My Two Census

Formerly the non-partisan watchdog of the 2010 US Census, and currently an opinion blog that covers all things political, media, foreign policy, globalization, and culture…but sometimes returning to its census/demographics roots.

Posts Tagged ‘mail response rates’

Fact Check: Is the mail participation rate a valid tool for tracking responses? Not until the following questions are answered.

Monday, April 26th, 2010

On Friday, the Commerce Department released a statement, “U.S. Commerce Secretary Gary Locke today congratulated the nation for its strong participation in the 2010 Census to date, as the Census Bureau released the latest mail participation data showing that 72 percent of U.S. households have mailed back their 2010 Census forms so far — the same rate the nation achieved at the end of the mail-back period during the 2000 Census.”

But what validity does this have? None, until the Census Bureau answers the following essential questions:

What data are used to adjust the mail response rates?  Who in the Postal Service supplies these data? To meet what specifications?  What distinguishes between unoccupied housing and Census address list errors? At what level of geographic detail?  The Census Bureau has stated the “participation rate” is “fairer”?  How is fairness defined?   Does the Postal service guarantee data consistency between and among all postal delivery service areas of the country? Or, are there big differences in what is returned to Census as undeliverable based on the quality of the address list used by the Census Bureau in each postal service area ? How does the Postal service distinguish between a bad address from the Census Bureau and a vacant house?  How does any of this get calculated in dense urban areas…..especially given the statement from the Census Bureau about “fairness” (For example, it is well known that delivery methods in multi-unit urban dwellings differ dramatically from suburban, single family residences — how does the proclaimed Census 2010 ”fairness” doctrine adjust for this)? When will the mail return rates for 2010 be calculated and how will this process differ from 2000?

D’Vera Cohn of the Pew Research Center (who covered the 2000 Census for the Washington Post) has tried to explain this process:

For the 2010 Census, the Census Bureau will use a new real-time metric, called the “mail participation rate,” to report the share of U.S. households-by state, city, county and neighborhood-that send back their completed forms. Why is this important?

The Census Bureau hopes to count every American in the coming months, but it has a hefty financial incentive to count them quickly. Census forms arrive in most home mailboxes next week. If a household sends back its postage-paid census form, the government spends less than 50 cents in mailing costs. If the completed form does not arrive back by late April, the Census Bureau will send an enumerator to knock on the non-respondent’s door, which costs $57.

As part of its promotional campaign to encourage households to return their forms fast, the Census Bureau plans to release mail participation rates down to the neighborhood level each weekday, from March 22 to April 26. Knowing where the problems are could help the bureau and its partner organizations—such as local governments and community groups—steer their census-encouragement efforts to the areas that could benefit most.

The 2010 mail participation rates will be displayed daily on a recently launched Census Bureau mapping tool, where users already can see 2000 data for states, counties, cities and census tracts (neighborhood-level units of about 4,000 people). For the Bureau’s publicity campaign, the mail participation rate replaces the “mail response rate” used in the 2000 Census because, for reasons described below, officials believe the new measure will give a truer picture in places with large numbers of foreclosed and vacant homes.

Three Different Mailback Rates

The mail response rate, the mail participation rate and a third measure of response, the “mail return rate,” are calculated for areas where household residents are asked to mail back forms that were mailed to their homes or dropped off by a census worker. These areas include almost all of the nation’s more than 130 million households.

The mail response rate is an unrefined measure —the percent of forms sent to households in these mailback areas that are returned to the Census Bureau. It is a preliminary measure that Census officials say somewhat understates participation, though, because many forms sent out by the Bureau cannot be mailed back — for example, those sent to vacant housing units and those where census forms could not be delivered, such as non-existent or non-residential addresses. In 2000, the final national mail response rate was 67%. (The initial mail response rate, over the first few weeks, was 65%.)

The mail participation rate is a refined version of the mail response rate–the percent of forms sent to households in these mailback areas that are returned to the Census Bureau,  after removing from the denominator addresses where census forms are determined by the U.S. Postal Service to be “undeliverable as addressed.” Nationally, the final census mail participation rate was 72% in 2000.

The mail participation rate is intended to exclude vacant and foreclosed homes, which have grown in number as a result of the national economic downturn. The mail participation rate also may provide an improved real-time measure of participation for areas with large numbers of seasonal homes that are unoccupied on Census Day, April 1.

However, the new metric will not eliminate all sources of error. For example, if the owner of a vacant or seasonal home has a friend who picks up the mail, the form may not be returned as undeliverable. Some forms may be sent to home addresses whose occupants get their mail from postal boxes, and those forms may be returned by the Postal Service as undeliverable even though the home is occupied. These kinds of addresses will be on the Census Bureau’s to-do list, however, and census-takers would make sure they are properly accounted for during follow-up visits, according to Census Bureau officials.

The mail return rate, the most precise measure of census participation, is the number of households returning a questionnaire from mailback areas mail divided by the number of occupied housing units that received questionnaires in those areas. It cannot be calculated until the end of the census counting process. At that point, officials will use data from census-takers’ follow-up visits and other sources to total the number of occupied home addresses in areas where residents mail back their forms. Once addresses are excluded from the denominator—mainly for being unoccupied, non-residential or non-existent—the rate will rise. In 2000, the mail return rate was 78%.