0
$\begingroup$

I would appreciate your help with a question I have.

I'm creating a Difference-in-Difference study to examine how a conditional cash transfer to individuals 18 years of age to be spent in sport activities can encourage participation in sports in the long run. I'm thinking about using data from the government which polls people about their participation in sports. It offers data in age groupings such 18–19 years, 20–24 years, and 25–29 years. I would contrast the changes in sport consumption among 18–19-year-olds (treatment group) with the trends of 25–29-year-olds (control group) along 5 years.

Are there any possible problems when comparing the combined data of 25-29 years cohort with the data from the 18-19 years old?

Thanks so much for your help

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

I think we should start with the identification problem. Suppose you have a basic DID model with post dummy( 1 for periods after the transfer) and treatment dummy (1 for the 18-19 age group and 0 for the 25-29 age group). Then the coefficient of the interaction term is our interested parameter.

You can always estimate it, but to give it a causal interpretation, we need additional assumption, which is the parallel trend assumption.

This assumption basically says your treatment group and control group must evolve in parallel absent the treatment. One violation I could think of is for the 25-29 age group, they enters college or labor market, thus have less time than 18-19 age groups. Thus you could see the two groups have different sport consumption even absent of the cash transfer.

The 25-29 group has lower consumption itself would not be a problem if the trend is parallel to the trend of 18-19 group, but would be a problem if their difference already show converge or diverge trend before the treatment.

There might of course be other problem, but this is the one I can think of now, hope it helps.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Tnx for your input, Eileen! Do you believe I'm making the right choice by opting for DiD over pursuing RD? RD could be an option given the policy's clear age thresholds and the ability to demonstrate the absence of strategic manipulation but it's challenging to obtain a sufficiently large sample exclusively around the age threshold (eg. 17yrs 11mtn and 15 days). Additionally, considering the focus on long-term effects and the availability of a substantial dataset, DiD seems to offer greater adaptability and reliability in addressing temporal trends. What are your thoughts on my reasoning? Tnx $\endgroup$ Commented Feb 24, 2024 at 19:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.