A few weeks ago, the eternal debate about the supposed meaningfulness of homeopathy was reignited. It's well-known that the positions on this topic are deeply entrenched.
Statistically speaking, it cannot be proven either that homeopathic medicine heals or that it doesn't.
The trigger for the homeopathy debate was a failed tweet by the Techniker Krankenkasse (TK), a German health insurance company. When asked by a Twitter user to present studies proving the efficacy of homeopathy, the health insurance company replied somewhat brusquely, stating that the user should provide studies disproving the efficacy of homeopathy instead. The backlash was swift and far from homoeopathically diluted. Amidst the usual exchanges, even supporters of the "You can prove anything with statistics" faction didn't take long to join in. Correctly stated: With statistics, you can't prove nothing.
Wait – is this a case of ruthless overestimation by the statistician? Quite the opposite. Demanding scientific evidence to prove the opposite (namely, lack of efficacy) simply because one cannot prove one's own case (efficacy) is not only highly controversial in medicine – it's statistically impossible. The concept of statistical testing is based on starting from the null hypothesis ("there is no effect") – similar to a court case, where ideally the defendant is presumed innocent until the guilt, the alternative hypothesis ("there is an effect"), is proven. Just as the high court can only use excerpts from the surroundings of the crime for judgment and practically never all (potential!) evidence, the statistician must also rely on the informativeness of samples and needs clear rules for this. Why? Because people occasionally recover by chance, and in hindsight, we have to decide whether this happened despite, because of, or completely independent of the administered therapy.
Conversely, this conclusion does not work the other way around. In statistics, there is only a "second-class acquittal" for lack of evidence. Thus, the TK's demand for studies proving the lack of efficacy of homeopathy is scientific nonsense.
But why do scientists constantly demand that efficacy must be proven first – isn't it just the pharmaceutical lobby trying to defend its turf against the gentle homeopaths? This is also far from the truth. Let's imagine instead that it's about chemotherapy. If one were to follow the argument that people should be treated with this therapy because the medication might work (and we just don't know it yet), then patients would be unnecessarily exposed to strong side effects. Statistical testing is thus damage limitation. It reduces the "Type I error," where an alleged effect does not actually exist, to a minimum in favor of the "Type II error," where an actual effect is not recognized. General life experience – think of the medieval "healing practices," which probably killed more people than cured – shows that this principle works as a basic scientific rule (not in every single case) for decision-making under uncertainty.
The senselessness of their response eventually became apparent to the TK. But by then, it was too late for reactions, because it was all about the money. After all, homeopathic treatments are paid for from the contributions of all TK insured persons. Homeopathy opponents protested that they were forced to finance "ineffective sugar pills."
To "prove" the alleged inefficacy, people regularly consume entire bottles of globules for demonstration purposes, just to be able to say afterwards, "I'm still alive." No one would dare to do that with a packet of ibuprofen. And it's unscientific to boot because, once again, it boils down to individual cases instead of statistical proof.
However, homeopathy sees itself as a doctrine whose specific approach depends on the individual. Just because something doesn't work for some doesn't mean it's ineffective for others. Indeed, various studies, much to the displeasure of many homeopaths, have recently found that there is little difference between homeopathic globules and placebos. So, there is a bit of "effect" after all.
The effect of placebos is often mocked in medicine. Patients are treated without active ingredients and instead rely on the belief in the efficacy of the "medication." The fact that faith can move mountains is more than just a comforting saying. Many studies, where patients only received placebos but believed they were receiving real medication, documented improvements in the illness. From migraines or coughs from viral infections to depression and bipolar affective disorders, the application areas of such sham medications range widely. So, even if the success of homeopathy is purely based on the placebo effect, why is it wrong for people to seek this treatment as long as they believe in its efficacy?
It's not wrong at all, as long as homeopathy recognizes its limits. Treating a tumor with globules instead of chemotherapy should never be supported by health insurance companies. Compared to the costs of premature prescriptions of antibiotics without a secure prognosis and the unwanted side effect of increasing antibiotic resistance, the "sugar pills" are not only cheaper but also healthier.