search

worldbank / dime-data-handbook

Development Research in Practice: The DIME Analytics Data Handbook. By Kristoffer Bjärkefur, Luíza Cardoso de Andrade, Benjamin Daniels, and Maria Jones
https://worldbank.github.io/dime-data-handbook/
Other
63 stars 26 forks source link

ch 3 - balance checks #337

Closed kbjarkefur closed 4 years ago

kbjarkefur commented 4 years ago

https://github.com/worldbank/d4di/blob/43b19d12da22e76d861f6eb3c004a4decf6f60d0/chapters/research-design.tex#L264

@mariaruth specify that these need to be done with pre-intervention data.

for that reason it's a little weird that this comes up in the section on cross-sectional data.

@bbdaniels This does not need to be the case. You can always check balance using other characteristics to make some assertion about the equivalence of groups even in a single period (https://www.sciencedirect.com/science/article/pii/S2214109X19300312)

kbjarkefur commented 4 years ago

Explain balance checks should be brought up as a general point before this, for example in the RCT section. Then in this cross sectional data section we can bring up what is different when you do not have baseline data

kbjarkefur commented 4 years ago

Do the same with:

https://github.com/worldbank/d4di/blob/43b19d12da22e76d861f6eb3c004a4decf6f60d0/chapters/research-design.tex#L266

@luizaandrade This is the only reference to quasi-experimental designs in this paragraph. I think it would read better if we start listing the requirements for RCTs, then moved on to how quasi-experiments differ from them.

@mariaruth Agreed, structured like the previous section

@bbdaniels Should I remove that from this section? I really don't want to list out for each design how these things differ since that is a general idea addressed in the previous and not intended to be the point of this section