Guide on Gender Analysis of Census Data Full Draft of 6 December 2012 Contents
Chapter 7: Households and Families
Download 4.06 Mb.
|
- Navigate this page:
- 2. Why is it important
- Country Examples 10: Co-Resident Domestic Servants in Kuwait and Colombia
- Country Example 11: The Changing Nature of Household Composition and its Link to Changes in Gender Roles in France
- 3. Data issues
Chapter 7:
|
|
Country Examples 10: Co-Resident Domestic Servants in Kuwait and Colombia In some countries, such as Kuwait, co-resident domestic servants correspond to a considerable proportion of household members, as indicated by the study conducted by Shah et al. (2002). Using data from a nationally representative survey on households, they found 17.3 per cent of the 14,835 individuals residing in the investigated households were domestic servants who were unrelated to the Kuwaiti residents. The prevalence of co-resident domestic servants was particularly high in households with elderly persons; about 90 per cent of all households with an older adult had at least one co-resident domestic servant. These numbers, however, are exceptionally high and owe a lot to the availability of cheap domestic servants from abroad. By comparison, in the 2005 census of Colombia less than 1 per cent (i.e. only 0.40 per cent) of all household residents were found to be co-resident domestic servants. The percentage was somewhat higher in larger households, with a maximum of 0.55 per cent in households with 5 members and a minimum of 0.30 per cent in households with only 2 members. |
278. Finally, analyses based on different time measurements, such as comparisons between successive censuses, may not only indicate changes in the social structure of the households but also changes in gender relations. Data analyses indicate, for instance, that increases in the number of women living alone are related to increasing numbers of female university students and female workers. This was noticed in urban areas of several, -especially- developed countries. In line with this, changes in household size may also be associated with increased access to sexual and reproductive services, which are recognized to be crucial for the promotion of gender equality (UNFPA/IPEA, 2007).
|
Country Example 11: The Changing Nature of Household Composition and its Link to Changes in Gender Roles in France Ogden and Hall (2004) studied household change in France during the 1990s, using data from the 1999 French population census and comparing them with data from earlier censuses. Their results indicate a decline in the mean household size and a rise in the number and proportion of people living alone. These observations may be associated, among other factors, with changes in gender roles and sexual norms (e.g. fertility trends, feminization of the labour force, growing student population). The tendency for women to be heads of one-parent households goes along with the fact that men during young adulthood and early middle age (25-49 years) are more likely to live alone. Nevertheless, the rise in the number of young women in their twenties and thirties who live alone is also remarkable. From 1975 to 1999 the number of men aged 25-39 who were living alone rose by a factor of 1.7, while the number of women living alone rose by a factor of 2.0. At older ages, the overwhelming predominance of women in one-person households was found to be due to earlier male deaths. Of the 1.7 million people living alone over the age of 75 in 1999, 81 per cent were women. Another finding was that of the 57 million people recorded as living within households in the 1999 census, just a little over 30 million were living as ‘traditional’ families (i.e. as a couple with children), whereas almost 27 million did not fit that description. |
3. Data issues
279. The use of census data for gender analysis has both strengths and weaknesses with regards to household and family composition and structure. In terms of strengths, censuses are perhaps the most complete source of information on household and family composition, because they enumerate the highest number of people in a population or country. Related to this, due to the universality of coverage, family structures can be divided into much finer categories than would be possible with most surveys. Also, because of the same universality, the census is less susceptible to the sampling biases that may affect surveys such as the Demographic Health Surveys (DHS), which are designed to capture women living in unions. Finally, another advantage is that censuses allow for international comparability of household and family composition on a regular basis for gender issues when the internationally agreed upon recommendations on definitions and classifications are used (e.g. the Principles and Recommendations).
280. Censuses collect data by enumerating all the individuals that normally live in households (i.e. de iure census) or that happen to be there at the time of enumeration (i.e. de facto census).21 In addition, the census collects data on the physical structure of living quarters and on the people living in institutional or collective-household arrangements, such as army barracks, prisons, hotels or pensions or on the street. Among the weaknesses, differences in methodologies of census taking, notably between de facto and de iure censuses, may affect the definition of a household and who is considered to be a member of it or not. In a strictly de facto census any person who did not spend the night of the census with the household would be considered to be a member of another household. In de iure censuses absentee household members continue to be considered as part of the household for up to six months. Interviewers often have difficulties with household membership in the face of recurring absences of persons that may stretch over several years (e.g. students). Another limitation of census data for gender analyses is that the relationships between members are usually only described with respect to the head of household or reference person. This limits the extent to which all the relationships between household members can be mapped. For instance, in a household where there are two daughters of the reference person, each in their twenties, and one man described as “son-in-law,” it may not be possible to determine to whom this son-in-law is married. The marital status of the daughters may give a clue, but this will not always work, especially if the marriage is consensual.
281. To organize household data, almost all censuses require that one of the household members be selected as the “head of household” or “reference person.” There are a number of reasons why this is done:
-
To assign a person as primary informant, responsible for providing information on all household members, including those who are not present at the time of the interview; -
To detect the relations within the household, i.e. to identify family nuclei, and describe the household structure (one-person household, nuclear household, extended or composite household)); -
As a key informant, to ensure that all household members are enumerated ; -
As a source person to check information pertaining to other household members; and -
To “summarize” the characteristics of the household by identifying them with those of the head of household, e.g. to analyse the living conditions of household members according to the occupation or the income of the head or reference person. This last use of the head of household concept is quite controversial, because there is no guarantee that the characteristics of the head of household or the reference person will adequately represent those of the household.
282. Using headship of household - female against male headed households - to analyse gender issues is hampered by a number of theoretical shortcomings and practical problems. Even though statistics differentiated by female or male household head can be useful for defining and planning policies for vulnerable population segments, statisticians and gender researchers note that this type of analysis may have its limitations, and these limitations should be taken into account when interpreting data results and planning policies. The following describes five problems related to the use of female/male headship to study gender differences.
-
The first problem lies in the fact that the definition and operationalization of the notion ‘head of household’ is vague and in no way uniform. Contrary to many other variables that are strictly defined, the Principles and Recommendations for Population and Housing Censuses, Rev. 2 (United Nations, 2008 a), leave the definition and appointment of the head of household wide open: ‘Countries may use the term they deem most appropriate to identify this person (household reference person, head of household, householder, among others) as long as the person so identified is used solely to determine relationships between household members. It is recommended that each country present, in published reports, the concepts and definitions that are used’. Appendix 1 illustrates the variety of definitions of head of household that a number of countries used in the 2010 round of censuses. As Hedman et al. (1996: 6) rightly state, the term ‘head of household’ “is used to cover a number of different concepts referring to the chief economic provider, the chief decision maker, the person designated by other members as the head”. Despite the diversity of approaches that may be assumed, questionnaires often do not properly inform which of the term of household head is specifically referred to. At least five different concepts of head of household can be found in censuses: 1) main breadwinner, 2) householder, 3) main authority, 4) reference person, and 5) questionnaire respondent. These concepts are discussed below in points a) through e), also the relative advantages and disadvantages of each possibility are discussed.
a) Main breadwinner. This economic approach to household headship may be useful for gender analysis if clear definitions are presented in the census taking process and in data tabulation and dissemination. It identifies the primary person responsible for the economic maintenance of the household, so data could be presented on the main contributors to the household’s economy, by sex. Further cross-tabulations could include marital status, family composition, owning or renting house, number of family nuclei, and other socio-economic characteristics of the head of the household (e.g. job sector). One country that used the notion ‘main breadwinner’ was Anguilla in the 2001 census, in which the head of the household was the major economic provider. Using the breadwinner metric allows the analysis to avoid under-representing women’s household contributions in practice, as may be the case if relying on authority and traditional gender stereotypes. However, one disadvantage to the breadwinner option is that the question about an economic head may not be adequate for societies where spouses share economic responsibilities more or less equally. In such cases, the Principles and Recommendations recommend the use of the term ‘reference person,’ with no implication of headship or, alternatively, that provision be made for designation of joint headship. A second disadvantage of using the breadwinner definition is that a gender bias may underlie this definition as it does not capture women’s hidden household economic contributions; as female labour is often unpaid, they may not be defined as “breadwinner” even in cases where the market value of their labour exceeds that of their male partners or if her activities are the main source of family livelihood. Additionally, female respondents may answer that they have no ‘occupation’ or ‘work’ although they have been working on craft activities at home or in agriculture. Also, using the breadwinner or economic head for identifying economic conditions in the household may be a pitfall, because household members may not equally share resources and may present different socioeconomic characteristics (Hedman et al, 1996). Finally, ownership of resources does not necessarily imply access to and use of resources (United Nations, 2008).
b) Householder. This term refers to the person who owns or pays the rent of the home, and may be defined as the person in whose name the household dwelling is registered, adding that “this approach is more objective than household head and may relate in some ways to power relationships in the household” (UNECE/World Bank 2010: 16). Australia, New Zealand and American Samoa use some of the countries that use the concept of householder. In terms of advantages, the householder option is similar to the breadwinner criterion, but offers greater specificity. In terms of disadvantages, the householder concept may be ambiguous in the context of developing countries, where homes are often rented without formal contracts, where occupants pay no rent at all, or where people live in homes that have been temporarily borrowed or that are makeshift temporary accommodations. Also, the person paying the rent may not always be the one earning the money with which the rent is paid.
-
Main authority. This approach identifies the person that the other household members recognize as the main authority for all sorts of household decisions. This approach has become quite popular, particularly because it is viewed as an indicator of decision-making (UNECE/ World Bank, 2010). In terms of advantages, the main authority approach can provide a measure (i.e. using sex ratios) on the main authority in the household. This can also be used in gender analysis to identify the ways in which power relations are featured in the household structure. In terms of disadvantages, the main authority approach carries the assumption that a hierarchical relationship exists between household members that may not represent reality. Related to this, authority and decision-making typically are not formal but continually negotiated processes. Further, in practice, the census taker may confuse the decision-maker with breadwinner, and decision-making may differ by the type of decision being made (UNECE/World Bank, 2010). Another limitation to the main authority approach is that authority, responsibilities and decision-making may be shared, especially considering the changes in gender and family relations in the last decades. Decision-making power may be shared by a couple or even several household members. Finally, household members may not have the same opinion on who is the main decision-maker or who exercises the authority to run the household. Moreover, because authority is defined as “perceived authority,” it may not be possible to come up with an objective criterion. The Brazilian country example below illustrates how the main authority approach can be used in practice and how some of these perceived disadvantages can be surmounted.
d) Reference person. The reference person approach implies selecting an adult household member at random. It has the advantage of making the process explicit in the sense that it does not allow any undue interpretations of what the headship criterion actually represents. Because it is random, it is not necessarily the main income earner, nor the person taking the major decisions regarding the household or the person that owns or pays the home. This concept focuses on the second function of the household head assignment, a reference for mapping family relationships, more than the third, a breadwinner or person in authority who can represent the household. The 1996 census of New Zealand, for example, did not use the term ‘household head,’ but replaced it by a sentence which referred to the person responsible for the answers. In the census form, it stated: “In every dwelling, one person must take responsibility for this form. An adult who lives here would be best, but any person can be the one who must: fill in this form; make sure that everyone fills in an individual form (the blue one); and have all the forms ready for the enumerator to collect.” The 2001 and 2006 census forms also used similar expressions. In some cases, this interpretation is made explicit by choosing the reference person randomly among the adult household members. This practice is currently used in Denmark, the UK, Canada, and several other countries. In terms of disadvantages, the reference person approach may generate confusion. Further, choosing a reference person randomly increases the chance that this person may not be related to the other household members, in which case, information on related persons and family nuclei will be lost.
e) Questionnaire respondent. This conceptual approach means that the ‘household head’ is the person responsible for answering the census form no matter what his/her status may be within the household. For instance, in the United Kingdom Population Census of 2011 a ‘householder’ is defined in order to indicate who is responsible for the completion of the questionnaire. It is not defined to produce any outputs as there is nothing on the questionnaire to identify this person. The advantages and disadvantages of the ‘questionnaire respondent’ are roughly the same as just discussed for the ‘reference person’ concept, i.e. the notion may be confusing and chances are that the person is not related to other members of the household.
This conceptual vagueness or ambiguity of the headship concept may lead to divergences between census workers’ assumptions and the responses given by household members. Each one will tend to define the head according to their own cultural representations and, naturally, there is a great risk that premises differ substantially from one another, or may not adequately reflect household realities. A census may supposedly use a “breadwinner” definition, but household members/enumerators may decide simply to put down the oldest male, without questioning the situation. As enumerators, respondents and analysts may not be talking about the same concept, and there is no guarantee of what is being measured. Without a clear definition, would two census workers define the same head? And would household members designate the same person as the head? Without adequate orientation and interview training, internal/external reliability becomes quite unlikely. Indeed, if the concept is vague or ambiguous, it is very likely that data and analysis will not adequately represent reality or provide accurate and consistent evidence for policy-making.
-
A second problem of using the female/male headship dichotomy in gender studies is that gender inequality may take place at the intra-household level (e.g. unequal distribution of earnings and consumption among members of the household). Therefore, focusing on female-headed households may not capture these inequalities and be misleading. The shortcoming of using female/male headship of household is well illustrated in a study on 11 Latin American countries by Deere, Alvarado and Twyman (2010). Drawing on the recent Living Standard Measurement Studies for Latin America and the Caribbean, they present baseline indicators of the degree of gender inequality in individual asset ownership. Disaggregated data on housing ownership suggests that the distribution of asset ownership by sex within households is much more equitable than a headship analysis would suggest. The authors estimate that in Nicaragua women own 36-41 per cent of household physical wealth. However, if the analysis of household wealth were conducted by sex of the head, female-headed households would own only 20 to 23 per cent of household wealth, significantly less than the share of female-headed households in that country. This different vision of relative female poverty is largely due to the fact that women in male-headed households often own property, either in their own right or as joint property with their spouses. As will be argued in more detail in the Chapter 8, the differences in poverty rates between male and female-headed households, if not broken down into finer categories, are typically small and tend to be associated with other demographic differences between these households, such as the number of children and adult male and female household members.
-
A third problem has to do with the limited possibilities for analysis and cross-country comparisons. Obviously, if the researcher is not sure of what the ‘household head’ variable of a given census is actually measuring, this may put into question how the results should be interpreted. Further confusion arises when census data on household headship from different countries are compared, with a high likelihood of putting side by side different measurements in cross-national comparisons.
Even when the concept is clear in the original questionnaire form, the definition may not be presented in census tables or other census reports. In this case, particularly when making use of secondary sources, analysis and comparisons must thus be conducted with caution to ensure that assumptions made by data users correspond to the respondents’ approach. Sometimes, misguidedly vocabulary changes also do take place. The original questionnaire form may have used the term ‘reference person’, while in tabulations or in analyses this was replaced by ‘household head’. This may lead to biased interpretations. In this regard, UNDESA warns that:
‘It is important to recognize that many countries use the concept of reference person in listing household members and that this person may or may not be the “household head”. Where this is the practice, the “household head” identified in tabulations is, in reality, the reference person and should be treated with caution’. (United Nations, 2005: 135)
-
Focusing on female-headed households may lead to biased policy priorities. In the words of Sylvia Chant (2003: 30), “Placing excessive emphasis on the economic disadvantage of female heads misrepresents and devalues their enormous efforts to overcome gender obstacles.” The ‘feminization of poverty’ thesis “...precludes an analytical consideration of the social dimensions of gender and poverty...” and “tends to translate into single-issue, single-group policy interventions.” These narrow policy interventions may in turn fail to affect and reshape the embedded structures of gender inequality found in the home, the labour market and other institutions.
-
A fifth consequence of the methodological bias is the reproduction or reinforcement of gender stereotypes. Stereotypes may not only be guiding questionnaire formulation and census execution, but also respondents’ answers, and the researcher’s analysis.
Participants of the Second Global Forum on Gender Statistics expressed concern that “using the conventional classification of household headship (i.e. whether household is female- or male-headed) implies a kind of hierarchy within the household that suggests subordination” (United Nations, 2009 b: 17). Questionnaires may also be designed under the assumption that gender norms naturally place men as the ‘head;’ in this case, it is quite possible that questions are formulated in a way that leaves little room for respondents to indicate a female-headed household. Even if the questionnaire was well formulated, the training of enumerators may not adequately address this topic, allowing for biased interview approaches.
As Budlender (1997: 4) points out, in many African cultures “the household head will be the oldest male,” adding that “in a multi-generational household, he will often not be the highest income-earner, or even control the resources of the household.” He is recognized as the ‘head’ in terms of respect, but it would be inappropriate to say that he is the economic head.
The First World Conference on Women (1975) had already pointed out, in its Plan of Action, the risk of methodological bias relating the operationalization of the concept of household head: “among other data biased by preconceptions are those on heads of households or families, when it is assumed that a woman can be the head only in the absence of a man” (Par. 164). These concerns were further incorporated in census manuals. The Principles and Recommendations (Pars. 2.118, 2.119), for instance, state that “the most common assumption that can distort the facts is that no woman can be the head of any household that also contains an adult male. Enumerators and even respondents may simply take such an assumption for granted”. It is also argued that this gender stereotype “often reflects circumstances that may have been true in the past but are true no longer, insofar as the household and economic roles of women are changing.” The Principles and Recommendations recommend, in their paragraph 2.119, that “It is important that clear instructions be provided as to who is to be treated as the head of the household so as to avoid the complications of enumerator or respondent preconceptions on the subject. The procedure to follow in identifying a head when the members of the household are unable to do so should be clear and unambiguous and should avoid sex-based bias.”
Nepal provides an example of a country which is currently making efforts to avoid gender bias regarding household headship. In the 2001 census questionnaire, illustrative examples in the census instruction manual were revised in order to move away from the traditional type of responses: “descriptive examples like ‘female or male’ in place of ‘male or female’ in earlier census manuals or the example of ‘female name’ as the household head in place of only ‘male name’ in earlier census manuals were included”. Additionally, women were shown as household heads in specific publicity materials (posters), which also presented women performing various non-marketed economic activities, such as “food processing, agricultural cropping, fetching water and fuel wood collection” (UNFPA, 2004: 17).
283. An illustrative example of divergent data outcomes regarding different definitions of household headship is found in a study based on the 1997 LSMS (Living Standard Measurement Study) data for Panama (Fuwa, 2000). Here, female-headed households were classified according to three criteria: self-reporting of a woman as the head; a “potential” female head, if no working-age male was present; and a “working head” definition, in cases where more than half of the total hours worked were contributed by a single female member. With respect to poverty rates, “the study showed that the overlap between these three sets of households was low, around 40-60 per cent”. Also, “the corresponding poverty rates were different: 29 per cent for the self-declared female-headed households; 23 per cent for the “potential” female-headed households; and 21 per cent for the households headed by a “working female” (United Nations, 2010 a: 164). In this case, different definitions in fact measured different social situations.
Download 4.06 Mb.
Share with your friends: