[Script Info]
Title: 
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:06.91,0:00:11.53,Default,,0000,0000,0000,,One simple vitamin can reduce \Nyour risk of heart disease.
Dialogue: 0,0:00:11.53,0:00:15.52,Default,,0000,0000,0000,,Eating chocolate reduces \Nstress in students.
Dialogue: 0,0:00:15.52,0:00:20.22,Default,,0000,0000,0000,,New drug prolongs lives of \Npatients with rare disease.
Dialogue: 0,0:00:20.22,0:00:23.19,Default,,0000,0000,0000,,Health headlines like these \Nare published every day,
Dialogue: 0,0:00:23.19,0:00:26.31,Default,,0000,0000,0000,,sometimes making opposite claims \Nfrom each other.
Dialogue: 0,0:00:26.31,0:00:28.43,Default,,0000,0000,0000,,There can be a disconnect between broad,
Dialogue: 0,0:00:28.43,0:00:31.54,Default,,0000,0000,0000,,attention-grabbing headlines \Nand the often specific,
Dialogue: 0,0:00:31.54,0:00:34.98,Default,,0000,0000,0000,,incremental results of the medical \Nresearch they cover.
Dialogue: 0,0:00:34.98,0:00:39.36,Default,,0000,0000,0000,,So how can you avoid being \Nmisled by grabby headlines?
Dialogue: 0,0:00:39.36,0:00:42.47,Default,,0000,0000,0000,,The best way to assess a headline’s \Ncredibility
Dialogue: 0,0:00:42.47,0:00:45.81,Default,,0000,0000,0000,,is to look at the original \Nresearch it reports on.
Dialogue: 0,0:00:45.81,0:00:48.22,Default,,0000,0000,0000,,We’ve come up with a hypothetical research\Nscenario
Dialogue: 0,0:00:48.22,0:00:50.51,Default,,0000,0000,0000,,for each of these three headlines.
Dialogue: 0,0:00:50.51,0:00:53.36,Default,,0000,0000,0000,,Keep watching for the explanation\Nof the first example;
Dialogue: 0,0:00:53.36,0:00:56.45,Default,,0000,0000,0000,,then pause at the headline to \Nanswer the question.
Dialogue: 0,0:00:56.45,0:00:58.52,Default,,0000,0000,0000,,These are simplified scenarios.
Dialogue: 0,0:00:58.52,0:01:02.83,Default,,0000,0000,0000,,A real study would detail many more \Nfactors and how it accounted for them,
Dialogue: 0,0:01:02.83,0:01:04.84,Default,,0000,0000,0000,,but for the purposes of this exercise,
Dialogue: 0,0:01:04.84,0:01:08.84,Default,,0000,0000,0000,,assume all the information \Nyou need is included.
Dialogue: 0,0:01:11.05,0:01:14.10,Default,,0000,0000,0000,,Let’s start by considering the \Ncardiovascular effects
Dialogue: 0,0:01:14.10,0:01:17.07,Default,,0000,0000,0000,,of a certain vitamin, Healthium.
Dialogue: 0,0:01:17.07,0:01:19.87,Default,,0000,0000,0000,,The study finds that participants taking\NHealthium
Dialogue: 0,0:01:19.87,0:01:24.08,Default,,0000,0000,0000,,had a higher level of healthy cholesterol \Nthan those taking a placebo.
Dialogue: 0,0:01:24.08,0:01:28.04,Default,,0000,0000,0000,,Their levels became similar to those of \Npeople with naturally high levels
Dialogue: 0,0:01:28.04,0:01:29.94,Default,,0000,0000,0000,,of this kind of cholesterol.
Dialogue: 0,0:01:29.94,0:01:33.33,Default,,0000,0000,0000,,Previous research has shown that people \Nwith naturally high levels
Dialogue: 0,0:01:33.33,0:01:37.49,Default,,0000,0000,0000,,of healthy cholesterol have lower \Nrates of heart disease.
Dialogue: 0,0:01:37.49,0:01:39.99,Default,,0000,0000,0000,,So what makes this headline misleading:
Dialogue: 0,0:01:39.99,0:01:44.31,Default,,0000,0000,0000,,"Healthium reduces risk of heart disease."
Dialogue: 0,0:01:44.31,0:01:48.79,Default,,0000,0000,0000,,The problem with this headline is that the\Nresearch didn’t actually investigate
Dialogue: 0,0:01:48.79,0:01:51.56,Default,,0000,0000,0000,,whether Healthium reduces heart disease.
Dialogue: 0,0:01:51.56,0:01:53.65,Default,,0000,0000,0000,,It only measured Healthium’s impact
Dialogue: 0,0:01:53.65,0:01:56.79,Default,,0000,0000,0000,,on levels of a particular \Nkind of cholesterol.
Dialogue: 0,0:01:56.79,0:01:59.93,Default,,0000,0000,0000,,The fact that people with naturally high\Nlevels of that cholesterol
Dialogue: 0,0:01:59.93,0:02:01.95,Default,,0000,0000,0000,,have lower risk of heart attacks
Dialogue: 0,0:02:01.95,0:02:04.39,Default,,0000,0000,0000,,doesn’t mean that the same \Nwill be true of people
Dialogue: 0,0:02:04.39,0:02:07.73,Default,,0000,0000,0000,,who elevate their cholesterol\Nlevels using Healthium.
Dialogue: 0,0:02:07.73,0:02:10.14,Default,,0000,0000,0000,,Now that you’ve cracked the \Ncase of Healthium,
Dialogue: 0,0:02:10.14,0:02:13.58,Default,,0000,0000,0000,,try your hand at a particularly alluring \Nmystery:
Dialogue: 0,0:02:13.58,0:02:17.32,Default,,0000,0000,0000,,the relationship between eating chocolate\Nand stress.
Dialogue: 0,0:02:17.32,0:02:20.28,Default,,0000,0000,0000,,This hypothetical study \Nrecruits ten students.
Dialogue: 0,0:02:20.28,0:02:23.53,Default,,0000,0000,0000,,Half begin consuming a \Ndaily dose of chocolate,
Dialogue: 0,0:02:23.53,0:02:25.14,Default,,0000,0000,0000,,while half abstain.
Dialogue: 0,0:02:25.14,0:02:28.69,Default,,0000,0000,0000,,As classmates, they all follow \Nthe same schedule.
Dialogue: 0,0:02:28.69,0:02:32.48,Default,,0000,0000,0000,,By the end of the study, the chocolate \Neaters are less stressed
Dialogue: 0,0:02:32.48,0:02:35.45,Default,,0000,0000,0000,,than their chocolate-free counterparts.
Dialogue: 0,0:02:35.45,0:02:37.30,Default,,0000,0000,0000,,What’s wrong with this headline:
Dialogue: 0,0:02:37.30,0:02:41.37,Default,,0000,0000,0000,,"Eating chocolate reduces \Nstress in students"
Dialogue: 0,0:02:43.41,0:02:48.64,Default,,0000,0000,0000,,It’s a stretch to draw a conclusion about \Nstudents in general from a sample of ten.
Dialogue: 0,0:02:48.64,0:02:52.02,Default,,0000,0000,0000,,That’s because the fewer participants are\Nin a random sample,
Dialogue: 0,0:02:52.02,0:02:54.88,Default,,0000,0000,0000,,the less likely it is that the sample will\Nclosely represent
Dialogue: 0,0:02:54.88,0:02:57.93,Default,,0000,0000,0000,,the target population as a whole.
Dialogue: 0,0:02:57.93,0:03:02.73,Default,,0000,0000,0000,,For example, if the broader population of \Nstudents is half male and half female,
Dialogue: 0,0:03:02.73,0:03:04.78,Default,,0000,0000,0000,,the chance of drawing a sample of 10
Dialogue: 0,0:03:04.78,0:03:10.14,Default,,0000,0000,0000,,that’s skewed 70% male and \N30% is about 12%.
Dialogue: 0,0:03:10.14,0:03:15.69,Default,,0000,0000,0000,,In a sample of 100 that would be less than\Na .0025% chance,
Dialogue: 0,0:03:15.69,0:03:17.52,Default,,0000,0000,0000,,and for a sample of 1000,
Dialogue: 0,0:03:17.52,0:03:22.80,Default,,0000,0000,0000,,the odds are less than 6 x 10^-36.
Dialogue: 0,0:03:22.80,0:03:25.14,Default,,0000,0000,0000,,Similarly, with fewer participants,
Dialogue: 0,0:03:25.14,0:03:29.34,Default,,0000,0000,0000,,each individual’s outcome has a larger \Nimpact on the overall results—
Dialogue: 0,0:03:29.34,0:03:32.55,Default,,0000,0000,0000,,and can therefore skew big-picture trends.
Dialogue: 0,0:03:32.55,0:03:37.49,Default,,0000,0000,0000,,Still, there are a lot of good reasons for\Nscientists to run small studies.
Dialogue: 0,0:03:37.49,0:03:39.27,Default,,0000,0000,0000,,By starting with a small sample,
Dialogue: 0,0:03:39.27,0:03:42.04,Default,,0000,0000,0000,,they can evaluate whether the results are \Npromising enough
Dialogue: 0,0:03:42.04,0:03:45.14,Default,,0000,0000,0000,,to run a more comprehensive, \Nexpensive study.
Dialogue: 0,0:03:45.14,0:03:48.61,Default,,0000,0000,0000,,And some research requires very specific\Nparticipants
Dialogue: 0,0:03:48.61,0:03:51.100,Default,,0000,0000,0000,,that may be impossible to \Nrecruit in large numbers.
Dialogue: 0,0:03:51.100,0:03:54.33,Default,,0000,0000,0000,,The key is reproducibility—
Dialogue: 0,0:03:54.33,0:03:57.63,Default,,0000,0000,0000,,if an article draws a conclusion \Nfrom one small study,
Dialogue: 0,0:03:57.63,0:03:59.65,Default,,0000,0000,0000,,that conclusion may be suspect—
Dialogue: 0,0:03:59.65,0:04:03.10,Default,,0000,0000,0000,,but if it’s based on many studies \Nthat have found similar results,
Dialogue: 0,0:04:03.10,0:04:04.70,Default,,0000,0000,0000,,it’s more credible.
Dialogue: 0,0:04:04.70,0:04:06.92,Default,,0000,0000,0000,,We’ve still got one more puzzle.
Dialogue: 0,0:04:06.92,0:04:11.90,Default,,0000,0000,0000,,In this scenario, a study tests a new drug\Nfor a rare, fatal disease.
Dialogue: 0,0:04:11.90,0:04:14.16,Default,,0000,0000,0000,,In a sample of 2,000 patients,
Dialogue: 0,0:04:14.16,0:04:17.64,Default,,0000,0000,0000,,the ones who start taking the drug upon \Ndiagnosis
Dialogue: 0,0:04:17.64,0:04:21.05,Default,,0000,0000,0000,,live longer than those who \Ntake the placebo.
Dialogue: 0,0:04:21.05,0:04:23.64,Default,,0000,0000,0000,,This time, the question \Nis slightly different.
Dialogue: 0,0:04:23.64,0:04:28.01,Default,,0000,0000,0000,,What’s one more thing you’d like to know \Nbefore deciding if the headline,
Dialogue: 0,0:04:28.01,0:04:33.62,Default,,0000,0000,0000,,"New drug prolongs lives of patients\Nwith rare disease", is justified?
Dialogue: 0,0:04:34.90,0:04:36.47,Default,,0000,0000,0000,,Before making this call,
Dialogue: 0,0:04:36.47,0:04:40.79,Default,,0000,0000,0000,,you’d want to know how much the drug \Nprolonged the patients’ lives.
Dialogue: 0,0:04:40.79,0:04:43.16,Default,,0000,0000,0000,,Sometimes, a study can have results that,
Dialogue: 0,0:04:43.16,0:04:48.24,Default,,0000,0000,0000,,while scientifically valid, don’t have \Nmuch bearing on real world outcomes.
Dialogue: 0,0:04:48.24,0:04:52.94,Default,,0000,0000,0000,,For example, one real-life clinical trial \Nof a pancreatic cancer drug
Dialogue: 0,0:04:52.94,0:04:57.46,Default,,0000,0000,0000,,found an increase in life expectancy—\Nof ten days.
Dialogue: 0,0:04:57.46,0:05:00.28,Default,,0000,0000,0000,,The next time you see a surprising medical\Nheadline,
Dialogue: 0,0:05:00.28,0:05:04.03,Default,,0000,0000,0000,,take a look at the science \Nit’s reporting on.
Dialogue: 0,0:05:04.03,0:05:06.96,Default,,0000,0000,0000,,Even when full papers aren’t \Navailable without a fee,
Dialogue: 0,0:05:06.96,0:05:09.99,Default,,0000,0000,0000,,you can often find summaries of \Nexperimental design
Dialogue: 0,0:05:09.99,0:05:13.18,Default,,0000,0000,0000,,and results in freely available abstracts,
Dialogue: 0,0:05:13.18,0:05:16.36,Default,,0000,0000,0000,,or even within the text \Nof a news article.
Dialogue: 0,0:05:16.36,0:05:19.67,Default,,0000,0000,0000,,It’s exciting to see scientific research \Ncovered in the news,
Dialogue: 0,0:05:19.67,0:05:23.40,Default,,0000,0000,0000,,and important to understand \Nthe studies’ findings.