[Script Info]
Title: 
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:00.11,0:00:03.93,Default,,0000,0000,0000,,♪ [music] ♪
Dialogue: 0,0:00:20.88,0:00:22.08,Default,,0000,0000,0000,,- [Thomas Stratmann] Hi!
Dialogue: 0,0:00:22.08,0:00:24.27,Default,,0000,0000,0000,,In the upcoming series of videos,
Dialogue: 0,0:00:24.27,0:00:26.86,Default,,0000,0000,0000,,we're going to give you\Na shiny new tool
Dialogue: 0,0:00:26.86,0:00:30.41,Default,,0000,0000,0000,,to put into your\NUnderstanding Data toolbox:
Dialogue: 0,0:00:30.41,0:00:31.98,Default,,0000,0000,0000,,linear regression.
Dialogue: 0,0:00:32.88,0:00:34.67,Default,,0000,0000,0000,,Say you've got this theory.
Dialogue: 0,0:00:34.67,0:00:37.25,Default,,0000,0000,0000,,You've witnessed\Nhow good-looking people
Dialogue: 0,0:00:37.25,0:00:39.07,Default,,0000,0000,0000,,seem to get special perks.
Dialogue: 0,0:00:39.64,0:00:40.88,Default,,0000,0000,0000,,You're wondering,
Dialogue: 0,0:00:40.88,0:00:43.80,Default,,0000,0000,0000,,"Where else might we see\Nthis phenomenon?"
Dialogue: 0,0:00:44.13,0:00:45.64,Default,,0000,0000,0000,,What about for professors?
Dialogue: 0,0:00:45.64,0:00:48.26,Default,,0000,0000,0000,,Is it possible\Ngood-looking professors
Dialogue: 0,0:00:48.26,0:00:50.01,Default,,0000,0000,0000,,might get special perks too?
Dialogue: 0,0:00:50.35,0:00:53.90,Default,,0000,0000,0000,,Is it possible\Nstudents treat them better
Dialogue: 0,0:00:53.90,0:00:57.21,Default,,0000,0000,0000,,by showering them\Nwith better student evaluations?
Dialogue: 0,0:00:57.87,0:01:00.47,Default,,0000,0000,0000,,If so, is the effect of looks
Dialogue: 0,0:01:00.47,0:01:03.57,Default,,0000,0000,0000,,on evaluations really big \Nor really small?
Dialogue: 0,0:01:04.35,0:01:08.26,Default,,0000,0000,0000,,And say there is a new professor\Nstarting at a university.
Dialogue: 0,0:01:08.62,0:01:11.81,Default,,0000,0000,0000,,What can we predict\Nabout his evaluation
Dialogue: 0,0:01:11.81,0:01:13.37,Default,,0000,0000,0000,,simply by his looks?
Dialogue: 0,0:01:13.94,0:01:17.22,Default,,0000,0000,0000,,Given that these evaluations\Ncan determine pay raises,
Dialogue: 0,0:01:17.67,0:01:21.71,Default,,0000,0000,0000,,if this theory were true,\Nwe might see professors resort
Dialogue: 0,0:01:21.71,0:01:24.98,Default,,0000,0000,0000,,to some surprising tactics\Nto boost their scores.
Dialogue: 0,0:01:25.47,0:01:27.46,Default,,0000,0000,0000,,Suppose you wanted to find out
Dialogue: 0,0:01:27.46,0:01:30.80,Default,,0000,0000,0000,,if evaluations really improve\Nwith better looks.
Dialogue: 0,0:01:31.44,0:01:34.45,Default,,0000,0000,0000,,How would you go about\Ntesting this hypothesis?
Dialogue: 0,0:01:34.96,0:01:36.55,Default,,0000,0000,0000,,You could collect data.
Dialogue: 0,0:01:36.76,0:01:40.02,Default,,0000,0000,0000,,First you would have students rate\Non a scale from 1 to 10
Dialogue: 0,0:01:40.02,0:01:42.08,Default,,0000,0000,0000,,how good-looking a professor was,
Dialogue: 0,0:01:42.08,0:01:44.81,Default,,0000,0000,0000,,which gives you\Nan average beauty score.
Dialogue: 0,0:01:45.23,0:01:48.55,Default,,0000,0000,0000,,Then you could retrieve\Nthe teacher's teaching evaluations
Dialogue: 0,0:01:48.55,0:01:50.42,Default,,0000,0000,0000,,from twenty-five students.
Dialogue: 0,0:01:50.42,0:01:53.27,Default,,0000,0000,0000,,Let's look at these two variables\Nat the same time
Dialogue: 0,0:01:53.27,0:01:54.74,Default,,0000,0000,0000,,by using a scatterplot.
Dialogue: 0,0:01:54.98,0:01:57.42,Default,,0000,0000,0000,,We'll put beauty\Non the horizontal axis,
Dialogue: 0,0:01:57.85,0:02:00.59,Default,,0000,0000,0000,,and teacher evaluations\Non the vertical axis.
Dialogue: 0,0:02:01.46,0:02:03.17,Default,,0000,0000,0000,,For example, this dot\Nrepresents Professor Peate,
Dialogue: 0,0:02:03.17,0:02:06.17,Default,,0000,0000,0000,,- ["Star Wars"
Dialogue: 0,0:02:06.17,0:02:08.81,Default,,0000,0000,0000,,who received a beauty score of 3
Dialogue: 0,0:02:08.81,0:02:11.87,Default,,0000,0000,0000,,and an evaluation of 8.425.
Dialogue: 0,0:02:12.08,0:02:14.96,Default,,0000,0000,0000,,This one way out here\Nis Professor Helmchen.
Dialogue: 0,0:02:14.96,0:02:16.80,Default,,0000,0000,0000,,- [Ben Stiller, "Zoolander"]\NRidiculously good-looking!
Dialogue: 0,0:02:16.80,0:02:18.72,Default,,0000,0000,0000,,- [Thomas] Who got\Na very high beauty score,
Dialogue: 0,0:02:18.72,0:02:20.87,Default,,0000,0000,0000,,but not such a good evaluation.
Dialogue: 0,0:02:21.10,0:02:22.28,Default,,0000,0000,0000,,Can you see a trend?
Dialogue: 0,0:02:22.28,0:02:25.53,Default,,0000,0000,0000,,As we move from left to right\Non the horizontal axis,
Dialogue: 0,0:02:25.53,0:02:27.96,Default,,0000,0000,0000,,from the ugly to the gorgeous,
Dialogue: 0,0:02:27.96,0:02:31.19,Default,,0000,0000,0000,,we see a trend upwards\Nin evaluation scores.
Dialogue: 0,0:02:31.87,0:02:35.17,Default,,0000,0000,0000,,By the way, the data\Nwe're exploring in this series
Dialogue: 0,0:02:35.17,0:02:38.92,Default,,0000,0000,0000,,is not made up --\Nit comes from a real study
Dialogue: 0,0:02:38.92,0:02:40.90,Default,,0000,0000,0000,,done at the University of Texas.
Dialogue: 0,0:02:41.34,0:02:46.02,Default,,0000,0000,0000,,If you're wondering, "pulchritude"\Nis just the fancy academic way
Dialogue: 0,0:02:46.02,0:02:47.88,Default,,0000,0000,0000,,of saying beauty.
Dialogue: 0,0:02:48.40,0:02:51.47,Default,,0000,0000,0000,,With scatterplots\Nit can sometimes be hard
Dialogue: 0,0:02:51.47,0:02:55.59,Default,,0000,0000,0000,,to make out the exact relationship\Nbetween two variables --
Dialogue: 0,0:02:55.59,0:02:59.10,Default,,0000,0000,0000,,especially when the values\Nbounce around quite a bit
Dialogue: 0,0:02:59.10,0:03:01.32,Default,,0000,0000,0000,,as we go from left to right.
Dialogue: 0,0:03:02.00,0:03:04.91,Default,,0000,0000,0000,,One way to cut through\Nthis bounciness
Dialogue: 0,0:03:04.91,0:03:08.14,Default,,0000,0000,0000,,is to draw a straight line\Nthrough the data cloud
Dialogue: 0,0:03:08.14,0:03:10.78,Default,,0000,0000,0000,,in such a way that this line\Nsummarizes the data
Dialogue: 0,0:03:10.78,0:03:12.61,Default,,0000,0000,0000,,as closely as possible.
Dialogue: 0,0:03:13.30,0:03:17.18,Default,,0000,0000,0000,,The technical term for this\Nis "linear regression."
Dialogue: 0,0:03:17.67,0:03:20.89,Default,,0000,0000,0000,,Later on we'll talk about\Nhow this line is created,
Dialogue: 0,0:03:20.89,0:03:24.28,Default,,0000,0000,0000,,but for now we can assume\Nthat the line fits the data
Dialogue: 0,0:03:24.28,0:03:26.46,Default,,0000,0000,0000,,as closely as possible.
Dialogue: 0,0:03:27.09,0:03:29.54,Default,,0000,0000,0000,,So, what can this line tell us?
Dialogue: 0,0:03:30.07,0:03:32.60,Default,,0000,0000,0000,,First, we immediately see
Dialogue: 0,0:03:32.60,0:03:35.36,Default,,0000,0000,0000,,if the line is sloping\Nupward or downward.
Dialogue: 0,0:03:36.11,0:03:39.83,Default,,0000,0000,0000,,In our data set we see\Nthe [fitted] line slopes upward.
Dialogue: 0,0:03:40.79,0:03:43.81,Default,,0000,0000,0000,,It thus confirms what\Nwe have conjectured earlier
Dialogue: 0,0:03:43.81,0:03:45.59,Default,,0000,0000,0000,,by just looking at the scatterplot.
Dialogue: 0,0:03:46.07,0:03:50.24,Default,,0000,0000,0000,,The upward slope means\Nthat there is a positive association
Dialogue: 0,0:03:50.24,0:03:53.03,Default,,0000,0000,0000,,between looks\Nand evaluation scores.
Dialogue: 0,0:03:53.54,0:03:55.91,Default,,0000,0000,0000,,In other words, on average,
Dialogue: 0,0:03:55.91,0:03:59.47,Default,,0000,0000,0000,,better-looking professors\Nare getting better evaluations.
Dialogue: 0,0:03:59.77,0:04:03.94,Default,,0000,0000,0000,,For other data sets we might see\Na stronger positive association.
Dialogue: 0,0:04:04.38,0:04:07.42,Default,,0000,0000,0000,,Or, you might see\Na negative association.
Dialogue: 0,0:04:07.86,0:04:10.76,Default,,0000,0000,0000,,Or perhaps no association at all.
Dialogue: 0,0:04:11.16,0:04:13.90,Default,,0000,0000,0000,,And our lines\Ndon't have to be straight.
Dialogue: 0,0:04:14.39,0:04:17.30,Default,,0000,0000,0000,,They can curve to fit the data\Nwhen necessary.
Dialogue: 0,0:04:17.77,0:04:21.26,Default,,0000,0000,0000,,This line also gives us\Na way to predict outcomes.
Dialogue: 0,0:04:21.58,0:04:25.57,Default,,0000,0000,0000,,We can simply take a beauty score\Nand read off the line
Dialogue: 0,0:04:25.57,0:04:28.43,Default,,0000,0000,0000,,what the predicted\Nevaluation score would be.
Dialogue: 0,0:04:28.61,0:04:30.55,Default,,0000,0000,0000,,So, back to our new professor.
Dialogue: 0,0:04:31.10,0:04:34.11,Default,,0000,0000,0000,,We can precisely predict\Nhis evaluation score.
Dialogue: 0,0:04:34.68,0:04:36.75,Default,,0000,0000,0000,,"But wait! Wait!" you might say.
Dialogue: 0,0:04:37.02,0:04:38.75,Default,,0000,0000,0000,,"Can we trust this prediction?"
Dialogue: 0,0:04:39.23,0:04:41.66,Default,,0000,0000,0000,,How well does\Nthis one beauty variable
Dialogue: 0,0:04:41.66,0:04:43.52,Default,,0000,0000,0000,,really predict evaluations?
Dialogue: 0,0:04:44.84,0:04:47.89,Default,,0000,0000,0000,,Linear regression gives us\Nsome useful measures
Dialogue: 0,0:04:47.89,0:04:49.77,Default,,0000,0000,0000,,to answer those questions
Dialogue: 0,0:04:49.77,0:04:52.04,Default,,0000,0000,0000,,which we'll cover\Nin a future video.
Dialogue: 0,0:04:52.84,0:04:55.44,Default,,0000,0000,0000,,We also have to be aware\Nof other pitfalls
Dialogue: 0,0:04:55.44,0:04:58.34,Default,,0000,0000,0000,,before we draw\Nany definite conclusions.
Dialogue: 0,0:04:58.83,0:05:00.43,Default,,0000,0000,0000,,You could imagine a scenario
Dialogue: 0,0:05:00.43,0:05:03.64,Default,,0000,0000,0000,,where what is driving\Nthe association we see
Dialogue: 0,0:05:03.64,0:05:06.90,Default,,0000,0000,0000,,is really a third variable\Nthat we have left out.
Dialogue: 0,0:05:07.34,0:05:09.96,Default,,0000,0000,0000,,For example,\Nthe difficulty of the course
Dialogue: 0,0:05:09.96,0:05:12.46,Default,,0000,0000,0000,,might be behind\Nthe positive association
Dialogue: 0,0:05:12.46,0:05:15.64,Default,,0000,0000,0000,,between beauty ratings\Nand evaluation scores.
Dialogue: 0,0:05:16.05,0:05:18.96,Default,,0000,0000,0000,,Easy intro. courses\Nget good evaluations.
Dialogue: 0,0:05:19.23,0:05:22.97,Default,,0000,0000,0000,,Harder, more advanced courses\Nget bad evaluations.
Dialogue: 0,0:05:23.66,0:05:27.67,Default,,0000,0000,0000,,And younger professors might\Nget assigned to intro. courses.
Dialogue: 0,0:05:28.08,0:05:32.10,Default,,0000,0000,0000,,Then, if students judge\Nyounger professors more attractive,
Dialogue: 0,0:05:32.10,0:05:34.34,Default,,0000,0000,0000,,you will find\Na positive association
Dialogue: 0,0:05:34.34,0:05:37.38,Default,,0000,0000,0000,,between beauty ratings\Nand evaluation scores.
Dialogue: 0,0:05:37.86,0:05:40.39,Default,,0000,0000,0000,,But it's really\Nthe difficulty of the course,
Dialogue: 0,0:05:40.39,0:05:43.54,Default,,0000,0000,0000,,the variable that we've left out,\Nnot beauty,
Dialogue: 0,0:05:43.54,0:05:45.85,Default,,0000,0000,0000,,that is driving evaluation scores.
Dialogue: 0,0:05:46.35,0:05:49.81,Default,,0000,0000,0000,,In that case, all the primping\Nwould be for naught --
Dialogue: 0,0:05:50.29,0:05:54.44,Default,,0000,0000,0000,,a case of mistaken correlation\Nfor causation,
Dialogue: 0,0:05:54.90,0:05:58.17,Default,,0000,0000,0000,,something we'll talk about further\Nin a later video.
Dialogue: 0,0:05:58.92,0:06:02.07,Default,,0000,0000,0000,,And what if there were\Nother important variables
Dialogue: 0,0:06:02.07,0:06:05.78,Default,,0000,0000,0000,,that affect both beauty ratings\Nand evaluation scores?
Dialogue: 0,0:06:06.63,0:06:09.58,Default,,0000,0000,0000,,You might want to add\Nconsiderations like skill,
Dialogue: 0,0:06:09.85,0:06:14.58,Default,,0000,0000,0000,,race, sex, and whether English\Nis the teacher's native language
Dialogue: 0,0:06:14.58,0:06:18.99,Default,,0000,0000,0000,,to isolate more cleanly the effect\Nof beauty on evaluations.
Dialogue: 0,0:06:19.41,0:06:21.76,Default,,0000,0000,0000,,When we get\Ninto multiple regression
Dialogue: 0,0:06:21.76,0:06:24.48,Default,,0000,0000,0000,,we will be able to measure\Nthe impact of beauty
Dialogue: 0,0:06:24.48,0:06:26.22,Default,,0000,0000,0000,,on teacher evaluations
Dialogue: 0,0:06:26.22,0:06:28.37,Default,,0000,0000,0000,,while accounting\Nfor other variables
Dialogue: 0,0:06:28.37,0:06:30.74,Default,,0000,0000,0000,,that might confound\Nthis association.
Dialogue: 0,0:06:31.76,0:06:35.51,Default,,0000,0000,0000,,Next up, we'll get our hands dirty\Nby playing with this data
Dialogue: 0,0:06:35.51,0:06:39.07,Default,,0000,0000,0000,,to gain a better understanding\Nof what this line can tell us.
Dialogue: 0,0:06:41.17,0:06:42.44,Default,,0000,0000,0000,,- [Narrator] Congratulations!
Dialogue: 0,0:06:42.44,0:06:45.25,Default,,0000,0000,0000,,You're one step closer\Nto being a data ninja!
Dialogue: 0,0:06:45.57,0:06:47.14,Default,,0000,0000,0000,,However, to master this
Dialogue: 0,0:06:47.14,0:06:48.70,Default,,0000,0000,0000,,you'll need\Nto strengthen your skills
Dialogue: 0,0:06:48.70,0:06:50.40,Default,,0000,0000,0000,,with some practice questions.
Dialogue: 0,0:06:50.86,0:06:53.98,Default,,0000,0000,0000,,Ready for your next mission?\NClick "Next Video."
Dialogue: 0,0:06:54.31,0:06:55.36,Default,,0000,0000,0000,,Still here?
Dialogue: 0,0:06:55.60,0:06:58.32,Default,,0000,0000,0000,,Move from understanding data\Nto understanding your world
Dialogue: 0,0:06:58.32,0:07:01.64,Default,,0000,0000,0000,,by checking out MRU's\Nother popular economics videos.
Dialogue: 0,0:07:01.89,0:07:04.41,Default,,0000,0000,0000,,♪ [music] ♪