using administrative data to create local-level child well-being indices robert goerge roopa...

16
Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Upload: andrea-osborne

Post on 15-Jan-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Using Administrative Data to Create Local-level Child Well-Being Indices

Robert GoergeRoopa Seshadri

Page 2: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Overview

Purpose

Background

Level of Aggregation

Data Description

Methods

Results

Use of the index

Page 3: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Purpose

• Build point-in-time indices of child well-being using administrative data

• Compare communities in Chicago• Compare indices with poverty rate – Is it

sufficient to use poverty rate in Chicago• How might it be used?

Page 4: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Level of aggregation

• Does it depend on purpose?• Chicago community area - 77 communities• Census tracts – 854 tracts• Or some combination of tracts?

• If interested in general well-being, 77 may be fine, but if we are trying to address the problems, 854 may be more appropriate.• Speaks to issues of resource allocation.

Page 5: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Chicago Community Areas

Variable Mean Std Dev Min 25th Pctl Median 75th Pctl Max

Area (sq. mi.) 2.94 1.81 0.59 1.75 2.85 3.55 110-5 population 3186 2124.1 94 1477 2772 4213 8980

0-17 population 8582 5621 577 4486 7859 11753 25946

• Described in the 1920s and divided Chicago into 77 areas.

• Collections of census tracts and roughly corresponded to neighborhoods.

• Although characteristics and homogeneity within the neighborhoods has changed over time, they continue to be used for research and policy planning purposes since their boundaries are static.

Page 6: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Data description• Poverty • Poverty rate, Head Start eligibility

• Birth• Birth rate, Births to single mothers, Birth weight

• Early Childhood• Elevated blood lead level,

• Childhood• Poverty rate, Educational outcomes, Child welfare

• Family/Neighborhood• MSF, Crime

Data are annual rates and range from 2007 to 2009, depending on availability.

Page 7: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Methods

• Robust centering and autoscaling of individual indicators. • Accounts for skewed distribution

• Standardized to a scale with mean=0 and SD=1.• Accounts for the different scales and ranges

• Composite scores calculated as the mean of the component indicators.• Equal weight to each indicator within an index

Page 8: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

IndicesIndex Mean Std Dev Min 25th Pctl Median 75th Pctl Max

Poverty rate 0 1 -1.72 -0.79 -0.02 0.64 2.27

Early Childhood Risk Index 0 0.77 -1.52 -0.6 0.07 0.47 1.7

Child Well-being Index 0 0.75 -1.18 -0.62 -0.2 0.51 2.08

• ECRI: Composite measure of 5 indicators of early childhood (under age 6) focusing on health and welfare. • Originally created to assess home visitation for early

childhood care need• CWI: Composite measure of 10 indicators spanning

the entire range of childhood and multiple domains.

Page 9: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Early Childhood Risk Index (ECRI)

Indicator Min 25th Pctl Median 75th Pctl Max

Birth rate -2.2 -0.76 0.05 0.66 2.33

% Medicaid-paid births -2.28 -0.44 0.34 0.83 1.14

% Births to single mothers -1.72 -0.79 -0.02 0.94 1.53

% Low Birth Weight -1.46 -0.76 -0.34 0.6 3.91

0-5 Abuse-neglect rate -1.24 -0.81 -0.29 0.79 3.9

Page 10: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Child Well-being Index (CWI)

Indicator Min 25th Pctl Median 75th Pctl Max

0-17 Poverty rate -1.72 -0.79 -0.02 0.64 2.27

Children in MSF -1.13 -0.81 -0.24 0.74 2.43

Head Start eligibility -0.41 -0.37 -0.31 -0.17 6.29

% Low Birth Weight -1.46 -0.76 -0.34 0.6 3.91

0-5 Abuse-Neglect rate -1.24 -0.81 -0.29 0.79 3.9

% Elevated blood lead level -1.4 -0.77 -0.14 0.63 2.81

3rd grade retention rate -1.43 -0.78 -0.35 0.84 2.27

High school dropout rate -2.42 -0.66 0.08 0.59 2.26

Violent crime rate -1.02 -0.71 -0.34 0.43 3.59

Property crime rate -1.06 -0.67 -0.25 0.27 4.85

Page 11: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Poverty rate vs. ECRI

r=0.71

Page 12: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Poverty vs. ECRI

CCA0-17 poverty rate

Birth rate

% Medicaid births

% births to single mothers

% Low birth weight

0-5 abuse-neglect rate

ECRI better than Poverty rate

Near North Side 0.59 -2.01 -1.61 -1.1 -0.07 -0.43

Near West Side 1.09 -0.95 -1.06 -0.52 0.03 0.95

Douglas 1.73 -1.64 -0.13 0.45 1.54 0.2

Washington Park 2.27 0.19 0.85 1.4 1.24 1.48

ECRI worse than Poverty rate

Montclare -1.21 0.66 0.24 -0.29 -0.41 -0.88

Fuller Park 0.24 1.05 0.98 1.44 3.91 1.13

Archer Heights -1.07 1.44 0.77 -0.02 -0.88 -0.7

West Elsdon -0.9 2.32 0.4 -0.28 -0.38 -0.86

West Lawn -0.93 2.01 0.46 -0.12 -0.63 -0.82

Page 13: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Poverty rate vs. CWI

r=0.85

Page 14: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Poverty vs. CWI

CCA Poverty rate

0-17 in MSF

Head Start eligibility

% LBW

0-5 abuse-neglect rate

% Elevated BLL

3rd Grade retention rate

High-school dropout rate

Violent crime rate

Property crime rate

CWI better than Poverty rateSouth Lawndale 1.09 -0.59 -0.24 -0.81 -0.29 0.56 -0.82 -0.38 -0.2 -0.49

Lower West Side 1.05 -0.56 -0.3 -0.55 -0.35 0.49 -0.91 0.03 0.15 0.19

Douglas 1.73 1.1 -0.02 1.54 0.2 -1.12 0.09 0.42 -0.68 -0.73

Riverdale 1.99 1.95 1.74 0.46 3.9 -1.4 0.7 0.59 -0.72 -1.03

CWI worse than poverty rate

Loop -1.72 0.05 -0.41 -0.78 -1.24 -1.4 -0.61 -1.96 -0.35 1.55

Fuller Park 0.24 2.22 6.29 3.91 1.13 2.74 1.19 1.72 -0.69 -0.9

Page 15: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

CWI vs. ECRI

r=0.8

Page 16: Using Administrative Data to Create Local-level Child Well-Being Indices Robert Goerge Roopa Seshadri

Use of the (an) index

• Policymakers• Service providers• Funders• Real estate agents• Researchers