Prepared by the North Dakota Kids Count Sept. 2010 1

Using the American Community Surveyfor Children’s Research

Dr. Richard RathgePolicy Analyst

North Dakota Kids Count

Kids Count Annual Meeting

Baltimore, Md

Sept. 24, 2010

Presentation Objectives:

1. Highlight some key challenges of using ACS for children’s research

2. Illustrate the need for a conceptual shift in the way we approach children’s research when using ACS data

3. Initiate dialogue of implications for research and policy

2Prepared by the North Dakota Kids

Count Sept. 2010

Key Challenges for children’s research

Effects of longer period estimates Consequences of smaller sample

size Rule changes and their impact on

trend analyses

3Prepared by the North Dakota Kids

Count Sept. 2010

Period Estimates … ACS vs. Census

Census (conceptually point in time April 1st)

ACS accumulates 12 months (1-year estimate) 36 months (3-year estimate) 60 months (5-year estimate)

Think of shutter speed on camera

4Prepared by the North Dakota Kids

Count Sept. 2010

5Prepared by the North Dakota Kids

Count Sept. 2010

Nebraska ACS Data

Thanks to Jerry Deichert for data

Omaha city was an ACS test sites: 1997-2004 Can examine 1, 3, 5-year data

6Prepared by the North Dakota Kids

Count Sept. 2010

7Prepared by the North Dakota Kids

Count Sept. 2010

8Prepared by the North Dakota Kids

Count Sept. 2010

9Prepared by the North Dakota Kids

Count Sept. 2010

10Prepared by the North Dakota Kids

Count Sept. 2010

11Prepared by the North Dakota Kids

Count Sept. 2010

Key Challenges for rural areas

Effects of longer period estimates Consequences of smaller

sample size Rule changes and their impact on

trend analyses

12Prepared by the North Dakota Kids

Count Sept. 2010

ACS sample size is smaller than Census

One-year sample is 2.5 % of HHs vs 16.7% for Census long form Over five-year period, ACS sample size

is 12.5 percent

Therefore ACS sampling error will be larger

13Prepared by the North Dakota Kids

Count Sept. 2010

Cass County, North Dakota (population 132,585)

14Prepared by the North Dakota Kids

Count Sept. 2010

People Moving to North Dakota Within the Past Year by State of Origin: 2006 ACS

-500

0

500

1000

1500

2000

2500

Mon

tana

Sou

th D

akot

a

Cal

iforn

ia

Col

orad

o

Ariz

ona

Mic

higa

n

Tex

as

Okl

ahom

a

Pen

nsyl

vani

a

Was

hing

ton

Ten

ness

ee

Mis

siss

ippi

Flo

rida

New

Yor

k

Neb

rask

a

Virg

inia

Idah

o

Wis

cons

in

Mis

sour

i

Wyo

min

g

Ore

gon

Ohi

o

Geo

rgia

Iow

a

Illin

ois

New

Mex

ico

Ken

tuck

y

Ala

ska

Nev

ada

Haw

aii

Loui

sian

a

Ala

bam

a

Mas

sach

uset

ts

Mar

ylan

d

Con

nect

icut

Sou

th C

arol

ina

Ark

ansa

s

Nor

th C

arol

ina

Kan

sas

Del

awar

e

Dis

tric

t of

Col

umbi

a

Indi

ana

Mai

ne

New

Ham

pshi

re

New

Jer

sey

Pue

rto

Ric

o

Rho

de I

slan

d

Uta

h

Ver

mon

t

Wes

t V

irgin

ia

State of Origin

Peo

ple

Mo

vin

g to

No

rth

Dak

ota

Comparison of 1-year, 3-year, and 5-year Estimates of the Percentage of Persons in Poverty for Omaha, NE: 1999-2005 ACS

1999 2000 2001 2002 2003 2004 2005

Single-year estimate

11.4 12.0 11.0 12.3 13.5 13.9 15.3

Single-year MOE

0.8 1.0 1.0 1.5 1.6 1.5 1.5

Three-year estimate

11.4 11.7 12.1 13.2 14.2

Three-year MOE

0.5 0.7 0.8 0.8 0.8

Five-year estimate

11.9 12.5 13.2

Five-year MOE 0.5 0.5 0.6

Suppression Issues

ACS using suppression similar to the 1970 and 1980 decennial census

Applied to base tables in 1 and 3-year data

Use “data release rules” to protect user from tables “whose reliability is unacceptable”

Suppresses entire table not just unreliable cells

17Prepared by the North Dakota Kids

Count Sept. 2010

41 cells which willcause suppressionfor smaller counties

18Prepared by the North Dakota Kids

Count Sept. 2010

Key Challenges for rural areas

Effects of longer period estimates Consequences of smaller sample

size Rule changes and their impact

on trend analyses

19Prepared by the North Dakota Kids

Count Sept. 2010

ACS income not compatible with decennial census data

Decennial census asks income in previous calendar year

ACS asks income in previous 12 months Both are inflation adjusted However, Bureau test results show decennial

income consistently lower than ACS (4.4% nationwide)

Bureau suggests users “exercise caution”

20Prepared by the North Dakota Kids

Count Sept. 2010

Residency rules very different

Decennial census- “usual place of residence “using 6+ month” rule.

ACS – current residence during the last 2-months. Attempting to better count seasonal

residents

21Prepared by the North Dakota Kids

Count Sept. 2010

What lessons have we learned thus far?

Need to be more mindful of our data users

Rethink how we interpret and disseminate data to the public

Examine ways in which we can better educate ourselves and data user

22Prepared by the North Dakota Kids

Count Sept. 2010

ACS Presentation 2010

Dr. Richard Rathge, Policy Analyst North Dakota Kids Count, Fargo, ND

NDSU, IACC 424, Fargo, ND 58108 Richard.Rathge@ndsu.edu Phone: (701) 231-8621 Fax: (701) 231-

9730 URL: www.ndkidscount.org

23Prepared by the North Dakota Kids

Count Sept. 2010