Die Wissenschaftslüge

A review of Ben Goldacre's "Bad Science", reread with an eye to what it teaches people who read and summarize research for a living.

When this book first appeared it was marketed as a funny, angry takedown of quack nutritionists, homeopaths and credulous journalists, and it is all of those things. Reading it again years later, what stands out is something less obvious: underneath the entertainment, it is one of the most accessible introductions to research methods ever written for a general audience. Goldacre uses the absurd claims of his targets as teaching cases, and by the end of the book a careful reader has absorbed the logic of controlled trials, blinding, placebo effects, regression to the mean, publication bias and systematic reviews without ever feeling lectured.

The structure

The book moves roughly from the ridiculous to the serious. Early chapters deal with products and practices whose claims collapse under the simplest scrutiny, such as detox foot baths and pseudoscientific exercises marketed to schools. These chapters are the funniest, and they establish a method: take a claim, ask what evidence would support it, and see whether that evidence exists.

The middle of the book turns to homeopathy, the placebo effect and the nutrition industry. Here Goldacre explains why a treatment can appear to work even when it has no specific effect at all. People often seek treatment when symptoms are at their worst, and symptoms tend to improve over time anyway. Expectation and ritual can change how people feel and what they report. Practitioners who believe in a treatment may unconsciously select or interpret results in its favour. Each of these explanations is presented with clear examples, and together they show why uncontrolled observation is so unreliable.

The later chapters become more serious and, at times, distressing. Goldacre examines how the media reported a supposed link between a vaccine and autism, how statistical evidence was misused in a high-profile criminal case, and how pharmaceutical research can be distorted. These sections move beyond amusement to show the real costs of poor reasoning about evidence.

What it teaches well

The chapter explaining the randomized controlled trial is a model of clear writing. Goldacre walks through why we need a comparison group, why allocation must be random rather than chosen by the investigator, why participants and assessors should be blinded where possible, and why all participants should be analysed in the groups to which they were assigned. Many students meet these ideas first as abstract rules in a methods course. Here they arrive as solutions to concrete problems, which makes them far easier to remember.

The discussion of systematic reviews and meta-analysis is equally valuable. Goldacre explains why looking at all the evidence, collected in a transparent and reproducible way, is better than picking studies that support a favoured conclusion. He introduces the forest plot and the idea that combining small studies can reveal an effect, or the lack of one, that no single study could detect. For readers who later work on literature reviews, this is an excellent conceptual foundation.

His treatment of publication bias is perhaps the most important section for anyone who relies on published research. If studies with positive results are more likely to be published than studies with negative or null results, then even a perfectly conducted review of the published literature will overestimate effects. Goldacre explains funnel plots and trial registration in plain terms and makes the case that missing data is a scientific and ethical problem, a theme he developed further in his later book on the pharmaceutical industry.

Limitations

The book's tone is combative, and some readers find the mockery of individuals uncomfortable. The targets are mostly public figures making commercial or influential claims, and Goldacre is careful to distinguish them from ordinary people trying to improve their health, but the sharpness may put off readers who would benefit from the ideas.

Some specific examples have dated. Particular media stories, companies and personalities that were prominent at the time of writing are less familiar now, and newer editions only partly update them. The underlying lessons are unaffected, but readers may need to look up background on some of the cases.

The book also focuses heavily on medicine and nutrition. Readers in education, psychology or policy research will recognize the same problems in their own fields, including small samples, uncontrolled designs, selective reporting and media exaggeration, but they will have to make the connections themselves.

Why it matters for research work

I now recommend this book to new members of any team that reads and synthesizes research, alongside more formal methods texts. It gives people an intuitive sense of why methods matter. Someone who has read Goldacre's account of regression to the mean is much less likely to be impressed by a before-and-after study with no control group. Someone who understands publication bias will ask whether a review searched trial registries and grey literature, not only journal databases.

These habits matter even more now that so much reading is mediated by software. Search engines, reference managers and AI research assistants such as Yila make it far easier to find and summarize large numbers of papers quickly. That speed is genuinely useful, but it does not tell us which papers deserve trust. The questions this book teaches readers to ask, such as compared with what, how were participants allocated, who was blinded, what happened to the missing data and what was not published, are exactly the questions that should accompany any summary, however it was produced.

Who should read it

It is suitable for interested general readers, for students in any discipline that uses empirical evidence, for journalists who report on science, and for researchers who want a lively way to introduce colleagues or students to core methodological ideas. Specialists in trial methods will find nothing new technically but may enjoy it as a demonstration of how to communicate these ideas.

Verdict

An entertaining, sharp and genuinely educational book. Its examples are sometimes dated and its tone sometimes harsh, but its explanations of controlled trials, placebo effects, systematic reviews and publication bias remain among the clearest available for non-specialists. It turns scepticism into a set of practical questions, and that is a lasting gift to anyone who reads research.

Four and a half stars.