Community Data Center Banner

Home >> Pre Katrina Home >> Articles >> Are Census numbers accurate?

Are Census numbers accurate?

The Census undercounts low-income people, children and minorities. But the Census is still one of the most important sources for data when you want to convince a funder about the need for your nonprofit's services.

by Allison Plyer

Nov. 26, 2001 | The U.S. Census counted 281,000,000 people in 2000. To accomplish this huge task, the Census hired some 860,000 temporary workers and spent more than $6 billion in total.

Despite their best efforts, the Census Bureau estimates that the 2000 Census missed 6.4 million people (1). Although this is a relatively small number of people (only 2% of the population), most of those not counted are minorities, children, and low-income people. There are many reasons why people might not get counted in the Census, including: privacy concerns, homelessness, low literacy levels and not enough time to fill out the forms.

The Census also estimates that they counted 3.1 million people twice (1). Most of these people were white and affluent. One reason for them being counted twice is that they may have received two Census forms to fill out because they own two homes.

Here's how estimates of the undercounts and overcounts look compared to the official count:

Census overcount and undercounts

The full analysis of the impact in Louisiana of the 2000 undercount is not out yet, but it's useful to look at what we've learned from 1990 (7):
Louisiana minorities undercounted in 1990

In the 1990 Census, Louisiana had the tenth highest rate of undercounting in the nation.

It's estimated that 94,112 people in Louisiana were not counted in the 1990 census. 61% of the uncounted people were minorities (mostly African Americans).

But only 34% of the state's population was minorities, which means that more minorities than whites were left out of the count in Louisiana in 1990.


Louisiana children undercounted in 1990

Louisiana was among the top 5 states for highest rate of undercounted children in the 1990 Census. And among cities, New Orleans was in the top 6 for worst rate of undercounted children in 1990.

The Census Bureau estimates that 9,103 New Orleans children were not counted in the 1990 Census.

Children are more often undercounted than adults for several reasons:

1) Single parents have less time for completing forms.

2) Children may be in foster care or homeless.

3) Census forms only provide enough space for 6 household members, which means larger families cannot report all their children.

Why does the Census undercount matter to nonprofit organizations?

The 6.4 million or so people who were not counted by the Census are also the type of people who are often recipients of nonprofit services (1). This means that the undercount is of particular concern for nonprofit organizations in writing needs assessments for grants.

In a single year (1998) more than $180 billion dollars in federal grants was distributed to the states based on Census counts. These formula grants help pay for education, employment training, public health promotion, housing, substance abuse treatment and environmental cleanup among other things. The U.S. Census Monitoring Board estimated that 59,063 people in Louisiana were not counted in the 2000 Census. And they estimate that because of Louisiana's undercount, our state will lose from $4 million to $6 million per year in federal funds. In the table below, you can see an estimate of the number of people not counted in the 5-parish area surrounding Orleans (2).

Net Undercount for 2000 Census by Parish

  St. Bernard
  St. Tammany
5-parish area

Source: U.S. Census Monitoring Board


What is the Census doing about the undercount?

Since 1940, the Census Bureau has acknowledged that its counts have been low and that minorities are more often undercounted than whites. The Census Bureau has been working hard to remedy this situation. For example, in preparation for the 2000 Census, the Bureau launched a paid advertising campaign to encourage minorities to fill out the Census form, and to help them overcome their fears about the Census. The Census has succeeded in reducing the net amount by which it undercounts the population (undercount minus overcount) from 5.4% of the total population in 1940 down to about 1% in 2000 (5).

Why should nonprofits rely on Census data?

Funding agencies, particularly Federal ones, make their decisions based at least partly on what the Census data reveals about different neighborhoods and cities. Every time a nonprofit applies for a grant, they should consider including Census information in their proposal because this is information funders often rely on to make funding decisions.

Here are a couple of local examples of how Census data might be used:

Example in Algiers Point

For example, if you are writing a proposal to HUD for funds to rehab vacant houses in Algiers Point, then HUD would likely want to see data that compares the percent of housing units that are vacant in Algiers Point, with the percent of housing units that are vacant nationwide.

Housing in Algiers Point (2000)

Algiers Point
Orleans Parish
United States
Housing units
Vacant housing units (%)

Source: U.S. Census 2000

This data sends a powerful message that Algiers Point has a high rate of vacant houses compared to the national average.

Example in Central City

If you were writing a proposal to start a childcare program in Central City targeting single moms, you might present Census data like this:

Children Living with Mother Only in Central City (2000)

Central City
Orleans Parish
United States
Children living with mother only (%)

Source: U.S. Census 2000

This data points out that Central City has a very high percentage of children living with single mothers compared to the state and parish averages. It suggests that childcare funding for single moms should be directed toward neighborhoods like Central City.

Census History

At the forming of the U.S. Constitution, our leaders decided that a head-count of the population should be taken every ten years to determine how many members of Congress each state will get. At that time, everyone counted except for slaves who were counted as three-fifths of a person, and Indians who were not counted because they paid no taxes (5).

After the Civil War, the Constitution was amended so that all people would be counted (excluding Indians). In 1999, the Supreme Court ruled that statistical sampling could not be used to redistribute Congressional seats because it would be unconstitutional. And in October 2001, the Census Bureau announced that it would not release statistically adjusted census information as a basis for the federal government to disperse funds, because there were too many errors in the data (4).

However, many argue that statistical sampling would provide a better estimate of the minority populations that are consistently undercounted. This issue has been the source of much debate among U.S. government officials in recent years (4). And as recently as October 9, 2002, a federal appeals court ruled that the Census Bureau had to release its statistically adjusted counts (8).

If the Census isn't accurate, wouldn't funders rather see nonprofits gather their own data?

Funders typically don't want nonprofits to collect their own data. There are several reasons for this:

  1. Funders usually think data collected by nonprofits is less reliable than Census data. Why? Because it's very difficult to gather data accurately. It's hard to get people to respond to surveys at all, and even if you can get people to answer your survey, it's hard to phrase questions so that people will provide the information you're looking for.

  2. Funders cannot compare data gathered by nonprofits to national data. Why? Because a nonprofit's data collection method has to be exactly the same as the Census' method or another national surveys' method to make results comparable.

  3. It's expensive and time consuming to collect your own data. Funders would rather see nonprofits rely on data that already exists so nonprofits can put their energies toward doing what they do best, serving people.

  4. Many folks in your target community may already feel like there has been too much research and not enough real change resulting from it. Being asked to respond to survey after survey magnifies this feeling in the community.

For these reasons and more, funders consider the Census to be one of the most trustworthy and appropriate sources for data to help them decide where their investment is most needed.


(1) An Evaluation of the 2000 Census
(requires Acrobat Reader)
This is from the U.S. Census’ Monitoring Board final report to Congress Sept 1, 2001.

(2 )Formula Grants: Affects of Adjusted Population Counts on Federal Funding to States (requires Acrobat Reader)
This is the February 1999 U.S. General Accounting Office's report on Formula Grants and the Effects of Adjusted Population Counts on Federal Funding to States. It refers primarily to adjusting the 1990 (not 2000) Census for the undercount.

(3) The Overlooked Undercount: Children Missed in the Decennial Census (requires Acrobat Reader)
This report from the Annie E. Casey Foundation presents issues and trends, and provides detailed statistics, related to the children missed in the 1990 Census.

(4) Presidential Members Report: Implications for Minority Voters in 2001
The U.S. Census Monitoring Board's report on the 1990 undercount and its demographic breakdown in Louisiana.

(5) Census History and 20th-Century Firsts
This site provides an easy-to-read history of the Census.

(6) Two Papers on Undercount Adjustment (requires Acrobat Reader)
"Straining Out Gnats & Swallowing Camels: the perils of adjusting for Census undercounts" and "Quantifying Measurement Error and Bias in the 1990 Undercount Estimates" are detailed and informative articles.

(7) A Social Justice Network
This site provides a great deal of information about the amount of undercounting in each state as well as the importance of the Census, the importance of undercounting, and how statistical sampling might reduce the problems associated with undercounting.

(8) Appeals Panel Tells Census Bureau to Release Adjusted Figures
A federal appeals court has ruled that the Census Bureau must release its statistically adjusted count for every state, county and neighborhood in the country, a decision that could affect how billions in government money is distributed.

Other Resources:

Census 2000 Undercount Could Cost States Billions
The U.S. Census Monitoring Board's report on the amount of federal grant money states will lose because of the 2000 Census undercount.

Census must release estimates of 2000 undercount
A federal appeals court has ruled that the Census Bureau must release its internal estimates of how many people were missed when the U.S. population was counted in 2000 -- a decision that could affect how billions of dollars in government money is distributed.

2000 Census Miss 1M Black, Latino Children,1282,-2226045,00.html
Black and Hispanic children, particularly those in urban areas, make up a disproportionately high percentage of the 1.1 million kids missed by the 2000 census, according to newly released government estimates.

Allison Plyer is a Senior Consultant with the Community Data Center.

Home >> Pre Katrina Home >> Articles >> Are Census numbers accurate?

tiny logo   

Greater New Orleans Community Data Center [email protected]
Copyright © 2000-2001 Greater New Orleans Community Data Center. All Rights Reserved.

Last modified: February 19, 2003