Do you think scientists are more susceptible to concluding that a false positive is a real positive or that a false negative is a real negative?
As I understand,
A) A false positive is when a scientist concludes that an experiment/assay succeeded when actually it failed.
B) A false negative is when a scientist concludes that an experiment/assay failed when actually it succeeded.
A scientist can waste a lot of resources on a false positive if (s)he doesn't notice that the material (s)he is testing is not what (s)he thinks it is. All observations are irrelevant to what you think you are studying as well as anything you publish about it.
A scientist could dismiss a powerful medical drug by falling for a false negative. The drug works. You just didn't test its properties well never to return to it ever again. Are theorists more susceptible to falling for false positives and experimentalists are more susceptible to falling for false negatives, for instance? For theorists, maybe we talking about models more than assays.
Also do you think scientists are conditioned to look out for false positives or false negatives more so? Which one is more dangerous to science?
Nearly all experiments and theories require optimization so false positives and false negatives are certainly possible occurrences.
I wouldn't speak for everyone but in my surroundings people don't conclude anything from a single experiment. In molecular biology, experiments are done several times, as a rule, but what's more, different experiments are done to corroborate one point.
For example, when a biologist looks for a certain gene in a certain cell line, he checks for RNA (e.g. qPCR), then for protein (e.g. western blot), then for specific localization of the gene product (e.g. immunostaining). Then, if all of the data fit together, the scientist can conclude - okay, this gene is there and it seems to work. You don't do one qPCR run and then "conclude" that, oh, the gene is/isn't there.
They are more susceptible to false positives when they want to rule in an association and to false negatives when they want to rule out an association. Since most experiments are aimed to rule in, the scientific community demands smaller type I error (false positives) than type II error (false negative) to counter balance such tendency (also because mis-information may be worse than missed information).
As we know that while standardizing or planning an experiment, we should have two replicates; 1. Experimental rep and 2. Biological rep. In this scenario, the chance of selecting false negatives and false positives may be less. When considering the effect of the two, both might have equal social, economical and health impacts. Susceptibility depends on two things; result oriented or research oriented.
Here are some thoughts about the topic,
Generally speaking statistical significance testing aims to reduce type I error = false positives more stringently than type II error = false negatives. For example in drug testing it is more safe to set inclusion criteria so high, that false positives are all ruled out and consequently some false negatives are included, i.e. real positives are being dropped out.
For example in our lab when performing ELISA, using positive control samples we set the threshold for positive so high, that we should not get false positives. By doing this, we will likely fail to detect some real positives, which were left under the threshold and so labeled as negative. We emphasize sensitivity more than specificity.
It also depends what is the effect size one wants to see. If the effect is modest and variability is great, the more samples/testing is needed to see the difference. Notable difference is more easily detected. Therefore small differences are more likely to be ignored.
Also in criminal justice cases it is more severe to jail an innocent person = false positive, than let the guilty walk free = false negative.
Doctors also are more likely to over treat healthy individuals = false negative, than left truly ill patient without care, possibly leading to death =false positive.
I work with experimental lab work.
- Badges
- Science topic
- Similar topics
- Mathematics