levels of analysis

SEED: save, earn, & enjoy deposits

green bank savings account with withdrawal restriction for rural communities

what should i measure?

measures are useful to inform how resources can be aligned with goals

potential data pitfalls

separate issue from measurement completeness is data quality

potential data pitfalls: survey

• reporting bias

• hard-to-measure outcomes

• attrition (drop-outs) or missing data

• surveyor bias

• selection bias

• regional bias

other potential data pitfalls

• institutional data

• “big data”

• data used as a proxy for harder to measure outcome

SEED program outcomes

• increased savings

• increased decision making power for women in households

as you are better able to estimate the counterfactual…

you move up the “ladder of inference” and have greater causal confidence

mission drives…

measures

measures drive…

strategy

consider ____ & ______

completeness of your measures & your data quality