1
00:00:02,316 --> 00:00:04,696
In today's class, I'm going to show you how

2
00:00:04,696 --> 00:00:07,436
to interpret the results from
a very simple experiment.

3
00:00:07,436 --> 00:00:12,096
But before we do that, let's
briefly review the previous lesson,

4
00:00:12,216 --> 00:00:14,456
where we introduced some new terminology.

5
00:00:15,586 --> 00:00:19,906
We learned what the word "outcome"
means, when talking about experiments.

6
00:00:19,906 --> 00:00:22,976
The word response is also
used instead of outcome.

7
00:00:23,766 --> 00:00:27,036
We learned about factors which
we sometimes called variables.

8
00:00:27,036 --> 00:00:31,686
Factors, or variables, can
either be categorical or numeric,

9
00:00:32,066 --> 00:00:35,376
and we discussed examples of both types.

10
00:00:35,376 --> 00:00:38,506
It might seem strange that I'm
showing you how to analyze the results

11
00:00:38,506 --> 00:00:43,086
of the experiment before I even show how
I planned the experiment, or designed it.

12
00:00:43,086 --> 00:00:47,666
Sometimes though, it's useful to see the end
and then work backwards to the beginning.

13
00:00:48,506 --> 00:00:52,116
The experiment I'm going to
describe here has one outcome.

14
00:00:52,766 --> 00:00:58,096
The profit made in one day when selling
a specific type of product in a store.

15
00:00:59,416 --> 00:01:02,686
Remember that profit equals
income minus expenses.

16
00:01:03,556 --> 00:01:09,276
So when the store owner calculates the total
income and then subtracts the total expenses,

17
00:01:09,676 --> 00:01:15,176
that number is called the profit and
is the outcome of our experiments.

18
00:01:15,906 --> 00:01:19,836
The objective of this experiment
is to increase profit.

19
00:01:20,786 --> 00:01:24,976
We defined what an objective
was in the prior class.

20
00:01:24,976 --> 00:01:27,086
Now this experiment has two factors.

21
00:01:27,776 --> 00:01:30,916
The first factor is the amount
of light in our store.

22
00:01:31,986 --> 00:01:35,656
We can use a dimmer knob to control
just how much light is in the store.

23
00:01:36,456 --> 00:01:41,136
We can put that knob at 50%
or at 75%, for example.

24
00:01:41,136 --> 00:01:49,976
This gives a low amount of
light or a high amount of light.

25
00:01:50,106 --> 00:01:55,476
It is always good practice to predict what the
effect of the factor will be on your outcome.

26
00:01:56,556 --> 00:02:01,286
In this example, do we expect that
more lighting will increase the profit?

27
00:02:01,836 --> 00:02:05,126
Maybe more lighting will decrease the profit.

28
00:02:05,856 --> 00:02:09,946
Or do we expect that more lighting really
has no effect on the profit at all?

29
00:02:11,216 --> 00:02:12,646
We'll look at the results in a minute.

30
00:02:12,646 --> 00:02:19,486
The second factor we're going to consider in
the experiment is the price of the product.

31
00:02:20,316 --> 00:02:27,386
We could sell the product for $7.79 or
we could sell the product for $8.49.

32
00:02:27,386 --> 00:02:29,236
This again is a numeric variable.

33
00:02:31,626 --> 00:02:36,326
Try to predict what the effect of
price is going to be on the outcome.

34
00:02:37,596 --> 00:02:40,776
Do we expect a higher price
to increase the profit?

35
00:02:41,766 --> 00:02:44,896
Do we expect a higher price
to maybe decrease the profit?

36
00:02:46,376 --> 00:02:50,346
Or it is quite possible that higher
price has no effect on the profit.

37
00:02:52,406 --> 00:02:54,986
Pause the video for a moment and write

38
00:02:55,046 --> 00:03:00,126
down what you predict the outcome will
be for changing those two factors.

39
00:03:09,376 --> 00:03:11,066
OK, let's look at those results.

40
00:03:11,786 --> 00:03:15,876
We have to consider all combinations
in our experiments.

41
00:03:16,036 --> 00:03:19,756
The easiest way to consider this
is with a simple visualization.

42
00:03:21,136 --> 00:03:26,246
Let's start with a horizontal axis for the
dimmer amount, where we have a low amount

43
00:03:26,246 --> 00:03:30,656
of light on the left and a high
amount of light here on the right.

44
00:03:32,346 --> 00:03:36,156
Then, add a vertical axis
for the next factor, price.

45
00:03:37,136 --> 00:03:41,286
We have a low price here at the bottom
and then a high price here at the top.

46
00:03:42,616 --> 00:03:45,156
So there are three experiments here so far.

47
00:03:46,146 --> 00:03:48,776
There is actually a fourth
experiment we must run.

48
00:03:49,376 --> 00:03:53,836
It is the combination of high
amount of light with high prices.

49
00:03:55,346 --> 00:04:00,306
Let's assume you have four days and you could
run these four experiments: one per day.

50
00:04:00,776 --> 00:04:06,296
On the first Monday, you run the experiment
with high amount of light and low price.

51
00:04:06,916 --> 00:04:09,926
And let's say you get a profit of $570.

52
00:04:12,146 --> 00:04:16,326
Then the next Monday, you run the
combination of low amount of light

53
00:04:16,326 --> 00:04:19,826
and high price and the profit is $370.

54
00:04:21,186 --> 00:04:26,186
The following Monday, you have prices that
are high and with high amount of light.

55
00:04:26,526 --> 00:04:29,606
And the profit there might be $450.

56
00:04:31,176 --> 00:04:36,736
Then the last Monday you have the final
experiment with low light and low prices

57
00:04:37,196 --> 00:04:40,866
and your outcome is a profit of $490.

58
00:04:41,676 --> 00:04:43,386
Here are the results in table form.

59
00:04:44,036 --> 00:04:47,436
We are going to see this
format regularly in the course.

60
00:04:47,436 --> 00:04:50,876
I'm going to explain in a future
module why we ran the experiments

61
00:04:50,876 --> 00:04:53,716
in a different order to that
shown here in the table.

62
00:04:53,716 --> 00:04:56,656
This table order is called standard order.

63
00:04:57,146 --> 00:05:02,036
The simplest way to analyze this
data is to use a visual analysis.

64
00:05:02,066 --> 00:05:04,586
No math and no statistics
are required to do this.

65
00:05:04,706 --> 00:05:07,896
Start by drawing a square
of the four experiments.

66
00:05:08,296 --> 00:05:12,266
As before, we have the lighting
factor on the horizontal axis

67
00:05:12,646 --> 00:05:15,196
and pricing factor on the vertical axis.

68
00:05:16,616 --> 00:05:20,426
Next add a minus sign and
a plus sign on the axes

69
00:05:20,426 --> 00:05:23,096
to help indicate the low
levels and the high levels.

70
00:05:23,596 --> 00:05:27,626
We sometimes also write what the
low levels and the high levels are.

71
00:05:29,586 --> 00:05:36,526
Transfer the values of the outcome
variable from the table onto the square.

72
00:05:37,466 --> 00:05:42,836
In other words, copy those profit
values, the outcome variable, across.

73
00:05:43,226 --> 00:05:44,726
We call this a cube plot.

74
00:05:45,516 --> 00:05:49,486
Later on, when we have three factors,
you will see that it really is a cube.

75
00:05:51,486 --> 00:05:56,216
So $490 is written at the bottom left
because that was the outcome profit

76
00:05:56,216 --> 00:05:59,106
when we used low lighting and had lower prices.

77
00:05:59,896 --> 00:06:08,116
We add the $570 at the lower right, $370 at
the top left and finally $450 at the top right.

78
00:06:09,236 --> 00:06:11,976
Now we are ready to analyze the results.

79
00:06:13,406 --> 00:06:16,106
Let's analyze the effect of lighting amount.

80
00:06:16,466 --> 00:06:19,276
We have two chances to judge
the effect of lightning.

81
00:06:20,376 --> 00:06:25,056
The first is to consider the difference
between the high level of lighting compared

82
00:06:25,056 --> 00:06:29,996
to the low level of lighting, but
keeping prices at their low value.

83
00:06:31,556 --> 00:06:37,086
When we do this, we see the
difference is $570 minus $490.

84
00:06:37,136 --> 00:06:38,946
That's a difference of $80.

85
00:06:40,676 --> 00:06:44,526
Now let's look at the difference in
lighting when pricing is at the high value.

86
00:06:45,386 --> 00:06:49,646
This time, it's $450 minus $370.

87
00:06:49,766 --> 00:06:51,846
That's a difference of $80 again.

88
00:06:52,776 --> 00:07:01,246
So we can see then, the effect of lighting
causes our profit to increase by $80 as we move

89
00:07:01,246 --> 00:07:04,546
from a low level of light
to a high level of light.

90
00:07:05,406 --> 00:07:10,956
This effect is consistent both
at low prices and at high prices.

91
00:07:12,426 --> 00:07:14,766
Now let's consider the effect
of adjusting price.

92
00:07:15,876 --> 00:07:22,996
At low amount of light, we can compare
$370 in profit to $490 in profit.

93
00:07:23,916 --> 00:07:25,406
We interpret this as follows.

94
00:07:26,326 --> 00:07:32,326
As prices are increased the profit
will, in fact, decrease by $120.

95
00:07:33,626 --> 00:07:38,246
We can also consider the effect of adjusting
prices at the high level of lighting.

96
00:07:39,176 --> 00:07:43,546
This time we have $450 minus $570.

97
00:07:43,636 --> 00:07:46,146
That's a decrease of $120 again.

98
00:07:46,536 --> 00:07:47,786
So let's recap.

99
00:07:48,506 --> 00:07:52,516
The effect of lighting is that it
increases our outcome variable.

100
00:07:52,896 --> 00:08:00,136
Our profit goes up by $80 when we move from
50% on the dimmer to 75% on the dimmer.

101
00:08:00,856 --> 00:08:03,406
So it's a good thing to use
more light in our store.

102
00:08:04,656 --> 00:08:06,616
Pricing though, has an interesting effect.

103
00:08:07,056 --> 00:08:12,776
It shows that our profit decreases
by a $120 when we raise prices

104
00:08:12,776 --> 00:08:14,966
from the lower level to the higher level.

105
00:08:14,966 --> 00:08:18,416
This example was in fact quite simple.

106
00:08:19,156 --> 00:08:22,746
Most experiments will not have the
same difference in values on the left

107
00:08:22,746 --> 00:08:26,806
and the right hand sides, or on the
top side and bottom side of the square.

108
00:08:28,086 --> 00:08:31,226
We will get into more complex
examples in the next videos.

109
00:08:31,226 --> 00:08:34,496
Before we end the class I want you to
consider what could have gone wrong.

110
00:08:34,496 --> 00:08:36,696
What if, on the day that we did our
experiments, there was bad weather

111
00:08:36,696 --> 00:08:38,096
and we had fewer customers come to our store.

112
00:08:38,096 --> 00:08:41,766
What if we repeated the experiment
a second time,

113
00:08:42,486 --> 00:08:44,996
would we get the same profit outcome value?

114
00:08:45,366 --> 00:08:49,476
In other words, is our experiment reproducible?

115
00:08:49,476 --> 00:08:52,556
Those are important points we have
to bear in mind and we're going

116
00:08:52,556 --> 00:08:54,466
to consider them in the coming classes.

117
00:08:54,846 --> 00:08:55,926
See you next time.