1
00:00:00,600 --> 00:00:05,070
Hello, guys, and welcome back to another class of our course, I want the complete introduction to

2
00:00:05,070 --> 00:00:08,040
data science and python programming.

3
00:00:08,940 --> 00:00:15,480
So in this class, we are going to have a basic introduction to some statistical concepts showing the

4
00:00:15,480 --> 00:00:16,080
best class.

5
00:00:16,080 --> 00:00:21,660
We talked about what our statistics and in this class, we are going to talk about some basic statistical

6
00:00:21,660 --> 00:00:23,850
concepts as we see the statistic.

7
00:00:23,850 --> 00:00:28,620
Concepts that we are going to cover in this class would be the sampling.

8
00:00:28,920 --> 00:00:32,020
And this is exactly what we are going to talk about.

9
00:00:32,040 --> 00:00:35,950
So let's jump right into it right now.

10
00:00:36,420 --> 00:00:37,970
So basically, it's possible.

11
00:00:38,190 --> 00:00:42,680
Well, in statistics or in the world in general, there is a population.

12
00:00:42,690 --> 00:00:46,710
So let's say, for example, we are studying, we want to study a certain population, want to study,

13
00:00:46,710 --> 00:00:48,840
for example, the population of the United States.

14
00:00:49,530 --> 00:00:54,110
So to know, for example, if they love pizza, once again, it's just an example.

15
00:00:54,360 --> 00:01:02,850
So we can ask each citizen of the US where each person who lives in the US, if you love pizza by sending

16
00:01:03,090 --> 00:01:07,450
them a survey, by just asking them a question or whatever.

17
00:01:07,830 --> 00:01:13,830
So basically you can ask the whole population, but it's not possible to ask the whole population because

18
00:01:13,830 --> 00:01:21,420
first of all, it's going to take too much resources to monitor control and will be able to ask the

19
00:01:21,420 --> 00:01:28,410
whole population this question if they like pizza, for example, and also because logistically it's

20
00:01:28,410 --> 00:01:34,310
almost impossible and there is no well, there will be no outcome.

21
00:01:34,320 --> 00:01:40,180
So basically, you will not make more money by asking the whole population if they love pizza.

22
00:01:40,650 --> 00:01:42,340
So what exactly would we do?

23
00:01:42,630 --> 00:01:48,780
So instead of asking the whole population to ask a sample of the population and this is where we have

24
00:01:48,780 --> 00:01:51,210
the difference between a population and a sample.

25
00:01:51,450 --> 00:01:59,610
So basically, as you can see here, the population will be the whole well, all the US citizen, all

26
00:01:59,610 --> 00:02:06,750
the US population, and the sample will simply be a part of this population to who will ask the question.

27
00:02:07,260 --> 00:02:10,770
So first of all, having a sample is way easier.

28
00:02:11,610 --> 00:02:16,230
So in this case, when you guys have a sample to who you will ask a certain question.

29
00:02:17,070 --> 00:02:23,280
It's easier to monitor, it's easier to control, it's easier to find a sample than to ask the whole

30
00:02:23,280 --> 00:02:23,880
population.

31
00:02:24,120 --> 00:02:25,530
Logistically, it's way easier.

32
00:02:25,830 --> 00:02:29,190
And usually the results are pretty much good.

33
00:02:29,190 --> 00:02:36,270
But there is a lot of different sampling techniques and then not all of them are well are perfect,

34
00:02:36,270 --> 00:02:39,270
but usually they work pretty well.

35
00:02:40,560 --> 00:02:44,250
So you have two types of sampling methods.

36
00:02:44,520 --> 00:02:51,570
So they have probability sampling method and no probability sampling method and basically the main difference

37
00:02:51,570 --> 00:02:53,340
between both of them.

38
00:02:53,340 --> 00:03:01,410
So probability and no probability is that the probability sampling method will be based on selecting

39
00:03:01,410 --> 00:03:02,520
random people.

40
00:03:02,910 --> 00:03:11,700
So you will try to select the much random people as possible and then no probability sampling method

41
00:03:12,660 --> 00:03:16,260
will not necessarily select random people always.

42
00:03:16,260 --> 00:03:23,190
So yes, we'll have random people, but you will understand the next few slides why it's not as well

43
00:03:23,460 --> 00:03:31,620
as random randomized as the sample sampling, the probability sampling method, which is based really

44
00:03:31,620 --> 00:03:33,210
on random people.

45
00:03:34,650 --> 00:03:42,090
Does it make more doesn't make the probability sampling method worse than the probability sampling method

46
00:03:42,090 --> 00:03:45,710
because the people are not necessarily 100 percent random?

47
00:03:47,010 --> 00:03:49,800
No, but at the same time it's not the best.

48
00:03:49,800 --> 00:03:55,860
So usually experts prefer using probability sampling method because it's based more on random people

49
00:03:55,860 --> 00:04:00,660
and it's really well, it's more randomized than the non the non probability sampling method.

50
00:04:01,590 --> 00:04:08,490
But once again, sometimes it's way more practical to work with non probability sampling method, even

51
00:04:08,490 --> 00:04:12,300
if it's not the best then to use the probability sampling method.

52
00:04:12,990 --> 00:04:23,160
So let's understand what is exactly the same thing method and what are all the the the sampling methods

53
00:04:23,160 --> 00:04:23,670
right here.

54
00:04:25,470 --> 00:04:29,850
So basically the probability sampling method we have for sampling methods.

55
00:04:30,060 --> 00:04:35,760
So we have the sample random samples, the systematic samples, the stratified samples and finally the

56
00:04:35,760 --> 00:04:36,960
cluster sampling.

57
00:04:37,710 --> 00:04:43,830
The sample, a random sample is basically, let's say, for example, we have a population and of one

58
00:04:43,830 --> 00:04:45,060
thousand people, for example.

59
00:04:45,300 --> 00:04:50,340
And you want to select, let's say, ten people, you need ten people for your sample.

60
00:04:50,700 --> 00:04:55,530
So instead of saying, let's say, I don't know, isolate selected, I will select really random people

61
00:04:55,530 --> 00:04:57,570
of those one thousand persons.

62
00:04:57,840 --> 00:04:59,460
So I will select for example.

63
00:04:59,460 --> 00:04:59,730
I will.

64
00:05:00,190 --> 00:05:08,710
Really, ten random people, and this would be really simple, random sample, and this is really simple

65
00:05:08,710 --> 00:05:11,520
to do, so you just select 10 persons.

66
00:05:12,220 --> 00:05:14,920
The next one would be the systematic sample.

67
00:05:15,430 --> 00:05:17,340
It's pretty much the same.

68
00:05:17,530 --> 00:05:20,110
Well, it's not the same thing as the simple random sample.

69
00:05:20,320 --> 00:05:24,770
It looks like the simple random sample, but it's a bit more different.

70
00:05:25,030 --> 00:05:30,730
So let's say, for example, we have once again our population of one thousand people, and that will

71
00:05:30,740 --> 00:05:33,240
give we'll give a number to each person.

72
00:05:33,250 --> 00:05:37,120
So let's say, for example, each person will be number one, number two, number three until one thousand.

73
00:05:38,140 --> 00:05:42,790
Then from that moment, let's say, for example, you select the person number five.

74
00:05:43,180 --> 00:05:49,640
And from this person, you say, OK, each plus, then I would be used inside of my sample.

75
00:05:49,990 --> 00:05:55,060
So in other words, you select person number five, then the next person in your sample would be the

76
00:05:55,060 --> 00:05:57,370
person number 15 to take the person number five.

77
00:05:57,370 --> 00:05:58,420
Fifteen twenty five.

78
00:05:58,420 --> 00:05:59,990
Thirty five, forty five.

79
00:06:00,250 --> 00:06:04,600
So as you can see in the picture here, you will select the person number five, your person.

80
00:06:04,600 --> 00:06:13,580
Twenty five, fifteen, twenty five is going to be the exact same jump each and every time for the stratified.

81
00:06:13,600 --> 00:06:14,130
So the next.

82
00:06:14,140 --> 00:06:16,330
So this is for this systematic sample.

83
00:06:17,080 --> 00:06:18,830
As you can see, it's not really complicated.

84
00:06:19,510 --> 00:06:22,100
So the next one would be the stratified sampling.

85
00:06:22,810 --> 00:06:24,220
This one is a bit different.

86
00:06:24,230 --> 00:06:26,470
So once again, it's based on the random people.

87
00:06:26,470 --> 00:06:34,420
But you want to let's say, for example, you want to to have a clear how you said you want to have

88
00:06:34,420 --> 00:06:39,100
a clear representation of each population inside of your sample.

89
00:06:39,400 --> 00:06:43,540
So let's say, for example, you have once again your population of one thousand people and you have

90
00:06:43,540 --> 00:06:50,390
two hundred people who wear a green t shirt and 800 people who wear, let's say, a blue t shirt.

91
00:06:50,950 --> 00:06:54,540
So what you will do, you will select two and you need a sample of 10 persons.

92
00:06:54,910 --> 00:07:01,390
So you will select two persons who wear a green t shirt and you select eight persons, wear a blue t

93
00:07:01,390 --> 00:07:06,700
shirt just to be able to represent properly each population inside of your sample.

94
00:07:06,910 --> 00:07:13,360
So you will not select just random people in this population, because let's say you have five persons

95
00:07:13,360 --> 00:07:19,150
who wear a green t shirt and five persons who wear a blue T-shirt, that this will not be representative

96
00:07:19,150 --> 00:07:24,280
of your population because you have way more persons with a with a blue t shirt.

97
00:07:25,210 --> 00:07:30,280
So in this case, let's say you select eight persons with a blue t shirt, but you will select random

98
00:07:30,280 --> 00:07:35,960
people inside of this population of eight hundred people with a blue t shirt.

99
00:07:36,520 --> 00:07:43,750
So basically the stratified sample is really selecting is really selecting people well, the right amount

100
00:07:43,750 --> 00:07:45,170
of people in each group.

101
00:07:45,190 --> 00:07:51,910
So basically, if you have a bigger group of people with a blue t shirt necessarily, you will select

102
00:07:52,450 --> 00:07:58,940
a well, a higher proportion of people in the in the group of people who have blue t shirts.

103
00:07:59,410 --> 00:08:01,510
So this is for the stratified sample.

104
00:08:01,900 --> 00:08:08,890
The last type of sample that is in the probability sampling method would be the cluster.

105
00:08:09,880 --> 00:08:16,090
The cluster sampling and basically this method is really well involves dividing the population into

106
00:08:16,090 --> 00:08:23,950
subgroups and that each subgroup would have similar characteristics, characteristics so to the whole

107
00:08:23,950 --> 00:08:24,550
sample.

108
00:08:24,700 --> 00:08:30,730
So in other words, instead of sampling individual individuals from each subgroup, you randomly select

109
00:08:30,730 --> 00:08:31,810
entire subgroups.

110
00:08:32,170 --> 00:08:33,740
So basically, what does this mean?

111
00:08:33,760 --> 00:08:36,440
It means you will work with that.

112
00:08:36,460 --> 00:08:37,960
Well, small you will.

113
00:08:38,410 --> 00:08:41,050
Let's say you have a population of one thousand people once again.

114
00:08:41,710 --> 00:08:46,350
And this population, you will divide it into groups of, let's say, five people.

115
00:08:46,630 --> 00:08:51,340
So instead of selecting one person, you will select a whole group of five people.

116
00:08:52,340 --> 00:08:55,390
So it's like the random sampling, the sample, random sample.

117
00:08:55,630 --> 00:09:00,430
But instead of working with one person, you will work with small groups.

118
00:09:00,460 --> 00:09:02,830
So basically this would be groups.

119
00:09:04,150 --> 00:09:06,510
So this is for the probability sampling methods.

120
00:09:06,520 --> 00:09:10,830
As I said, those methods are really into random.

121
00:09:10,840 --> 00:09:19,570
So you work with random people and as those are really the most used by experts because they are more

122
00:09:19,570 --> 00:09:25,870
representative of the population and you have more representative numbers in general with those sampling

123
00:09:25,870 --> 00:09:26,290
methods.

124
00:09:27,250 --> 00:09:32,200
So the second type of sampling methods will be the non probability sampling method.

125
00:09:32,500 --> 00:09:38,680
And that in this case, as I said, this one is not well, it's a bit different from the sampling method

126
00:09:38,980 --> 00:09:44,530
because you will work with people not who you know, but that are more accessible to you.

127
00:09:45,100 --> 00:09:51,910
So basically, you will select the people with who who you ask or who you will conduct the survey or

128
00:09:51,910 --> 00:09:52,950
whatever you want to do.

129
00:09:52,960 --> 00:09:53,860
So you study.

130
00:09:53,860 --> 00:09:59,620
Let's say you want to study people who love pizza so you will decide to who you ask.

131
00:10:00,100 --> 00:10:07,000
The question, so let's see those for sampling method, so first of all, we have the convenience sample,

132
00:10:07,000 --> 00:10:13,150
which is the first one right here, then we have the involuntary response sample, the purposive sample,

133
00:10:13,150 --> 00:10:15,070
and finally the snowball sample.

134
00:10:15,940 --> 00:10:17,920
So first of all, we have the communion sample.

135
00:10:17,930 --> 00:10:22,810
This one is, well, the simplest basically how it works.

136
00:10:22,810 --> 00:10:24,550
It's pretty simple.

137
00:10:24,880 --> 00:10:32,680
It's imagine you and you need a sample of 10 people and you don't select random people.

138
00:10:32,680 --> 00:10:34,460
You just select people who you know.

139
00:10:34,480 --> 00:10:40,480
So in this case, let's say, for example, you have friends and this is way more convenient for you

140
00:10:40,660 --> 00:10:45,880
to ask your friends to fill up the survey than to go and ask some strangers somewhere.

141
00:10:46,240 --> 00:10:48,100
So basically, this is the convenience sample.

142
00:10:48,100 --> 00:10:53,560
So you ask, let's say, for example, your friends, you tell your friends, you have your sample of

143
00:10:53,560 --> 00:10:55,810
10 people and here we go.

144
00:10:55,810 --> 00:11:00,250
You have well, you have your sample and you conduct your study just with your 10 friends.

145
00:11:00,610 --> 00:11:04,570
But let's say, for example, the ten of your friends, like pizza once again, is not representative

146
00:11:04,570 --> 00:11:08,100
of the whole population because once again, you controlled the sample.

147
00:11:08,500 --> 00:11:11,320
So this is why it's not always a good thing.

148
00:11:12,760 --> 00:11:16,060
The second thing would be the voluntary response sample.

149
00:11:17,200 --> 00:11:20,020
This one is pretty much as the convenience sample.

150
00:11:20,020 --> 00:11:24,580
But instead of asking your friends, you ask, well.

151
00:11:26,450 --> 00:11:31,970
You will ask people to answer, well, with surveys and everything, you ask people to volunteer to

152
00:11:31,970 --> 00:11:33,580
respond to your sample.

153
00:11:33,650 --> 00:11:39,110
So basically you let's say, come somewhere, you drop your survey and people come and fill up your

154
00:11:39,110 --> 00:11:39,560
survey.

155
00:11:40,460 --> 00:11:47,570
Once again, this is really limited and this is not really random and representative of the whole population,

156
00:11:47,570 --> 00:11:52,970
because let's say, for example, you're at the university, you come at your campus somewhere, you

157
00:11:52,970 --> 00:11:57,590
sit down, you put on your survey, you will not necessarily have access to the whole population or

158
00:11:57,590 --> 00:11:59,270
to a representative sample.

159
00:11:59,660 --> 00:12:05,870
You will have people, let's say, from one department who are studying at that department, at that

160
00:12:05,870 --> 00:12:06,590
university.

161
00:12:06,830 --> 00:12:12,800
Who will Majali answer your survey, which is not really representative of the whole population.

162
00:12:12,800 --> 00:12:17,510
It's just representative of the department where you conducted the survey in question.

163
00:12:18,950 --> 00:12:19,210
All right.

164
00:12:19,220 --> 00:12:24,980
So this is for the voluntary response sample and then after that will have the purpose of sampling.

165
00:12:25,160 --> 00:12:30,650
So this method of sampling will simply involve the researcher using their judgment, judgment to select

166
00:12:30,650 --> 00:12:34,800
a sample that is the most useful to the purpose of the research.

167
00:12:35,660 --> 00:12:43,110
In other words, you select the people to who you will to you will conduct the research.

168
00:12:43,490 --> 00:12:48,110
So once again, this is not representative at all, in my opinion.

169
00:12:48,110 --> 00:12:54,260
Once again, of in this case, if you want to if you want to stay as random as possible, this is not

170
00:12:54,260 --> 00:12:55,270
representative at all.

171
00:12:56,030 --> 00:12:56,330
Why?

172
00:12:56,340 --> 00:13:03,250
Because once again, the researcher will select the people who for him are fit to conduct the research.

173
00:13:03,560 --> 00:13:10,280
So let's say, for example, you want to look for people who, I don't know, luff be used and the researcher

174
00:13:10,460 --> 00:13:18,050
wants to have a positive, let's say, for example, a positive positive answers in his in his, I don't

175
00:13:18,050 --> 00:13:18,670
know, his survey.

176
00:13:19,130 --> 00:13:24,310
So he starts asking people who he knows love pizza to answer his service.

177
00:13:24,320 --> 00:13:29,440
Once again, the results will not will not necessarily represent the population in general.

178
00:13:29,810 --> 00:13:32,640
So basically purpose of sampling is just that.

179
00:13:32,650 --> 00:13:41,670
Well, it could work in some cases in cases that you really need some exact people, some really you

180
00:13:41,750 --> 00:13:46,390
need subjects that are that fit, that fits certain criteria.

181
00:13:46,670 --> 00:13:50,780
But if you want to stay as random as possible, it's not the best way to conduct the research.

182
00:13:51,170 --> 00:13:58,540
This is the third type of non probability sampling method, and the last one would be the snowball sampling.

183
00:13:59,180 --> 00:14:02,390
So in this case, you will do the exact same thing.

184
00:14:02,390 --> 00:14:04,580
So it's going to be the way it works.

185
00:14:04,580 --> 00:14:08,720
It's like convenience sampling, but just with a snowball effect.

186
00:14:08,960 --> 00:14:13,850
So I'd say, for example, you ask two of your friends to answer your survey and you ask your friends

187
00:14:13,850 --> 00:14:20,030
to share your survey with some other people so your friends will share your survey with people who will

188
00:14:20,030 --> 00:14:23,180
share your survey with other people, etc..

189
00:14:23,510 --> 00:14:30,050
So basically, as I said, it's like a convenience sample, but just with a snowball effect, in some

190
00:14:30,050 --> 00:14:32,300
cases, this could be really effective.

191
00:14:32,300 --> 00:14:37,400
So let's say, for example, you are studying a I don't know, you are studying a certain community

192
00:14:37,410 --> 00:14:41,900
somewhere in the country and, you know, just one person inside of this community.

193
00:14:41,900 --> 00:14:48,380
So maybe this person, you can ask the person to fill up your survey and ask this person to share,

194
00:14:48,380 --> 00:14:52,880
for example, this survey in his or her community.

195
00:14:53,300 --> 00:14:57,800
And in this case, you will be able to conduct your study on this community in general.

196
00:14:58,640 --> 00:15:05,240
Once again, this is not necessarily representative of well, of the community because it's maybe representative

197
00:15:05,240 --> 00:15:08,790
of the community in a certain place.

198
00:15:08,790 --> 00:15:13,010
So, for example, you are studying, I don't know, you are studying population of people who wear

199
00:15:13,010 --> 00:15:15,950
a blue t shirt in the United States.

200
00:15:15,950 --> 00:15:20,060
But you are conducting your research, let's say, in New York City.

201
00:15:20,360 --> 00:15:23,480
So you ask someone to fill up your survey and to ask.

202
00:15:23,660 --> 00:15:29,090
And since you don't know people who are wearing blue t shirts, so this person who is wearing a blue

203
00:15:29,090 --> 00:15:34,250
t shirt will simply hand your survey to other people who are wearing blue t shirts.

204
00:15:34,370 --> 00:15:37,760
Once again, this is just an example for you guys to understand.

205
00:15:38,930 --> 00:15:43,340
So in this case, it could be representative.

206
00:15:43,340 --> 00:15:50,060
But until a certain point, and it's not like the best way to represent well, to to represent and conduct

207
00:15:50,060 --> 00:15:54,650
a random study to represent the population in general and to conduct a study.

208
00:15:55,550 --> 00:15:58,400
So this is for these sampling methods that exist.

209
00:15:58,970 --> 00:16:01,850
So, as I said, you have eight of those sampling methods.

210
00:16:02,300 --> 00:16:05,000
You have two groups of sampling methods.

211
00:16:05,000 --> 00:16:11,150
You have the probability sampling method and the non probability sensing method to understand the difference

212
00:16:11,150 --> 00:16:12,110
between both.

213
00:16:12,110 --> 00:16:14,060
The first one is really randomized.

214
00:16:14,060 --> 00:16:19,520
So you ask some random people and you work really with random people, which is more representative

215
00:16:19,520 --> 00:16:20,900
of the population in general.

216
00:16:20,900 --> 00:16:25,310
And the second one would be the non probability sampling method, which is not.

217
00:16:25,390 --> 00:16:29,750
Necessarily worst or which is not necessarily.

218
00:16:29,920 --> 00:16:34,870
Well, it's more efficient and easier to do because you will ask, for example, friends.

219
00:16:34,870 --> 00:16:40,260
You ask people who you know, once again, it's not necessarily representative of the whole population.

220
00:16:40,270 --> 00:16:40,840
It could work.

221
00:16:40,840 --> 00:16:41,940
It could bring you numbers.

222
00:16:42,190 --> 00:16:44,590
Once again, those numbers are not necessarily exact.

223
00:16:45,470 --> 00:16:46,750
So that's it for this part of the course.

224
00:16:46,790 --> 00:16:53,500
Guys, right now, you know the difference between all the sampling methods and all in our next class.