Previously you have written proposals to conduct observational and survey-based studies. This week we have assessed how randomized controlled experiments can be used to draw inferences, assert causal relationships, and identify mechanisms for changes in behavior.
In a short paper (1200-1500 words), propose a research study based on a question you find interesting and explain how you would use a computationally-enhanced experiment to answer your question. You should (briefly) state your topic of interest and the specific question you seek to answer in this research project. Next, identify the basic structure and format of the experiment.
The bulk of your proposal should be a critical assessment of why you believe this project will be a good example of digital research. This includes:
- A clearly stated research question
- A clearly stated plan for using a computationally-enhanced experiment
- Your experiment does not have to be fully digital. At minimum, you need at least one of the four steps (recruit participants, randomize, deliver treatments, and measure outcomes) to utilize digital systems.
- An assessment of the experimental design
- This should include a discussion of the experiment's validity and its ability to measure heterogeneity of treatment effects and causal mechanisms
- Your experiment does not have to maximize all of these components. If you decide to concentrate on just one specific aspect (for instance, only estimating the average treatment effect) or have to make trade-offs (e.g. sacrificing external validity for internal validity), justify these decisions.
A suggestion for your research question: use the same question from your observational study and survey design papers. Then you can compare and contrast the benefits and drawbacks of implementing an experimental design vs. an observational study or survey.
See here for instructions on submitting course assignments.
Submit your paper as a Markdown document (.md).
Submit your pull request before class on Monday, October 30 (11:30 am).