1 of 17

Dosage Variability in School-Based Programs: �The Roles of Individual- and Classroom-Level Attendance in Refugee Student Intervention Outcomes

Abigail C. Keim, The Pennsylvania State University

Ha Yeon Kim, Global TIES for Children, New York University

2 of 17

The Education of Syrian Refugee Children in Lebanon�

  • As of 2018:
      • An estimated 1 in 5 people in Lebanon is a refugee
      • Approximately 58% of Syrian children ages 3-18 in Lebanon were out of school (UNHCR, 2018)
  • Refugee students enrolled in Lebanese public schools are high risk for low attendance and dropout for various reasons including:*
      • financial strains
      • high mobility
      • transportation access
      • negative social experiences at school

* Jalbout, 2015 , Dryden-Peterson, 2015

https://data2.unhcr.org/en/documents/details/65859

2

3 of 17

Cook, et al., 2013

​

Classrooms as Multilevel Systems

Institutions, social forces

Dyadic or small group interactions

Individual Experiences

3

4 of 17

Average Attendance

Student Level:

  • Individual Dosage

Classroom Level:

  • Group Dosage

Attendance Variability

Student Level:

  • Dosage Consistency

​

Classroom Level:

  • Contextual Stability

​

A Multilevel Approach to Attendance

​

​

More exposure to curriculum and skill-building activities

Less consistent practice

More chaotic classroom environment

4

5 of 17

Learning in Healing Classrooms�

Cycle 1 Treatment Conditions:

    • Learning in a Healing Classroom (LIHC)
    • LIHC + Mindfulness
      • Daily mindfulness activities that focus on breathing and mind-body regulation.

3EA Lebanon 2016-17 Program

​

IRC LIHC

Learning in a Healing Classroom

5

6 of 17

The Present Study�

  • Question 1: Do student and classroom-level attendance rates and attendance variability have unique effects on student outcomes?
    • Higher average attendance rates and lower attendance variability will predict more positive outcomes
      • Student Level: stronger effects for average monthly attendance
      • Classroom Level: stronger effects for attendance variability

​

  • Questions 2: Do the effects of student and classroom-level attendance rates and variability differ depending on treatment condition?
    • The added component in the LIHC + Mindfulness condition will protect against the negative effects of lower and more variable attendance rates.

​

​

​

6

7 of 17

Student and Classroom Characteristics�

Characteristics

N

% / M (SD)

Range

Student Level (L1)

​

​

​

L1 Average Attendance Rate

1859

0.62 (0.26)

0.02-1

L1 Attendance Variability

1859

0.21 (0.12)

0-0.64

Age

1859

9.75 (2.33)

6-17

Female

1859

49.76%

0-1

Grade

1859

2.59 (1.62)

1-9

Child Labor

1859

15.06%

0-1

Classroom Level (L2)

​

​

​

L2 Average Attendance Rate

150

0.63 (0.13)

0.25-0.90

L2 Attendance Variability

150

0.31 (0.05)

0.17-0.41

LIHC + Mindfulness

150

48.67%

0-1

Mean Student Age

150

10.01 (1.54)

7.47-12.85

Number of Students in Class

150

26.19 (6.23)

12-53

Proportion Female Students

150

0.50 (0.12)

0.12-0.79

School In Tent

150

27.33%

0-1

Region: Bekaa

150

51.66%

0-1

7

8 of 17

Measures

​

    • Student Monthly Attendance Rate (November – March)
    • Strengths and Difficulties Questionnaire (Goodman, 2007)
      • Prosocial Behavior
      • Emotional/Behavioral Problems
    • Brain Games Executive Function Survey (Jones, unpublished)

​

​

​

​

​

​

Baseline

​

Midline

Measure

Scale

M (SD)

​

M (SD)

Prosocial Behavior

0-2

1.08 (0.49)

​

1.13 (0.52)

Emotional/ Behavioral Problems

0-2

0.65 (0.38)

​

0.59 (0.37)

Executive Function

0-4

2.19 (0.78)

​

2.32 (0.84)

8

9 of 17

Analysis

  • Multilevel Models (student < classrooms)
  • DVs: Midline Student Outcomes
    • Prosocial Behavior
    • Emotional/ Behavioral Problems
    • Executive Function
  • Key Predictors:
    • Student Level:
      • Average Monthly Attendance Rate
      • Attendance Variability Across Months
    • Classroom Level:
      • Average Student Attendance Rate Across Months
      • Variability in Student Attendance Across Months
      • Treatment Group (LIHC vs. LIHC + Mindfulness)
  • Control Variables:
    • Student Level: Age, Female, Grade, Child Labor, Baseline Score
    • Classroom Level: Mean Age, % Female, Class Size, School in Tent, Region

​

​

​

​

9

10 of 17

Results of Main Effects Models

*** p < .001, ** p < .01, * p < .05, + p < .1

 

Prosocial Behavior

 

Emotional/ Behavioral Problems

 

Executive Function

​

b

​

SE

​

b

​

SE

​

b

​

SE

Student Level (L1)

 

 

 

 

 

 

 

 

 

 

 

Average Attendance

0.21

***

0.05

​

-0.21

***

0.03

​

0.51

***

0.08

Attendance Variability

-0.07

​

0.09

​

0.15

*

0.07

​

0.13

​

0.15

Age

-0.01

​

0.01

​

0.01

**

0.01

​

-0.01

​

0.01

Female

0.15

***

0.02

​

-0.11

***

0.01

​

0.22

***

0.03

Grade

0.05

***

0.01

​

-0.04

***

0.01

​

0.12

***

0.02

Labor

0.07

*

0.03

​

-0.07

**

0.02

​

0.12

*

0.05

Baseline Score

0.21

***

0.02

​

0.23

***

0.02

​

0.37

***

0.02

Classroom Level (L2)

 

 

 

 

 

 

 

 

 

 

 

Average Student Attendance

0.16

​

0.26

​

-0.06

​

0.18

​

0.85

*

0.36

Student Attendance Variability

0.48

​

0.68

​

0.63

​

0.47

​

0.22

​

0.94

Group: LIHC + Mindfulness

-0.02

​

0.05

​

0.04

​

0.04

​

0.06

​

0.07

Mean Student Age

0.00

​

0.02

​

0.01

​

0.01

​

0.01

​

0.03

Proportion Female

-0.19

​

0.21

​

0.17

​

0.15

​

-0.42

​

0.29

Class Size

0.00

​

0.00

​

0.00

​

0.00

​

0.00

​

0.01

School in Tent

-0.18

**

0.06

​

0.00

​

0.04

​

-0.10

​

0.08

Region: Bekaa

0.00

​

0.06

​

-0.15

***

0.04

​

0.16

+

0.09

10

11 of 17

      • Student-level average attendance rates consistently predicted more positive outcomes (more prosocial behavior, lower rates of emotional/behavioral problems, higher executive function scores).
      • Classroom-level average attendance predicted greater gains in executive function .

​

Question 1: Do student and classroom-level attendance rates and attendance variability have unique effects on student outcomes?

​

11

12 of 17

Results: Impact Variation Models

*** p < .001, ** p < .01, * p < .05, + p < .1

Note: All models additionally controlled for: age, female, grade, child labor, and baseline scores at the student level as well a mean student age, the proportion of female students, class size, school in tent, and region at the classroom level.

​

 

Prosocial Behavior

 

Emotional/ Behavioral Problems

 

Executive Function

​

b

SE

​

b

​

SE

​

b

​

SE

Student Level (L1)

 

 

 

 

 

 

 

 

 

 

 

Average Attendance

0.31

***

0.07

​

-0.20

***

0.05

​

0.64

***

0.12

Attendance Variability

-0.15

​

0.13

​

0.18

​

0.09

​

0.26

​

0.21

Classroom Level (L2)

 

 

 

 

 

 

 

 

 

 

 

Average Student Attendance

0.01

​

0.31

​

0.03

​

0.21

​

1.01

**

0.42

Student Attendance Variability

-0.37

​

0.86

​

1.36

*

0.58

​

-1.51

​

1.17

Treatment: LIHC + Mindfulness

0.00

​

0.05

​

0.04

​

0.04

​

0.04

​

0.07

Interactions

 

 

 

 

 

 

 

 

 

 

 

L1 Average Attendance:Treatment

-0.16

+

0.09

​

-0.02

​

0.07

​

-0.22

​

0.15

L1 Attendance Variability: Treatment

0.19

​

0.18

​

-0.08

​

0.13

​

-0.23

​

0.30

L2 Average Attendance: Treatment

0.61

​

0.50

​

-0.42

​

0.34

​

0.06

​

0.68

L2 Attendance Variability: Treatment

2.10

+

1.23

​

-1.73

*

0.84

​

3.70

*

1.68

12

13 of 17

Interaction Effects

13

14 of 17

      • Classroom-level attendance variability predicted more negative outcomes for students in the LIHC program.
      • The LIHC + Mindfulness program protected against the effects of classroom-level attendance variability on emotional/behavioral problems and executive function.

​

Questions 2: Do the effects of student and classroom-level attendance rates and variability differ depending on treatment condition?

14

15 of 17

Future Directions

  • Exploring Selection Effects
    • What predicts variability in student attendance?
    • Profiles of classrooms with higher attendance variability

​

    • Analyzing Cycle 2 Data (LIHC + Brain Games)

​

    • Analyzing Effects on Academic Outcomes

​

15

16 of 17

Acknowledgements

Training Interdisciplinary Educational Scientists

  • The first author was supported by the Department of Education IES training grant (R305B090007).

RC Lebanon 2016-2017:

  • In alphabetical order: Samer Houshaimi, Michele Boujikian, Mahmoud Bwary, & AnasTahhan

NYU 3EA Team 2016-2017

  • Mayari Montes De Oca, Sima Shabaneh

IRC HQ Team:

  • Drs. Jeannie Annan, Paul Frisoli, and Sarah Smith, & Autumn Brown, Rebecca Fernandes, Bryan Plummer, Jamie Weiss-Yagoda

NYU Global TIES Colleagues:

  • Drs. Peter Halpin, Alice Wuermli, Hiro Yoshikawa, & Shirley Archer-Fields, Marian Haji-Mohammed, Nikki Hill

Funders 2016-2017

  • NYU Abu Dhabi, Dubai Cares, US State Department (PRM)

​

16

17 of 17

 

Prosocial Behavior

 

Emotional/ Behavioral Problems

 

Executive Function

​

b

SE

​

b

​

SE

​

b

​

SE

Student Level (L1)

 

 

 

 

 

 

 

 

 

 

 

Average Attendance

0.31

***

0.07

​

-0.20

***

0.05

​

0.64

***

0.12

Attendance Variability

-0.15

​

0.13

​

0.18

​

0.09

​

0.26

​

0.21

Age

-0.01

​

0.01

​

0.01

**

0.01

​

-0.01

​

0.01

Female

0.15

***

0.02

​

-0.11

***

0.01

​

0.22

***

0.03

Grade

0.06

***

0.01

​

-0.04

***

0.01

​

0.12

***

0.02

Labor

0.06

*

0.03

​

-0.07

**

0.02

​

0.11

**

0.05

Baseline Score

0.21

***

0.02

​

0.23

***

0.02

​

0.37

***

​

Classroom Level (L2)

 

 

 

 

 

 

 

 

 

 

 

Average Student Attendance

0.01

​

0.31

​

0.03

​

0.21

​

1.01

**

0.42

Student Attendance Variability

-0.37

​

0.86

​

1.36

*

0.58

​

-1.51

​

1.17

Treatment: LIHC + Mindfulness

0.00

​

0.05

​

0.04

​

0.04

​

0.04

​

0.07

Mean Student Age

0.00

​

0.02

​

0.01

​

0.01

​

0.02

​

0.03

Proportion Female

-0.24

​

0.22

​

0.21

​

0.15

​

-0.43

​

0.29

Class Size

0.00

​

0.00

​

0.00

​

0.00

​

0.01

​

0.01

School in Tent

-0.18

**

0.06

​

0.00

​

0.04

​

-0.12

​

0.08

Region: Bekaa

0.00

​

0.06

​

-0.16

***

0.04

​

0.17

+

0.09

Interactions

 

 

 

 

 

 

 

 

 

 

 

L1 Average Attendance:Treatment

-0.16

+

0.09

​

-0.02

​

0.07

​

-0.22

​

0.15

L1 Attendance Variability: Treatment

0.19

​

0.18

​

-0.08

​

0.13

​

-0.23

​

0.30

L2 Average Attendance: Treatment

0.61

​

0.50

​

-0.42

​

0.34

​

0.06

​

0.68

L2 Attendance Variability: Treatment

2.10

+

1.23

​

-1.73

*

0.84

​

3.70

*

1.68

17