of Development Programs and Projects Class 7 Randomization how to design an RCT that accommodates real world constraints Randomly sample from area of interest Clarification Random ID: 628315
Download Presentation The PPT/PDF document "Monitoring and Evaluation" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Monitoring and Evaluation of Development Programs and Projects
Class
7
–
Randomization*
*(how to design an RCT that accommodates real world constraints)Slide2
Randomly
sample
from area of interest
Clarification:
Random
sampling
vs.random
assignmentSlide3
Randomly
sample
from area of interest
Randomly
assign
to
treatment
and
control
Random sampling and random assignment
Randomly
sample
From both
treatment and controlSlide4
BEFORE YOU EVEN TALK ABOUT RANDOMIZATION DESIGN…Explain why randomization is necessaryTalk about attribution
A
rticulate
why
(non-random) comparison group may be different and specify way this may bias resultsExplain that the intervention may have unintended (good or bad) consequences worth measuring
Can measure spilloverCost-benefit analysesLecture OutlineSlide5
Lecture OutlineUnit of RandomizationWays to RandomizeSimple Randomization (eg
: lottery)
Randomization “in the bubble”
Phase-inRotationEncouragement Design
Multiple treatments, 2 stage(if time) StratificationSlide6
Unit of Randomization?
Options
Individual
ClusterWhich level to randomize?ConsiderationsWhat unit does the program target for treatment?
What is the unit of analysis?Slide7
Unit of Randomization: Individual?Slide8
Unit of Randomization: Individual?Slide9
Unit of Randomization: Clusters?
“Groups of individuals”: Cluster Randomized TrialSlide10
Unit of Randomization: Class?Slide11
Unit of Randomization: Class?Slide12
Unit of Randomization: School?Slide13
Unit of Randomization: School?Slide14
School Tutoring programIn the example discussed last week, students were tutored after school how did we randomize last week?
What if students are taken out of class to be tutored, say by a teaching assistant?
how would you randomize?Slide15
Randomizing at the child-level within classesRandomizing at the class-level within schoolsRandomizing at the community-level
Think about:
logistics, fairness
, politicsSlide16
Nature of the TreatmentHow is the intervention administered?What is the catchment area of each “unit of intervention”How wide is the potential impact?
Aggregation level of available data
Do you want to measure spillover?
Power requirementsGenerally, best to randomize at the level at which the treatment is
administered.How to Choose the Level / UnitSlide17
Sometimes a program is only large enough to serve a handful of communitiesPrimarily an issue of statistical powerWill be addressed
next week
Think about: sample sizeSlide18
Lecture OutlineUnit of RandomizationWays to RandomizeSimple Randomization (
eg
: lottery)
Randomization “in the bubble”Phase-inRotation
Encouragement DesignMultiple treatments, 2 stage(if time) StratificationSlide19
Simple Randomization of bank branches:
Interest Rate Elasticity
“high rate”
“
low rate”
Final
sample included 132 branch offices in 80 geographic
clustersSlide20
Simple Randomization: LotterySchool Voucher Program in Colombia (PACES)Student
must be entering 6th grade and under 15 years old
Students must provide evidence that they live in poor
neighborhoodRenewable
through graduation unless student is retained in a gradeVouchers awarded by lottery if demand exceeds supplyCovered about 60% of feesKey Findings (after 3 years) on Voucher Recipients:
Increased Usage of Private SchoolsHigher Educational AttainmentNo Difference in Drop-out RatesLess Grade RepetitionHigher Test Scores
Less Incidence of Teen-age EmploymentFurther research on peer effectsSlide21
Lotteries are simple, common and transparentRandomly chosen from applicant poolParticipants know the “winners” and “losers”Simple lottery is useful when there is no a priori reason to discriminate
Perceived as fair
Transparent
LotteriesSlide22
What if you have 500 applicants for 500 slots?Could increase outreach activities(but think of external validity)Sometimes screening mattersSuppose there are 2000 applicants for 500 slots
Screening of applications produces 500 “worthy”
candidates
A simple lottery will not
work - What are our options?Lottery (continued)Slide23
What are they screening for?Which elements are essential?
Selection procedures may exist only to reduce eligible candidates in order to meet a capacity constraint
If certain filtering mechanisms appear “arbitrary” (although not random), randomization can serve the purpose of filtering
and
help us evaluateConsider the screening rulesSlide24
Randomization “at the margin”Organizations may not be willing to randomize among eligible people.But
might be willing to randomize
those at the margin –
ie, those who are borderline in terms of eligibility
Just above the threshold not eligible, but almostWhat treatment effect do we measure? What does it mean for external validity?(hint: review RDD from last week)Slide25
Randomization at the margin: Impact of consumer creditDean Karlan and Jonathan
Zinman
worked with a bank in South Africa
Loan officers ranked applicants as “egregiously uncreditworthy” or “marginally
uncreditworthy”Randomly selected marginal applicants to be reconsidered53% of those were offered a loanLook at impact of credit on those randomized into the “reconsider group” Not just those offered loan… more on this next week. Found that access to credit increased likelihood that clients would retain job, increase income, feel less food insecurity
Marginal loans are profitable, but less than regular loansSlide26
Take advantage of operational constraintsTypical during expansion phaseEveryone gets program eventuallyFigure out what determines order of expansion
Examples:
Progresa
(Mexico), Deworming
(Kenya)Phase-inSlide27
Phase-in design
Round 1
Treatment
: 1/3
Control
:
2/3
Round 2
Treatment
: 2/3
Control
:
1/3
Round 3
Treatment
: 3/3
Control
:
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Round 1
Treatment
: 1/3
Control
:
2/3
Round 2
Treatment
: 2/3
Control
:
1/3
Randomized evaluation ends
Randomized evaluation endsSlide28
AdvantagesEveryone gets something eventually
Provides incentives to maintain contact
Concerns
Can complicate estimating long-run effects
Care required with phase-in windows
Do expectations of change actions today?Phase-in designsSlide29
Round 1
Treatment
: 1/2
Control
: 1/2
Rotation design
Round 2
Treatment
from Round 1
Control
——————————————————————————
Control
from Round 1
Treatment
Round 1
Treatment
: 1/2
Control
:
1/2Slide30
Groups get treatment in turnsGroup A gets treatment in first periodGroup B gets treatment in second period
Advantages
Perceived as fairer; easier to get accepted
Concerns
If people in Group B anticipate they’ll receive the treatment the next period, they can have a different behavior in the first periodImpossible to measure long-term impact since no control group after first period
RotationSlide31
Extra Teacher program: Rotation with schools
Group
Year
1
Year 2
Year 3
AGrade 3
Grade 4Grade 3
B
Grade 4Grade 3
Grade 4
Schools in Varodara, India were divided into two groups: ( A and B)and teaching assistant offered in schools according to the following schedule:Slide32
Sometimes it’s practically or ethically impossible to randomize program accessBut most programs have less than 100% take-upRandomize encouragement to receive treatment
Encouragement design: What to do
when you can’t randomize accessSlide33
Something that makes some folks more likely to use program than othersNot itself a “treatment”
For whom are we estimating the treatment effect?
Think about who responds to
encouragement
Do not choose an encouragement that affects those who are different than the entire populationWhat is “encouragement”?Slide34
Methods of randomization - recapSlide35
Lecture OutlineUnit of RandomizationWays to Randomize
Simple Randomization (
eg
: lottery)
Randomization “in the bubble”Phase-inRotationEncouragement DesignMultiple treatments, 2 stage
(if time) StratificationSlide36
Treatment 1
Treatment 2
Treatment 3
Multiple treatmentsSlide37
Manage or Measure SpilloverSpillover: when the control group, although “untreated”, is affected (positively or negatively) by the treatmentChoose unit that contains
spillover (
ie
randomize at school or village rather than individual level)Measure Spillover: TUP (
Bangladesh, Honduras, Peru, Pakistan, Ghana, Ethiopia, Yemen)Slide38
2 Stage Example: Targeting the Ultra PoorProgram which consists of targeting poorest families in a village and providing consumption support ($$ or food), asset transfer, livelihood training
Eventual graduation to
microfinanceSlide39
TUP ASSET TRANSFERSSlide40
TUP Evaluation Design I
Communities
(Total
= 80)
40 treatment
40 control
C 20 cont
Households(Total =
1600)
A 20 treat
Ho do we determine impact?
Program direct impact: A-C Slide41
TUP: INDIRECT EFFECTS
Villagers may benefit from neighbors being treated
Or they may be negatively affected
Why would it be desirable (from a policy perspective) to measure spillover?Slide42
TUP Evaluation Design II
Communities (Total
= 80)
40 treatment
40 control
C
20 cont
Households(Total = 2400)
A 20 treat
B 20
cont
Ho do we determine impact?
Program direct impact: A-C
Program indirect impact: B-C Slide43
StratificationObjective: when you have a small
sample, make sure key variables are balanced between Treatment and Control
What is it:
dividing the sample into different subgroups
selecting treatment and control from each subgroupStratify on variables that could have important impact on outcome variable (bit of a guess
)Stratify on subgroups that you are particularly interested in (where may think impact of program may be different)Can
get complex to stratify on too many variablesMakes the draw less transparent the more you stratifySlide44
Next WeekSampling!