Parental Incentives and �Early Childhood Achievement:�A field Experiment in Chicago Heights

(Rorland Fryer, Jr., Steven Levitt, John List, 2015)

​

September 4, 2023. Micro Reading Group

Hyunjun Cho, Jiwon Hwang

​

I. Introduction

Motivation and Design

• Importance of “parental involvement” in early childhood

​

What will be the effect of short-term incentive to enhance parental involvement?

​

• If low-income families lack parental involvement because …
• (1) Lack of motivation, information, discounted future 🡪 Effective measure
• (2) Structural reason (neighbors, nutrition, etc.) 🡪 Ineffective measure
• (3) Intrinsic motivation 🡪 Crowding-out effect (negative)

​

​

Parental Incentive Experiment (2011-2012)

• Chicago heights: low performing urban school district
• 257 families (Control vs. Treatment group)

​

• Treatment : provided financial incentive for …
• Attendance in Parental Academy sessions
• Proof of homework completion
• Performance on benchmark assessments (end-of-semester, …)

​

• Two different treatment groups:
• Paid with cash/direct deposits (“Cash”)
• Trust account, accessible only when children enroll in college (“College”)

​

• Children outcome for cognitive and non-cognitive outcomes

​

​

Brief Summary of Outcomes

Brief Summary of Outcomes (2)

Brief Summary of Outcomes (3)

II. Previous Literature

Previous Literature

• CCT (Conditional Cash Transfer)
• Reduce poverty, and to accumulate human capital
• Conditional on children’s attendance, grants incentive to parents
• Positive change in children’s outcome (health, education) observed
• PROGRESSA, Bolsa Familia, …

​

• PROGRESSA
• Mexico, 1998
• (1) Attendance: able to receive up to $62.50 for attendance
• (2) Emphasized actual achievement: ineligible if failed more than once

​

​

​

​

PROGRESSA

• Schultz (2000): Positive impact on school enrollment
• Primary school: increase by 1.1% point for boys, and 1.5% point for girls
• Secondary school: 2.1% point for boys, 2.3% point for girls

​

• For poor household, PROGRESSA achieved 10% increase in schooling attainment

​

• <Opportunity NYC> in similar context
• Aug, 2007 ~ Aug, 2010
• Riccio et al. (2013) : insignificant impact on school outcomes

​

​

​

​

​

III. Research Design

Chicago Heights

1. Total Population: 30,276 (2010)
• Majority of students were Black or Hispanic
• Population composition quite consistent across

treatment and control group

​

2) Per capita Income: $17,546

- Relatively low compared to other regions

- Average U.S. per capita income at 2010: $40,277

​

3) 90% of students receive free or reduced price lunch

​

Procedure

• Local marketing campaign to inform parents

​

• Curriculum for Cognitive and Non-cognitive skills

- Cognitive 🡪 Literacy Express

- Non-cognitive 🡪 Tools of the Mind

​

• Assessment for cognitive and non-cognitive skills
• Cognitive: Testing vocabulary, verbal ability, general intellectual and academic achievements (PPVT, WJ-III)
• Non-cognitive: Testing working memory, self-regulation in emotional, attentional, behavioral domains (Executive Function, Self-Regulation Assessment)

Research Design

• Simple, single draw, block randomization
• Total: 260 subjects
• Control: 99 children
• Treatment 1 (Cash): 74 children
• Treatment 2 (College): 84 children

​

• Treatment:
• 90 minutes Parent Academy sessions (every 2 weeks with 18 sessions)
• Maximum earning by attendance: $7,000

- Participate until their children entered kindergarten

- Differential distribution according to their attendance/tardiness

​

​

Research Design

[Parent Academy session]

• Assignments to reinforce learning objectives

- Videos working with their children

• Homework incentives according to their grades (A,B,C,Incomplete)

- “A” quality for every homework (17 total) receives $3,500

• Interim assessments: earn up to $1,800 per year
• End-of-semester assessments: earn up to $1,600 per year

​

• Control Groups were exempted from any rewards, but awarded $100 for end-of-semester assessments

​

​

​

​

IV. Data and Model

Data

• Data Collection

- Prior to randomization: Parental Demographic data

- Middle and End of Treatment: Children’s assessment scores

​

• List of Controls
• Pre-treatment cognitive/non-cognitive scores

- Pre-treatment testing before the treatment

2) Race dummies, child’s gender, child’s age, mother’s age, …

3) Mid/End-of-year parent investment surveys

- Parental investment: hours spent per weekday teaching children

- Belief about children’s rank in their age

[Table 3]

Estimates (1) : Intent-to-Treat (ITT)

Estimates (2) : �Local Average Treatment Effect (LATE)

Estimates (2): LATE

V. Experimental Results

Estimates Results

​

​

​

​

• Impact of parental incentives on non-cognitive skills is larger and statistically significant
• Consistent with Kautz (2014) that parental investment is important contributor to non-cognitive development.

​

• Similar result for LATE and ITT estimates implies that both methods provide consistent evidence that the parental incentives had a positive impact on children’s non-cognitive skills.

​

Race

Gender

Parent Income

Mother’s Age

Children in the Household

Estimates (2): LATE

• Race
• Blacks ; negative but statistically insignificant treatment effects for both cognitive and non-cognitive
• Hispanic and Whites; statistically significant at conventional levels
• Gender
• Point estimates are larger for girls but not statistically significantly so.

​

Estimates (2): LATE

• The sample was divided along three dimensions that may correlate with differential likelihoods of being at risk for parental underinvestment
• The experiment was more effective for children in higher risk groups.
• The effects are greater for young mothers and families with below median income.

​

Estimates (2): LATE

• Cognitive
• Same for the non-cognitive
• Below median cognitive scores derive a greater benefit in the cognitive domain
• Non-Cognitive
• Above the median experienced large gains in both cognitive and non-cognitive skills.
• Contrast from the below median who did not gain from this program
• Sufficiently developed non-cognitive skills are a necessary input for learning

​

• The greatest gain was on “high on non-cognitive and low on cognitive”
• This is consistent with our results above
• And those who had opposite characteristics “low on non-cognitive and high on cognitive” experienced significantly negative treatment effects on cognitive skills.

​

VI. Robustness to Attrition

Robustness check (1)

• (Lee, 2009) Calculates conservative bounds on the true treatment effects

​

• Similar attrition rates between treatment and control the impact on our estimates is small, and the pooled cognitive estimates are unaffected.

​

Robustness check (2)

“Different reasons for attrition across treatment and control groups”

​

• When lowest gaining children are systematically missing in treatment group and highest gaining children are unassessed in the control group, the attrition bias can be extreme.
• Interim test was conducted halfway through the program that compute the score changes looking for systematic differences in treatment and control groups.
• Although the sample sizes are small, there is no definitive pattern of systematic attrition.

VII. Understanding Racial Differences in Treatment Effectiveness

Possible explanation

1. Black parents may invest less heavily in the program
2. Race may be simply a proxy for other observable characteristics
3. Selection on unobservable
4. Pattern of outcomes across different components of the test may provide some explanation on racial differences
5. Home language theory may be applied

The “Home Language Theory”

• Early childhood intervention have had greater impact on households where English is not their home language

- Currie & Thomas (1999) and Wagner & Clayton (1999)

- Possible explanation for the difference (Hispanic vs. Black)

​

• Cognitive scores were consistent with “home language theory”
• Non-cognitive scores were inconsistent, yet insignificant differences
• Not account for Whites

VIII. Conclusions

Conclusions

• Financial incentives to families to engage in activities with their children stimulate both cognitive and non-cognitive growth.
• Home language theory supported that speaking Spanish at home is associated with large cognitive gains for Hispanics. Yet, it did not explain the racial differences in non-cognitive growth or impact on Whites.
• The program effects are concentrated on those with strong non-cognitive skills and poor cognitive skills, when entering the program
• The finding raise important public policy implication in education.