Ben Goldacre
speaker
65 appearances
3 recordings
1 series
first heard Dec 2010
last heard May 2012
Ben Goldacre’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsNo recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.
Appearances
Well, publication bias is the phenomenon whereby positive, exciting, interesting findings are much more likely to get published in academic journals than boring, negative findings.
And at face value, that sounds kind of trivially true.
It's a bit like the judgments that a newspaper editor might make.
You know, when a dog bites a man, that's not news, but when a man bites a dog, that is news.
But the problem with it is that overall it completely pollutes the academic literature because it means that what we're seeing is many, many, many more flukes and we're not seeing the full statistical context.
In the world of medicine, for example, findings aren't just kind of abstract.
Findings in medicine relate to the real human world of suffering and flesh and pain.
And so we can get ourselves in a lot of trouble if the published research findings aren't as true as we think they are.
But when it comes to treatments in the real world, we've got an even bigger problem.
So the best way we have to find out if an intervention like a pill, for example, works or doesn't work is by doing a randomised controlled trial.
And the problem is some randomised controlled trials will get published and some won't.
In fact, it turns out that only about half of all of the trials that get conducted get published.
And trials which report positive results for the pill being tested are about twice as likely to get published as studies reporting negative results.
So that means that we are dramatically over-exaggerating the benefits of the treatments that we give in medicine.
The effect in the final analysis in the clinic room where you're seeing a patient is we don't know the true effects of the medicines that we're prescribing.
So there is a solution to this problem and it's called a register.
So this has been done mostly in the area of clinical trials.
You write down the protocol of exactly what you're going to do in your experiment...
and you post it somewhere in public.
So the idea of that is we can then see all of the trials that have been conducted and completed, and then we can go back and we can say, well, are there any that have been completed but never reported?
Showing 1–20 of 65 · page 1 of 4
Next →