Workshop · 30 minutes · then your portfolio piece
Same job as the opener, harder cases. For each study, name the design and decide whether the authors' claim is one that design can support.
1 The studies
State Medicaid research office · published 2024
Analysts pulled the enrollment records of 42,000 adults who joined a state Medicaid managed-care program during 2015. From the enrollment file they recorded whether each person's plan included a behavioral health benefit at the time they joined — a feature that varied by which plan was available in their county, not by anything the person chose. None of the 42,000 had a depression diagnosis in the year before enrollment.
The analysts then read forward through claims data from 2016 through 2023, recording who received a first depression diagnosis and when. Over eight years, 6,900 people did. Enrollees whose plans had included the behavioral health benefit were diagnosed earlier in the course of their illness and were more likely to receive follow-up treatment within ninety days. The analysts note that claims data capture only care that was billed, and that people who left the program were followed until their last claim.
What the authors concluded
“Enrollees with a behavioral health benefit at enrollment went on to be diagnosed earlier and treated sooner than those without one.”
University nutrition department · published 2023
Researchers recruited 620 adults who had received a diagnosis of type 2 diabetes within the past year, through three endocrinology clinics. They also recruited 620 adults of similar age, sex and neighborhood who had been screened at the same clinics and had normal blood glucose. All 1,240 participants completed a food frequency questionnaire asking them to estimate how many sugar-sweetened beverages they had consumed per week over the previous five years.
Participants with diabetes reported an average of 8.2 sugary drinks per week over that period; the comparison group reported 4.6. The difference persisted after adjusting for body mass index at the time of interview and for household income. The study was funded by a university research fund with no industry support, and the authors state that dietary recall over a five-year window is subject to error.
What the authors concluded
“Sugary drink consumption is a major cause of the type 2 diabetes epidemic, and taxing these beverages would substantially reduce new cases nationwide.”
National health statistics center · published annually
Each year since 2014, the center has conducted a national household survey on mental health. Every year a new random sample of roughly 27,000 adults is drawn; no one is deliberately surveyed twice, and the samples are independent from year to year. Each respondent completes the interview once, answering questions about symptoms in the past thirty days.
The share of adults aged 18 to 25 reporting frequent anxiety symptoms rose from 8.1% in 2014 to 14.7% in 2023. Over the same period, the share among adults aged 50 to 64 was roughly flat. The center publishes the figures with confidence intervals and notes that the survey mode shifted from in-person to a mix of phone and web interviewing in 2019.
What the authors concluded
“As young adults have moved through their twenties over the past decade, their anxiety has steadily worsened — a generation is becoming more anxious as it ages.”
Clinical psychology journal · published 2022
Researchers at a private residential university recruited 280 undergraduate volunteers through the psychology department's participant pool, which grants course credit for participation. Volunteers had to be aged 18 to 22, enrolled full time, and free of a current psychiatric diagnosis. Participants were randomly assigned by computer to either an eight-week guided mindfulness program or a waitlist control condition.
Assignment was concealed from the research staff who scored the outcome measures. At eight weeks, 264 of the 280 participants completed a standardized stress inventory. The mindfulness group scored meaningfully lower than the waitlist group, and the difference was larger than chance would explain. The trial was preregistered and the authors report all prespecified outcomes.
What the authors concluded
“Mindfulness training reduces stress in adults. Employers and health systems should adopt eight-week mindfulness programs to address workforce stress.”
2 Your task
Groups of three or four · 30 minutes
| Design — and what told you | Does the claim hold? If not, what should they have said? | |
|---|---|---|
| E | ||
| F | ||
| G | ||
| H |
3 Portfolio piece
This is the second piece in your public portfolio. It is not an assignment write-up and it should not read like one. Someone who has never met you should be able to open it and see, in under two minutes, that you can look at a published health claim and say exactly where it goes wrong.
How an opening might sound
A county health survey reported last year that regular exercise reduces depression, and recommended that the county fund exercise programming as a mental health intervention. The finding behind that recommendation was real: adults who reported exercising four or more days a week were less than half as likely to report frequent depressive symptoms. But the survey asked both questions on the same form, on the same day, of the same people — and that is all it did…
This is a sample of tone and pace, not a template to fill in. Write it in your own voice.
Write the piece yourself first — all of it, in a document, finished. Then have Copilot build the page. The writing is yours; the formatting is not worth your time.
Step 1. Paste this prompt into Copilot, then paste your finished piece underneath it:
I have written a short analytical piece for my public portfolio site on GitHub Pages. Build the page for me. I do not write code. Give me exactly two things, clearly labeled, and nothing else: THING 1 - the file. Start by writing the line: "Save this as: cross-sectional.html" Then give me one complete HTML file in a single code block, from <!DOCTYPE html> to </html>. Do not shorten it or leave anything out. THING 2 - the link. Give me one line of HTML, in its own code block, that I can paste into my homepage so people can reach this page: <a href="cross-sectional.html">MY TITLE HERE</a> with my real title filled in. Rules for the file: - Fix my spelling, grammar and punctuation, and split a run-on sentence if you need to. That is the only editing you may do. - Keep my wording and my ideas. Do not rewrite sentences to sound better, do not add or remove points, do not add an introduction or a conclusion, and do not invent a new title. - At the very end, after everything else, list any sentence you changed by more than a word or two, so I can check it. - One <h1> with my title, then my paragraphs exactly as I wrote them. - Clean, readable typography. Generous line height. A single column no wider than about 700px, centered. It must read well on a phone. - All CSS inside a <style> tag in the same file. No external files, no frameworks, no JavaScript. - At the bottom, in small text: my name and "PH 320 · Applied Biostatistics in Public Health". - Add a link at the bottom back to my homepage: <a href="index.html">Home</a> Do not explain anything. Do not give me instructions. Just the two things above. Here is my piece:
Step 2. Check it before it goes anywhere.
Copilot will fix your typos and grammar, which is fine — that is proofreading. What it may not do is think for you. Read its version against your draft, and look at the list of changes it made at the end. If it reworded an idea, or smoothed a sentence into something you would not say, put yours back. Your judgment is the graded part.
Step 3. Put the file in GitHub.
In your repository, click Add file → Create new file. Name it exactly cross-sectional.html — all lowercase, no spaces. Paste in everything Copilot gave you under "Thing 1," then click Commit changes. Your site rebuilds itself in about a minute.
Do not name it index.html. That is your homepage, and you would erase it.
Step 4. Link it from your homepage. This is the step people forget.
Right now nothing points to your new page, so nobody can find it. Open index.html, click the pencil icon to edit, find the portfolio card that says “Coming soon,” and replace that text with the link Copilot gave you under “Thing 2.” Commit that too.
Step 5. Test it the way a stranger would.
Open username.github.io, and click through from the homepage to your new piece. If you can get there without typing a URL, you are done.
One rule about the AI
Proofreading is fine. Spelling, grammar, a run-on sentence cut in two — let it. What it may not do is think for you. If you hand it a topic and let it write the piece, you have published someone else’s judgment under your own name, and the judgment is the only part of this an employer is reading for. The words can be tidied. The thinking has to be yours.