1
00:00:00,025 --> 00:00:10,025
[MUSIC]

2
00:00:13,020 --> 00:00:19,010
In this video I will introduce
the Teachings Ratings Data.

3
00:00:19,010 --> 00:00:23,670
We have been working with teaching
ratings data from University of Texas.

4
00:00:23,670 --> 00:00:27,580
And the underlying question is,
if students teaching evaluations

5
00:00:27,580 --> 00:00:31,500
are influenced by the looks
of individual instructors.

6
00:00:31,500 --> 00:00:36,400
Or you can ask if they're teaching
evaluations differ by gender,

7
00:00:36,400 --> 00:00:41,430
or if good looking instructors
get higher teaching evaluations.

8
00:00:43,750 --> 00:00:49,220
I obtained this data from Professor
Daniel Hamermesh, who has written

9
00:00:49,220 --> 00:00:55,250
a paper about how beauty may impact in
an instructors teaching evaluation.

10
00:00:55,250 --> 00:00:58,480
In fact, he has written an amazing
book called Beauty Pays,

11
00:00:58,480 --> 00:01:00,780
in which he answers these questions.

12
00:01:00,780 --> 00:01:05,251
Such as do you think good looking
employees get higher pay or

13
00:01:05,251 --> 00:01:07,000
faster promotions?

14
00:01:07,000 --> 00:01:11,900
Do you think good looking instructors
get higher teaching evaluations and

15
00:01:11,900 --> 00:01:13,560
the data comes from University of Texas?

16
00:01:13,560 --> 00:01:15,470
It's a survey.

17
00:01:15,470 --> 00:01:20,115
And data obtained from 463 courses.

18
00:01:21,990 --> 00:01:26,110
So the data is first
referenced in the book,

19
00:01:26,110 --> 00:01:31,520
Getting Started with Data Science, making
sense of data with Analytics in Chapter 4.

20
00:01:31,520 --> 00:01:36,050
They're variables that essentially
define the attributes of instructors and

21
00:01:36,050 --> 00:01:37,600
characteristics of the courses.

22
00:01:37,600 --> 00:01:42,030
Some variables are continuous, others that
dichotomous or categorical variables.

23
00:01:42,030 --> 00:01:46,350
So the primary two
variables of our interests

24
00:01:46,350 --> 00:01:50,710
are beauty score,
which is basically the physical appearance

25
00:01:50,710 --> 00:01:55,140
of an instructor which was ranked
by a panel of six students.

26
00:01:55,140 --> 00:01:59,190
And I think that they have
normalized the beauty

27
00:01:59,190 --> 00:02:03,830
score such that it had the mean
of zero and variance of one or

28
00:02:03,830 --> 00:02:08,259
a standard deviation of one
the same as Z transformation.

29
00:02:08,259 --> 00:02:13,850
The dependent variable,
the variable of interest is evaluation,

30
00:02:13,850 --> 00:02:18,710
which is basically the teaching evaluation
ranging between the scale of 1 to 5,

31
00:02:18,710 --> 00:02:21,150
one being very unsatisfactory and

32
00:02:21,150 --> 00:02:25,830
the student found the course
to be excellent, then it's 5.

33
00:02:25,830 --> 00:02:29,210
And then there are other dichotomous or
binary or

34
00:02:29,210 --> 00:02:32,460
categorical variables such as minority.

35
00:02:32,460 --> 00:02:37,555
If the instructor was non Caucasian,
age is a continuous variable

36
00:02:37,555 --> 00:02:42,590
as the professors age or instructors age,
gender being male or female.

37
00:02:42,590 --> 00:02:45,640
Native stands for native English speaker.

38
00:02:45,640 --> 00:02:48,970
If the instructor with a native speaker
of English language10, otherwise.

39
00:02:48,970 --> 00:02:53,930
If the professor was tenured,
10 otherwise.

40
00:02:53,930 --> 00:02:58,150
I have produced some descriptors
statistics for your reference.

41
00:02:58,150 --> 00:03:02,170
For instance for the continuous
variables such as age and beauty and

42
00:03:02,170 --> 00:03:06,230
teaching evaluation and the number of
students who are enrolled in the course

43
00:03:06,230 --> 00:03:09,388
on the number of students who
performed the teaching evaluations.

44
00:03:09,388 --> 00:03:13,530
I produce the descriptive statistics
such as the minimum, the maximum mean,

45
00:03:13,530 --> 00:03:14,944
and standard deviation.

46
00:03:14,944 --> 00:03:19,242
For categorical variables such as gender,
female yes or no,

47
00:03:19,242 --> 00:03:24,463
visible minority yes or no person being
a tenured professor or otherwise.

48
00:03:24,463 --> 00:03:27,541
I produced the frequency distributions and

49
00:03:27,541 --> 00:03:32,210
percentage of individuals falling
in one category or otherwise.

50
00:03:32,210 --> 00:03:37,585
Notice that the teaching evaluation score
is 3.99, with the standard deviation of

51
00:03:37,585 --> 00:03:42,560
0.55 and
let's see if I were to produce a histogram

52
00:03:42,560 --> 00:03:47,730
of this variable teaching evaluation,
how will it look like with raw data?

53
00:03:47,730 --> 00:03:52,060
And if I were to use the normal
distribution and feed the two parameters,

54
00:03:52,060 --> 00:03:56,640
that is the mean and
the standard deviation, how will the same

55
00:03:56,640 --> 00:04:00,260
distribution look like using
a normal distribution?

56
00:04:00,260 --> 00:04:05,630
I am presenting here the distribution
of the raw data on left side and

57
00:04:05,630 --> 00:04:08,740
the presentation of the same data with
the same parameters of the mean and

58
00:04:08,740 --> 00:04:11,220
standard deviation using
normal distribution.

59
00:04:11,220 --> 00:04:14,080
You could see that the data not
exactly following a Bell curve.

60
00:04:14,080 --> 00:04:16,332
This is the raw data and it seldom does.

61
00:04:16,332 --> 00:04:19,615
But then the theoretical
distribution looks like this.

62
00:04:19,615 --> 00:04:23,853
Essentially the same data set with
a mean of 3.998 in standard deviation

63
00:04:23,853 --> 00:04:27,634
is presented here and then a normal
distribution drawn from these two

64
00:04:27,634 --> 00:04:30,453
characters appear step,
so it's much smoother.

65
00:04:30,453 --> 00:04:31,007
Theoretical distributions are much
smoother than the raw data.

66
00:04:31,007 --> 00:04:31,507
[MUSIC]