For example, van der Naald et al. looked at seven years’ biomedical research using animals at the University Medical Center Utrecht and found that only 60% of the studies lead to one or more publications, and that of the 5590 animals used in the studies they looked at, only 26% ended up in published research.[25] (For small animals, which made up the majority of those used, the number was a mere 23%.[26]) The main reasons for not publishing were “lack of statistical significance, technical problems and objections from supervisors and peer reviewers”
How many studies that use non-animal models end up in published research? Lack of statistical significance leading to non-publication is an issue that is not specific to research on animals, but fairly widespread.
He notes that very few drugs that make it through animal trials also succeed in human trials. That is usually not because they end up being unsafe for humans, but because they end up being ineffective
Again the counterfactual is really unclear. If we didn’t use animal models even fewer drugs would succeed in human trials.
those are both good questions. i tried to find base rates with a cursory search but came up empty-handed. maybe i just didn’t use the right search terms, though. but even if the numbers here are the same as the base rate, i would argue that’s still pretty bad, because the costs involved in animal testing are higher. i think it makes sense to judge animal-testing research more stringently than other research. though base rates would be useful to see e.g. how difficult it would be to improve methodologies and reduce the amount of unproductive research.
one thing i didn’t make clear in the post but which i now realise i should’ve is that an experiment not getting published due to lack of statistical significance (or more precisely rejection of the null hypothesis) doesn’t mean that research wasn’t valuable—it could have been rejected due to publication bias.
How many studies that use non-animal models end up in published research? Lack of statistical significance leading to non-publication is an issue that is not specific to research on animals, but fairly widespread.
Again the counterfactual is really unclear. If we didn’t use animal models even fewer drugs would succeed in human trials.
those are both good questions. i tried to find base rates with a cursory search but came up empty-handed. maybe i just didn’t use the right search terms, though. but even if the numbers here are the same as the base rate, i would argue that’s still pretty bad, because the costs involved in animal testing are higher. i think it makes sense to judge animal-testing research more stringently than other research. though base rates would be useful to see e.g. how difficult it would be to improve methodologies and reduce the amount of unproductive research.
one thing i didn’t make clear in the post but which i now realise i should’ve is that an experiment not getting published due to lack of statistical significance (or more precisely rejection of the null hypothesis) doesn’t mean that research wasn’t valuable—it could have been rejected due to publication bias.