August 11, 2013

Chilean census problems

A neat bit of snark from the New York Times

“I humbly apologize to Chileans for those mistakes,” Mr. Piñera said in a statement Thursday. “When I was informed of these errors, I felt as upset and outraged as millions of Chileans throughout the country.”

It is not clear, however, how many millions of Chileans there are.

Last year, the Chilean census went from a one-day event run by volunteers to a much more ambitious operation taking months, but missing a lot more people — the census review commission (PDF, Spanish)estimates that more than 9% were missed. Worse, based on the projections from other data, the proportion of people missed seems to have varied by age and gender (and so presumably by other factors as well).  The review commission recommended using the data as little as possible, and trying to run a new census in 2015 for a subset of the information. For the next full census, in 2022, they recommended using adminstrative records, online forms, and other approaches to reduce the need to visit every household in person.

The commission also had fairly strong recommendations about the national statistical office

se estima indispensable que el INE fortalezca sus competencias, recursos, atribuciones y niveles de autonomía para que se convierta en una oficina de estadísticas de excelencia, propia del nivel de desarrollo que ha logrado el país. también se requiere que sea una institución abierta al escrutinio público, disponible para la participación de otras instancias públicas y privadas en las actividades que les competan, y que las estadísticas producidas se transformen en bienes públicos (las bases de datos y documentación asociada estén disponibles a todo público sin cargos financieros y en forma expedita, bajo los resguardos de confidencialidad necesarios).

or in English, that they improve their skills, resources, and competencies; become more transparent and independent; and make results publicly available for free in a useful format.

This is probably worse than the Canadian census debacle, though at least this one wasn’t deliberate. 

(via Luis Apiolaza)

avatar

Thomas Lumley (@tslumley) is Professor of Biostatistics at the University of Auckland. His research interests include semiparametric models, survey sampling, statistical computing, foundations of statistics, and whatever methodological problems his medical collaborators come up with. He also blogs at Biased and Inefficient See all posts by Thomas Lumley »