One of the privileges of contributing to Development Impact blogs is the opportunity to use it as a platform to advocate for doing things differently, as a bit of a soapbox. Here I am going to revisit a topic I blogged about before, inspired by discussions I had last week with a great group of Econ graduate students.
Back in 2019, I blogged with Dominique van de Walle that comparing male and female headed households is a shotty approach to studying differential outcomes between men and women. Dominique also has a CGD blog on this here with Agnes Quisumbing. I am revisiting this topic, and specifically make a case for dropping “household head’ forever from future questionnaires.
Let’s start with the points related to the case for dropping the household head from surveys:
(1) Within surveys, it is almost always (if not always) not defined. Rather, it is left to judgement of which ever household member is responding, or to the survey enumerator. This raises concerns not only within a survey on its meaning across households, but about comparability across surveys. See this paper on South Africa by Michael Rogan.
A brief anecdote: While a graduate student, a female peer (who I shall not name) was doing field work in a low-income country (that I shall not name). At the time the peer shared lodging with a junior male graduate student. When the census enumerators came knocking, the junior colleague declared he was the head – of course he was, he is a guy. Her response was “wait, what, are you kidding?”
If one wants to know about who holds power or authority in households, who makes decisions on this, that, and the other, then insert such questions. Do not substitute a single, sloppy, and vague roster “who is head” question for such information.
Possibly (but I doubt it) there are contexts where one can make a case that it is a simple-yet-well-understood measure for that specific setting – sure, maybe, possibly. But unless you do the work to know that ahead of time, do not assume it.
(2) It is often used to draw inferences on gender (in)equality, and is a very flawed for that purpose --see the aforementioned blog. It may not be intended to be used as such, but people just can’t seem to help themselves when they get their hands on the data where the roster tells us neatly who the head is.
(3) We do not need this noisy measure to identify critical household types to inform policies and programs (for example, to target programs). A better approach is to look at the demographic and earnings profile of household members as done here and here. See also UN Women report on Families in a Changing World (Box 2.2 and elsewhere).
(4) Consider that it is obsolete in high income economy national surveys (at least the major ones that I checked). Harriet Presser, sociologist and demographer, as well as President of the Population Association of America^, wrote a lovely piece from 1998 called Decapitating the U.S. Census Bureau's "Head of Household". She outlines the history of the campaign and efforts that lead to the removal of the term in the US Census 1980 and then (mostly) onwards.
It’s 45 years later: time for consider updating our household roster approach in surveys in LICs/MICs?
….But, you may wonder, are there valid reasons to ask who is the head of the household in the household roster? Frankly, I am hard pressed to think if any reason to keep the concept in the roster, besides inertia. I have had people tell me it is too complex to otherwise identify relationships. I think this is not the case. One can easily replace “head” with the idea of “reference person” (who is not someone with any specific position necessarily) in that singular “relationship to” question, as is done in HIC surveys. And good household roster will also specifically identify couples and parent-child relations, among other links.
Others has said it is used to screen into other sections of the questionnaire. But this has not been the case in the surveys I have worked on (as far as I can recall), including large-scale LSMS surveys. Rather, some sections will select the “most knowledgeable household member” as a respondent but not assume the respondent is the head. Rather, the simple “head” information from the roster proves useless to the completion of the rest of the questionnaires.
Please let me know if I am missing some valid arguments for keeping our precious household head, which Presser notably described as “an ambiguous concept which implied an authority structure imputed… but not measured, and offensive to many people.”
^ Speaking of the PAA: I hope to see some of you at the upcoming annual meetings of the Population Association of America in Washington DC in April 2025. It is billed as “the premier conference of demographers and social and health scientists from the United States and abroad.”, and you can also find some development economists there. For those of you who work on “households” in developing country contexts, it is a great venue to learn from other social scientists and broaden your research view of the world to more than the work of fellow economists. The program is posted online.
Join the Conversation