Recipes (40-50 min)
Look at the following two recipes. Even if you don’t cook, try to think about which one you prefer and why. Then, think about the possible purpose of the one recipe compared to the other recipe.
Share with a partner for 1-2 minutes.
What did you all think? What did you prefer? Why?
What do you think the purpose of Recipe 1 was (the complete text)? Recipe 2 (the complete text)?
Activity: With a partner or by yourself, transform the methods section from either the sleep study or the methods section from the CDC technical report into a recipe (NOTE: the CDC technical report does not have an explicit methods section, but from second paragraph of page 913 to page 914 before first heading).
On CourseWeb, I have a template for what this could look like. The main consideration, though, is to turn a methods section into a series of steps or instructions that one would have to take in order to replicate the study. If you prefer, you need not divide it into data/ingredients and data prep/methods/directions as the template is divided. If you go with the latter framing as set up in the template, the first section is more about data-to-get and the second section is about data-to-prepare and data-to-analyze.
NOTE: Here is the full link to the sleep study if you go with that. What I have on CourseWeb does not include the appendix that contains the survey.
Use the following questions based on the Miller chapter to think about what is (and is not) present in both methods sections and to help you write your recipe in individual steps.
THINGS TO LOOK FOR:
Data
- When were data collected? (all at same time? Over multiple events?)
- Where were data collected? (location, geography)
- How were data collected. Study design (case studies, experiment, observational [most of you are doing this], surveys, randomized control trial)
- How was sample or population collected? Method of randomization? Convenience sample?
- Technique for collecting: survey, focus group, surveillance (e.g., cancer registry), admin records (e.g., census), physical exam. Done via phone, in person, online? Talk about potential issues: sampling error, coding mistakes, selection bias, recall bias, potential self-reporting issues
- Who collected data? Researcher? Trained assistants? Automated methods? How?
- Study response rate? Population or sample? Did you lose people over time? How many?
- What variables were measured and how? How? (e.g., a specific question to mothers about their children’s asthma, ages were transformed to categorical age ranges, combining variable values into aggregate [e.g., total family income])
- What were units of measurement?
- Were their outliers? How’d you deal with them?
- Any missing values? Did you drop them or keep them? Did those values impact measurement in any way? How? (e.g., drop, imputation [inputting an ‘average’ value or ‘like’ value])
- Did you have any item non-response?
- What was the sample or population size? If sample, how representative was it of the population?
- Don’t rely on Miller only: What else do you think is important to say about a paper’s data?
Methods:
- What statistical methods were used? Why? (invisible: meet assumptions of all techniques and tests!)
- Were sampling weights used? Especially with random samples, sometimes “weights” are applied to underrepresented aspects of your sample to better account for population proportions. For example, if the population of a town included about 20% of people working in health care and social assistance (roughly the proportion of Pittsburgh!) and only 8% of your sample had this background, depending on your research question, you might want to apply a sampling weight to that 8% to make them count more. See this for more on sampling weights. This can get kind of complicated, especially with more advanced statistical techniques.
- Bring up equations if you are using a new method, otherwise, people will know common statistical techniques, so you don’t have to go into great detail.
- Don’t rely on Miller only: What else do you think is important to say about a paper’s methods?
Recipe Swap
Team up with another person or pair to compare your recipes.
What parts were fairly easy to write? Why? (e.g., had all the information you needed, clear language, good order in methods section)?
Were there any parts that were hard to write? Why (e.g., didn’t understand enough about the technical material, language wasn’t clear, they left out important information)?
Do you think there was enough stated about the data? About the methods?
Was anything “left out” that made sense to leave out? If so, what?
Close Out
What makes for a well-written methods section if one of the goals is to write in such a way that anyone could replicate your study? Use examples from the methods section you looked at today.
How much information should be in a methods section? How much should be left out? Did you feel like there should have been more or less information in either methods section we looked at? What?
Discussion section: Data and Methods Return
Talk about the strengths and limitations. Connect to existing literature. Explain how strengths and limitations should impact interpretation of your results. Talk about directions for future research.
How did either example do this?
Your Methods Section
Take some time to get started on the methods section of your project. How are you going to talk about your data? About your analysis?
Sci/Tech Writing Project Index (20 min)
To help with getting organized as you get to work, see below to help get you ready with some important aspects of the project:
- What genre of writing are you using?: Scientific or Academic paper for an academic journal, Technical Report, Grant Proposal, Something else? _____________________.
- If you are writing for an academic journal, which one (check out resource from 10/17 lesson plan)? If a technical report, what organization are you writing for? If a grant proposal, whom are you representing and/or whom are you writing to? If other, what publication/organization/audience? Find several examples (at least 5) of your genre of writing for that publication/organization to get a sense of the patterns in layout, style, etc. you’ll need to adapt to.
- What advanced calculation are you using? Are you performing your own analysis or are you citing something from secondary research? If performing your own analysis, do the assumptions hold? See here for assumptions for t-test, see here for assumptions for Pearson’s correlation coefficient, see here for assumptions for confidence interval.
- What public piece option are you choosing for the advanced calculation? If option 1, do you know what you’ll revise? If option 2, do you know what genre you are choosing and other important details (audience, organization)? Sample genres for option 2: press release, tweet thread, public statement, summary tech report on public-facing website). If going with option 2, find several examples (at least 5) of your genre of writing for that publication/organization to get a sense of the patterns in layout, style, etc. you’ll need to adapt to.
- In data-driven technical and academic writing, there will be lots of incorporation of secondary research in the introduction and usually also in the concluding section (and sometimes elsewhere, too). What research are you doing on your topic to show what others have said on your topic? What keywords would help you find things? What academic journals talk about this subject?
- What initial thoughts do you have for visuals in this piece?
Next Time (2-5 min)
-Read chapter 6 in Miller
-Work on your sci/tech writing project (first draft due 11/7)
-We will start transitioning into talking more about data visualization, and we will start by talking about tables and how they are utilized in technical and academic writing.