"Science Fictions" by Stuart Ritchie is a riveting exploration of the scientific process, its pitfalls, and its potential for misdirection. Ritchie's analysis is both enlightening and unsettling, shedding light on the vulnerabilities within scientific research while also celebrating its merits. The book is meticulously researched, eloquently written, and deeply engaging, making complex topics accessible to a general audience. A must-read for anyone interested in the intricacies of scientific integrity, "Science Fictions" serves as a clarion call for transparency, rigor, and ethical responsibility in research.

Science has become the new norm: from makeup through nutrition to sports, an increasing number of companies seem to rely on science to present their products and services. The intent and the practice are great. However, taking scientific findings at face value can lead to wrong conclusions and, if used in products and services, may not meet customer expectations.

The book, Science Fictions, is an in-depth investigation of the main reasons why scientific findings end up presenting information that, put simply, are misleading. Some of the reasons include failing to check the final paper for numerical accuracies, the endorsement of publishing papers that show something ‘groundbreaking’ (as opposed to confirming the non-existence of something), and self-serving biases. The bulk of the book addresses these reasons thoroughly, citing evidence for where science turned into science fiction.

An important chapter of the book is dedicated to teaching us how to be able to judge whether a scientific paper’s findings are indeed reliable, valid and therefore, can be used in various claims. The 10 key suggestions may seem daunting, but with practice, one can easily scan a scientific paper to judge its credibility.

Most of these credibility-checking exercises do not require any knowledge of science, whereas some of them may require a little familiarity with experimental design and very basic statistics.

The basics

Is the author from a reputable university, company or lab?

Does the journal look professional (e.g. judging by its website)?

Transparency

Are the data and other materials available online?

Design

Is the study well-designed?

Sample

Is it big enough for the purposes of the experiment?

Effect

How big is the effect - check for significance (p-values)

Inferences

Are the inferences appropriate? Do they not confuse correlation with causation?

Do they overgeneralise? The study was conducted on a small sample size and yet the conclusions are presented as telling us something about people in general?

Bias

Could there be any bias, eg.g political or social implications that may have caused the results seem less partial?

Was the study funded by a company that appears in the experiment?

Common sense

Is it really plausible? Do you sense that something is off with this experiment?

Did the study answer the question it set out to answer?

Replication

You are probably really familiar with this (as one of the current debates in a few disciplines is around the lack of replicability): simply put, try not to rely on individual studies - the more studies running similar experiments, finding similar results, the more reliable the paper is

Other scientists’ opinion

Don’t just rely on peer-reviews: Check out Twitter, and other sources where you can find reviews in a different, and hopefully unbiased way

Science Fictions is a highly approachable analysis of the current model of scientific review in academia. Fans of Ben Goldacre will appreciate this new addition to the literature.