Scott Alexander analyzes online drug ratings, finding patients prefer older antidepressants while doctors prefer newer ones, and explores potential explanations for this paradox. Longer summary

Scott Alexander analyzes patient ratings of antidepressants from online databases, finding that older drugs like MAOIs are rated higher than newer ones. He then compares this to doctor ratings, discovering a negative correlation between patient and doctor preferences. The post explores possible explanations for these paradoxical results, including confounding factors and the hypothesis that newer antidepressants may actually be less effective. Scott extends the analysis to other drug classes, finding the negative doctor-patient correlation holds broadly, while the preference for older drugs is specific to psychiatric medications. Shorter summary

Scott examines how rare but severe drug side effects often lead to underprescription compared to drugs with common but less severe side effects, potentially resulting in suboptimal patient care. Longer summary

This post discusses the issue of rare but severe side effects in psychiatric medications versus common but less severe side effects. Scott uses two main examples: nefazodone vs SSRIs for depression, and modafinil vs Adderall for stimulants. He argues that drugs with rare but spectacular side effects (like nefazodone and modafinil) are often underprescribed compared to drugs with more common but less severe side effects (like SSRIs and Adderall). The post explores the reasons for this, including doctors' risk aversion, lawsuit concerns, and media coverage of rare side effects. Scott suggests that this pattern may lead to suboptimal treatment choices for patients. Shorter summary

Scott Alexander examines evidence on marijuana legalization's effects on usage, health, and drug war costs, finding uncertainties too large for firm conclusions but suggesting traffic accidents may be the key factor. Longer summary

Scott Alexander examines the evidence around marijuana legalization, looking at its effects on usage rates, health impacts, and costs of the drug war. He finds that decriminalization and medical marijuana don't seem to increase youth usage much, but full legalization might increase adult use by 25-50%. Health effects are unclear but may include IQ declines and slightly increased psychosis risk. The costs of the drug war are significant in terms of arrests and spending, but relatively few people are in prison solely for marijuana. Scott attempts a utilitarian analysis comparing costs and benefits of legalization, but finds the uncertainties too large to draw firm conclusions. He suggests the most important factor may be how legalization affects traffic accidents. Shorter summary