What a corner survey can count
Day labour is hard to count because it does not stay still. Workers assemble before dawn at a kerb, a car park or a hardware store entrance, are hired by strangers in vehicles, and are gone by mid-morning. There is no employer list, no payroll and no register. Any figure attached to this kind of work is therefore a figure produced by a survey design, and the design decides a great deal.
The study
The most widely cited attempt remains the National Day Labor Study, published in 2006 as On the Corner: Day Labor in the United States by Abel Valenzuela Jr., Nik Theodore, Edwin Meléndez and Ana Luz Gonzalez. Fieldwork ran through the summer of 2004. Researchers first canvassed metropolitan areas to locate hiring sites, then drew a sample of sites and a sample of workers within them, producing interviews with roughly 2,660 workers at some 264 sites across twenty states and the District of Columbia.
The results described a labour market with recognisable structure rather than an unorganised crowd. Hiring sites clustered where residential construction and landscaping demand overlapped with settlement patterns of recent migrants. Some sites were run as organised hiring centres with rules and a queue; most were informal kerbside gatherings with neither. Non-payment and underpayment were reported by a large share of respondents for the weeks preceding the interview — an unusually direct measurement of a phenomenon that rarely appears in official statistics at all, since a worker with no contract and no pay stub has little to report to anyone.
What the design fixed in advance
The population studied was defined by the method. A survey conducted at visible outdoor hiring sites can only describe people who look for work at visible outdoor hiring sites. Workers hired by telephone through a foreman they know, workers who pass through temporary help agencies, and workers in long-running informal arrangements with the same household or contractor are all doing short-term hired work, and none of them stand on a corner. Whether they resemble the people who do is an open question the study could not answer, and reasonable readers have taken it in opposite directions.
Time of day cuts the same way. Hiring peaks early; the workers picked up first are, by construction, absent when interviewers do their rounds. Whoever remains at nine in the morning is disproportionately whoever has not been hired. Field teams worked across hours to reduce this, but the direction of the remaining bias is not obscure: a sample taken from those still waiting will tend to describe a harder market than the average worker experienced that day.
The earnings and non-payment figures rest on recall, without documents, over a window of weeks, in interviews conducted by strangers with obvious sympathies. Recall error in wage reporting is well documented in survey research generally, and there is no reason to assume it vanishes here. One further limit applies to the national headline figure — the estimate of how many people seek day labour on a given day — which was built by extrapolating site counts beyond the areas actually canvassed, which makes it the least robust number in a study whose local observations are its strongest part.
Why it still matters
None of this makes the study weak. It makes it a survey, with the properties surveys have. The useful reading is narrow and specific: at hiring sites of this kind, in that summer, workers reported these conditions. Stretched into a statement about short-term hired work in general, the same numbers become something the design never supported. Much of the subsequent literature has been an argument about exactly how far that stretch can go.
References
- Valenzuela Jr., A., Theodore, N., Meléndez, E. and Gonzalez, A. L. (2006). On the Corner: Day Labor in the United States. UCLA Center for the Study of Urban Poverty.
- Valenzuela Jr., A. (2003). Day labor work. Annual Review of Sociology, 29.
Corrections to this article can be sent to support@daywork.pro with the passage in question.