top of page
Search

I Work Both Sides of the Room. I Am Not Neutral About Science.

Writer: Galen Cole
Galen Cole
Sep 22
12 min read

Updated: Sep 28

By Galen E. Cole, PhD, MPH, LPC @ www.galencole.com


I’m not here to change your vote.


I have sat with conservatives, liberals, and people who would rather not be either. I have sat with people who treat politics like a sacrament and people who treat it like a migraine. I have sat with couples who can love each other and still hear the same headline as two different emergencies.


That work has required political neutrality. Not because convictions do not matter. Because the room only works if the person across from me does not have to wonder whether I am recruiting them.


I am not neutral about science.


That is not a slogan. It is the method I was trained in, the method I used for more than two decades at the Centers for Disease Control and Prevention, and the method I still use when someone brings me a claim that is supposed to settle an argument.


Both sides of the political spectrum have taken different stands on science. Sometimes they have used it well. Sometimes they have used it as a costume. The costume is getting better. Deepfake audio, synthetic video, and AI-written “studies” can now look finished before a single measurement has been taken. If you care about truth that can be brought about by the mechanism of science, you need the basics again: what science is, how it is conducted when it is helping, how it is manipulated when it is selling, and how a non-specialist can tell the difference.


That is the whole argument.Not “trust the experts” as a blank check.


Not “do your own research” as a license to collect only the papers that already agree with you.


Stewardship of evidence.


Political neutrality is a professional requirement. Method is not optional.

A psychotherapy office is not a caucus room.


If a client is conservative, I do not need them to become liberal in order to sleep, stay sober, or stop detonating their marriage. If a client is liberal, I do not need them to become conservative in order to do the same work. Feelings follow focus. If I make the session a referendum on the news cycle, I am not treating the person in front of me. I am rehearsing a fight they already have at home.


I wrote about that habit in You Feel What You Focus On: Why a 24-Hour News Habit Can Wreck a Person. The point there still stands. You can care about the country and still let political rumination wreck sleep, blood pressure, and the people who actually live with you.


Science is a different category.


Science is not a team jersey. It is a way of asking a question so that the answer can survive contact with people who do not already like it. When a culture treats science as a loyalty test, two things happen at once. People who should be cautious become certain. People who should be curious become cynical. Both of those postures feel like strength. Neither one is a method.


I have watched this from both directions. One camp treats a preliminary finding as settled law if it confirms the moral story. The other camp treats a settled body of evidence as propaganda if it threatens the moral story. The content changes. The move does not.


That move is older than this news cycle. It is just faster now.


What science is — and what it is not

Science is not a pile of studies. It is not a lab coat. It is not a press release with a graph. It is not a man on a podcast saying “the data is clear.” It is not an influencer holding a bottle and saying “clinically proven.”


Science is a public method for reducing the chance that we are fooling ourselves.


At minimum, a scientific claim has to do five things:

  1. State a question that can be wrong. If nothing could count as disconfirming evidence, you are not doing science. You are defending a position.

  2. Say how the question will be tested before the result is known. Methods first. Conclusions later. If the method is invented after the data is in hand, the data is being dressed, not examined.

  3. Compare the thing of interest with a fair alternative. A control group. A placebo. A baseline. A comparison that is not the researcher’s hope.

  4. Expose the work to people who did not have a stake in the outcome. Peer review is imperfect. It is still better than publishing only to your own list.

  5. Survive repetition. One study is a clue. A body of independent replications is a finding. A systematic review of that body is closer to knowledge.


Notice what is missing from that list: the political usefulness of the result. A true finding can be inconvenient to your side. An untrue finding can be useful to your side. Usefulness is not a method.


This is why I am not neutral about science even when I am politically quiet in the room. Neutrality about parties is courtesy. Neutrality about method is how error becomes doctrine.


How good science is actually conducted

When science is working, it is slower and more modest than the version that shows up in a feed.


A serious study names the population. It says who was included and who was left out. It says how many people were studied, not just that “research shows.” It distinguishes a primary outcome — the thing the study was designed to measure — from a secondary finding that appeared after the fact. It reports effect size, not only whether a p-value crossed a fashionable line. It admits limitations in plain language. It tells you who paid for the work.


When I led the Health Communications Branch at CDC’s Office on Smoking and Health, it often felt like we were at war with an industry that had vast resources, sophisticated messaging, and decades of normalized harm on its side. Our side had measurement. We did not get to win an argument by producing one favorable paper and calling it a day. We had to look at the totality of the evidence, test messages, evaluate campaigns, and keep going after the first press cycle ended.


The tobacco playbook is worth remembering because it did not die with cigarettes. It migrated.


Deny. Delay. Sow doubt. Fund friendly research. Highlight the one study that refuses to replicate. Attack the messenger. Then sell a product that looks like health.


Alcohol was one of the first places the playbook put on a white coat.


How the playbook moved to the dinner table

A generation of adults can still recite the headline: a drink a day is good for the heart. Red wine became a health food. Moderate drinking became part of a “Mediterranean” identity. Dinner guests who declined a glass were treated as if they had missed a medical appointment.


Most people who repeated that line never read a methods section.


What they were usually repeating was an observational pattern called a J-curve. In large population studies, people who reported light or moderate drinking often looked healthier on some heart outcomes than people coded as abstainers. The graph dipped, then rose again among heavy drinkers. The dip was sold as proof that daily alcohol protects the heart.


The graph was never as clean as the dinner-table version.


First, many “abstainers” were not lifelong non-drinkers. They were former heavy drinkers who had stopped because they were already sick. That is the sick-quitter problem. If you put the damaged former drinkers in the zero-alcohol column, the people still sipping look healthier by comparison. That is not a benefit of ethanol. That is a sorting error.


Second, moderate drinkers in those datasets were, on average, different people. They often had more education, more money, better diets, more exercise, and better access to care. Alcohol got credit for a lifestyle it traveled with.


Third, the same molecule that was being praised for the heart is classified as a Group 1 carcinogen. A finding that looks friendly on one outcome can still be costly on another. A study designed to hunt for heart benefit and underpowered to see cancer is not a complete health story. It is a partial camera angle.


Fourth — and this is the part most people never heard — the alcohol industry did not merely enjoy the J-curve. Parts of the industry helped keep it in circulation.


Reviews written by authors with alcohol-industry funding have been more likely to conclude that drinking protects the heart. Industry-linked forums have spent years praising papers that found benefit and picking apart papers that found harm. That is not how an independent literature behaves. That is how a product category protects its reputation.


The clearest public demonstration came in 2017 and 2018. The National Institutes of Health launched a large trial meant to test whether one drink a day actually improves cardiovascular health. On paper, that is the study the observational literature had been begging for: randomize people, follow them, stop arguing from dinner-table graphs.


Then the funding and the framing became public.


Five of the world’s largest beer and spirits companies — including Anheuser-Busch InBev, Diageo, Pernod Ricard, Heineken, and Carlsberg — pledged most of the money, on the order of sixty-eight million dollars, through a foundation that was supposed to keep donors away from the science. An NIH investigation later found that officials and researchers had courted those companies, that industry actors had been given a look at the design, and that the scientific premise had been steered toward showing benefit. Correspondence discussed the need for a result that could support recommending alcohol as part of a healthy diet. Critics noted the trial was better built to catch possible heart effects than to catch cancer and other harms.


NIH shut the trial down in June 2018.


That is not a rumor from a comment section. It is a cancelled federal study, a published investigation, and a case study in what happens when a product category funds the question it most wants answered.


None of this means every paper on alcohol is a lie. It means the sentence you remember — “science says a drink a day is good for you” — was never the totality of the evidence. It was a convenient slice, amplified by an industry that sells the slice, and insufficiently interrogated by a public that wanted the slice to be true.


The World Health Organization has since stated that no safe level of alcohol use can be established for health. Other reviews still wrestle with mixed observational data. That disagreement is not a license to go back to the dinner-table slogan. It is a reminder that one famous curve is not a method.


If you have spent any time in the vitamin and supplement aisle — or on the internet that has replaced the aisle — you have seen the next migration of the same playbook.


The supplement aisle is a teaching clinic

I am not anti-supplement. Some deficiencies are real. Folate before pregnancy is not a vibe. Vitamin D in a person who is actually deficient is not a conspiracy. Omega-3s, creatine, and a short list of other compounds have human evidence that is not a marketing department’s daydream.


That is not the market you are walking through.


Dietary supplements in the United States are not approved by the FDA the way drugs are approved. Under the Dietary Supplement Health and Education Act, a company can generally bring a product to market without first proving to the government that it does what the advertisement implies. The company is responsible for safety and for not making disease claims. Structure/function language — “supports immunity,” “promotes cognitive health,” “helps maintain healthy inflammation” — can be legally used with a disclaimer. The sentence sounds like medicine. It is often closer to poetry with a citation.


Then comes the “science.”


A brand funds a small study. The sample is tiny. There is no meaningful control, or the control is weak. The product is compared with nothing, or with a version of nothing that makes the product look busy. The researchers measure twenty outcomes and report the two that moved. The paper lands in a low-bar journal, or it never leaves a company website. A press release turns a possible signal in twelve people into “clinically proven.” An AI-generated explainer repeats the press release in the cadence of a professor. A deepfake clip of a doctor you almost recognize nods along.


Industry-involved supplement research has been shown, in more than one analysis, to be far more likely to report favorable findings than research without that involvement. That does not mean every funded study is a lie. It means you do not outsource your skepticism to the person who needs the bottle to sell.


Tobacco taught the method. Alcohol dressed it for the dinner table. Supplements put it on a wellness label. Deepfake AI now makes the appearance cheaper than the measurement.


The method is not “they used a study.


The method is “they used the appearance of a study to end the conversation.”


A framework: eight questions that separate science from “scientific”

You do not need a doctorate to use these. You need the willingness to slow down for two minutes before you believe, buy, pour, or forward.


1. What, exactly, is being claimed? “This compound was associated with a small change in one lab marker in a specific group” is a claim. “This bottle will fix your brain” is a different claim. “A drink a day is good for you” is larger than “some observational datasets showed a J-shaped association with one class of heart events after imperfect adjustment.” If the headline is larger than the methods, believe the methods.


2. Can the claim be wrong? If every possible outcome is treated as proof — it worked, it almost worked, it works in a different way than we thought — you are watching a belief protect itself.


3. What kind of evidence is this? A test-tube study is not a human trial. A mouse study is not a middle-aged man in Georgia. A case story is not a sample. An expert on a podcast is not a dataset. An AI summary of an expert on a podcast is not peer review. An observational J-curve is not a randomized trial. Rank the design before you rank the conclusion.


4. Was there a fair comparison? No control group, no randomization, no blinding, and a company-chosen endpoint is not a courtroom. It is a commercial. A comparison group stuffed with sick former users is not a fair zero.


5. How many people, and who were they? Twelve employees of the manufacturer are not “clinical research.” A study of healthy college students may not apply to your thyroid, your medications, or your kidneys. If the paper does not say who was studied, it is not finished speaking.


6. Who paid, and what happens if the result is disappointing? Funding does not automatically falsify a paper. Hidden funding, a product to sell, and a design that cannot fail should turn your confidence dial down. Independent replication should turn it up. If the people who sell the drink helped frame the question, read the answer standing up.


7. Is this one paper or a body of evidence? Cherry-picking is the most common scientific sin in public argument. One supportive trial plus nine silent failures is not “the science.” Look for systematic reviews and meta-analyses from groups that do not sell the product. Then read whether those reviews are themselves current, independent, and honest about mixed findings.


8. Has anyone tried to break it? Science that cannot be criticized is not science. Look for pre-registration, data availability, replications, corrections, and — when they exist — retractions. A paper that has never been allowed to meet an enemy has not yet been tested. A trial that is shut down because the premise was steered is data too. It tells you how the question was built.


If you cannot answer most of those questions, you do not have a finding. You have a story that borrowed a lab coat.


Where to find good science papers without moving into a library

You will not become a journal editor this afternoon. You can stop getting your biology from a comment thread.

Start here:


A practical sequence when someone sends you a “study”:

  1. Find the original paper, not the reel.

  2. Read the abstract last. Read the methods and the conflicts first.

  3. Ask whether humans were studied.

  4. Ask who was in the comparison group.

  5. Ask who paid, and what the design could not see.

  6. Search the same question on PubMed and Cochrane. See what else exists.

  7. Only then decide what the paper is allowed to change in your life.


If that sounds slower than a feed, good. Science that can be consumed at the speed of outrage is usually marketing.


What this has to do with the people in my office

I stay politically neutral because I treat people, not parties.


I do not stay methodologically neutral because method is how we keep from baptizing our preferences and calling them discoveries. In a deepfake environment, the feeling of certainty is cheap. A well-lit clip can give you the sensation of having checked. The nervous system does not automatically distinguish a real finding from a finished-looking fake. That is not a character defect. It is how attention works. You feel what you focus on. If you focus on tribal science, you will feel tribal certainty. If you focus on method, you will feel something less dramatic and more useful: provisional confidence.


Provisional confidence is not weakness. It is how adults handle evidence.

You can keep your convictions. You can keep your vote. You can keep the issues that matter to you. What you cannot keep — not if you want a mind that can still be updated — is the habit of deciding first and shopping for studies second.


Science, conducted honestly, is one of the few tools we have that does not care which side of the room you sit on. That is precisely why both sides keep trying to own it. It is also why I will not.


Learn more and connect:

Phone: 678-629-2578


Copyright notice:

© 2026 Galen E. Cole, PhD, MPH, LPC. All rights reserved.

This essay is an original work first published at galencole.com. No part of this work may be reproduced, redistributed, scraped, or used to train commercial models without prior written permission, except for brief quotation with clear attribution to the author and a link to the original post.


Publication: September 22, 2026

Author: Galen E. Cole Official site: https://www.galencole.com



 
 
 

Comments

bottom of page