Gendering Peasant Household Surveys: Who to interview?



The use of household surveys to analyse the economic activities of peasants has become much more commonplace since Alain de Janvry and I published ‘A Conceptual Framework for the Empirical Analysis of Peasants’ in 1979. However, an issue that has received insufficient attention until recently is that of who should be interviewed in a household survey.

Most commonly, respondents to household surveys, as in the 1973 Cajamarca Household Income Survey that we drew upon, are household heads, the majority of whom tend to be male. Underlying this practice is the unlikely assumption that household heads have near perfect information on the activities of all household members.

Given the gendered division of labour, men and women often carry out different activities, with different responsibilities, networks, and relation to the market. What they are likely to know best and report most reliably are their own activities. We had sensed this, and in our follow-up survey in Cajamarca, the 1976 Peasant Family Survey, which focused on the gendered division of labour, we had interviewed only women. However, we failed to report this in the article, or to consider that all of our results might depend on who was interviewed.

Feminist economic analysis is increasingly demonstrating how the social construction of gender influences men and women’s perceptions of economic activity. Due to gender socialization, survey respondents sometimes report what is expected of men and women in gender-conforming roles, roles that may embody or reflect gendered power relations. For example, a study of household income in Malawi found that when interviewed separately, spouses disagreed over the amount of income generated by the women in 94 per cent of households. Overall, husbands tended to underestimate their wives’ income compared to the reporting by women.

In the national household asset survey that I directed in Ecuador in 2010, we found that, when interviewed separately, 31 per cent of couples disagreed over who owned their agricultural parcel.  Moreover, 70 per cent of the couples disagreed on the value of the same parcel, with men significantly more likely to provide a higher estimate than that given by their wives.  Couples also disagreed on women’s role in agricultural decision-making, even when the wife was the landowner or a joint owner. Compared to the reporting by wives, husbands tended to significantly underestimate their participation in four major decisions, what to plant, whether to purchase inputs, how much to sell, and how to use the income so generated.

If husbands consistently underestimate their wives’ income or assets, this could distort analyses of poverty and inequality, overestimating the former and potentially, exacerbating the latter. Similarly, if there is a tendency for men to exaggerate the value of what they earn or own and for women to be more circumspect in their estimates, this will influence the degree of estimated income or wealth inequality found, depending on who was interviewed in the household survey. In addition, questions on decision-making are often used as proxies for women’s empowerment in household bargaining models and analyses. If men and women have different perceptions of participation in decision-making as well as of outcomes, the results may very well depend upon who was interviewed in the survey.

With respect to class analysis, our 1979 article emphasized the multiple income-generating activities in which peasant households engage, facilitated by the gendered division of labour, and how each activity is subject to its own form of surplus extraction. It was only implicit, however, that the class position of peasantries could not be simply read off the occupation of the male household head and must be based on an analysis of the activities of each of its members. Hence, the importance of the issue of who to interview. An analysis based on a survey of predominantly male respondents will likely underestimate the full range of activities carried out by women, potentially underestimating the multiple class relations in which household members engage.

This discussion suggests the importance of interviewing more than one person in household surveys, ideally, both the husband and wife, and perhaps in some cases, interviewing them jointly. There is growing consensus that direct reporting is always preferable to reporting by proxy, a practice now relatively standard, for example, in labour force surveys. Having two or more responses to the same question regarding asset ownership or income-generating activities, of course, complicates the analysis. It requires a bit of ingenuity to develop the appropriate rules for the reconciliation of conflicting responses to produce household-level aggregate estimates of income or wealth. A gender-informed practice, however, should improve the reliability of these estimates. As a minimum, researchers need to report the gender distribution of survey respondents.


Carmen Diana Deere is Distinguished Professor Emerita of Latin American Studies and Food & Resource Economics at the University of Florida, USA. Since 2015, she is Distinguished Professor of the Latin American Faculty of Social Science (FLACSO-Ecuador) in Quito where she is teaching periodically. Carmen is also a member of the Journal of Agrarian Change’s International Advisory Board. 

Comment from Pauline Peters:

I welcome Carmen Diana Deere’s reminder of the serious mistake in interviewing only one person, usually the male ‘head of household’, in the surveys that still tend to provide the main sources of data on people’s economic and other activities. But it is disconcerting that this reminder has to be made in 2020 after at least fifty years of critique not only of interviewing one person – the male ‘head’, but also of taking ‘the household’ as a sole, discrete unit. Not only have these techniques obscured the role of gendered relations but more broadly the processes that produce and reproduce the identified ‘households’. In the 1980s, Jane Guyer and I, along with other feminist colleagues in several disciplines, pointed to these problems with special but not exclusive reference to Africa (Guyer 1981, Peters 1983, Guyer and Peters 1987). Overall, our conclusion for analyzing economic life was that, rather than assuming ‘the household’ to be the sole (and unitary) unit of analysis, one should identify ‘what are the units and processes of production, consumption and distribution’. In most African studies, households (understood as a conjugal unit though in some places they can be a sibling unit, hence some studies specify only the presence/absence of a man) have to be understood in relation to the encompassing units and networks that shape those processes.

A virtual issue of Development and Change in 2013 allowed Bridget O’Laughlin to revisit the problematic of ‘Conceptualizing households in rural Africa’. She started from the 1987 Guyer and Peters Introduction and the associated papers and took the arguments further. But even she had to conclude that there was still an ‘unsettled debate’ about the use of ‘the household’ in “the ways we now measure, compare and, most importantly, understand, poverty and well-being.” Perhaps we should spend time asking why this continues to be the case.

[Marxist Methods Corner: I]