31
AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Quasi Experimental Methods I Florence Kondylis

AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Embed Size (px)

Citation preview

Page 1: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

AADAPT Workshop Latin AmericaBrasilia, November 16-20, 2009

Non-Experimental Methods

Quasi Experimental Methods I

Florence Kondylis

Page 2: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

What we know so far

Aim: We want to isolate the causal effect of our interventions on our outcomes of interest

Use rigorous evaluation methods to answer our operational questions

Randomizing the assignment to treatment is the “gold standard” methodology (simple, precise, cheap)

What if we really, really (really??) cannot use it?!

>> Where it makes sense, resort to non-experimental methods

Page 3: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

3

Objective

Find a plausible counterfactual Every non-experimental method is

associated with a set of assumptions The stronger the assumption, the

more doubtful our measure of the causal effect

▪ Question our assumptions▪ Reality check, resort to

common sense!

Page 4: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

4

Principal Objective▪ Increase maize production

Intervention▪ Fertilizer Vouchers distribution▪ Non-random assignment

Target group▪ Maize producers, land over 1 Ha &

under 5 Ha Main result indicator

▪ Maize yield

Example: Fertilizer Voucher Program

Page 5: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Before After0

2

4

6

8

10

12

14Control GroupTreatment Group

5

(+) Impact of the program

(+) Impact of external factors

Illustration: Fertilizer Voucher Program (1)

Page 6: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Before After0

2

4

6

8

10

12

14Control GroupTreatment Group

6

(+) BIASED Measure of the program impact

Illustration: Fertilizer Voucher Program (2)

“Before-After” doesn’t deliver results we can believe in!

Page 7: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Before After0

2

4

6

8

10

12

14Control GroupTreatment Group

7

« After » difference btwparticipants andnon-participants

Illustration: Fertilizer Voucher Program (3)

« Before» difference btw participants and nonparticipants

>> What’s the impact of our intervention?

Page 8: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

8

Difference-in-Differences Identification Strategy (1)Counterfactual: 2 Options

1.Non-participant maize yield after the intervention, accounting for the “before” difference between participants/nonparticipants (the initial gap between groups)

2.Participant maize yield before the intervention, accounting for the “before/after” difference for nonparticipants (the influence of external factors)

1 and 2 are equivalent

Page 9: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Difference-in-DifferencesIdentification Strategy (2)Underlying assumption:Without the intervention, maize yield for participants and non participants’ would have followed the same trend

>> Graphic intuition coming…

Page 10: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

10

Data -- Example 1

Average maize yield

(T / Ha)2007 2008 Difference

(2007-2008)

Participants (P) 1.3 1.9 0.6

Non-participants (NP)

0.6 1.4 0.8

Difference (P-NP) 0.7 0.5 -0.2

Page 11: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

11

Data -- Example 1

Average maize yield

(T / Ha)2007 2008 Difference

(2007-2008)

Participants (P) 1.3 1.9 0.6

Non-participants (NP)

0.6 1.4 0.8

Difference (P-NP) 0.7 0.5 -0.2

Page 12: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

12

NP2008-NP2007=0.8

Impact = (P2008-P2007) -(NP2008-NP2007)

= 0.6 – 0.8 = -0.2

2007 20080

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Participants Non-Participants

P2008-P2007=0.6

Page 13: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

13

2007 20080

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Participants Non-Participants

P-NP2008=0.5

Impact = (P-NP)2008-(P-NP)2007

= 0.5 - 0.7 = -0.2

P-NP2007=0.7

Page 14: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Assumption of same trend: Graphic Implication

2007 20080

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Participants Non-Participants

Impact=-0.2

Page 15: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Summary

Negative Impact: Very counter-intuitive: Increased input use should

increase yield once external factors are accounted for!

Assumption of same trend very strong

2 groups were, in 2007, producing at very different levels

➤ Question the underlying assumption of same trend!➤When possible, test assumption of same trend

with data from previous years

Page 16: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

2006 2007 20080

0.5

1

1.5

2

2.5

participantsnon-participants

Questioning the Assumption of same trend: Use pre-pr0gram data

>> Reject counterfactual assumption of same trends !

Page 17: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

17

Data – Example 2

Average maize yield

(T / Ha)2007 2008 Difference

(2007-2008)

Participants (P) 1.5 2.1 0.6

Non-participants (NP)

0.5 0.7 0.2

Difference (P-NP) 1.0 1.4 0.4

Page 18: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

182007 20080

0.5

1

1.5

2

2.5

participantsnon-participants

P08-P07=0.6

NP08-NP07=0.2

Impact = (P2008-P2007) -(NP2008-NP2007)

= 0.6 – 0.2 = + 0.4

Page 19: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Assumption of same trend: Graphic Implication

2007 20080

0.5

1

1.5

2

2.5

participantsnon-participants

Impact = +0.4

Page 20: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Conclusion

Positive Impact: More intuitive

Is the assumption of same trend reasonable?

➤ Still need to question the counterfactual assumption of same trends !➤Use data from previous years

Page 21: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Questioning the Assumption of same trend: Use pre-pr0gram data

>>Seems reasonable to accept counterfactual assumption of same trend ?!

2006 2007 20080

0.5

1

1.5

2

2.5

participantsnon-participants

Page 22: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Caveats (1)

Assuming same trend is often problematic No data to test the assumption Even if trends are similar in the past…

▪ Where they always similar (or are we lucky)?

▪ More importantly, will they always be similar?▪ Example: Other project intervenes in our nonparticipant villages…

Page 23: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Caveats (2)

What to do?

>> Be descriptive! Check similarity in observable characteristics

▪ If not similar along observables, chances are trends will differ in unpredictable ways

>> Still, we cannot check what we cannot see… And unobservable characteristics might matter more than observable (ability, motivation, etc)

Page 24: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Matching Method + Difference-in-Differences (1)

Match participants with non-participants on the basis of observable characteristics

Counterfactual: Matched comparison group

Each program participant is paired with one or more similar non-participant(s) based on observable characteristics

>> On average, participants and nonparticipants share the same observable characteristics (by construction)

Estimate the effect of our intervention by using difference-in-differences

24

Page 25: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Matching Method (2)

Underlying counterfactual assumptions

After matching, there are no differences between participants and nonparticipants in terms of unobservable characteristics

AND/OR

Unobservable characteristics do not affect the assignment to the treatment, nor the outcomes of interest

Page 26: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

How do we do it?

Design a control group by establishing close matches in terms of observable characteristics Carefully select variables along which to

match participants to their control group So that we only retain

▪ Treatment Group: Participants that could find a match

▪ Control Group: Non-participants similar enough to the participants

>> We trim out a portion of our treatment group!

Page 27: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

Implications

In most cases, we cannot match everyone Need to understand who is left out

Example

Score

NonparticipantsParticipants

MatchedIndividuals

Wealth

Portion of treatmentgroup trimmed out

Page 28: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

28

Conclusion (1)

Advantage of the matching method Does not require randomization

Page 29: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

29

Conclusion (2)

Disadvantages: Underlying counterfactual assumption is

not plausible in all contexts, hard to test▪ Use common sense, be descriptive

Requires very high quality data: ▪ Need to control for all factors that influence

program placement/outcome of choice Requires significantly large sample size

to generate comparison group Cannot always match everyone…

Page 30: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

30

Summary

Randomized-Controlled-Trials require minimal assumptions and procure intuitive estimates (sample means!)

Non-experimental methods require assumptions that must be carefully tested

More data-intensiveNot always testable

Get creative: Mix-and-match types of methods!

Page 31: AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis

31

Thank You