View
3
Download
0
Category
Preview:
Citation preview
Long-term effect of Michigan’s state-funded preschool program
Tomoko Wakabayashi
Oakland University
Lawrence J. Schweinhart
HighScope Educational Research Foundation
Zongping Xiang
SRCD--Baltimore, MDMarch 21, 2019
Michigan Great Start Readiness Program (GSRP)
•Established: 1985
•The Premise: High-quality preschool can mitigate risk factors and support the school readiness and subsequent achievement for children with demographic factors that predispose them to poorer school and life outcomes.•Evaluation: 1995-(HighScope 1995-2017)
•Recent Expansion: 2013-2015
What are the program elements and implementation context?
NIEER Benchmark (2017)
What is the research design and key findings on impact?
GSRP Evaluation Study SamplesStudy Year N
Longitudinal Evaluation 1995-2011 Treatment=338
Comparison=258
GSRP Lansing Regression
Discontinuity Design (RDD)
2011-2014 Treatment=317
Comparison=333
Preschool Variation Study 2004-2011 GSRP=190
Head Start=143
Tuition-based=70
Pre-Post Design
• Urban
• Rural
• West
2011-2017
2012-2017
2014-2017
Cohort 2=483
Cohort 3=463
Cohort 2=467
Cohort 1=525
Cohort 2=598
GSRP Longitudinal High School Graduation Study
Schweinhart, L. J., Xiang, Z., Daniel-Echols, M., Browning, K. & Wakabayashi, T. (2012, March). Michigan Great Start Readiness Program evaluation 2012: High school graduation and grade retention findings. Technical report prepared for Michigan Department of Education. https://www.michigan.gov/documents/mde/GSRP_Evaluation_397470_7.pdf
A study of 595 low-income children entering kindergarten in 1996
•In six Michigan school districts – Detroit, Grand Rapids, Grayling, Kalamazoo, Muskegon, and Port Huron.
•338 children attended GSRP.
•257 children did not attend a preschool program but had family incomes under $30,300, low enough to qualify for GSRP.
Similar Group BackgroundsCharacteristic GSRP No GSRP
Age at kindergarten entry 5.3 5.3
% female 51% 51%
Fathers in home 62% 61%
Persons in household 4.5 4.7
Mothers’ years of schooling 12.1 12.0
Fathers’ years of schooling 12.1 11.7
Average annual income $17,882 $18,022
High School Graduation on Time
57%59%
56%
43%
37%
47%
0%
10%
20%
30%
40%
50%
60%
70%
All Non-White White
GSRP
No GSRP
****
Grade Retention by Grade 12 by GSRP Status
37%
49%
0%
10%
20%
30%
40%
50%
60%
Repeated a grade
GSRP
No GSRP
**
Graduation Timing by Grade Retention
80%
6%
3%
28%
17%
67%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
Never retained Ever retained
Graduated on Time
Graduated a Year Later
Has not graduated
How can the findings be used to improve quality in the field at large?
Michigan GSRP Expansion
FY2014 $65Million budget
FY2015 $65Million budget
TOTAL: $130 Million expansion
GSRP enrollment increased by 14,891 children annually.
GSRP Benefit-Cost Analysis
Wakabayashi, T., Claxton, J., & Hardin, B. A. (2018, February). Examining costs and benefits of Michigan Great Start Readiness Program expansion 2013-2016: Phase 2 report. Technical report prepared for Michigan Department of Education.
Wakabayashi, T., Hardin, B. A., Claxton, J., & Grace, E. (2017, September). Examining costs and benefits of Michigan Great Start Readiness Program expansion 2013-2016: Phase 1 report. Technical report prepared for Michigan Department of Education.
Legislative decision to expand GSRP was based on:
The evidence that attending GSRP programs can effectively increase eligible children’s school readiness and later school success;
The assumption that more funding will increase GSRP access, i.e., more four-year-olds eligible for GSRP will enroll in the program; and
The expansion will not compromise the essential components that contribute to the effectiveness of the GSRP model; hence more children will reap the benefit of GSRP.
GSRP Total Enrollment FY 2013-2016
State-wide GSRP PQA Mean Scores FY2013-2016
PQA Scale 2012–13 2013-14 2014-15 2015-16
Classroom Level (Form A) N=1042 N=1537 N=2034 N=2245
Total Score for Form A 4.20 4.29 4.30 4.35
I. Learning environment 4.15 4.20 4.23 4.28
II. Daily routine 4.15 4.31 4.33 4.37
III. Adult-child interaction 4.23 4.27 4.23 4.27
IV. Curriculum planning and assessment 4.37 4.49 4.54 4.59
Center Level (Form B) N=392 N=333 N=480 N=659
Total Score for Form B 4.35 4.32 4.41 4.43
V. Parent involvement and family services 4.50 4.38 4.48 4.49
VI. Staff qualifications and development 4.02 4.07 4.18 4.24
VII. Program management 4.47 4.48 4.55 4.55
Benefit-Cost Analysis
Total ADDITIONAL
cost (FY2014-2015)
(C)
$195,104,004 $95,015,650 $54,629,121
Yield (number
children NEWLY
served during the
expansion year)
20,265 9,869 5,674
HS Graduates
over HS Dropouts
(Newly Served)
Female only
(48.7% of Total)
Black only
(28% of Total)
Total benefit $458,410,330 $311,237,798 $219,547,725
Net Benefit (B-C) $263,306,326 $216,222,148 $164,918,604
Benefit/Cost Ratio 2.35 3.28 4.02
Legislative changes in GSRP program elements and implementation
• Administrative/financial oversight
• Community-based organizations
• Income eligibility
• Transportation funds
Issues raised➢Regional differences
➢Teacher salaries and benefits
➢Rent/facility usage and maintenance
$23,020 $21,635 $31,488
$20,609
$40,788
$59,290 $56,800 $62,720
$- $10,000 $20,000 $30,000 $40,000 $50,000 $60,000 $70,000
*
Questions?
Tomoko Wakabayashi Twakabayashi@Oakland.edu
Oakland University
@twakabayashi264
For more information about GSRP:Michigan.gov/gsrp
Thank you!
EXTRA SLIDES--ONLY USE IF NEEDED DURING Q & A--
GSRP Risk Factors
1. Low income;
◦ 90% of enrollment below 250% Federal Poverty Level;
◦ 10% may attend GSRP w/ sliding scale tuition based on income (from 2013)
2. Diagnosed disability/developmental delay;
3. Severe or challenging behavior;
4. Primary home language other than English;
5. Parent(s) with low educational attainment;
6. Abuse/neglect of child or parent;
7. Environmental risk.
Kindergarten Teacher Ratings of Retained and Non-Retained Students
74%
80%
61%
80%
80%
90%
22%
33%
16%
39%
61%
75%
0% 20% 40% 60% 80% 100%
Literacy
Math
GOOD STUDENT
Retains learning
Completes assignments
SOCIAL
Gets along with children
Gets along with adults
Not Retained
Retained
GSRP Evaluation 2012-2017 Study Sites
Kent ISD (2014 – 2017)
•Pre-Post Risk Factor Analysis (2 cohorts) with focus on Spanish-English Dual Language Learners
Thumb Area (Huron, Tuscola & Sanilac; 2012- 2017)• Pre-Post Risk Factor Analysis (2
cohorts)
Lansing School District (2011- 2017)
•Regression Discontinuity Study (2011-2014)
•Pre-Post Risk Factor Analysis (2 cohorts)
GSRP Evaluation 2012-2017GSRP has significant impact on increasing children’s early literacy and math skills (GEE-RDD*).
*Group Equivalency Enhanced Regression Discontinuity Design—a supplementary data collection and analysis method to enhance the equivalency between treatment and control that we added to a traditional RDD.
Wakabayashi, T., & Xiang, Z. (2014, October). Michigan Great Start Readiness Program evaluation: Lansing school district 2011-2014—Group Equivalency Enhanced Regression Discontinuity Design. Technical report prepared for Michigan Department of Education.
Regression Discontinuity Design (RDD)
GEE-RDD SampleTreatment Comparison
N=333 N=317Female 167 53% 162 52%
Male 166 47% 148 48%Risk Factors
Extremely low-income 228 72% 222 72%Primary language other than English 50 16% 43 14%
Low parental education 66 21% 81 26%Ethnicity/Race
White 120 38% 96 31%
Black/African American 65 21% 86 28%Asian 7 2% 10 3%Hispanic 47 15% 65 21%Other 93 24% 60 17%
Group Equivalency Enhanced RDD Sample
Type of Participants Included in the Estimation
NEstimated Treatment
Effect at CutoffTreatment (K-entry)
Comparison (Pre-K entry) PPVT Math
Letter-Word
Traditional RDD (All participants tested)
Adjusted for SES & demographic status 303 363 2.89 1.57** 4.20**
Group Equivalency RDD (Controlled for differential attrition)Participants with & without sufficient treatment
Adjusted for SES & demographic status 333 317 4.50+ 1.73** 4.55**Participants with sufficient treatment only
Adjusted for SES & demographic status 316 310 4.86+ 1.79** 4.47**
+ p<.10; ***p<.01
Estimated Treatment Effects (in raw score
GSRP Evaluation 2012-2017
➢ Significant achievement gap exists between higher-risk and lower-risk children at GSRP entry (Pre).
➢ GSRP helps reduce the achievement gap between higher-risk children and lower-risk children (Pre-Post).
Wakabayashi, T., & Xiang, Z. (2012, October). Michigan Great Start Readiness Program evaluation: Lansing school district 2011-2012. Technical report prepared for Michigan Department of Education.
Wakabayashi, T., & Xiang, Z. (2015, October). Michigan Great Start Readiness Program evaluation: GSRP programs in Huron, Tuscola, and Sanilac Counties (the Thumb area) 2012–2013. Technical report prepared for Michigan Department of Education.
Wakabayashi, T., Xiang, Z., & Kantor, P. (2013, January). Preschool Variation Study: Follow-up through fifth grade (2004-2011). Technical report prepared for Michigan Department of Education.
Vulnerability Category Status N PPVTLetter Word
Spelling Math
Extremely low family incomeYes 271 98.58 96.42 93.46 101.71
No 92 109.03 102.11 97.39 106.36
Diagnosed disability/developmental delayYes 114 102.90 97.57 94.48 103.37
No 249 100.46 98.06 94.47 102.65
Severe or challenging behaviorYes 7 105.86 103.29 93.00 108.57
No 356 101.14 97.79 94.50 102.76
Primary home language other than EnglishYes 59 96.00 96.54 95.19 102.32
No 304 102.24 98.17 94.34 102.98
Parent(s) with low educational attainment
Yes 77 98.51 93.77 92.24 99.81
No 286 101.96 98.98 95.08 103.73
Abuse/neglect of child or parentYes 19 106.32 101.37 95.65 104.05
No 344 100.95 97.71 94.40 102.81
Environmental riskYes 303 101.00 97.84 94.08 102.80
No 60 102.40 98.20 96.36 103.28
Minority (non-White)Yes 229 98.02 96.47 93.91 100.59
No 114 106.72 100.26 95.41 106.71
Mean Program Entry and Gain Scores by Vulnerability Index (Urban)
Gain Scores
0 risk vs. 1-3 risks: p < .05
112.1
105.21
97.96
93.43
80
85
90
95
100
105
110
115
n=40 n=113 n=166 n=44
0 1 2 3
Vulnerability index
Stan
dar
d S
core
s
1.07
4.6 4.36 4.09
0123456
n=40 n=113 n=166 n=44
0 1 2 3
Vulnerability index
Mean Program Entry and Gain Scores by Vulnerability Index (Rural)
Gain Scores
Low vs. High risks: p < .05
106.83 105.76
101.74
80
85
90
95
100
105
110
115
n = 89 n = 182 n = 47
Low Medium High
Vulnerability Index
Stan
dar
d S
core
s
1.932.59
5.32
0123456
n = 89 n = 182 n = 47
Low Medium High
Vulnerability Index
Preschool Variation Study
Original versus Longitudinal follow-up Sample Size
Preschool group
N
Original
N
Follow-up
%
Retention
GSRP 190 160 84
Head Start 143 116 81
Tuition-Based 70 58 83
Total 403 334 83
PPVT and WJ Math Scores at Program Entry by Vulnerability Level
Vulnerability Category nPPVT WJ Math Problem
mean p mean p
Number of economic risks 4 15 91.8
<.001
94.1
<.0013 75 91.5 98.32 59 97.9 100.91 88 98.8 101.10 158 105.3 105.9
Mother’s educationLower than high school 24 92.9
<.00193.6
<.001High school diploma 184 96.3 99.62-4 years of college 116 100.6 103.9Bachelor degree or higher 71 108.9 109.1
Minority statusNon white 99 91.8 <.001 97.0 <.001White 300 102.1 103.9
Overall vulnerability level3 risks 70 88.7
<.00195.3
<.0012 risks 117 97.8 100.51 risk 107 101.4 102.70 risk 105 106.7 108.1
Preschool Variation Study—Preschool Entry Scores
102.2
94.8
103.2
94.5
103.1
98.8
104.9
98.7
889092949698
100102104106
LowerVulnerability
HigherVulnerability
LowerVulnerability
HigherVulnerability
GSRP Head Start
Stan
dar
d S
core
s
Program Entry
PPVT
Math
Preschool Variation Study—Preschool Gain Scores
1.9***
5.9***
1.6 1.41.9
3.5
1.71
01234567
LowerVulnerability
HigherVulnerability
LowerVulnerability
HigherVulnerability
GSRP Head Start
Stan
dar
d S
core
s
Gain Scores
PPVT
Math
***p<.001
Preschool Variation Study—MEAP Scores
67.1 67.9 66.2
87.277.8
83.878.1 68.5
74.3
55.2**52.1* 48.5**
0102030405060708090
100
Grade 3 Grade 4 Grade 5 Grade 3 Grade 4 Grade 5
GSRP Head Start
% P
rofi
cie
nt
Reading
LowerVulnerability
HigherVulnerability
*p<.05; **p<.01
Preschool Variation Study—MEAP Scores
73.7 71.663.7
89.7 91.781.182.2 80.8
72.9
58** 59.2**
43.5**
0102030405060708090
100
Grade 3 Grade 4 Grade 5 Grade 3 Grade 4 Grade 5
GSRP Head Start
% P
rofi
cie
nt
Mathematics
LowerVulnerability
HigherVulnerability
**p<.01
Recommended