NONSEXIST RESEARCH: A METATHEORETICAL APPROACH1 MARGRIT EICHLER ...
NONSEXIST RESEARCH: A METATHEORETICAL
APPROACH1
MARGRIT EICHLER
How to identify the sexist bias embedded in most scientific knowledge produced in human societies? A simple
diagnostic tool is presented here to identify seven forms of sexism. Every component of the research process
including the title, language used, concepts employed, the research design, methods, data interpretation and
the ensuing policy recommendations is combed for its sexist manifestation. The culminating value of this paper
is in the solution it offers in eliminating sexist bias in all disciplines, irrespective of the subject matter or of the
cultural context within which it exists.
Dr. Margrit Eichler is Professor of Sociology at the Ontario Institute for Studies in Education, Canada. A leading
author of many books and articles on sexism in scholarship and nonsexist alternatives, she has specially contributed
this paper to the readers of IJSW, adapting it from a set of exhaustive guidelines developed within the framework
of the Canadian Women's Studies Advisory Committee of the Secretary of State's Women's Programme.
Introduction
Knowledge production, including the process of research, is a social process, as we well know. As
such, it is subjected to various social influences, one of the most important of which is sexism.
Thus we have known for some time that most, if not all, scientific knowledge in the human
sciences suffers from a sexist bias2.
If sexism in research is the problem, developing tools to identify and criticise it in existing research
and avoiding it in contemporary research is the solution. The starting point and the envisioned goal
are therefore clear, but how to get from one to the other is neither easy nor straightforward. This
paper outlines an approach that allows researchers with a simple diagnostic tool, to identify
various forms of sexism in existing research. It also provides means to avoid sexism in on-going
research, in all of its stages, from inception over data collection to its conclusion.
It is a metatheoretical approach since it is applicable to all human sciences, irrespective of subject
matter or of the cultural context within which it exists. It consists of a set of "yes" or " n o "
questions a researcher can ask of research — someone else's or one's own. Depending on the
answer, a problem is indicated or not. If there is a problem, a solution on how to address it is
provided. Although this sounds simple, the implementation may be highly complex.
Sexism in research is only one manifestation among many of sexism in general. Ultimately, the
solution will lie in eliminating sexism from our social structures. This is clearly not a short-term
project. Feminist movements across the world have been actively working towards this end for
decades (Morgan, 1984). The various forms of sexism support each other. Efforts to reduce
sexism in one element — such as research — are hampered by the existence of sexism in other
aspects of the society (e.g. the low numbers of women in scientific and research positions and the
overall male-oriented structure of most research organisations). By the same token, if we weaken
sexism in one area, it will have some spill-over effect into other areas, precisely because of this
linkage. Eliminating sexism in research is therefore important for its own sake, as well as for the

330 Margrit Eichler
effect it can have in helping reduce sexism in other aspects of social life. The approach presented
below will make it simpler to address the problem, at least in research.
History of this Approach
Identifying sexism even within one single concept may be a task that takes one or more
researchers many years of work. For instance, one of the most important sexist concepts concerns
the common definition of the term "productivity". This is related to our definition of the term
"work", which for the longest time was defined as paid work only, thus excluding most of the work
performed by women across the world, although this work is life sustaining and therefore of utmost
importance (Shiva, 1989).
Following this, the general definition of "productivity" excludes unpaid work that does not
generate surplus value on the market place and specifically, all reproductive work. The United
Nations System of National Accounts (UNSNA), which forms the basis for economic and policy
planning across the world, is thus premised on a sexist concept that excludes the necessary work
(without which the world economy would utterly collapse) performed by women outside the market
place — and therefore fails to take into account, one of the very bases on which our current social
and economic systems rest (Waring, 1988).
In order to be able to make such a statement, an enormous amount of critical analysis of the
concepts of "work", "production", "reproduction" on the part of many people needs to be done.
We are, therefore, dealing with a very slow, laborious and difficult set of thought processes.
Until recently, every analysis of a sexist concept, a sexist frame of reference, a sexist
interpretation of results, etc. started virtually from scratch. Researchers in many countries and
many disciplines were performing similar types of analyses for every single problem of sexism. In
such a process, knowledge cumulation works only to the degree in which sexism presents itself in
the same form. The question then posed itself whether it was possible to analyse the logic of
sexism in such a manner that would make it possible to transfer insights derived from one set of
problems to another, as yet unanalysed problem.
In order to determine whether it was possible to identify the structure of questions underlying
existing analyses of sexist problems in research, such analyses were carefully read and analysed
in terms of the underlying (implicit) questions they asked. This process continued over many
years. Whenever an analysis of sexism of any aspect of the research process was found, it was
analysed in terms of its underlying logic and approach.
In order to be able to engage in this endeavour, several preliminary questions had to be settled
first. One question concerned terminology. The concepts "sexist", "androcentric", "patriarchal"
and others were often used in an interchangeable manner in the literature, as were the concepts
"feminist", "nonsexist", "woman-centred" and others.
In order to come up with a good understanding of nonsexist research, it was necessary to separate
the concept from feminist research. This, it turned out, was both possible and useful. Briefly,
feminist research (of which woman-centered research is one sub-aspect) is informed by a
commitment to social justice for women. One aspect of this is to identify sexism in research and to
develop nonsexist tools. Nonsexist research, by contrast, is not bound to any topic, and uses the

Nonsexist Research 331
tools developed by feminist researchers, without a necessary specialisation in feminist
approaches (Eichler, 1986).
Second, it was necessary to determine whether sexism in research occurred primarily or
exclusively in one, some or all components of the research process. By analysing analyses it
became clear, after a while, that sexism can (but need not) be present in all components of the
research process, but may be present in only one or some, depending on the piece under
examination, (Eichler, 1991).
Third, the question arose whether there were different forms of sexism which were non-reducible
to each other, or whether there was one overriding form of sexism that contained various subforms
within itself. The former turned out to be the case. This is a very significant finding. If there is
basically one problem of sexism, one solution would do. If there are several problems which are
nonreducible to each other, one might inadvertently substitute one problem of sexism with another
and have the illusion that one has actually solved the sexist problem first identified.
A simple example of this is the initial response to the problem of sexist language. When the
argument was first made that the use of so-called "generic" male terms (the generic " h e " , "man"
etc.) was sexist, some people thought that the answer to the dilemma was to use truly generic
terms (people, humans, individuals etc. instead of man, mankind, etc.), regardless of the context.
This meant that generic terms were used even if only one sex was considered (e.g. using
"spouses" instead of "wives" or "husbands" in cases where only one of them was affected). All
this achieved was that one type of sexist problem — androcentricity — was replaced by another
sexist problem — gender insensitivity.
Each type of sexism was therefore assigned a specific name (taken from the existing literature on
the topic) and basic questions were matched with each identified problem. Each type of problem
was then systematically applied to each component of the research process (e.g., What form would
gender insensitivity take in the title, or a double standard in data collection, etc.). The end result
was a schema which identified seven types of sexism3 which could occur in any part of the
research process.
With the schema of analysis in place, the next step was to go to the library and take recent issues
from various social science disciplines, using always the most established journals in case there
were several journals to choose from. This analysis was performed in 1985, and regrettably, in
each issue that was scrutinised, many examples of sexism were found. There was no exception.
The examples were all from the social sciences, and claims for applicability were only made for the
same. Once the analytical framework was available, it was used for medical research as well, and
it turned out that it was transferable to that type of research in its entirety (Eichler, Reisman and
Borins, forthcoming and Williams and Borins, forthcoming). It is therefore obviously a robust and
versatile approach.
It is also culturally nonspecific. However, the utility of this approach is limited by the amount of
knowledge a researcher has about a specific subject area. Some of the diagnostic questions can
be answered without specialised knowledge about a subject area (e.g. does the research present
itself as general while addressing the concerns of one sex only?) while other questions
•presuppose an intimate knowledge of the topic under consideration (e.g. is this quality part of a

332 Margrit Eichler
superordinate quality that is present in both females and males but that has been conceptually
linked with one sex only? or: are the implications of this statement in fact different for females and
males?).
A Metatheoretical Approach to Diagnosing Sexism in Research
Obviously, there are at least two logical ways to present this approach: by type of problem or by
the component of the research process that is affected. Given the space limitations, I will
concentrate on the second option, but will present a brief definition of each of the seven problems
first. (Examples will be provided in the next section.)
Types of Sexist Problems
Androcentricity can be defined as the adoption of an overall male perspective.
Gender insensitivity, in brief, denotes ignoring sex as a socially important variable in contexts
within which it is important. Given contemporary society, it seems safe to proceed on the
assumption that sex is socially important until it has been shown not to be.
Sexual dichotomism involves treating the sexes as two entirely discrete social, as well as
biological groups, rather than as two groups with overlapping characteristics. It is, in many ways,
the mirror image of gender insensitivity. The one ignores the importance of sex, the other
exaggerates it.
Familism consists in treating the family as the smallest unit of analysis in cases where it is, in fact,
individuals within families (or households) that engage in certain actions, have certain
experiences, suffer or profit from particular costs or benefits, etc.
Overgeneralisation occurs when a study deals with only one sex but presents itself as if it were
applicable to both sexes.
Double standards are used when identical situations, behaviours, or traits are evaluated
differentially on the basis of sex.
Finally sex appropriateness becomes a sexist problem when descriptive terms for sex differences
are used in a prescriptive manner, or are assigned ontological status, such as when the absence
of an historically grown "appropriate" gender identity is seen as a psychological disorder.
Examples of Sexism in the various Components of the Research Process
Space prohibits providing an example of all types of problems for all components of the research
process. I shall therefore restrict myself in the number of examples offered. A complete list of
questions to be asked is contained in the appendix. A complete discussion of all types of problems
in the various research components can be found in Eichler, 1991.
Although titles seem like a convenient starting point, and they are listed first in the appendix, I shall
defer their discussion to the end, since they draw on the discussion concerning other sexist
problems.

Nonsexist Research 333
This brings us to the issue of language. Sexist language is sometimes misunderstood as the sole
problem. This is, of course, not the case by a far stretch of the imagination, nevertheless, it is
impossible to generate a nonsexist piece of scholarly work with sexist language. Unfortunately, it
is quite possible to generate a sexist piece of work while using entirely non-sexist language. When
authors use male terms (e.g. "man", "mankind", "he") for people in general, they are using a
sex-specific term for generic purposes (a problem of overgeneralisation). The solution is, of
course, to use generic terms for generic purposes (e.g. people, humanity, they, one). Conversely,
sometimes generic terms are used when only one sex is discussed, e.g. "parent" often stand for
"mother"; "citizen" or "entrepreneur" often stands for male citizen or male entrepreneur. Clearly,
the solution here is to use sex specific terms for sex specific purposes.
These problems, and the others identified in the appendix, are important because scientific
language needs to be exact. When the language is so imprecise as to make it guesswork whether
an author is talking about people in general or only about men (or women, although this usually
occurs with issues related to fertility and caring work), it is unsuited for scholarly purposes.
Recognising sexism in concepts is not as straightforward and may, in fact, be exceedingly difficult.
We are here dealing primarily with concepts that are not sexist at the level of the language
employed, but in terms of their assigned meaning within a given context. One question to ask is
whether there is a discrepancy between the theoretical and empirical referent. For instance,
entrepreneurs can and have been of either sex. The theoretical referent of the concept is,
therefore, of either sex. If an article, then, defines an entrepreneur as "one who would start an
industry or take to a new enterprise which is a deviation from his traditional family occupation or
profession" (Setty, 1990: 87) and focuses on "... the man and the group behind him" (Setty, 1990:
89) we have found a discrepancy (a problem of androcentricity). This discrepancy can be resolved
either by making the concept sex specific (male entrepreneurs) or by changing the content and
including potential female entrepreneurs.
When a concept expresses a relational quality from the perspective of one sex only (e.g. polygyny
or polyandry), it can either be matched with a concept that introduces the other perspective
(multiple wives/husband sharing and multiple husbands/wife sharing) or by creating a superordin-
ate concept, where that is possible (e.g. the concept "parental deprivation" which appears in two
subforms: "maternal deprivation" and "paternal deprivation").
Another question is whether ego is constructed as male. For instance, when "intergroup warfare"
is interpreted as "a rational means of gaining livestock, women and slaves" (Shaw, 1985: 166) —
who is defined as the actor? Who wages war? If "inclusive fitness" increases with greater
cohesiveness, and cohesiveness, in turn, is seen as a function of relatedness, who is defined as a
member of the group, if the women are won by warfare and presumably unrelated? Are they — or
are they not — part of a cohesive group? A new set of questions emerges by introducing the
perspective of women that have not been posed in the article due to an overall androcentric bias
which relegates women to the status of objects to be fought about.
A third version of a sexist concept involves a human trait that is represented as being present in
one sex only. For instance, if researchers try to isolate the effect of family members on
schizophrenic children, and in the process coin the concept of a "schizophrenogenic mother"4 but
fail to examine the "schizophrenogenic father" they are applying a double standard to mothers
and fathers. The solution is, obviously, to examine the influences of both parents (provided one
wants to go that route at all!).

334 Margrit Eichler
Sexism in the research design deals with the overall frame of reference and the overall approach
taken. It is easier to identify than to remedy, because it is likely to involve an entire reshifting of any
given project.
Coming back to an example used previously, a recent article on "Developing Entrepreneurship in
Remote Rural-Tribal Communities" (Setty, 1990) provides an example of an androcentric frame of
reference. Women may fall into all the potential segments from which potential entrepreneurs may
be drawn: they may belong to artisan groups, may be high-school drop-outs, may be educated
unemployed, belong to merchant-trading groups, be farmers of farm-labourers (p: 95), yet no
attention is paid to what would be involved in mobilising women to become entrepreneurs. For
instance, when discussing problems, there is no mention of institutionalised sex roles as an
impediment to developing entrepreneurship, or male domination of women which may prevent
women from participating in programmes, or differential male and female access to family
resources, including time, or family functions that need to be replaced while the candidate is being
trained, such as childcare and cooking.
This particular article contains a proposal for a programme, to be implemented in five phases, all
of which are dealt with in a gender insensitive manner. The phases are first, identification of
natural resources, second, training of training personnel, third, training of potential entrepreneurs,
fourth, interagency collaboration, and fifth, evaluation. Nowhere is there a recognition that women
and men are differentially socially placed and that, therefore, all policies need to be sensitive to
such differences, to attract sufficient numbers of women.
The article, which proposes an otherwise attractive approach to developing entrepreneurship thus
suffers from sexism in all of its components, thereby severely limiting its utility. A gender sensitive
approach might have doubled the number of potential candidates by including women within its
frame of reference and thus doubled its effectiveness.
There are, of course, many other ways in which sexism within the research design may manifest
itself. This component overlaps partially with the methods used. With respect to methods, any
method can be used in a sexist manner. Whether or not any method can also be used in a
nonsexist manner depends partially on how broadly or narrowly one defines a particular method —
that is, whether it is still given the same name when lifted out of a particular application.
To provide just a few examples of sexism in methods, one way in which it can enter is through the
questions asked — both in the figurative sense of "What research question is being asked in this
study?" as well as in the literal sense of the formulation of questions in survey research, for
instance. If we approach a historical study by asking only what role public men played within a
given period, without asking what the direct or indirect contributions (positive or negative) of
women were to the phenomenon under consideration, we have asked different questions of the
sexes. The same applies, on a smaller scale, when men are asked about issues which are seen as
of public import and women about issues which are seen as of private import.
This distinction of public and private is, of course, problematic in and of itself. Why should it be a
public matter if a woman is attacked by a stranger, but a private matter if she is continually abused
by her husband? Why is military spending seen as a high public priority in many countries, while
child care is seen as only of private importance? Which of the two are more important for the
ultimate survival of a society? The current definition of a public vs. a private matter is an example
of a double standard.

Nonsexist Research 335
Another problem occurs when males are taken as the norm. For instance, if a study asks
respondents (of whatever sex).
Do you think that women perform as pilots (fire fighters/brain surgeons/entrepreneurs/
supervisors/etc.)... worse than men ... as well as men?
without adding as an option "... better than men" women are measured against a male standard,
and the formulation of the question fails to offer the whole spectrum of theoretically possible
answers. It is another example of an androcentric approach.
With regard to sexism in data interpretation, it can take a wide variety of forms. One
manifestation is the failure to analyse data on the basis of sex although data from both sexes have
been collected. A particularly drastic example of this is the famous aspirin study which examines
the effects of aspirin on the likelihood to experience heart attacks. Although the study collected
data from both females and males, it is unclear what the exact proportions are, and the data are
not analysed by sex (Fields et at., 1977). Since other studies (American-Canadian Co-operative
Study Group, 1985) have suggested that the benefit of aspirin in thromboembolism varies by sex,
the conclusions concerning the efficacy of aspirin must be questioned for both sexes. In medical
research, sexism (in this case, gender insensitivity) may literally be a question of life and death.
To stress just one more point, when we are dealing with victims, e.g. the recipients of abuse,
mutilation, even death, we need to ask ourselves whether there is any implicit justification of such
abuse (e.g. in the name of cultural traditions, religious precepts, etc.). This may take seemingly
neutral formulations, such as when wife battering is discussed as a "mechanism for conflict
resolution". Battering, of course, is (part of) the problem, never a solution. It may take more
obvious forms, as when the focus is on how and why the victim "provoked" the abuse, rather than
how and why the perpetrator chose to administer the abuse.
Turning to policy evaluations and recommendations, this is a component that is present only in
a minority of studies. It is, however, particularly important that policy evaluations and
recommendations be nonsexist, since they may have more immediate effects on people than
other studies. Any policy study that fails to take sex as a socially important variable into account is
gender insensitive. Unless we explicitly consider the consequences of any given policy for females
and males, we stand to limit the utility of such studies, sometimes to the point of failure.
Lastly, to return to the titles — they can be sexist by employing sexist language or concepts, by
overgeneralising, for instance by evoking the image that an entire phenomenon is treated when in
reality it is considered for one sex only, or they can be sexist in other ways.
Conclusion
This has been a very rapid, and therefore incomplete, review of some of the problems of sexism in
research and how to overcome them. It is neither a simple process, nor will it be a quick one.
However, if we wish to conduct research that is accurate and useful, making sure that it is
nonsexist is a necessary (although not sufficient) aspect.
APPENDIX
Diagnostic questions for identifying the presence of sexism in research
(1) Sexist Titles:
1.1 Does the title evoke the image of applicability to both sexes? If yes, is it, in fact, applicable to both sexes?

336 Margrit Eichler
1.2 Does the title contain a sexist concept?
1.3 Does it contain sexist language?
If the answer is "yes" to any of the questions, either the title or the content need to be changed.
(2) Sexist Language:
2.1 Are any male or female terms used for generic purposes?
If yes, use sex specific terms only for sex specific purposes.
2*.2 Are any generic terms employed when, in fact, the author is speaking only about one sex?
If yes, use generic terms only for generic purposes.
2.3 Are females and males in parallel situations described by nonparallel terms?
If yes, use parallel terms.
2.4 When both sexes are mentioned together in particular phrases, does one sex consistently precede the
other?
If yes, alternate in some manner.
2.5 Are the two sexes consistently discussed in different grammatical modes?
If yes, identify the passive or active portion that is missing and incorporate to the degree possible.
(3) Sexist Concepts:
3.1 Is ego constructed from only a male (or female) perspective?
If yes, change either concept or content.
3.2 Does the concept refer to a relational quality expressed from the viewpoint of one sex only?
If yes, express from the viewpoint of both sexes, either by creating a conceptual pair, or a. superordinate
concept.
3.3 Does the concept demean one sex?
If yes, replace with a non-demeaning concept.
3.4 Is there a difference between the theoretical and the empirical referent of a concept?
If yes, match the theoretical and empirical referent.
3.5 Is the concept premised on an attribute that is present in both sexes but is operationally defined in such a
manner that it will categorize females and males differently?
If yes, categorise females and males equally if they display equal attributes.
3.6 Does the concept refer, in a sex-linked manner, to a situation, trait, or behaviour that exists for both sexes?
If yes, change concept so that it expresses human attributes in non-sex specific terms.
3.7 Does a conceptual pair correspond largely to a sexual division in an unjustified value-laden manner?
If yes, reformulate so that equal value is assigned to male- and female-dominated attributes.

Nonsexist Research 337
3.8 Is the concept premised on the notion that certain human behaviours, traits, or attributes are appropriate
for one sex only?
If yes, change concept so that it is descriptive, not prescriptive.
3.9 Does the concept attribute individual properties, attributes, or behaviours to families or households?
If yes, substitute concepts that identify individual properties, attributes or behaviours as such.
(4) Sexism in the Research Design
4.1 Is the frame of reference androcentric? Test by substituting "woman", "girl" for generic terms such as
"individual", "person", "citizen", "child".
If the result is nonsensical, explore the females' situation and incorporate, or revamp the study as
pertaining to one sex only. The latter is not appropriate when the study deals with an issue that by
necessity involves both sexes.
4.2 Is the frame of reference embryonically5 gynocentric?
If yes, give attention to the role of males.
4.3 Are men consistently treated as actors, women as acted upon?
If yes, explore the role of women as actors and of men as acted upon.
4.4 Is male behaviour taken as the norm, and female behaviour as the deviation that needs to be explained?
If yes, either expand the framework by assessing male behaviour against female behaviour, or establish a
genuinely sex-unrelated behaviour as the norm. The latter is not feasible if the behaviour in question is
strongly sex-differentiated.
4.5 In cases in which victims and perpetrators are involved, are women blamed — with or without justification?
Are women or men perpetrators? Are perpetrators held personally responsible for their deeds? Is the role
of male participants in the overall process adequately considered? Are overall structural factors
adequately taken into account?
Perpetrators need to be held responsible for their actions, the role of males needs to be considered as
carefully as that of the female participants, and overall structural factors need to be taken into account.
4.6 Does the phenomenon under consideration affect both sexes? If so, does the literature give adequate
attention to the roles of both sexes? (In particular, in studies concerning family roles and reproduction, has
the role of men been given adequate attention? In all other studies, has the role of women been given
adequate attention?)
Include excluded sex in your study design. However, where a field of study has largely excluded one sex, a
one-sex study of the excluded sex may be highly appropriate. A issue is the overall balance of the
collective research effort.
4.7 Are both sexes asked the same questions?
If not, do.
4.8 Is the same research instrument used for both females and males? If not is the use of a differential
instrument justified by physical differences between the sexes?
If different instruments are used without compelling reason, develop an instrument that is applicable to
both sexes, if different instruments are necessary, justify them in detail.

338 Margrit Eichler
4.9 Does the research instrument divide the sexes into two discrete groups when, in fact, they have
overlapping characteristics?
If yes, develop a new instrument that does not dichotomize overlapping distribution of traits.
4.10 Of the major variables examined in the study, are they equally relevant to women and men? Is there an
imbalance of variables pertaining more to one sex than the other?
If so, correct imbalance by including variables that affect the underrepresented sex.
4.11 Is the sex of the relevant participants in the research process reported and controlled for?
If not, report and control where possible and necessary. Where not possible, potential distorting effects of
the sex of the participants need to be acknowledged and discussed.

4.12 When dealing with family issues, is it possible that the event (attribute, traits, experience, behaviour) under
consideration may be different for different family members?
If yes, using the family as the smallest unit of analysis is inappropriate. Individual actors within the unit
need to be identified, and studied separately to observe potential different effects. This may involve a
drastic revision of the research design.
4.13 Are explicit or implicit comparisons made between the sexes? If so, are the sex groups being compared
equivalent on all those variables that are likely to influence the outcome under investigation?
if no, create comparable groups. Where this is impossible for practical reasons, carefully list and discuss
the variables that differentiate the two groups.
(5) Sexism in Methods:
5.1 Has the research instrument been validated on one sex only but is used on both sexes?
If so, use for sex on which it has been validated only or validate for both sexes.
5.2 Is the sex composition of the sample adequately reported?
If not, do.
5.3 Does the question posed use sexist language or concepts (see above)?
If so, replace with nonsexist language and concepts.
5.4 Does the question take one sex as the norm for the other, thus restricting the range of possible answers?
If so, reformulate the question to allow for the theoretically possible range.
5.5 Is the question implicitly or explicitly premised on the notion of sex appropriate behaviours, and therefore
fails to ask equivalent things for the other sex?
If so, reformulate to probe for the existence of such behaviours in both sexes.
5.6 Does this particular method categorise males and females into discrete groups on the basis of attributes
that can be found in both sexes?
Categorise nondiscrete traits in nondiscrete ways.
5.7 When opinions are asked of one sex about the other (including in indirect form, such as by using historical
information), are they treated as opinions of one sex about the other rather than as fact?
If no, reinterpret other-sex opinions as statements of opinions and no more.

Nonsexist Research 339
5.8 Are identical coding procedures used for females and males?
If no, make coding procedures identical.
(6) Sexism in Data Interpretation:
6.1 Are the implications of findings for both females and males explicitly considered?
If not, do.
6.2 Are biases in the data collection process explicitly acknowledged and their implications discussed?
If not, do.
6.3 Is there any justification of female subjugation or male dominance? Is any form of bodily mutilation, death,
or other abrogation of human rights justified in the name of a supposedly higher value?
If yes, describe and analyse such practises, but under no circumstances excuse or justify them.
6.4 Is there a clear victim? If so, is the victim blamed for her (occasionally his) victimization?
If so, identify the circumstances (or individuals) that led to victim-blaming and eliminate the blame from the
interpretation.
6.5 Is only one sex considered? If so, are conclusions drawn in general terms?
Make conclusions sex specific where only one sex is considered, or change the research design and
consider both sexes.

6.6 Are data collected on both sexes? If so, are they analyzed by sex? Is the difference or lack thereof between
the sexes considered?
If not, do.
6.7 Does the particular situation or event under consideration have potentially different implications for the two
sexes? Have these been explicitly considered and discussed?
If nor, do.
6.8 Are sex roles (or sex identities) seen as normatively appropriate?
If yes, acknowledge sex roles (and sex identities) as socially important and historically grown, but make it
clear that they are neither necessary, natural, nor normatively desirable.
(7) Sexism in Policy Evaluations and Recommendations :
7.1 Does this policy affect both sexes? Is the position of the sexes comparable with respect to the important
factors that inform and are governed by this policy? Is the effect of this policy positive for both sexes?
If the policy affects both sexes but has differential effects, it is biased. This does not necessarily mean it
should be abandoned. If it is meant to right an old unfair situation, there may be reasons for maintaining it.
Or if it is disadvantageous for one sex, but nevertheless highly desirable for other reasons (e.g.

environmental considerations), it may be more appropriate to develop compensatory policies. If there is no
justification for a biased policy, or if compensatory policies cannot be enacted, it should be reevaluated in
terms of its impact on both sexes.
7.2 Are the same circumstances evaluated differently on the basis of sex?
If so, reevaluate so that the sexes are treated in the same manner.

340 Margrit Eichler
7.3 Is there a division that corresponds largely to a division by sex and for which differential treatment is
recommended?
Make the sexual division visible and treat the sexes equally.
NOTES
1. I would like to thank Sujata Ramachandran for her useful comments on an earlier version of this paper.
2. The literature is too voluminous to cite in its entirety. I am therefore restricting myself to some books: Bleier,
1984, Eichler, 1980, Eichler and Lapointe, 1985, Gergen, 1988, Harding, 1986, Keller, 1985, Lowe and
Hubbard, 1983, Sherman and Beck, 1979, Tancred-Sheriff, 1989, Tomm, 1989, Tomm and Hamilton, 1989,
Vickers, 1984.
3. The determination as to how many superordinate problems exist is partially determined on the basis of logic,
and partially a pragmatic decision, depending on how widespread a problem is. In my earlier work, I chose to
deal with four superordinate problems (androcentricity, gender insensitivity, overgeneralisation, and double
standards), and three particularly important subsidiary ones (Eichler, 1988). On the basis of work that I have
since done, I am here working with seven problems, in order to give sufficient visibility to the three
particularly important derived problems of familism, sex appropriateness, and sexual dichotomism.
4. It should be noted that this may be a problematic concept not solely because it is sexist, but also for other
reasons.
5. No true gynocentric framework can emerge in an overall sexist society that is male-dominated. However,
certain questions, particularly those concerning reproduction and family matters, are usually studied by
focusing on women only — e.g. fertility patterns. This is simply the reverse form of sexism than that involved
in an androcentric frame of reference.
REFERENCES
American-Canadian Co-operative
"Persantine Aspirin Trial in Cerebral Ischemia Part II: Endpoint
Study Group
Results", Stroke, Vol. 16, Number 3: 406-415.
1985
Bleier, Ruth
Science and Gender: A Critique of Biology and Its Theories on
1984
Women, New York: Pergamon Press.
Eichler, Margrit
The Double Standard. A Feminist Critique of Feminist Social Science,
1980
London: Croom and Helm.
1986
"The Relationship between Sexist, Nonsexist, Woman-Centered, and
Feminist Research", in Thelma McCormack (ed.) Studies in Commu-
nications (Vol.3), Greenwich, Conn. : Jai Press, pp.37-74.
1991
Nonsexist Research. A Practical Guide, New York: Routledge.
Eichler, Margrit and
On the Treatment of the Sexes in Research, Ottawa: Social Sciences
Jeanne Lapointe
and Humanities Research Council.
1985
Eichler, Margrit, Anna L. Reisman
"Gender Bias in Medical Research", in E.D. Rothblum, E.C. Cole,
and Elaine F. Borins
E.F. Borins and G. Feldberg (eds.), Gender, Health and Research,
Binghampton, N.Y.: forthcoming.
Fields, William S. et. al.
"Controlled Trial of Aspirin in Cerebral Ischemia", Stroke, Vol. 8:
1977
301-316.

Nonsexist Research 341
Gergen, Mary McCanney (ed)
Feminist Thought and the Structure of Knowledge, New York : New
1988
York University Press.
Harding, Sandra
The Science Question in Feminism, Ithaca : Cornell University Press.
1986
Keller, Evelyn Fox
Reflections on Gender and Science, New Haven : Yale University
1985
Press.
Lowe, Marian and
Woman's Nature, Rationalizations of Inequality, New York
Ruth Hubbard
Pergamon Press.
1983
Morgan, Robin (ed.)
Sisterhood is Global: The International Women's Movement
1984
Anthology, Garden City, N.Y.: Anchor Books.
Setty, Desingu E.
"Developing Entrepreneurship in Remote Rural-Tribal
1990
Communities", Indian Journal of Social Work, Vol. 51, Number 1:
87-98.
Shaw. R. Paul
"Humanity's Propensity to Warfare: A Sociobiological Perspective",
1985
in Canadian Review of Sociology and Anthropology, Vol.22, Number
2: 158-183.
Sherman, Julia A. and
The Prism of Sex: Essays in the Sociology of Knowledge, Madison,
Evelyn Torton Beck
Wisc: University of Wisconsin Press.
1979
Shiva, Vandana
Staying Alive: Women, Ecology and Development, London: Zed
1989
Book.
Tancred-Sheriff, Peta
Feminist Research: Prospect and Retrospect, Kingston:
1988
McGill-Queen's University Press.
Tomm, Winnie (ed.)
The Effects of Feminist Approaches on Research Methodologies,
1989
Waterloo: Wilfrid Laurier University Press.
Tomm, Winnifred and
Gender Bias: The Pervasive Prejudice, Waterloo: Wilfrid Laurier
Gordon Hamilton (eds.)
University Press.
1988
Vickers, Jill McCalla (ed.)
Taking Sex Into Account: The Policy Consequences of Sexist
1984
Research, Ottawa : Carleton University Press.
Waring, Marilyn
If Women Counted: A New Feminist Economics, New York: Harper
1988
and Row.
Williams, Karen and
"Sexism in a Leading Medical Journal: A Quantitative Measure",
Elaine F. Borins
Journal of American Medical Women's Association, forthcoming.