Workshop 4

Analysis Plan

What should you produce?

  1. A written data analysis plan. This should be 1-3 paragraphs (depending on how complicated your research question and analysis is). It should describe what statistical model you’re going to use to address your research question. You can/should use equations when relevant (i.e., write out your regression model).
  2. Your analysis plan needs to describe what software you’re going to use. You can provide example code for your model. You should state what commands or routines you’re going to use (i.e., regress in Stata, lme4 in R, etc.)
  3. Your analysis plan needs to be clear about what parts of your model provide information about your research question (e.g., if you are doing a treatment study, what parameter is the treatment effect?). If your research question has multiple parts, then make sure you show each part is covered.
  4. Finally, your analysis plan should include a brief example of a table and a figure you’re going to make.

Steps to take

  1. Find 2-3 papers in the subject area. Describe (bullet points are fine) how the researchers analyzed their data? Did they provide an analysis plan? Did they explicitly connect their analysis model to their research question? Are there alternative approaches to analyzing that kind of data? Did they miss anything important?
  2. Identify a statistical model that you want to use in for your main research question. Be specific. What are the main parts of the model and which parts will be used to address your research question? For example, if you were going to use regression, be specific about what your outcome variable is, what your predictors are, and how you’ll interpret each coefficient. Which coefficients would address your main research question?
  3. As you consider your analysis plan, be sure to consider potential issues that could come up with your analysis. For example, how are you going to deal with any missing data? How are you going to deal with multiple comparisons?
  4. What software are you going to use to estimate your model? What command/routine within the software are you going to use? Write out the syntax for your main model. For the purposes of this class, I want you to include example syntax in your paper.
  5. What other analyses are you going to do? What kind of descriptive statistics, etc.?
  6. How will you present your results? Look at how other papers in your area present their results – graphs and tables. What are the strengths and weaknesses of their presentation? Design 1 figure and 1 table to present your main results. For the table, just create an empty table. For the figure, describe what kind of figure it will be and what the elements within the figure will be.
  7. Write your analysis plan (i.e., bring everything together).

Parts of the Workshop

Complete the relevant steps prior to the due date. See Learning Suite for the due dates.

  • Workshop 4a: Steps 1-5
  • Worskhop 4b: Steps 6-7

Feedback to Colleagues

These are suggestions – they aren’t the only the things you should consider when evaluating your colleagues’ work. Be thoughtful and remember to provide actionable feedback.

  1. Can you clearly state what analysis your colleagues are going to use? If yes, say what it is to make sure that your colleagues can be confident that they have described their analysis well. If not, what is missing? What do they need to provide more information about?
  2. Do your colleagues connect their analysis to their research questions? Are you clear what parts of the model/analysis address the main research questions? Do you know what the outcome variable is in each model? Do you know what the predictors are? Do you understand the variables used are relevant to the research question?
  3. Have your colleagues stated what software and what commands they are going to use to perform their analysis?
  4. What concerns do you have about the analysis? Are they going to deal with multiple comparisons in some way? Missing data? P-hacking?
  5. Are the proposed tables and figures for presenting the results reasonable? If so, why? If not, why not?

What will you turn into to me?

Turn in your completed analysis plan (4a + 4b) that takes into account your colleagues’ feedback. You should also include the comments your received from colleagues as well your brief summary of how you dealt with their feedback.

%d bloggers like this: