In May 1988 I graduated from Loyola College with an undergraduate degree in mathematics. Shortly thereafter I began working at the Bureau of the Census as a mathematical statistician. I do not work on the Decennial Census; rather I work on a few of the ongoing surveys that the Census Bureau conducts.
The survey on which I currently work is the Consumer Expenditures Survey. This survey gathers information from individuals pertaining to items that they purchase, everything from toothpaste to cars. After the data are collected, they are sent to another agency to produce the Consumer Price Index (CPI).
My initial question was, “So how does statistics come into play?” Believe me, it does. The first year opened up a new dictionary of terms: sampling, clustering, variance estimation, and more. It quickly became apparent that there was an entire world of statistics beyond what I had previously experienced.
As a mathematical statistician, my main responsibility is to ensure that the sample selected is representative, so that the “best” data are obtained. This procedure includes projects in sample selection, tracking non-response, testing new questionnaires, and computing variances on the data collected.
We are always researching new ways of collecting accurate data for less money. This sort of cost modeling was the idea behind a very interesting project which I ultimately presented at an American Statistical Association meeting. I developed, with input from many colleagues, a computer simulation that mimics interviewer travel in a geographic region.