WEBVTT 00:00.560 --> 00:10.170 Welcome to the data hall youtube channel. In this video, we are going to discuss how do we analyze the 00:10.170 --> 00:19.380 categorical latter. So specifically, we are going to see how do we generate descriptive statistics and the graphs 00:19.380 --> 00:25.530 for categorical variables. So normally when we are working with categorical variables 00:25.530 --> 00:30.490 we are mainly interested in in the frequencies 00:31.310 --> 00:40.390 in percentages. Um the count of the data, the account of the different categories or were interested in the bar charts or the 00:40.390 --> 00:49.490 pie charts. So this is what we are going to discuss in this video and see how do we generate these statistics 00:49.490 --> 00:59.310 these descriptive statistics and graphs are using stata. So mainly if you click on statistics steps before 00:59.310 --> 01:07.190 that, let me show you the data. This is the question of data that we had been using with our previous videos where we work 01:07.190 --> 01:11.760 with the questionnaires. You would remember we have serial number, occupation, age 01:11.760 --> 01:17.020 gender and a couple of questions. If I assure you it in the data view 01:17.020 --> 01:27.190 you would see we have already labeled them and we have the the the the values are 01:27.190 --> 01:36.900 labeled values for each, for each question. Okay. So mainly we would be working with this summary 01:36.900 --> 01:44.100 statistics. We click on statistics, summary statistics. Then within summary statistics, we have this frequency tables 01:44.100 --> 01:53.560 Remember when we are working with categorical variables, we the statistics that we are interested in in frequency 01:53.560 --> 01:59.360 distributions or frequencies. Um So, so we have a couple of options 01:59.360 --> 02:04.070 We have five options over here and out of these five options 02:04.070 --> 02:08.480 we would, we would cover the first four options in this video 02:08.480 --> 02:15.130 And, but maybe later on, we would, we would see how this stable calculator would work 02:15.130 --> 02:18.950 But for now in this video, we are going to cover these four 02:18.950 --> 02:26.500 Um So when we work with, with categorical variables, we have to generate tabulations 02:26.500 --> 02:30.610 right? So we have a one way tabulation or to vote tabulations 02:31.610 --> 02:38.510 So we either generate one way tabulation in one way tabulation. We generate a category 02:38.510 --> 02:47.500 a frequency distribution of a single variable or X, right? In multiple or two week tabulation 02:47.500 --> 02:54.940 we generate a tabulation for different different variables. So let's start with the one way tabulation 02:55.550 --> 02:59.750 the simple command S tablet and then we write the variable name 02:59.750 --> 03:06.530 So if you tablet occupation, we would know that we have three categories within occupation 03:06.770 --> 03:16.030 Um and 24% of the the observations that we have our solid 03:16.030 --> 03:25.500 person and 40 to our students, we also have the cumulative this uh the cumulative percentages and 03:25.500 --> 03:29.440 the frequencies we know there are 50 observations in our data set 03:29.440 --> 03:34.280 So this tells us that within the occupation variable, we do not have any missing value 03:34.580 --> 03:39.870 Similarly, we can tablet the gender variable. So again, this is the one way to abolition 03:39.870 --> 03:45.830 We are only generating a table based on the categories of just one variable 03:45.830 --> 03:55.170 Uh or we can tablet through a single question. So we have different categories 03:55.170 --> 03:58.690 or different, you know, it is a liquid scale for question one 03:58.690 --> 04:06.210 either a person would strongly disagree, disagree or I would strongly agree 04:06.210 --> 04:14.420 So these are five different categories in this. But if you wanted to generate a tabulation for all these variables 04:14.420 --> 04:21.690 for multiple variables, it would be, you know, we would have to write 33 commands 04:21.690 --> 04:27.370 But let's just say we wanted to generate one way tabulation, you know 04:27.370 --> 04:32.800 all these these single tables that we just generated but using a single command 04:32.800 --> 04:38.510 then we would use tab one. So we write tab one, then all the variables that we want 04:38.510 --> 04:47.960 So I have also added Q four, we just hit run and it has generated the one way a 04:47.960 --> 04:55.600 tabulation, right? For, for each of the, of the variable. So in essence 04:55.600 --> 05:02.140 what it is doing, it is running the same tabulation command four times for each variable 05:02.140 --> 05:09.070 So if I wanted to do this individually, then I would have to write for tablet commands 05:10.280 --> 05:14.550 The way it can be used from the menus is weekly con statistics 05:14.610 --> 05:20.810 then assembly statistics, frequency tables. And if you wanted to generate a one way tabulation 05:20.810 --> 05:30.190 You would click on this option and give the name of the variable that we had to submit and it would generate 05:30.190 --> 05:36.020 the one word abolition. This was a simple, simple, simple tablet command 05:36.020 --> 05:39.210 If you wanted to have 05:40.200 --> 05:46.090 multiple one way tabulations, the tab one command that we just saw 05:46.090 --> 05:50.610 this is the option that we can use. We would select all the categories 05:50.610 --> 05:57.180 sorry, all the variables that we want it okay. And it would generate the tables that we want 05:57.240 --> 05:58.100 Okay 06:00.000 --> 06:06.420 From this analysis, from the stable, we would notice that all the variables have 50 observations 06:06.420 --> 06:13.490 But instead this Q4 would have 40 observations. We would want to know how many missing values are there 06:13.490 --> 06:18.380 And we want to adjust these percentages according to those missing values 06:18.380 --> 06:22.720 So currently it is generating these percentages with respect to the valid responses 06:22.720 --> 06:28.480 See we have these valid responses, but we also want But instead of this 06:28.480 --> 06:35.330 we want to have percentages that would also, you know, account for the missing values 06:35.330 --> 06:42.730 So we use this tab one instead of having all these categories 06:42.730 --> 06:49.890 So all these variables we just want are interested in Q4. We would have probably 10 tablet 06:49.890 --> 06:57.470 but it's the same thing tab one for only one variable is essentially tablet 06:58.280 --> 07:04.640 command. We introduced the missing option. So if we execute this command 07:04.640 --> 07:14.330 we would see that it is somewhat similar to this one. But now we have a role for the missing observations and these percentages are 07:14.330 --> 07:22.640 adjusted with respect to, you know, now, with respect to this total number of observations rather than the violent responses 07:24.180 --> 07:29.130 Now, we have started this discussion of wild responses and you know 07:29.130 --> 07:38.420 percentages with respect to the total number of observations. It there is a user written command which is called F 07:38.420 --> 07:45.790 R E. If you haven't installed, you can simply write SSC installed FRE because I have already installed it 07:45.790 --> 07:50.940 So I won't execute this command. That's why I have commented it out 07:51.680 --> 07:57.130 So what this free would do, it was provide the percentages of violent responses 07:57.130 --> 08:03.480 So in addition to this table, it would give us additional statistics 08:03.480 --> 08:08.250 So let's execute this because Q one and Q four have missing observations 08:08.250 --> 08:13.910 So I'm just executing this command for the Q one and Q four variable 08:15.070 --> 08:19.950 Now we have this for the Q one variable and this for the Q four variable 08:20.520 --> 08:26.200 The difference that you would see is all the 41 we didn't have missing values 08:26.200 --> 08:31.950 This table is essentially similar to the table that we had with the tablet command 08:31.950 --> 08:36.880 But this would be different here. We would have the missing observations 08:36.880 --> 08:46.410 We would have percentages, you know, accounting for the missing observations of percentage with respect to this 50 same that we generated 08:46.410 --> 08:49.860 you know, using the missing option 08:50.880 --> 08:55.140 And then we would have these percentages with respect to the valid responses 08:55.140 --> 09:01.310 The table for Q for that we generated without using the missing missing option 09:01.310 --> 09:10.570 So you have got, you know, the both the values by just executing the one command 09:10.570 --> 09:20.280 The one more thing that I like about this, this free command is that uh this tablet command will just give you the 09:20.280 --> 09:24.360 the value labels. It wouldn't give you the pneumatic values, right 09:25.950 --> 09:33.510 But with the free command, it would give you both right and sorted based on the numbers 09:34.060 --> 09:42.280 So it is kind of more, you know, easy to analyze, easy to visualize 09:42.280 --> 09:48.210 Um probably we would want to have this kind of, you know 09:48.210 --> 09:53.520 presentation with our graph. But let's just say if you wanted to use 09:54.510 --> 10:03.680 the same tablet command, but now you want to have these value labels and the values by 10:03.680 --> 10:09.730 both of these two. But with the tablet command, you do not want to use the free command 10:09.730 --> 10:14.040 you want to use the tablet command and you want both of these 10:14.040 --> 10:20.370 So if you use the tablet command without this no label option 10:20.490 --> 10:29.970 it generates the default is to have value labels. But if we use the no label 10:29.970 --> 10:37.030 the default, the the the the the tablet command would generate only the values 10:37.890 --> 10:43.200 we want both of them. So the command for this would be numb label all 10:43.200 --> 10:46.080 So it would add numbers to the labels 10:47.080 --> 10:48.790 So let's execute this command 10:50.120 --> 10:52.370 Now let's establish gender 10:53.610 --> 11:01.090 You will see that it is somewhat similar to the representation that we have with the free command 11:01.090 --> 11:10.320 we can do tablet one, right? And we would have the table for gender as 11:10.320 --> 11:14.780 well as for question number four. If you wanted to remove the value labels 11:14.780 --> 11:22.850 sorry, the value numbers because they are actually added in the data view 11:22.850 --> 11:29.600 right? So, so the value labels are actually changed. So what this command had done 11:29.600 --> 11:36.040 it had added the value um as a prefix of the value label 11:37.670 --> 11:47.120 If we wanted to go back to the original data, then we would simply use num label and would use the option remove and 11:47.900 --> 11:50.400 our data would look like the way it was 11:53.500 --> 11:58.660 okay. So that was just the one way tabulation. Uh 11:59.460 --> 12:07.760 If you wanted to see how many, you know, businessmen's are there that are male 12:08.020 --> 12:17.110 So we want categorization based on two different variables. So you want frequency based on two different variables for 12:17.110 --> 12:24.130 that, we would use the same tablet command, but now we would add two variables 12:24.130 --> 12:29.440 So if we saw previously with tablet command, we would just have one variable 12:29.440 --> 12:34.620 This would generate one way tabulation. But if we wanted to have to a tabulation 12:34.620 --> 12:40.250 then we would add multiple variables we can keep on adding 12:42.310 --> 12:48.620 Okay. So now this would give us the idea of how many salaried male 12:48.620 --> 12:57.710 how many male students do we have in our data? How many total number of obviously male and female 12:57.710 --> 12:58.870 students do we have 13:01.720 --> 13:06.840 There is one more command which is tapped to and what the stab to will do 13:06.840 --> 13:13.200 it will generate all two way tables, all possible paired to wait tables 13:13.870 --> 13:22.150 So we have four variables over here. So it would generate a two way table for occupation and gender 13:22.150 --> 13:28.190 then it would generate a table for occupation and Q one, then for occupation in Q four 13:28.260 --> 13:33.150 then for gender and Q one gender and Q four. And lastly for Q one and Q four 13:34.900 --> 13:42.010 So it generated the table as uh I have explained for occupation 13:42.010 --> 13:51.050 gender occupation, Q one and so on and so forth. So all the possible to wait tables that 13:51.050 --> 14:00.280 can be generated had been generated before moving towards the graphs. Let me again show you these commands from 14:00.280 --> 14:06.220 the, from the menus. We have discussed this one way table and multiple one way tabulation 14:06.220 --> 14:10.780 If you want to wait tables, you can use this command 14:12.100 --> 14:21.400 And if you want all the possible two tables, the last command tap to that we just executed and 14:21.400 --> 14:23.290 this is the option that you want to use 14:24.140 --> 14:30.140 Okay. Let's see. How do we generate graphs for categorical variables 14:30.950 --> 14:38.460 Normally we would use either pie charts or bar chart to present categorical variables 14:38.460 --> 14:47.730 If we had continuous variable, we would probably use line chart. But for categorical pie chart and bar chart is the way to go 14:48.680 --> 14:54.010 I will first show you from the command menu. We click on graph 14:54.010 --> 14:57.050 then we click on pie chart 14:58.470 --> 15:07.420 And for now, I want to use the pie chart for the queue for variable. So I would 15:08.270 --> 15:13.890 if I have different categories within a single variable, then I would use this option 15:13.890 --> 15:20.590 If we have categories, um you know, being converted into different variables 15:20.590 --> 15:28.300 we have a data into wide format. I would probably use this this option But I have my data in long format 15:28.300 --> 15:37.030 I have categories within a variable in single variable. So I would select the variable Q4 and I click submit 15:39.040 --> 15:44.260 Okay. It would generate a pie shot, nice looking pie shot 15:46.030 --> 15:54.160 Uh And but what I want is I want the smallest, you know, category 15:55.020 --> 15:59.950 being over here and then it should move towards the largest category 15:59.950 --> 16:03.740 So I wanted to be somewhat sorted 16:05.620 --> 16:14.830 The option that I would use is the size order. It would sort the variables based on sort the 16:14.830 --> 16:20.940 the the the pies in this chart based on its size 16:23.540 --> 16:27.700 So now we see the smallest buyers over here and the largest one is over here 16:27.700 --> 16:33.730 So the largest is the agreed and the smallest is stronger, disagree for the question number four 16:33.730 --> 16:35.260 whatever the questionable four is 16:36.230 --> 16:45.560 probably I would want to exclude the missing observation. I would use this option And it would give me the by chart excluding the 16:45.560 --> 16:53.770 missing observation, the sort option you have just seen. And we for to remove the missing observation 16:53.770 --> 16:55.810 we will use C4 16:56.590 --> 17:00.780 But probably graphs can be easily, you know 17:01.720 --> 17:07.990 created using the menus rather than from, from command 17:10.310 --> 17:13.260 Next in date, a bar chart 17:18.150 --> 17:26.230 We would have a bar chart for frequencies. So we'll use this option and the categorical variable 17:26.230 --> 17:30.500 let's use Q1 this time, let's generate this graph 17:32.620 --> 17:33.690 And 17:35.110 --> 17:40.460 it's a nice looking graph. But we want to play with some 17:40.460 --> 17:47.600 you know, we can, instead of having these value labels as horizontal 17:47.600 --> 17:53.620 we can probably have them vertical. It's, you know, it's a bad idea in this case 17:53.620 --> 17:59.190 but sometimes you might need it, we can change the label size 17:59.190 --> 18:06.480 probably would have small label sizes and we can play with other options over here 18:06.480 --> 18:15.690 But the idea of this video is to give you some knowledge of how do we work with the the categorical variables 18:15.690 --> 18:17.140 Thanks for watching the video