Racial Disproportionality in Washington’s Child Welfare System:

An Update

FCAP SeminarApril 14, 2010

Marna Miller, Ph.D.Washington State Institute for Public Policy

www.wsipp.wa.gov

WASHINGTON STATE INSTITUTE FOR PUBLIC POLICY

• Created by State Legislature in 1983

• Mission: Carry out non-partisan research on projects assigned by the Legislature or Board of Directors

• Board of Directors:

Representative Glenn Anderson Robin Arnold-Williams, Executive Policy Office Representative Mary Lou Dickerson Director Victor Moore, OFMSenator Karen Fraser Sandra Archibald, University of WashingtonRepresentative Phyllis Gutierrez Kenney James L. Gaudino, Central Washington Univ.Senator Jeanne Kohl-Welles Robert Rosenman, Washington State UniversityRepresentative Skip Priest Les Purce, The Evergreen State CollegeSenator Pam Roach Ken Conte, House Office of Program ResearchSenator Mark Schoesler Richard Rodger, Senate Committee Services

Background

Disproportionality: The over-representation of children belonging to racial/ethnic minorities in the child welfare system. Always defined as comparison with White children.

For example, in 2004, American Indian children comprised less than 4 percent Washington’s children, but accounted for 14 percent of children removed from their homes in that year.

Today

• Review findings from our earlier study on racial disproportionality

• Update on racial disproportionality in the period between 2004 and 2008.

• Ongoing work on effects of Structured Decision Making on racial disproportionality

Study DirectionESSB 1472, Laws of 2007

Established the Racial Disproportionality Advisory Committee: “to analyze and make recommendations on the disproportionate representation of children of color in Washington's child welfare system.”

Directed WSIPP to: “… serve as technical staff for the advisory committee.”

2 of 26

Study DirectionESSB 1472, Laws of 2007

To Examine:• Level of involvement of children of color in the child

welfare system (CWS) at entry, exit, and all points at which treatment decisions are made

• The number of children of color in low-income or single-parent families involved with the child welfare system

• The family structures of families involved in the state's child welfare system

• Outcomes for children in the existing child welfare system

Measuring Disproportionality

Disproportionality Index:The rate of occurrence of an event for children in a racial group divided by the rate of the same event for White children.

Measuring Disproportionality

Example:

5,612 American Indian children were referred in 200455,872 Indian children in Washington in 2000Rate of CPS referrals for Indian children:

5,612 ÷ 55,872 = 0.100Or

100 children referred per 1,000 Indian children in Washington

Measuring Disproportionality In the same year, 2004, the rate of referral to CPS

for White children was 34 per 1,000 children.

Disproportionality Index:Rate for Indian children ÷ Rate for White children

100 ÷ 34 = 2.92

Defining Race

WSRDAC specified the hierarchical rules for classifying multi-racial/ethnic children.

American Indian: Any Indian heritageBlack: Any non-Indian Black heritageAsian: Any non-Indian, non-Black Asian/Pacific

Islander heritageHispanic: Hispanic heritage, not in any of the other

racial groupsWhite: Non-Hispanic, White only

Decision Points In Child Protective ServicesAnd Time in Care, 2004 Cohort

Referral to CPS

Accepted

Initial High Risk

Removed From Home

Over 60 Days

Over Two Years

58,005

43,423

35,474

4,744

3,194

1,476

3.053.31

4.56

4.96

6.29

1.00 1.04 1.13

1.70

2.15

2.92

1.56

0

1

2

3

4

5

6

7

Referrals Accepted Initial High-risk

Placed Over 60 days

Over 2 years

Dis

pro

po

rtio

nal

ity

Ind

ex (

Ind

ian

vs

Wh

ite) DI

DI After Referral

White children

Previous Findings for Referrals in 2004 : American Indian Children

Previous Findings for Referrals in 2004 : Black Children

2.02 2.17 2.29 2.24

2.79

1.00 1.07 1.15 1.21 1.181.48

1.89

0

1

2

3

4

5

6

7

Referrals Accepted Initial High-risk

Placed Over 60 days

Over 2 years

Dis

pro

po

rtio

nal

ity

Ind

ex (

Bla

ck v

s W

hit

e) DI

DI After Referral

White children

Previous Findings for Referrals in 2004 : Hispanic Children

1.44 1.41 1.48 1.45 1.371.34

0.921.031.031.051.00 1.07

0

1

2

3

4

5

6

7

Referrals Accepted Initial High-risk

Placed Over 60 days

Over 2 years

Dis

pro

po

rtio

nal

ity

Ind

ex (

His

pan

ic v

s W

hit

e) DI

DI After Referral

White children

Previous Findings for Referrals in 2004: Asian Children

1.00 1.06 1.05 1.020.85 0.86

0.410.410.490.500.510.480

1

2

3

4

5

6

7

Referrals Accepted Initial High-risk

Placed Over 60 days

Over 2 years

Dis

pro

po

rtio

nal

ity

Ind

ex (

Asi

an v

s W

hit

e)

DI

DI After Referral

White children

Previous Findings for Referrals in 2004: All Races

0

1

2

3

4

5

6

7

Referrals Accepted Initial High-risk

Placed Over 60 days

Over 2 years

Dis

pro

po

rtio

nal

ity

Ind

ex (

Co

mp

ared

to

Wh

ite) Indian Children

Black Children

Hispanic Children

Asian Children

White children

16 of 26

Who Refers to Child Protective Services?

Mandated Reporters: Certain professionals are required by law to report suspected child abuse or neglect. These include:

Medical professionalsSchool personnelDSHS employees and other social service

professionalsChild care providersLaw enforcement

Non-mandated Reporters: RelativesFriends/NeighborsParentOther

Who Refers to Child Protective Services?

WSIPP, 2008

CPS Referrals

0%

20%

40%

60%

80%

100%

AmericanIndian

Black Asian Hispanic White

Per

cen

t o

f A

ll R

efer

rals

Mandated Not Mandated

Who Refers to Child Protective Services?Disproportionality Index by Type of Referrer

WSIPP, 2008

0

1

2

3

4

AmericanIndian

Black Asian Hispanic White

Dis

pro

po

rtio

nal

ity

Ind

ex(C

om

par

ed t

o W

hit

e C

hild

ren

)

Mandated Not Mandated

Outcomes For Children Following a CPS Referral 2004 Cohort

Referral to CPS

Accepted

Initial High Risk

Removed From Home

Over 60 Days

Over Two Years

58,005

43,423

35,474

4,744

3,194

1,476

Reunification

Guardianship

Adoption

Age Out

Outcomes for Children Following a Placement Linked to a CPS Referral

(As of November 1, 2007)

WSIPP, 2008

0%

20%

40%

60%

80%

100%

Indian Black Asian Hispanic White

Reunified Guardianship Adopted Reached Majority Placement Still Open

Permanency for Children in PlacementsLinked to a CPS Referral

(As of November 1, 2007)

WSIPP, 2008

0%

20%

40%

60%

80%

100%

Indian Black Asian Hispanic White

Permanency Reached Majority Placement Still Open

Children Living with Single Parents

0%

20%

40%

60%

80%

100%

Indian Black Asian Hispanic White All Races

Pe

rce

nt

of

Ch

ild

ren

All Children Children In Out-of-Home Placements

Legal Outcomes Children Removed from HomePlacements Linked to CPS Referrals in 2004

0%

20%

40%

60%

80%

100%

American Indian Black AsianHispanic White All races

Perc

ent o

f Chi

ldre

n

If dependent

Other Factors Affecting Referrals and Placements for the 2004 Cohort

Poverty: Children in families receiving food stamps were more likely to

be referred to CPS and removed from home

Geography (DSHS Administrative Region): Some regions placed more children than others.Region 4 (King County) had markedly greater

disproportionality than other regions for American Indian children.

Type of Reporter: For example, removals were more likely if law enforcement

reported the alleged maltreatment

Other Factors Associated with Referrals and Placements

History of the Intake Worker: Regardless of race, the average risk tag assigned by a

worker over the previous two years was the strongest predictor of:

• Whether the referral was accepted• The risk tag assigned to the referral

That is: Some workers appear to be “harder graders” than others. Given similar case characteristics, some workers will assign higher risk tags than others.

Regression Analysis

Regression analysis is a fancy algebra that allows us to take into account many factors simultaneously. We can ask the question:

All things equal EXCEPT for race, is race a significant factor?

Regression Analysis

Controlling for other case characteristics:

For Indian children, about 30 percent of disproportionality after referral could be explained by known case characteristics.

Disproportionality was unexplained by case characteristics for other races.

Main Conclusions from the 2004 Cohort

Disproportionality exists in WA’s child welfare system:

Greatest for Indian and Black children Most of the disproportionality occurs at the point of

referral to CPS Mandated reporters are only part of the story Disproportionality for Indian children can be partly

explained by different case characteristics

CurrentStudy

Current Study DirectionESSB 5882, Laws of 2009

“…the Washington state institute for public policy shall evaluate the department of social and health services' use of structured decision-making practices and implementation of the family team decision-making model to determine whether and how those child protection and child welfare efforts result in reducing disproportionate representation of African-American, Native American, and Latino children in the state's child welfare system.”

Study Approach for SDM Analysis

Structured Decision Making: • Actuarial risk assessment tool.• In Washington, it is used with CPS cases only, after

the investigation

If SDM can affect disproportionality, we should observe less disproportionality at the point of removal from home.

Study Approach for SDM AnalysisTimeline

SDM Statewide

27-Oct-07

Pre-SDM CPS Referrals

Post-SDM CPS Referrals

Pre-SDM Placements

Post-SDM Placements

Jan-07 Jul-07 Jan-08 Jul-08

Comparing Referrals2007 to 2008

Referrals Accepted Initial High-risk

Placed Over 60 days

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6

Indian Children

20072008White

Dis

prop

ortio

nalit

y In

dex

After

Ref

erra

l

Comparing Referrals2007 to 2008

Referrals Accepted Initial High-risk

Placed Over 60 days

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6

Black Children

20072008White

Dis

prop

ortio

nalit

y In

dex

After

Ref

erra

l

Multi-Year ComparisonsDisproportionality After Referral

Referrals January through June

Referrals Accepted Initial High-risk

Placed Over 60 days

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6Indian Children

20042005200620072008White

Dis

prop

ortio

nalit

y In

dex

After

Ref

erra

l

Multi-Year ComparisonsDisproportionality After Referral

Referrals January through June

Referrals Accepted Initial High-risk

Placed Over 60 days

0.80

0.90

1.00

1.10

1.20

1.30

1.40

1.50

1.60 Black Children

20042005200620072008White

Dis

prop

ortio

nalit

y In

dex

After

Ref

erra

l

Multi-Year ComparisonsDisproportionality After Referral

Referrals January through June

Referrals Accepted Initial High-risk

Placed Over 60 days

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

Asian Children

20042005200620072008White

Dis

prop

ortio

nalit

y In

dex

After

Ref

erra

l

Multi-Year ComparisonsDisproportionality After Referral

Referrals January through June

Referrals Accepted Initial High-risk

Placed Over 60 days

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

Hispanic Children

20042005200620072008White

Dis

prop

ortio

nalit

y In

dex

After

Ref

erra

l

Multi-Year ComparisonsDisproportionality Index

Referrals January through June

Referrals Accepted Initial High-risk

Placed Over 60 days

0

1

2

3

4

5

6Indian Children

20042005200620072008WhiteD

ispr

opor

tiona

lity

Inde

x

Multi-Year ComparisonsDisproportionality Index

Referrals January through June

Referrals Accepted Initial High-risk

Placed Over 60 days

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

Black Children

20042005200620072008White

Dis

prop

ortio

nalit

y In

dex

Tentative Findings: Regression Analysis

Effect of SDM on removal from homeReferrals from Jan through June 2006, 2007 and 2008

Controlling for child age, sex and race, alleged CAN type, risk tag at intake, CPS history, DSHS region and reporter type (mandated or not), intake worker history

Beta P-value Odds ratioAll children -0.0306 0.3138Indian children -0.0816 0.3504Black children 0.3054 0.0003 1.36Asian children 0.0053 0.9753Hispanic children -0.0523 0.541

Tentative Findings: Regression Analysis

Effect of SDM on removal from homeReferrals from Jan through June 2006 and 2008 (omit 2007)

Controlling for child age, sex and race, alleged CAN type, risk tag at intake, CPS history, DSHS region and reporter type (mandated or not), intake worker history

Beta P-value Odds ratioAll children -0.00108 0.9755Indian children -0.0652 0.5178Black children 0.2599 0.0073 1.30Asian children 0.3183 0.1311Hispanic children -0.0211 0.8298

Conclusions• The pattern of disproportionality for Black

children varies markedly from year to year• Apparent effects of implementing of Structured

Decision Making in October 2007:– No effect on overall rates of removal, except for Black

children– Following implementation of SDM, increased rates of

placement for Black children leading to greater disproportionality at the point of removal from home.

• The story is not over.

What’s Next

• Obtaining additional information on families, particularly receipt of food stamps in the year of the referral

• Refine our estimates for the effects of SDM

• Analyze effects of Family Team Decision Making

• Report due in September 2010

Thank you!

Questions?