1
00:00:00,025 --> 00:00:07,200
[MUSIC]

2
00:00:07,200 --> 00:00:11,440
In this video we will illustrate how one
can use regression models in place of

3
00:00:11,440 --> 00:00:14,660
tests conducted for correlation analysis.

4
00:00:14,660 --> 00:00:18,300
We will return to the basics.
There are two types, or

5
00:00:18,300 --> 00:00:23,240
mostly two types of variables. First,
are the categorical variables for which we

6
00:00:23,240 --> 00:00:28,445
use chi-square tests to determine if
there is an association between the two.

7
00:00:28,445 --> 00:00:31,540
And second are the categorical variables,
or

8
00:00:31,540 --> 00:00:36,480
we could have continuous variables where
we use the Pearson correlation test.

9
00:00:36,480 --> 00:00:41,170
We will focus on just the continuous
variables. We can plot two continuous

10
00:00:41,170 --> 00:00:46,620
variables in a scatter plot. The teaching
evaluation scores are on the Y axis and

11
00:00:46,620 --> 00:00:51,690
the normalized beauty scores are on the X
axis. You could sort of see a relationship

12
00:00:51,690 --> 00:00:56,310
between the two variables. It's an upward
sloping type of a relationship.

13
00:00:56,310 --> 00:01:01,630
We see that as the beauty score increases,
so does the teaching evaluation score.

14
00:01:01,630 --> 00:01:06,140
Remember, we used the Pearson correlation
test to determine the relationship and

15
00:01:06,140 --> 00:01:08,050
its significance level.

16
00:01:08,050 --> 00:01:12,470
Now let's do the same in regression.
Just like we did with the T test and

17
00:01:12,470 --> 00:01:17,170
the F test, we will fit a linear model for
both the beauty, an evaluation score,

18
00:01:17,170 --> 00:01:25,090
values and print out the models summary.

19
00:01:25,090 --> 00:01:30,340
Taking a closer look,
it prints out a P value of 4.25 * 10

20
00:01:30,340 --> 00:01:35,720
raised to power negative 5,
which is less than 0.05..

21
00:01:35,720 --> 00:01:39,330
That is very similar to when
we run the Pearson r function.

22
00:01:39,330 --> 00:01:45,970
It will also give us the R-squared- value,
that is, if we took square root 0.036.

23
00:01:45,970 --> 00:01:46,751
It will give us 0.189
which is the some value as

24
00:01:46,751 --> 00:01:47,686
the correlation coefficient
from computing the Pearson R.

25
00:01:47,686 --> 00:01:48,186
[MUSIC]