive of

the Canadian Sociology and Anthropology Association and is

now on the editorial boards of Perception and of Australia/

Canada Studies .

Ron Bourgeault currently works for the Association of Metis

and Nor:- status Indians of Saskatchewan (AMNSIS) and the

Gabriel Dumont Institute of Native Studies and Applied

Research, Regina, Saskatchewan. His current research deals

with the modern history of the Indian and Metis of western

and northern Canada.



Roberta Hamilton received her B.A. (sociology ) from Carleton

University in 1963 and her M.A. from Concordia University in

1975. She is now completing a doctoral thesis in humanities

at Concordia University and has just been appointed assistant

professor in the Faculty of Arts and Sciences at Queen's

University . She is the author of The Liberation of Women:

a Study in Patriarchy and Capitalism .

Cerise Morris worked for several years as an activist in the

women's movement in Montreal , co-ordinating a women's

centre and focussing on women ' s issues in social work

practice . Her graduate work was done at McGill University

where she obtained a MSW in 1971 and a Ph.D. (sociology ) in

1982. Her research interest is in how the demands and issues

of feminism get translated into policy.

Eileen Saunders is presently employed as an assistant

professor in the School of Journalism, at Carleton University ,

She obtained her M.A. at Queen's and recently completed her

Ph.D. (in sociology) at Carleton University. Her doctoral

dissertation was entitled Feminist Theory and Socialist

Revolution.



Carleton University

Ottawa, Ontario

offers programs of study leading to the Master's
degree in

• Anthropology
• Geography
• International Affairs

• Public Administration
• Social Work
• Soviet and East European Studies
• Canadian Studies
• Management Studies

and to the Master's and Ph.D. degrees in

• Economics
• Political Science
• Psychology
• Sociology

Generous support is available to qualified candi-

dates. Further information may be obtained
from the Graduate Studies and Research Office,

15th Floor, Arts Tower, Carleton University,

Ottawa, Ontario K1S 5B6.



A Critical Review Vol. 6, 1983

Contents

Pat and Hugh Armstrong 1

Beyond Numbers: Problems with
Quantitative Data

Roberta Hamilton 41
The Collusion with Patriarchy:
a Psychoanalytic Account

Eileen Saunders 61

Socialist Feminist Theory: The Issue
of Revolution

Cerise Morris 87

Pressuring the Canadian State for Women's
Rights: the Role of the National Action Committee
on the Status of Women

Ron Bourgeault 109

The Development of Capitalism and the Subjugation
of Native Women in Northern Canada



Graduate Studies in Sociology and Anthropology
Carleton University

The Department of Sociology and Anthropology
offers the following programs of advanced study
and research:

M.A. and Ph.D. degrees in Sociology, with
specialization in

Comparative Social Organization

Social Demography - Ecology
Theory - Methodology

M.A. degree in Anthropology, with specialization in

North American Studies
Theory - Methodology

Information regarding admission requirements

and financial support may be obtained from:

Coordinator of Graduate Studies
Department of Sociology and Anthropology

Carleton University

Ottawa, Canada K1S 5B6



Beyond Numbers: Problems with
Quantitative Data

Pat and Hugh Armstrong

Using Statistics Canada as an example , the authors argue that

statistical data collection must be sex-conscious where it is

now sex-blind. It must take the different work, location

,

bodies and experiences of women and men into account

.

The theoretical assumptions behind the numbers must be examined

from the perspective of women; the claim to truth and the

monopoly of fact challenged. Supported by examples from

interviews conducted by them for other research , the authors

go on to argue that qualitative techniques can expose the

holes in, as well as complement and check, the data collected

by more quantitative means.



Introduction

As C. Wright Mills pointed out many years ago, "Social

research of any kind is advanced by ideas; it is only disciplined by

fact" (1959:71). Ideas, particularly the systematic organization of

ideas into theory, are, however, too often forgotten in social

research. Facts are frequently identified with numbers and numbers

with neutrality or, even, truth. This is especially the case when

the numbers are collected by a state agency such as Statistics

Canada.

In this paper, we argue that numbers, and the methodology

used to gather them, cannot and should not, be separated from their

social context, nor from the ideas that direct their selection,

toreover, numbers cannot be viewed as the sole, nor even necessarily

the most accurate and relevant kind of fact. The inadequacy of

numbers for a full understanding of the position of women is

camouflaged by reference to "scientific" methodology, by the

adoption of non-sexist language, by the use of similar questions for

both sexes and by the tabulation, according to sex, of most data.

But for the most part, women and men do different work, in different

places, for different pay. Given this segregation, as well as the

different reproductive capacities of the sexes, many questions are

women's questions, many answers are women's answers, many categories

are women's categories, many omissions primarily affect women.

From this perspective, we begin with a critical examination

of Statistics Canada, the chief source of the numbers so often taken

as the facts. Explicit and, more often, implicit assumptions are

built into the collection and tabulation of these data, assumptions

which may serve to conceal or to distort important aspects of

- 2 -



women's work. Moreover, some information which is crucial for

understanding the nature and conditions of women's work is

inaccessible to, or lost from, quantitative measures, however

carefully structured the questions and explicit the ideas. Building

on this critique of Statistics Canada in particular and quantitative

approaches in general, we go on to suggest how qualitative

techniques can expand and correct our view of women's work,

complementing and acting as a check upon information gathered by

quantitative means. In conclusion, we recommend, first, that more

funds be allocated to theory development, that is, to the ideas

which advance social science research. Secondly, we recommend that

a thorough examination of the assumptions, methodology and

presentation of data by Statistics Canada be initiated from the

perspective of women, with a recognition of the sex-segregated

nature of women's work. Finally, financial support should be

directed towards research which goes beyond surveys and the limited

data available from Statistics Canada to connect these with the

qualitative aspects of women's work, to the nature and conditions of

their jobs in and out of the labour force. In what follows,

examples from Statistics Canada are used only to illustrate the

argument. A comprehensive and systematic examination of this

important agency's approach still needs to be done.

Assumptions and Concepts

According to Mills (1959:67), all empirical research involves

"a choice as to what is most real." But the basis of these choices,

the theoretical assumptions inherent in the collection and

tabulation of data, are often obscure, sometimes even to the

- 3 -



statisticians themselves. As John Irvine, Ian Miles and Jeff Evans

(1979:5) explain in their introduction to Demystifying Social

Statistics :

It is not enough, however, merely to make the case that
statistical data cannot be interpreted without reference to
their underlying theoretical assumptions. All data, whether
produced in the course of academic research or by state
bureaucracies, are structured by the conceptual framework that
is applied as well as by the technical instruments used in their
production. It is the precise nature of these practical and
theoretical commitments that needs investigating for each set of
data.

While the technical instruments are usually clearly described in

Statistics Canada material, the theoretical commitments are seldom

discussed. Few publications explore the assumptions behind the research

or the specific selection of questions, responses and techniques of

analysis. [1] This alone suggests a perspective which considers its

theoretical approaches unimportant or absent, its methods scientific,

and its numbers objective reflections of reality. As a result, it is

not always a simple task to determine what is being measured and why,

let alone what is being ignored and for what reasons.

Explicit statements outlining theoretical approaches are rare but

an examination of concepts and how they are operational ised does reveal

some of the assumptions behind the data. For women, probably the most

obvious example is the definition of work:

It includes working for wages, salary, tips or commission,
working in their own business, farm or professional practice, or
working without pay in a family business or farm owned or
operated by a relative in the same household. "Work" excludes
housework or other work around the house, and volunteer work
(Statistics Canada, 1982a :17).

What Statistics Canada is interested in counting is not all the ways

people spend their days or their energy, not how they survive on a daily

basis, not how food gets to their tables, nor even how they improve

- 4 -



their standard of living. Ch the other hand, interest is not limited to

paid employment, to work that brings individuals money income. Rather,

the focus of concern is work which results directly in profit and

exchange. As the Guide to Labour Force Survey Data explains:

The concepts of employment and unemployment are derived from the
concept of the supply of labour as a factor of production
measured over a short interval of time. The production referred
to is in turn defined as those goods and services included in
the National Accounts (Statistics Canada, 1979:15).

The interests served are not those of women, since the work most women

do at home for their families, as well as the important volunteer work

they do in hospitals, schools, playgrounds and a wide variety of other

institutions disappears.

This is not to suggest that counting housework or volunteer work

would change the nature of that work. That they are not considered

reflects the perspective of Statistics Canada. The omission indicates

that there are assumptions behind the data, assumptions which produce a

picture that excludes much of the work most women do and dramatically

underestimates the labour involved in daily survival.

Less obvious than the assumptions about the kind of work that

counts are the assumptions related to the structure of families. Mien

collecting data on households, Statistics Canada seems to assume that

each household has one person primarily responsible for financial

support, that this person is normally male if there is a man in the

house and that women by themselves have the babies. Here too, the

concern seems to be with the connection of the household to the economy

outside the home rather than with what happens within this domain.

Largely in response to feminist protests, Statistics Canada has

dropped the term "Head of the Household" and replaced it with "Person

1". For the 1981 Census, respondents were instructed to:

- 5 -



Choose one of the following as Person 1:
- either tre husband or wife in any married couple living here
- either pertner in a common-law relationship
- the parent, where one parent only lives with his or her

never-Ferried son(s) or daughter(s) of any age
If none of the above applies, choose any adult member of this

household.

But in spite of the introduction of non-sexist terminology, little

change in the data was anticipated. Indeed, alterations in other

sections of the Census helped ensure that this would be the case.

According to the 1981 Census Dictionary :

In 1981, the criterion for determining family type was changed.
A new question was added to the census questionnaire to
determine a person responsible for paying the rent, or mortgage,
or taxes, or electricity, and is used to identify primary and
secondary families. In previous censuses, the primary family
was defined as the family of the head of the household. While
we anticipate that in the majority of cases the person
responsible for household payments will also be considered as
the household reference person (Person 1) , this will not always
be the case (Statistics Canada, 1982a:63).

Research in the united States (Vickery, 1979) and in Canada (Armstrong

and Armstrong, 1983) suggests that the income of employed women who live

with men pays for food, clothing and services rather than shelter and

energy. Thus, the criteria selected to determine the family maintainer

favour men, even though the actual amount allocated for mortgages, taxes

and electricity may be lower than that spent on food (Canada Department

of Finance, 1981:13). The primary family, then, becomes another way of

getting men beck as the head of the household.

Men's headship is further reinforced by the decision to record

only one family maintainer. If respondents write in that these expenses

are jointly paid, one will be selected by the statistician.

Maintenance, the Census assumes, cannot be shared. According to a

Statistics Canada spokesperson, it is not possible to obtain the

financial contributions and expenditures of each member of the household

- 6 -



because "household income is not important data." Not important for

whom? Once again, it is women's contribution which is the most likely

to disappear.

One aspect of the household, however, does seem to be recorded

for women alone. Census question number 37 asks:

For WOMEN [emphasis in the original] who are married or have
ever been married: How many children were EVER born to you?

Furthermore, according to the 1981 Census Dictionary , "Childless

families refers to those families where the wife has reported that no

children were ever born to her" (Statistics Canada, 1982a:62).

Biological reproduction does require male participation, ferriage is

not required for conception. Yet the Census records the fertility of

ever-married women only. Wnile women always know how many babies they

produce and while some men may be in doubt, the lack of male precision

hardly seems large enough to warrant their exclusion from these

parenting records. In the absence of a clear justification for this

approach, it is tempting to assume that women's childrearing

responsibilities — the social extension of biological reality — are

the focus of concern. However, given that some men retain alone the

responsibility for their children and given that some never-married

women also keep their children, the data obscure actual child care

arrangements.

The same is true of the practice of counting only never-married

children living at home.f2] Divorced and separated offscrinq may,

however, be returning to their parents' homes, seeking financial and

emotional support as well as babysitting and other family care services.

Failing to record these dependencies, like the omission of data on the

financial and work contributions of each member, hides the actual

- 7 -



economic relationship within the family. It also conceals the

additional work such dependencies may mean for women. Moreover, given

the growing unemployment and inflated housing costs facing young people,

it seems plausible that more and more young marrieds will be looking to

their parents for financial support and for accommodation. Statistics

Canada seems ready to provide little assistance in monitoring these

growing dependencies.

The conceptualisation and operational isation of work, maintainer,

and fertility illustrate the importance of assumptions to findings, of

choices to the selection of facts. The development of the problem, of

the hypothesis, takes place at least as much before the information is

collected as after Statistics Canada publishes its numbers. However, in

the absence of explicit theoretical justification for these choices, it

is possible only to speculate about the theoretical framework guiding

the data collection. Consequently, debate has focused on the

techniques, which are explicit, rather than on the hypotheses, which are

obscure. But, as E. P. Thompson (1978:231) said of historical research,

scientific investigation "consists in a dialogue between concept and

evidence." Wien one half of the dialogue is missing, it is difficult to

evaluate the results. In Canada, theoretical work has been divorced

from, and has taken second place to, empirical research. Both suffer in

isolation. The theoretical assumptions behind the data should be

brought to light for examination and the development of theory should be

strongly supported as legitimate social science work.

Collection of Data

The methods used to gather data both reflect the assumptions

behind the research and affect the information collected. Techniques



cannot be fully evaluated nor understood outside the social and economic

structure or their particular historical period. Who is interviewed

,

how they are interviewed, the information sought, the alternative

responses offered and the consequences of the answers are all

significant factors in the production of statistics. Here, too, choices

are necessary and the choices have important consequences for our view

of women's work as is shown below. The point, of course, is not to

argue that everything is relative; it is not to deny the existence of

the 'real' world in some phenomenological sense. Rather, it is to

advocate theoretically-informed and sceptical caution in the use of data

as we, quite reasonably, seek more reliable knowledge about the world.

1. Deskilling

Harry Braverman (1974) , in his perceptive analysis of work in the

twentieth century, describes how the pursuit of profit has encouraged

the separation of manual and mental labour as well as the progressive

deskilling of both white collar and blue collar jobs. In these times of

government restraint, a similar process seems to be taking place in the

collection of statistics. Many of the tasks are simplified to allow

their completion by untrained persons. While the Labour Force Survey

continues to employ trained enumerators, the Census now depends heavily

upon self-enumeration. However, as Statistics Canada (1980a:14) points

out in the Social Concepts Directory , "The questions are usually phrased

to be suitable for self-enumeration for a reasonably well-educated

respondent." The assumption of "reasonably well-educated" respondents

can be made neither about the population covered by the Census nor about

those included in the more comprehensive questions of the 1/5 sample.

Not only does this procedure encourage the simplification, and perhaps

- 9 -



even the omission of important questions, it also means that errors in

the interpretation and responses to questions are much more likely.

Those many women who have not had the opportunity to learn to read and

write one of our major languages cannot be sure that they are counted in

(and many of those viio have, in previous censuses, earned money

administering the questionnaire are counted out of those paid jobs) .

2. Restraining Legislation

Even if the respondents clearly comprehend the questions, they

may be unwilling to expose their position to government scrutiny, in

spite of assurances about confidentiality. According to the women we

interviewed, a range of government programmes and regulations discourage

them from reporting their employment and earnings (Armstrong and

Armstrong, 1983). For example, veterans' allowances, disability

pensions and welfare payments set a low maximum on family income.

Reporting employment and pay often means, then, a reduction in much

needed benefits. Low maximum earnings set for day care subsidies have a

similar effect. For those relatively low-income families, the effective

tax rate on paid employment may well be 100 per cent. And the income

tax regulation Wiich sets a low maximum on the amount a spouse may earn

and still be claimed as a deduction also frightens some away from

declaring their employment to any government representative conducting a

survey. Employers may discourage their employees from reporting income,

wishing to evade regulations and contributions to government programmes.

Rules related to contributory plans such as unemployment insurance and

Canada/Quebec Ftension also encourage women to hide their employment,

sacrificing their future and current protection in order to take home a

little more money at the end of the week. Many of these people are

- 10 -



unwilling to take the chance on promises of secrecy. That such factors

can have a significant effect on data collection has not been entirely

ignored by Statistics Canada. The 1981 Census Dictionary (Statistics

Canada, 1982a :13) notes that tax changes allowing for the deduction of

spouse's wages as expenses "may have changed the status of some

self-employed persons from 'without paid help' to 'with paid help' if

they decided to pay wages to their spouses." Few of those disappearing

from and appearing in the unpaid spouse category are male.

That this hidden economy, this invisible workforce, is composed

primarily of women is, of course, difficult to prove. Our research

suggests, however, that many of them are women who do hairdressing or

typing at home, look after children in their ovn or other people's

houses, scrub floors and do errands for pay, sew baby clothes and stick

labels on bottles, or sell their sexual services on the street. It

seems likely, then, that labour force data fail to capture many of the

women vino work for pay, especially those who are employed in the lowest

paid and least protected work.

Employers as well as individuals may be influenced in their

responses by government regulations. For example, the Statistics Canada

monthly publication Employment , Earnings and Hours which provides the

current information in this area, is based on employers' answers to

survey questions. But laws related to minimum wages and equal pay, to

hours and overtime, to piece rates and benefits may discourage some

employers from accurately filling in the blanks. It should also be

noted that the survey includes only those establishments with twenty or

more employees. Since many women are employed in smaller workplaces,

this information may not only count some women inaccurately but also

exclude entirely many of those in the lowest paid jobs.

- II -



3. Sampling

Financial and geographical considerations frequently make

sampling techniques necessary. However, important data can be lost

through the choice of sample over comprehensive methods, or through the

choice of a small sample size. When many people are in a category, they

are accurately picked up by sampling techniques but the method is much

less precise when few people fit into the group. With women highly

concentrated in some occupations and industries and virtually absent

from others, the use of samples in collecting labour force data in both

the monthly surveys and the 1981 Census means that some women will not

show up in non-traditional occupations and industries, because the

estimates are unreliable. Trie female electricians in Prince Edward

Island, for example, are extremely unlikely to become visible, although

the female professors in Ontario have a better chance.

4. Machine Processing

When large numbers of responses must be processed quickly, the

tendency is to rely on those questions and responses which can be read

by machine and easily categorised into neat tables. As C. Wright Mills

(1959:57) pointed out, "The kinds of problems that will be taken up and

the way in which they are formulated are quite severely limited by The

Scientific Method. Methodology, in short, seems to determine the

problems." It is difficult, of course, to tell what is left out as a

result of being difficult to measure numerically, or to determine the

extent to which questions are structured by methodological as against

theoretical considerations. But the two kinds of consideration are

clearly linked, and the consequence is that we miss out on some

important issues. For example, the absence of questions on the labour

- 12 -



process is perhaps a result of both theoretical and technical

considerations. The absence of statistics on the nature and conditions

of work influences our understanding of the employment situation of

women since it is on the shop floor, and in the bank, the restaurant,

the hospital and the grocery store that sex segregation and inequality

become reality.

5. The Wbrding of Questions

Statistics Canada argues in its Social Concepts Directory , that

"The key consideration to be borne foremost in mind, when constructing

questions, is the viewpoint of the respondent with the needs of the

survey sponsor in the background" (1980a:15). While the viewpoint of

the respondent is certainly important, sponsor needs are seldom

explicitly revealed to those using the data. Moreover, it is not simply

the viewpoint of the respondent which should be thought about but also

the structural constraints influencing the answers. In developing the

wording of questions, more than their clarity, brevity and simplicity

should be considered; sex segregation of work should also be taken into

account

.

The Census asked, for example: "Last week, how many hours did you

work (not including housework or other work around the house)?" CXir

research suggests that such questions are answered in a variety of ways,

depending upon the nature and conditions of the work. Women who work

part-time frequently have short daily hours of steady work, with few if

any paid breaks. The hours they report working are fully spent at the

desk, behind the counter or on the line. With few, if any, paid

holidays, the days per week they report may also not be comparable to

those with full-time jobs. Moreover, those who work in non-unionised

- 13 -



establishments often work without breaks, have short lunch periods and

stay late to finish up. Yet they calculate their hours from when they

are required to be at work until the formal quitting time. Not only are

they not paid overtime, but they do not view these extra minutes as time

on the job. Finally, those who work in homes as domestics, babysitters,

cleaners and piece workers normally have very long, but inaccurately

recorded, hours.

The point, of course, is that these part-time, non-unionised

and/or home workers are disproportionately women. The recorded sex

differences in hours worked are only in part the result of actual

differences. They also reflect the way in which the question is asked

and answered. To ignore the sex segregation of the labour force is to

produce faulty data on issues such as hours worked.

6. Excluded Questions

It is not only the questions that are asked but the ones that are

not asked which should be considered in light of women's structural

segregation in and out of the market. For instance, the monthly survey

of Employment, Earnings and Hours requests information on the total

number of female employees as well as the proportion that are wage and

salary earners. Employers are not questioned, however, on how much they

pay these female wage and salary workers. Nor are they asked about

turnover rates, about which people leave which jobs after what period of

time for what reasons. Once again, it is women who are most concerned

about revealing wage differentials. It is women who are accused (in the

absence of evidence) of having high turnover rates resulting from their

low level of commitment to their paid jobs.

fofriile people are asked about their formal education, they are not

- 14 -



asked about their skills. In assuming competence can be measured by

formal certification, Statistics Canada fails to measure the actual

skills people employ in their jobs. Women's skills, in particular,

disappear because as one woman explained to us about the requirements of

her job, "It's what every woman knows how to do." Furthermore, the

absence of such a question means that there is no real assessment of the

resources available for coping with the new micro-electronic technology.

Because these skills are new and constantly changing, many are learned

on the job and cannot, therefore, be measured by questions on formal

education. And many more women are learning how to use this equipment

on the job than are acquiring these skills in credit courses.

Although fringe benefits are an increasingly important part of

workers' incomes, the sex differences in this field are virtually

ignored by Statistics Canada. Various factors — such as the
ineligibility of some part-time, salaried and young workers, not to

mention the wage and salary basis of most benefits — may mean that
female employees receive fewer rewards in this area as well. Yet the

monthly survey reported in Employment , Earnings and Hours does not cover

fringe benefits. And between 1967 and 1975 several surveys of labour

costs in various industries were conducted and then jointly published by

Statistics Canada and the Department of Labour, but with no data on the

benefits by sex. Finally, only in Saskatchewan are workers'

compensation records published by sex of claimants, and these figures

indicate significantly lower claims for women (Statistics Canada,

1980b:27). In part because Statistics Canada produces so little

information on occupational health and safety, it is difficult to learn

just why this is the case.

The failure to ask a question about relatives requiring special

- IS -



care has a similar effect. Information is collected on the number and

ages of children but not on aging grandparents or disabled younger

people who make up part of the "personal or family responsibi] ities"

fulfilled by women as part of the invisible work that reduces their

labour force participation. In general, by not asking these questions,

and by not collecting information by sex, significant aspects of

women's, and men's, work are hidden from view.

7. Selecting Alternative Responses

The data are not only conditioned by the questions asked, or not

asked, but also by the choices of response that are offered. In the

Census question on major source of income, for instance, one choice was

"working for wages, salaries, tips or commissions." That Statistics

Canada asks employers answering their monthly survey to distinguish

between wage and salary earners clearly indicates that these differences

are seen to be significant, yet they disappear from the Census, from

what is supposed to be the more complete source of information. Workers

may well be concentrated by sex into each of these categories, but this

is impossible to tell, given the choices offered.

Perhaps the problem is more obvious in another possible response

to this question. The "other income from government sources" choice

includes, "alimony, child support, periodic support from persons not in

the household, net income from roomers and boarders, non- investment

income such as retirement income from abroad, scholarships, etc."

(Statistics Canada, 1982a:24). By lumping together these diverse

sources, Statistics Canada does not facilitate the analysis of income

sources by sex and occupation.

In not offering technological change as a reason for leaving last

- 16 -



job, the Labour Force Survey is missing the opportunity to monitor the

effect of micro-technology on employment. Since the research to date

(eg. Menzies, 1981) suggests that women will be hit first and most,

omitting this alternative conceals important factors influencing female

labour force work.

8. Denying History

If the purpose is to understand what is happening to women today

and what will happen to them in the future, it is necessary to examine

the changes that have been taking place, to take an historical

perspective that exposes the factors influencing these changes. But

Statistics Canada makes such analysis difficult by failing to ask

historical questions and by changing questions or categories.

Few historical questions can be found in the Census. In the

questionnaire distributed to the 1/5 sample, those designated "Person 1"

are asked how long they have lived in their current dwelling. Everyone

is asked where they lived five years ago. Ever-married women are asked

how many children were ever born to them. Labour force information is

collected only for the last year or at most for the last year and a

half, and job search for the last four weeks. For personal history,

then, the only information available from the Census is ever-married

women's life-long fertility, most people's residence change from five

years ago, and some people's recent labour force experience.

Nor does the monthly Labour Force Survey collect much historical

data, although some occasional publications do look at labour force

activity over time. For example, in 1976 Statistics Canada published

Earnings and Work Histories of the 1972 Canadian Labour Force . Based on

the 1973 Survey of Consumer Finances, the study reported data on male

- 17 -



and female earnings differentials and work history between 1967 and
1972. Respondents to the sample were asked what their main activity was

while not working for periods of one year or more, with "kept house,

raised children" as one possible answer. They were not asked why they

left their last job. The analysis of the data suggests that the higher

turnover and lower wages reported for women were related to their

decision to stay home with the house and children, to their choice of

dropping out of the labour force to do their unpaid women's work. Yet,

since most women continue to do the housework and child care work

whether or not they have paid employment, it seems unlikely that many

women without a paid job would not be mainly keeping house and raising

children. With the rare historical investigations based on such

questions and alternative responses, it is difficult to find out what is

happening to women over time and what factors influence these patterns.

Little, if any, information is available on the historical impact

of new technologies, particularly of the micro-electronic technology now

so important to women's work. Nor is it possible, from Statistics

Canada data, to examine the effect over time of different kinds of work

on the health of men and women. It is also extremely difficult to

relate alterations in the resources and structure of families to labour

force changes. If the data followed some families, recording the hours,

duration, type and wages of work for both spouses and for children, as

well as their health situation, their skills and their reasons for

entering or leaving specific jobs, it might be possible to understand

the factors structuring people of different ages and sexes into and out

of labour force jobs. Without such information, any explanation is

necessarily partial and superficial, focused on individuals outside

their historical, familial and even much of their economic context.

- 18 -



Changes in the collection and categorization of data also make

historical investigation difficult. Of course, the alterations may

reflect primarily changes that are taking place in the labour force.

This was the case, for example, in the separation of file clerks and

receptionists from the secretarial category and in the new definition of

part-time work (a reduction in weekly hours from less than 35 to less

than 30). In other cases, the changes may result in part from new

government policies and programmes. For farm workers, (Statistics

Canada, 1982a: 12) :

The data between 1971 and 1981 may not be strictly comparable
because of small changes in definitions. Females who were

unpaid family workers, worked as farm labourers and did less

than 20 hours of unpaid work a week were excluded from the
labour force according to the 1971 definitions. These persons

are now included in the employed labour force. Cue to changes

in the tax laws and Census procedures, some persons formerly
identified as unpaid family workers may now be classified as
paid workers. The tax changes permitted for the first time, in
the 1980 tax year, the deduction of a spouse's wages as
expenses. This may have resulted in some changes in the status
from unpaid family workers to paid workers. As well, Census
editing of this category was more stringent in 1981.

Some alterations, such as the switch from "Head of household" to "Person

1," may be related to pressure from feminist groups. But some changes

seem unrelated to any of these concerns. For example, the Census

Dictionary reports that:

In all the 1981 tables on "Occupation Based on the 1980
Classification", the 1971 "Not stated" category will no longer

appear. This category has been redistributed by imputation into
new "imputed groups" within each minor group (Statistics Canada,
1982a: 37).

Since clerical employees, part-time workers, unpaid employed spouses,

workers with unstated occupations and people not heading households are

usually women, these new definitions and categories primarily affect the

analysis of women's work. Whatever the reasons for the changes, the

need for historically comparable data should be a central concern. In

- 19 -



explaining the objectives of the Labour Force Survey Revision Project,
Statistics Canada (1977:19) indicated an awareness of the problem,

noting a

...demand for new information on the dynamic aspects of the
labour force to supplement the monthly "snapshots" the new
survey provides, much more information is required on the gross
movements between various labour force categories each month,
along with somewhat longer term longitudinal data covering the
six month period over which the respondents are surveyed.

The problem is, however, far from solved.

9. Comparing Data Sources

The problem of comparable data is not confined to history. Many

government agencies and departments collect statistics but the

definitions, categories, responses, techniques, time periods and

tabulations are often different, making comparisons difficult. Even

within Statistics Canada, these differences sometimes exist. Miile

Statistics Canada is working to eliminate these discrepancies, some do

remain, obscuring the actual situation of women at work. For example:

For the unemployed and the not in the labour force groups, data
from the Labour Force Survey refer to the last job held by
persons who have worked in the last five years. Since no
question is asked on the most important activities or duties in
the job held, as in the Census, the Labour Force Survey may
classify more persons to the "Managerial" group on the basis of
job titles than would be the case with the Census (Statistics
Canada, 1982a: 37) .

Itfiile the numbers are probably not large, the discrepancy makes

comparison difficult. And once again, it is most likely women who are

the extra persons classified by the Labour Force Survey into the

managerial group, since they are the ones most likely to have jobs that

are labelled managerial but whose actual tasks would place them lower on

the hierarchy of Census occupational categories.

- 20 -



Tabulation of Data

Not only the collection but also the tabulation of data can

influence the view of women at work. As Statistics Canada points out in

its Guide to Labour Force Survey Data (1979:13) "virtually any

combination of questionnaire values can be cross-tabulated on request,

subject to a 'reliability' criterion which is related strictly to the

sampling variability and size of the estimate." However, such extra

tabulations often cost money, limiting their access to those with

research funds. And of course, the information available for tabulation

is restricted by the problems with data collection cited above.

Furthermore, the tabulations that are published, since they are the most

accessible, are the most widely used, especially by the media. Thus,

although most information is now collected by sex and could be tabulated

for men and women, the published compilations form the basis for most of

the public commentary on women as well as much of the research. These

tabulations therefore deserve particular attention for the view of women

they present.

Tabulations may both reflect and confirm assumptions about women.

The proposed tables on census families to be drawn from the 1981 Census

are a case in point. According to Statistics Canada's Products and

Services of the 1981 Census of Canada (1982b:32-36) , tables will be

published on husband-wife families in private households by the wife's

age and highest level of schooling. These are the only listed tables on

husband-wife families in private households which use the wife's

characteristics as the basis for compilation. Data will be tabulated

for these households by both the husband's and wife's highest level of

schooling, labour force activity and age. The rest of the tables for

these households will be compiled primarily on the basis of the

husband/pa rent's characteristics. In particular, this is the case for

- 21 -



the income data which will be tabulated by husband/parent's age, highest

level of schooling, work activity and period of immigration. That

income tables are not based on the wife's characteristics indicates a

particular view of women's place as well as reinforcing the perspective

which sees the economic contribution of wives as secondary. Perhaps

their childbearing and childrearing responsibilities account for the

concern about their age and schooling.

The categories used in tables also influence the perception of

women's work. The Labour Force Survey's "managerial, professional,

etc." category, for example, lumps together laboratory technicians and

neurosurgeons, library technicians and full professors, managers of

small shops and senior executives from General fotors.r3] Trie wide

differences in power and pay, as well as the sex segregation, disappear.

Indeed, the growth in the proportion of women in this category has

helped to create the false impression that there has been a significant

improvement in the position of women.

In publishing tables by occupation and industry separately, the

Labour Force Survey data also creates the impression that women have

been moving into prestigious management jobs and into traditional male

fields such as construction work. Cross-tabulations of industry and

occupation data indicate, however, that, in 1980, three-quarters (74.8%)

of female managers worked in the service industries while less than half

(42.8%) of the male managers were employed there. In the construction

industry, less than ten per cent (9.3%) of the women actually did

construction work; over seventy per cent of the women employed in that

industry held clerical jobs (Armstrong and Armstrong, 1983: Tables 3

and 10). Wien such cross-tabulations are unpublished, segregation is

camouflaged. Furthermore, change may be suggested where little has, in

- 22 -



fact, taken place.

Theory and Methodology in Quantitative Research

Counting women in means much more than including a question on

sex and tabulating the results separately for women and men. It

requires not only an awareness that people come in two sexes but, also,

a recognition that the experiences of women and men are segregated in

many ways, that policies, programmes, concepts and questions have a

differential impact on women and men, that some questions are women's

questions while others ignore or denigrate their contribution. The

intention here is not to attack Statistics Canada. Indeed, it should be

noted that Statistics Canada has been willing to alter some concepts to

bring them more in line with the actual contributions of women and men.

Note, for example, its abandonment of the Census policy permitting

respondent's paternal ancestry only to be recorded. Nor is the

intention to draw up a shopping list of changes required in established

techniques. The examples provided here are designed merely to

illustrate problems with its methodology and thus to indicate that

official data collection and the funding of research need to be

re-examined if women's work is to be counted in.

But to begin such a re-examination, it is necessary to start with

theory, with the systematic organization of ideas, both because theory

is the major way of advancing our understanding and because methodology

is inseparable from the perspective which guides the choice of what is

real. The assumptions behind the data should be drawn out for critical

appraisal. However concretely the questions surrounding the collection

and tabulation of statistical information are posed, the methodological

answers must be theoretically informed if they are to be useful. In

- 23 -



order for statistical information to be improved, theoretical work, as

well as theoretically informed methodological work, must be undertaken.

This in turn entails that theoretical work be financially supported. In

particular, problems of deskilling, restraining legislation, sampling,

machine readability, the wording and avoidance of specific questions,

the selection of possible response choices, the denial of history, the

comparability of data sources, and the choices about what to tabulate

and what tabulations to publish need to be examined from a theoretical

perspective which takes the sexual division of labour into account. And

that theoretical perspective itself needs to be refined.

No matter how finely tuned and sex conscious the theory and

techniques are, however, some information will continue to evade

quantitative approaches, whether based on sample or universal surveys.

Some of the information collected through such methods, no matter how

carefully the questionnaires are structured and the data collected and

tabulated, may create an inaccurate view of women's work. At best, such

techniques can produce only a partial picture of the daily working lives

of Canadian women. Quantitative data miss, and may misrepresent, many

aspects of women's experiences in and out of the labour force. In the

final section of this paper, we draw on material and lessons learned

from lengthy interviews conducted with women in working class jobs to

argue that qualitative data are at least as important as quantitative

data in developing an accurate view of women's work.

Qualitative Research

From Horace Miner's 1939 study of a French Canadian parish to V>eg

Luxtor ' s 1980 examination of women's work in the home, qualitative

research has provided valuable insights into women's daily experiences.

- 24 -



Such studies are rare, however, in part because they are expensive and

time-consuming, but mainly because they are considered somehow less

legitimate than statistical analysis. As a recent letter to the public

from the 1981 Census Manager makes clear, Statistics Canada views itself

as producing "facts," profiles which are "useful, convenient, complete."

And many social scientists would agree that these facts, these numbers,

are more reliable and accurate than data collected from lengthy hours

spent and conversations with women in Flin Flon. But the bias in favour

of 'hard' data is seldom based on a rigorous comparison of the findings

of quantitative methods with those of qualitative research, or of their

value in explaining or predicting women's actions. Nor is the bias

often based on respondents' assessnent of the accuracy of the facts, on

the ability of these facts, in Lillian Rubin's (1976:13) phrase, "to

generate an 'aha' experience." Nor is the bias gender-neutral , as

Pauline Bart has so eloquently pointed out:

We speak of hard data as being better than soft data, hard
science better than soft science, hard money better than soft
money. In the fifties, one was criticised for being "soft on
communism." This is of course a male sexual metaphor, so since
discovering this, I have substituted a metaphor based on female
sexual experience and refer to wet and dry data (quoted in
Roberts, 1981:22)

.

The bias is not, however, simply a matter of metaphor. It reflects the

priority given to the collection of some kinds of data, those which are

easily counted, frequently repeated and amenable to statistical

analysis. But, as feminists have long made clear, the personal is

political. Daily work and family relationships have a profound impact

on work activity; physical and social conditions influence ideas and

actions, lb leave them out — and many are difficult or impossible to
include in sample survey and statistical analysis — is to create not
only huge gaps in the data but, often, inaccurate information as well.

- 25 -



Our experience in interviewing women about the nature and conditions of

their work suggests that gualitative research provides a necessary

complement to quantitative analysis because it can test the reliability

of quantitative data, can collect information not accessible to survey

techniques and can indicate areas that should be included in

quantitative analysis. T4]

1. Checking the Facts

Keeping questions clear, simple and direct does not necessarily

guarantee the accuracy of responses. Because the approaches used in

gathering qualitative information take many aspects of the individual's

life and relationships into account, and because the less rigid and more

informal style of data collection often encourages a more open exchange,

these qualitative "facts" can frequently be used to test the results of

quantitative research. Interview procedures which are flexible,

unstructured or partially structured, and wide ranging, permitting

people to offer their own responses and views, frequently draw out quite

different information than that collected through highly structured

,

official sounding questionnaires which allow only machine-readable

choices.

In our experience, direct questions may elicit responses which

are often contradicted by information gathered through more probing and

indirect techniques. For example, when we asked women if they knew of

any health dangers in their jobs, some simply said no. Yet, when

describing in detail an average day on the job, various work related

problems emerged. A clerical worker who uses visual display terminals

(VDT's) in a trust company responded negatively to the health question

but earlier had told us that during the day, "Sometimes I'll go dovn to

- 26 -



the lounge and fall asleep, my eyes are so sore." A factory worker who

offered a similar response also told us that, with her right foot

operating the sewing machine, her right hand adding cloth, and her left

hand guiding the material as quickly as possible all day to ensure a

survival wage from her piece work job, her arms, feet and legs hurt:

"Sometimes when I come home, I can't make more job in my house."

Various explanations were offered for this discrepancy. They expect

work to hurt, so sore arms, legs, heads and hands may not be defined as

health hazards but simply as part of the job. Many of the dangers women

face are invisible and have a slow, cumulative effect. Thus, they are

less likely to be associated with the job than are the broken toes or

burnt hands that men more often suffer. And a few do not want to think

about the dangers, because they have to go back there to work tomorrow.

Wiatever the reasons for these seemingly contradictory responses, they

indicate that simple, direct questions will not always reveal the simple

truth.

Like the questions on health and safety, those on sexual

harassment also suggest that direct questions provide only limited, or

inaccurate, pictures of the situation. V^en we asked women if they had

ever been sexually harassed, some had difficulty answering the question

— even when examples were given — often, because they had come to
accept this behaviour as part of the job. Some defined such treatment

as a problem only if they could not resist. One woman offered a very

definite no to the question . When reminded, however, that she had said,

in describing her work history, "That was something else — if you could
keep out of the boss's hands," she explained that she had made it clear

to "knock it off." Therefore, she felt "you don't have to write a book

about it." Some said it was unlikely, given they worked with women only



or had a lone male boss.

What this evidence suggests is that, if a question on sexual

harassment were asked of all employed women, there would probably be an

overwhelmingly negative response because of the way women perceive and

respond to advances and because many women are segregated into jobs

where there are few men to do the harassing. I^oreover, if the only

answers recorded were those given for this question — as is the case
with most machine-readable questionnaires — then much of the
counter-evidence would disappear.

There are also often quite valid reasons for women to hide

information from official government statisticians or from written

questionnaires they fill out for the government . When we asked one

woman for her age, she requested that the tape recorder be turned off so

she could tell us her real age. Her boss, she explained, thought her

below retirement age. She could not live without her employment income

and thus she could not risk exposure. This woman not only had our

assurances that we would not link her age to her name, job or employer,

she also lived in a large metropolitan area where any identification was

extremely unlikely (as was the possibility that her employer would read

a book on work by a couple of sociologists writing for the Canadian

Advisory Council on the Status of Women) . However, she was not alone.

Barely making ends meet, one woman collects a small unemployment

insurance cheque, takes in foster children, babysits on weekends and

does odd clerical jobs while the children are in school. Although

willing to exchange this information over coffee in the kitchen, she

made it very clear that, since her survival was at stake, there was no

way she would write that information down on a questionnaire. Thus,

even on such seemingly straightforward questions as age and employment,

- 28 -



the facts may not be accurate. The qualitative date can help us become

more accurate and become more sceptical about the supposedly hard data.

In other cases, the answers given may be literally correct but

hide a significantly different reality. One waitress, for instance, is

paid the required minimum wage, but her employer then forces her to sign

the cheque back to him. The taxes are paid, the law is obeyed to the

letter; however, her real income is low and irregular. A similar

problem is evident in data on segregation and wage differentials. As

one clerical worker explained, the difference in wages can be justified

on the basis of different job categories but conceals both segregation

and unequal pay:

("There used to be] rnale accounting clerks. They promoted all
those male accounting clerks to management and left the women in
accounting, and they hired nothing but women in accounting
since. So they're hiring women to work for lower jobs than the
men and the excuse is that [the men] have a family. Women don't
have families? Plus, when there were men in accounting they
always got paid higher than the women and the excuse they use is
because they're men. Women don't live any more cheaply than
men... It was a pretty sick excuse but to fight it wasn't worth
it. ..The manager was called in and Twe] bluntly asked , "Why
are the men getting paid more than us?" We're not saying we're
doing any more work but we are equal in work, we should be equal
in pay. We know as much, maybe even sometimes more. He said,
"If the men are married and if they're not, men always get paid
higher." I couldn't believe it! At least find a decent excuse.
Don't use that one on me.

Whatever the excuse, such inequality is unlikely to appear in the survey

data. Rather, evidence for equality and for conformity to the law is

more likely to show up.

These examples are designed to illustrate how qualitative data

can be used to check quantitative data. But this methodology can do

more than suggest where the figures do not measure what they are

supposed to measure. This approach can also indicate where there are

holes in the data, where information could be, but is not, collected in

- 29 -



quantitative research.

2. Exposing the Holes

Whole areas of women's lives remain virtually untouched by

government and other quantitative research and it would be impossible to

do an inventory here. In the course of our interviews, however, many

conditions were exposed which could be measured by existing quantitative

techniques, and thus serve to illustrate the oft-complementary nature of

these approaches. For example, some women indicated that they could not

join the company pension plan until they had been with the organization

for a specified period of time or until they had reached a certain age.

General questions on persons covered by pensions overlook policies like

these which have the effect (and perhaps the purpose) of preventing

women from adequately protecting themselves in their old age.

The wcmen we interviewed also drew attention to sick pay

policies. Simple to measure but seldom enquired about, sick leave is

not covered by most labour codes. For many women, "if you sick, you

sick." No work means no pay. Given their meagre earnings, most simply

cannot miss work if they are ill. "Clerks cough in your face because

they can't afford to be sick. I know I do." The significance of

workers' health to the transmission of disease was recognized many years

ago when special labels were attached to homework goods to warn

potential customers that they might spread contagious diseases from

workers' homes (Johnson and Johnson, 1982:47). But today, women who are

not well may process food, sew clothes, care for other people's

children, and serve meals in restaurants because they cannot miss a

day's wages. Gnce again, the interview response indicated an important

issue easily measured by quantitative methods but all too often ignored.

- 30 -



Another example of the contribution qualitative research can make

to the collection of quantitative data comes from the investigation of

procedures for promotion. Niany of the women we interviewed told us, "I

don't think there's such a thing there as promotion," suggesting that

enquiries into women's upward mobility should begin by examining whether

or not such movement is possible. Although many people have pointed out

that women are often locked into dead end jobs, few have exposed another

difficulty mentioned by these women. Some refused promotions because a

move to a higher position meant little, if any, increase in power or pay

and because it would eliminate the only satisfying aspect of their work

— companionship with other women. Oie woman turned down a promotion

because it meant working at night, with more responsibility and

virtually no increase in pay. Here too, questions on what rewards are

offered by promotion, if promotion is possible, are not difficult to get

at through survey techniques but the researcher must first be aware of

such possibilities.

In quantitative research, the problems revealed are likely to be

technical. In qualitative research, the difficulties exposed are more

frequently substantive, suggesting new areas for investigation, ones

which are often amenable to quantitative techniques. The examples of

pensions, sick pay and promotions serve to indicate some ways

qualitative data can improve the collection of facts in quantitative

research, and can aid in the selection of what is real.

3. Collecting Efeta Inaccessible to Quantitative Techniques

Qualitative research can do more than merely indicate ways to

improve quantitative techniques. It can provide complementary (or

contradictory) information that brings life to the statistics, fills in

- 31 -



the numbers and explores relationships, Moreover, such methods can look
at the meanings of experiences in a way that cannot be measured by

multiple-choice questionnaires that generate machine- readable answers.

They can permit the investigation process to be an exchange which allows

those being studied to participate actively in the description and

definition of their lives.

How could a survey of worker skills take into account the kind of

conscientious performance revealed in a fish packer's description of her

work?

For my job, I don't think it requires too much skill. It takes
energy. It takes common sense to try to keep doing your work
properly so it doesn't slow down production. If you have no
interest and no common sense... you know that the bosses are
going to stop you and the machine is not going to take it and
there you're going to stop production. There are some girls
that just don't care. I'm not saying I'm perfect 'cause I'm not
and I do lots of blunders myself but I do respect my job. After
all, it can do something to your job if you're not going to pack
good fish that the public will want to buy, and if the people
are not going to buy from the plant that's going to be bad for
your job.

Or the pride of a home worker who is paid by the number of skirts she

sews?

My sister-in-law says, "I don't know how you work so fast," but
I have to do fast. Everything I have to do fast. On Thursday I
have 74 skirts. My husband said, "You don't finish this week."
I said, "Sure".. . .and I finish on Friday night.

Or the sound of an office?

The noise level is unreal because we have— 11 phones in a
small area. They ring constantly. All of the typewriters
going... 10 typewriters and usually 4 or 5 more for the
casuals. ..By 3 o'clock, I can't stand it and my nerves are
really bad.

Or the reality of working on a line?

Basically, I stand there all day and slash the necks of the
chickens. You make one slash up on the skin of the neck and

then you cut around the base of the neck so the next person

beside you can crop it... The chickens go in front of you on the
line and you do every other chicken or whatever. And you stand
there for 8 hours on one spot and do it. On a regular day Tfive
people] can do maybe about 30,000 fryers.

- 32 -



Or the dangers that many women face?

One day the propane dryer caught on fire. They never evacuated
the plant. They just tried to put the fire out. I think when a
propane dryer that size catches on fire, it's a big fire. The
plant should have been evacuated. That's my opinion now. And
there are people working right by that dryer. If you die, you
die. If you don't, you get the work done.

Such things cannot be easily measured, tabulated and fed into a computer

for the analysis of statistical relationships.

It is not only descriptions that emerge from qualitative

analysis. Often, some of the most important processes at work, some of

the most potent forces, are missed in quantitative data collection. For

example, the implied threat of firing and the awareness of a huge

reserve of women desperately searching for work have a powerful

influence on women's performance on the job. According to one clerk:

To get fired from office staff, all you gotta do is look
sideways. It's got to the point where everybody knows that
you've got to do your work to keep your job. There's always
another person ready to take your job.

Close supervision also has an impact that is difficult to measure. A

supervisor of proof-readers ...

...used to time us with his stopwatch behind your back... you can
imagine what it was like. I came home on Friday night and my
husband, he had to give me two good slaps in the face because I
was knocking my head on the wall. I wanted to kill the foreman.

The structure of work, too, strongly influences performance, as a

chicken slasher explained:

Basically, the reason that you don't make too many mistakes is
because one job relies so much on the other. If you don't slash
it and if they can't crop it, then it can't be drawn. So you
can't miss a chicken because then further down the line they
bring it back or they holler, "Hey, what are you doing?" So
this pretty well means you stay awake.

These relations with other workers both ensure that the pace is

maintained and help women get through the day — that is, if they are

permitted to talk:
- 33 -



If you were standing there all day and you wouldn't talk, well
you go right bananas—A. friend by me, she always says, "For
goodness sakes...say something, say something," 'cause it breaks
up the time if you have a little conversation. You're not
allowed to stand and talk but. ..you can talk as much as you like
as long as your hands go.

Such research can also show that women are not merely passive but

actively participate in shaping their working conditions. From the

women who adjust the thermostat when the boss is not looking, to the

women who develop elaborate strategies to vary their tasks and eliminate

boredom, there are many examples of women individually resisting the

structure of their work. Women also get together to slowdown the line

when it is going too fast, or to force changes in their work

environment, as was the case in one plant:

When I'm working, it is hot. ..We don't have ventilation enough.
Where I'm working, there are three big fryers and the fat is hot
and when that fish strikes the fat, it gives smoke. And there's
times when you go in there, you can just about see through the
place. Now, we're inhaling this all day. Now, we don't have no
ventilation. There's no windows. When you're in there you
don't know if it's raining, snowing, hurricane or whatever
outside. When you go in, you're there; when you come out, then
you see what kind of day it is. They have fans and half the
time they are not working for the simple reason that it's the
fat that clogs them. We have walked out already and I was the
head one on our line too. I told my supervisor, "There's no way
I can stand it in there. I'm choking and I cannot stand it.
I'm going out 'till it's cleared up." And the girls followed
us... They got it fixed the next day.

Although justified by the conditions, this woman was frightened: "t*y

heart pounded like you'd never believe. I hate doing anything like that

'cause I'm scared of getting fired." It was not her sex but her lack of

job security that had discouraged this kind of resistance in the past.

Qualitative research can also capture, in a way that is

impossible in quantitative investigations, what it is like to be an

employed woman. For example, the women we interviewed reported that

they were frequently asked to do extra tasks that were unrelated to

- 34 -



their job description but fell into that all too broad category "women's
work." As a woman hired to fill mail orders explained, "I hate dusting

and cleaning. I don't go to work to do these things. I can do them at

home. ..Yeah, I do it." Most women go home to yet more of these same

kinds of tasks:

All I have to say is that a working mother, when she has to work

and leave her kids at home, it's pretty hard whether she's
married or single. A mother that works and raises a family has
to be pretty strong, very strong. I really feel sorry for the
mother with the little kids — have to bundle them up in the
morning and take them to the day care center and pick them up
again at night. I wouldn't go through that again that way.

It's too hard.

In our research, we were unable to explore the relationship between home

and work in any detail, but the recent case study of an east coast

fishing village (Cbnnelly and MacEonald, 1983) clearly illustrates the

importance of relating domestic and wage labour to each other and to the

wider economic structure. Such relationships and such an integrated

approach are missing from quantitative data which usually measure

individual activity, for short periods of time, in a restricted context.

And quantitative research misses what technological change means

to the workers: "Before, everyone was a typesetter. Now everyone works

on a typesetting machine." "I went through tests two years ago and they

found out that my hearing is poorer in one ear than the other. Mien I

started there they told me that my hearing was very good." Perhaps even

more elusive are the silent, psychological firings that never appear in

the statistics. Che long-time secretary described how she was refused

training on the new electronic equipment, how her job slowly

disappeared, how she spent hours at her desk with nothing to do, how she

had to stay at her desk in order to get paid and how staying there meant

it was difficult to look for alternative work. Finally she quit and was

therefore not defined as a victim of the new technology.

- 35 -



Qualitative data are often dismissed as anecdotal, as

unscientific, as unrepresentative because they do not include a

statistically selected sample of the target population. Researchers

often respond to this criticism by resorting to statistical analysis

alone or by transforming their information into numbers. Komarovsky

(1962:204), for instance, takes a group of fifty-eight marriages and

tells us that "among the older couples. . .only three per cent of the

less-educated husbands" complained that "we have nothing interesting to

talk about." Yet, the way our examples resonate in the lives of women

suggests that they tap many shared experiences. They may provide not

only a fuller, more humane, picture of women's daily lives, but a more

accurate one as well.

Conclusion

Theory, statistics and qualitative data are all necessary for

rigorous research. Theory guides the selection of the real, the

techniques employed to expose it, and the explanations used to

understand it. Statistics can help to establish the broad outline of

the picture, a picture that can only be fully developed with the aid of

qualitative data.

We have argued that, like all methodologies, those of Statistics

Canada reflect a perspective, one that better serves the interests of

men and the economy than those of women and the home. Available data do

not permit a thorough examination of the sex-segregated labour force;

they offer little assistance in examining the labour process, health and

safety, hiring, promotion, wages and benefits, job security — to

mention only a few issues — as they affect women. The priorities of

the agency thus allow only a partial, and sometimes distorted, view of

- 36 -



women's work.

We have also argued that statistics are not all they have been

cracked up to be. The way data are collected and tabulated, the way

questions are asked and not asked, the way government programmes and

policies are structured and the way in which history is considered, all

influence the data and in the process often leave out and sometimes

misrepresent the position of women. Qualitative data, while not free of

faults, can provide an effective complement to quantitative techniques,

checking the results, suggesting alternative areas and methods for

research and filling in the numbers with the actual experiences of

women.

What is required is, first, a more thorough evaluation of

Statistics Canada from a perspective which acknowledges that, given the

sexual division of labour, most structures, processes and programmes

have a particular impact on women. Second, more theoretical work should

be done to aid this critical evaluation and to promote the development

of other ideas. Third, more funding for qualitative research is

necessary if we are to understand, and change, what it is like to be a

woman in Canada today.

NOTES

* This is a slightly revised version of a paper appearing in Mary
Kinnear and Greg Mason (eds.) , Women and Work: Proceedings of a
Conference on Data Requirements to Support Research into Women and
the Canadian Economy (Winnipeg: University of Manitoba Institute
for Social and Economic Research for the Social Sciences and
Humanities Research Council of Canada, 1983). Our thanks to Pat
Grainger, Bob Stirling and Maryanne Webber for comments on the
original version.

[1] Statistics Canada does explain its methodology but the justification

- 37 -



for particular kinds of techniques rests on the assumption that
these are primarily scientific and practical rather than theoretical
issues. The tools are not related consistently to the theoretical
framework. For example, Statistics Canada's Methodology of the
Canadian Labour Force Survey (1977:23) offers the following
justification for the use of a stratified sample:

Though the main advantage of stratified sampling is the
possible increase in efficiency per unit cost in estimating
the population characteristics, stratification also introduces
considerable flexibility in the sense that, depending upon the
information available, sampling and estimation procedures may
differ from stratLm to stratum. Further, in a continuous
survey like the LFS, stratification provides an added
flexibility of updating or redesigning the sample of a
specified stratum or groups of strata, without affecting the
design in the remaining strata.

See also the tool-kit orientation of Statistics Canada's (1980a)
Social Concepts Directory . For a contrary example, however, see Kim
Farrell7^ (1975) article on the impact of Keynesian and
post-Keynesian developments in economic theory on labour force
survey measures.

[2] The Census does publish data on additional persons in the household
(see, for example, Statistics Canada, 1982c:Table 6) but the nature
of this relationship and, particularly, the consequences for the
household economy and domestic labour are difficult to determine
from such figures. The assumption seems to be that only
never-married children are dependent.

[3] Some LFS occupational data are now available at the minor group
(three-digit) level of detail, at which the broad "managerial,
professional, etc." category is broken down into 26 distinct
occupations. Shortcomings remain, however. This information can be
obtained in unpublished form only, and, because of the 'reliability'
criterion, as part of only a few, general cross-tabulations.

[4] The 65 interviews included in our study (Armstrong and Armstrong,
1983) were conducted by 7 researchers selected for their expertise
in interviewing techniques, knowledge of feminist issues and
familiarity with a particular area of Canada. The interview
schedule was designed as an outline of areas to be covered rather
than as specific questions to be asked. Interviewers were
instructed to gather as much information as possible on each area,
to record conversations on tape and to view the work site. We wish
to acknowledge the assistance of Sonja Greckol , Sharon Kirsh, Judy
Mosoff , Shirley Pettifer and Martha Tracey in the interview stage of
the project.

- 38 -



REFERENCES

Armstrong, Pat and Hugh Armstrong
1983 A Working Majority . Ottawa: Supply and Services Canada.

Canada Department of Finance
1981 Economic Review . Ottawa: Supply and Services Canada.

Connelly, M. Patricia and Martha MacEonald
1983 "Women's Work: Domestic and Wage Labour in a Nova Scotia

Community", Studies in Political Economy 10.

Farrall, Kim
1975 "Theories, Concepts and the Art of Measurement: Economic

Theories' Relationship to the Labour Force Survey Concepts,"
Notes on Labour Statistics 1973 (Statistics Canada Cat. no

72-207) 20-28. Ottawa: Information Canada.

Irvine, John, Ian Miles and Jeff Evans (eds.)
1979 Demystifying Social Statistics . London: Pluto Press.

Johnson, Laura C. and Robert E. Johnson
1982 The Seam Allowance: Industrial Home Sewing in Canada .

Tbronto: The Women's Press.

Komarovsky, Mirra
1962 Blue-Collar Marriage . New York: Random House.

Luxton, Meg
1980 More Than a Labour of Love. Three Generations of Women's Work

in the Home . Tbronto: The Women's Press.

Menzies, Heather
1981 Women and the Chip: Case Studies of the Effects of Informatics

on Employment in Canada . Montreal: The Institute for Research

on Public Policy.

Mills, C. Wright
1959 The Sociological Imagination . New York: Grove.

Miner, Horace
1939 St. Denis: A French-Canadian Parish . Chicago: University of

Chicago Press.

Roberts, Helen (ed.)
1981 Doing Feminist Research . London: Routledge and Kegan Paul.

Rubin, Lillian Breslau
1976 Worlds of Pain . New York: Basic Books.

Statistics Canada
Employment, Earnings and Hours , Monthly (Cat. no. 72-002).
Ottawa: Supply and Services Canada.

The Labour Force , Monthly (Cat. no. 71-001). Ottawa: Supply

and Services Canada.

- 3.9 -



1976 Earnings and Work Histories of the 1972 Canadian Labour Force .
(Cat. no. 15-55"). Ottawa: Industry, Trade and Commerce.

1977 Methodology of the Canadian Labour Force Survey 1976 (Cat. no.
71-526). Ottawa: Industry, Trade and Commerce.

1979 Guide to Labour Force Survey Data (Cat. no. 71-528). Ottawa:
Supply and Services Canada.

1980a Social Concepts Directory (Cat. no. 12-560). Ottawa: Supply
and Services Canada.

1980b Workmen's/Workers' Compensation: Social Security, Provincial
Programs (Cat, no. 86-501). Ottawa : Supply and Services Canada.

1982a 1981 Census Dictionary (Cat. no. 99-901). Ottawa: Supply and
Services Canada.

1982b Products and Services of the 1981 Census of Canada . Ottawa: Supply
Supply and Services Canada.

1982c 1981 Census, Volume 1. Census Families in Private Households .
(Cat. no. 92-955) . Ottawa: Supply and Services Canada.

Thompson, E.P.
1978 The Poverty of Theory . London: Merlin Press.

Vickery, Clair
1979 "Women's Economic Contribution to the Family," in Ralph E.

Smith (ed.) The Subtle Revolution: Women at Work . Washington:
The Urban Institute.

- 40 -