0:00:06.912,0:00:11.532
One simple vitamin can reduce [br]your risk of heart disease.

0:00:11.532,0:00:15.522
Eating chocolate reduces [br]stress in students.

0:00:15.522,0:00:20.222
New drug prolongs lives of [br]patients with rare disease.

0:00:20.222,0:00:23.187
Health headlines like these are published[br]every day,

0:00:23.187,0:00:26.307
sometimes making opposite claims [br]from each other.

0:00:26.307,0:00:28.427
There can be a disconnect between broad,

0:00:28.427,0:00:31.537
attention-grabbing headlines [br]and the often specific,

0:00:31.537,0:00:34.977
incremental results of the medical [br]research they cover.

0:00:34.977,0:00:39.357
So how can you avoid being [br]misled by grabby headlines?

0:00:39.357,0:00:42.467
The best way to assess a headline’s [br]credibility

0:00:42.467,0:00:45.807
is to look at the original [br]research it reports on.

0:00:45.807,0:00:48.217
We’ve come up with a hypothetical research[br]scenario

0:00:48.217,0:00:50.507
for each of these three headlines.

0:00:50.507,0:00:53.357
Keep watching for the explanation[br]of the first example;

0:00:53.357,0:00:56.447
then pause at the headline to [br]answer the question.

0:00:56.447,0:00:58.517
These are simplified scenarios.

0:00:58.517,0:01:02.827
A real study would detail many more [br]factors and how it accounted for them,

0:01:02.827,0:01:04.837
but for the purposes of this exercise,

0:01:04.837,0:01:08.837
assume all the information [br]you need is included.

0:01:11.048,0:01:14.098
Let’s start by considering the [br]cardiovascular effects

0:01:14.098,0:01:17.068
of a certain vitamin, Healthium.

0:01:17.068,0:01:19.868
The study finds that participants taking[br]Healthium

0:01:19.868,0:01:24.078
had a higher level of healthy cholesterol [br]than those taking a placebo.

0:01:24.078,0:01:28.036
Their levels became similar to those of [br]people with naturally high levels

0:01:28.036,0:01:29.936
of this kind of cholesterol.

0:01:29.936,0:01:33.326
Previous research has shown that people [br]with naturally high levels

0:01:33.326,0:01:37.486
of healthy cholesterol have lower [br]rates of heart disease.

0:01:37.486,0:01:39.991
So what makes this headline misleading:

0:01:39.991,0:01:44.311
Healthium reduces risk of heart disease.

0:01:44.311,0:01:48.791
The problem with this headline is that the[br]research didn’t actually investigate

0:01:48.791,0:01:51.561
whether Healthium reduces heart disease.

0:01:51.561,0:01:53.651
It only measured Healthium’s impact

0:01:53.651,0:01:56.791
on levels of a particular [br]kind of cholesterol.

0:01:56.791,0:01:59.931
The fact that people with naturally high[br]levels of that cholesterol

0:01:59.931,0:02:01.951
have lower risk of heart attacks

0:02:01.951,0:02:04.391
doesn’t mean that the same [br]will be true of people

0:02:04.391,0:02:07.731
who elevate their cholesterol[br]levels using Healthium.

0:02:07.731,0:02:10.141
Now that you’ve cracked the [br]case of Healthium,

0:02:10.141,0:02:13.581
try your hand at a particularly alluring [br]mystery:

0:02:13.581,0:02:17.321
the relationship between eating chocolate[br]and stress.

0:02:17.321,0:02:20.281
This hypothetical study [br]recruits ten students.

0:02:20.281,0:02:23.531
Half begin consuming a [br]daily dose of chocolate,

0:02:23.531,0:02:25.141
while half abstain.

0:02:25.141,0:02:28.691
As classmates, they all follow [br]the same schedule.

0:02:28.691,0:02:32.481
By the end of the study, the chocolate [br]eaters are less stressed

0:02:32.481,0:02:35.451
than their chocolate-free counterparts.

0:02:35.451,0:02:37.301
What’s wrong with this headline:

0:02:37.301,0:02:41.371
Eating chocolate reduces [br]stress in students

0:02:43.408,0:02:48.638
It’s a stretch to draw a conclusion about [br]students in general from a sample of ten.

0:02:48.638,0:02:52.024
That’s because the fewer participants are[br]in a random sample,

0:02:52.024,0:02:54.884
the less likely it is that the sample will[br]closely represent

0:02:54.884,0:02:57.934
the target population as a whole.

0:02:57.934,0:03:02.734
For example, if the broader population of [br]students is half male and half female,

0:03:02.734,0:03:04.784
the chance of drawing a sample of 10

0:03:04.784,0:03:10.144
that’s skewed 70% male and [br]30% is about 12%.

0:03:10.144,0:03:15.693
In a sample of 100 that would be less than[br]a .0025% chance,

0:03:15.693,0:03:17.525
and for a sample of 1000,

0:03:17.525,0:03:22.795
the odds are less than 6 x 10^-36.

0:03:22.795,0:03:25.137
Similarly, with fewer participants,

0:03:25.137,0:03:29.340
each individual’s outcome has a larger [br]impact on the overall results—

0:03:29.340,0:03:32.550
and can therefore skew big-picture trends.

0:03:32.550,0:03:37.490
Still, there are a lot of good reasons for[br]scientists to run small studies.

0:03:37.490,0:03:39.269
By starting with a small sample,

0:03:39.269,0:03:42.039
they can evaluate whether the results are [br]promising enough

0:03:42.039,0:03:45.139
to run a more comprehensive, [br]expensive study.

0:03:45.139,0:03:48.609
And some research requires very specific[br]participants

0:03:48.609,0:03:51.999
that may be impossible to [br]recruit in large numbers.

0:03:51.999,0:03:54.329
The key is reproducibility—

0:03:54.329,0:03:57.629
if an article draws a conclusion [br]from one small study,

0:03:57.629,0:03:59.649
that conclusion may be suspect—

0:03:59.649,0:04:03.099
but if it’s based on many studies [br]that have found similar results,

0:04:03.099,0:04:04.699
it’s more credible.

0:04:04.699,0:04:06.919
We’ve still got one more puzzle.

0:04:06.919,0:04:11.899
In this scenario, a study tests a new drug[br]for a rare, fatal disease.

0:04:11.899,0:04:14.156
In a sample of 2,000 patients,

0:04:14.156,0:04:17.636
the ones who start taking the drug upon [br]diagnosis

0:04:17.636,0:04:21.052
live longer than those who [br]take the placebo.

0:04:21.052,0:04:23.642
This time, the question [br]is slightly different.

0:04:23.642,0:04:28.012
What’s one more thing you’d like to know [br]before deciding if the headline,

0:04:28.012,0:04:33.619
New drug prolongs lives of patients[br]with rare disease, is justified?

0:04:34.902,0:04:36.472
Before making this call,

0:04:36.472,0:04:40.792
you’d want to know how much the drug [br]prolonged the patients’ lives.

0:04:40.792,0:04:43.163
Sometimes, a study can have results that,

0:04:43.163,0:04:48.243
while scientifically valid, don’t have [br]much bearing on real world outcomes.

0:04:48.243,0:04:52.941
For example, one real-life clinical trial [br]of a pancreatic cancer drug

0:04:52.941,0:04:57.457
found an increase in life expectancy—[br]of ten days.

0:04:57.457,0:05:00.277
The next time you see a surprising medical[br]headline,

0:05:00.277,0:05:04.027
take a look at the science [br]it’s reporting on.

0:05:04.027,0:05:06.957
Even when full papers aren’t [br]available without a fee,

0:05:06.957,0:05:09.987
you can often find summaries of [br]experimental design

0:05:09.987,0:05:13.177
and results in freely available abstracts,

0:05:13.177,0:05:16.361
or even within the text [br]of a news article.

0:05:16.361,0:05:19.671
It’s exciting to see scientific research [br]covered in the news,

0:05:19.671,0:05:23.951
and important to understand [br]the studies’ findings.