Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:00,004 --> 00:00:02,001
- [Instructor] Businesses run on data.
2
00:00:02,001 --> 00:00:04,006
Which is both good news and bad news.
3
00:00:04,006 --> 00:00:07,002
The good news is that it has never been easier to
4
00:00:07,002 --> 00:00:10,005
collect data about your customers, products and sales.
5
00:00:10,005 --> 00:00:12,009
The bad news is that you still have to decide
6
00:00:12,009 --> 00:00:15,002
what all that data means.
7
00:00:15,002 --> 00:00:16,004
As an example,
8
00:00:16,004 --> 00:00:19,003
suppose you ran a web advertising campaign
9
00:00:19,003 --> 00:00:21,005
with a goal of bringing in at least 70
10
00:00:21,005 --> 00:00:24,000
new qualified customers per week.
11
00:00:24,000 --> 00:00:27,000
You ran the ads on 25 different sites for four weeks.
12
00:00:27,000 --> 00:00:30,004
So you have a sample size of 100.
13
00:00:30,004 --> 00:00:31,003
On average,
14
00:00:31,003 --> 00:00:33,007
each ad brought in 73 new customers
15
00:00:33,007 --> 00:00:38,001
with a standard deviation of 15.1.
16
00:00:38,001 --> 00:00:39,002
The central question is though,
17
00:00:39,002 --> 00:00:41,004
whether the ads met your goal of bringing in
18
00:00:41,004 --> 00:00:44,001
70 new customers with each campaign.
19
00:00:44,001 --> 00:00:47,005
Your sample size is 100 and your mean is 73.
20
00:00:47,005 --> 00:00:49,005
But the standard deviation is pretty large.
21
00:00:49,005 --> 00:00:51,000
Over 15.
22
00:00:51,000 --> 00:00:53,005
So you have to ask yourself one question,
23
00:00:53,005 --> 00:00:55,006
did I just get lucky?
24
00:00:55,006 --> 00:00:57,003
I've used the numbers I collected
25
00:00:57,003 --> 00:00:59,001
to calculate my standard error
26
00:00:59,001 --> 00:01:01,008
and what's called a Z-score.
27
00:01:01,008 --> 00:01:04,002
The Z-score lets me calculate how likely it is
28
00:01:04,002 --> 00:01:06,007
my results were due to chance.
29
00:01:06,007 --> 00:01:08,008
And in this case, again,
30
00:01:08,008 --> 00:01:11,001
based on all my values,
31
00:01:11,001 --> 00:01:15,001
the probability is 2.35%.
32
00:01:15,001 --> 00:01:16,000
And
33
00:01:16,000 --> 00:01:17,009
2.35%, okay.
34
00:01:17,009 --> 00:01:19,000
Is that good?
35
00:01:19,000 --> 00:01:20,001
Bad,
36
00:01:20,001 --> 00:01:21,007
neither?
37
00:01:21,007 --> 00:01:24,003
Most analysts say that a result is significant
38
00:01:24,003 --> 00:01:26,001
if the probability your results were due
39
00:01:26,001 --> 00:01:28,009
to chance is below 5%.
40
00:01:28,009 --> 00:01:29,009
In other words,
41
00:01:29,009 --> 00:01:32,008
if your data would show that the conclusion will be wrong
42
00:01:32,008 --> 00:01:34,006
less than one time out of 20,
43
00:01:34,006 --> 00:01:38,003
then you can claim the result is significant.
44
00:01:38,003 --> 00:01:39,002
So in this case,
45
00:01:39,002 --> 00:01:42,001
97.65%,
46
00:01:42,001 --> 00:01:45,000
or 100% minus 2.35,
47
00:01:45,000 --> 00:01:49,000
appears to be significant past the 95% confidence level.
48
00:01:49,000 --> 00:01:50,007
Of course, there is still the possibility
49
00:01:50,007 --> 00:01:53,000
your results actually were due to chance
50
00:01:53,000 --> 00:01:56,006
and that your ads did not have a meaningful effect.
51
00:01:56,006 --> 00:01:58,005
That's where your knowledge of your business,
52
00:01:58,005 --> 00:02:01,005
intuition and willingness to continue analyzing
53
00:02:01,005 --> 00:02:03,000
comes in handy.
3737
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.