ISSS608 AY2020-21_T2 Visual Analytic DataViz Makeover2

Description: Visualisation and analysis of willingness of the public on Covid-19 vaccination

1. Overview

This post will show how to visualise the result of surveys regarding how people think about Covid-19 vaccination around the world. The original data can be found at Imperial College London YouGov Covid 19 Behaviour Tracker Data Hub Firstly, we will criticize the original charts then show how to make better visualization from the same data set.

• Original chart

Figure0 : Original visualisation

In this GitHub page, below information and data files were used for analysis.

• README.md
• codebook of the core questionnaire codebook.xlsx
• list of countries with data countries.csv
• individual country files with respondent-level data data/country-name.csv

2. Critique of the original charts

2.1. Clarity

• This chart is a visualisation for a question “If a Covid-19 vaccine were made available to me this week, I would definitely get it”. However, the title of this only show the answer of the question. We need to show proper title to clarify the result.
• This chart is sorted by country name in alphabetical order. Users cannot figure out which country is more pro-vaccine.
• Users cannot tell which country is most pro-vaccine or not because the results are visualised using 100% stack bars. It is difficult to show using stack bars, because users should compare all the sizes and values line by line.
• There is no indicator to compare and sort. We need to create a new value to compare among countries properly.
• There is a typo error in the title ‘pro-vacinne’. It need to be changed to ‘pro-vaccine’.
• “% of Total Record”, the title of x-axis should be changed to “Percentage of Total Reocrd”

• As we seen above, this chart also regarding a question “If a Covid-19 vaccine were made available to me this week, I would definitely get it”. However, the title of this only show the answer of the question. We need to show proper title to clarify the result.
• With this chart, we cannot see how many of respondents replied to the survey. Therefore, it is hard to say that percent of people in Sweden is higher than that in Canada, in terms of Confidence interval. We need to show the total number of respondents by country to show the result for clarity.

2.2. Aesthetics

• This chart used 5 different color to show the proportion of each answer. However, the colors are not consistent to show the scale of answers. We need to use the color scheme wisely, grouping the similar sentiment
• All the countries have lowercase letters. We need to change first letter of country name.

• We can see a number of countries in a form of bar chart with white background color, so it is not easy to compare by countries. We need to shade each rows to differentiate results.
• The numbers in x-axis have decimal point. We can eliminate all decimal points which are not needed.

• The title of legend is not appropriate, because users cannot understand properly with ‘Vac 1’.
• The legend for the charts only shows the scale of 1 and 5. To make the legend clearer, we need to show all the proper title of the legend

3. Alternative suggestion of the graph

3.1. Sktechs of graphical presentation and advantages

1) Sketch1 - Dashboard

• After finding several issues in clarity and aesthetic, we can design a new dashboard to show the survey result clearly. Below the left side chart clearly shows that average scores of answers comparing countries. Also the right side chart clearly shows that not only the percentage of respondents “strongly agreed” to a question, but also confidence interval ranges.
• Note that in final dashboard, we will add the total number of respondents by country next to the “% of strongly agreed to vaccination” chart.
• As seen below, we can put question filter upper position so that users can easily see what about the chart trying to show. Also we can locate legend of “Answer” and “Scale of answer” to show users to see easily.

2) Sketch2 - Tooltips

• The dataset contains several important attributes, such as gender, age, answered week, and employment status. Below charts are also important that show how the answers are organised. These charts will be displayed when the you hover over to the previous dashboard graph.
• By showing the detailed answers by different sub-categories, users can understand easily which factors were affected to the results.
• Also, we can add ‘Average score’ through the chart to compare easily by the sub-categories.
• Lastly, we can also include exact numbers of the results, which are hard to show in main dashboard due to space limitation.

3.2. Final dashboard of visualisation using Tableau.

All the charts we designed above were rearranged for one cleaned dashboard

4. Step-by-step preparation of visualization

4.1. Preparation for data sets

Some individual country files are compressed, so you have to decompress them in csv format.

4.2. Upload raw data set in Tableau

After decompressing all files, data sets are needed to be imported to Tableau.

1. In order to union all the 30 individual country csv files, you need to remove pre-set data set(germany.csv). Then, drag and drop “New Union” to the pane. Next, select all the csv files on the next and drag into the dialog box.

2. For the purpose of the analysis, we need to hide unnecessary fields first. Click “Manage Metadata” button then select unnecessary fields. Next, right click and hit “Hide”. Leave required fields only as below.

3. Change name of field of “Table Name” to “Country”. Then click “Aliases” and exclude ‘.csv’. in all Values on the right side.

4. After changing all the name of fields regarding questions, select all the six question fields and right click to pivot the values.

5. Then change pivoted fields’ name to “Question” and “Score”.

6. Click “Create Group..” in “Age” field. Select all the age instances that you want to group and click “Group” button.

7. Go to Sheet1 and drag and drop “Score” into “Filters” pane. Un-check Null values in Score because the values are not necessary to visualise.

8. Click “Analysis” > “View data” menue and click “Extract All” to get necessary data.

4.3. Upload cleaned data set in Tableau and create necessary calculated fields for analysis

1. Open new tableau file and connect to the extracted file created above.

2. Create “Answer” field using “created calculated field” function as below. By using letters not digits, users may not be confused easily.

3. It is more intuitive to express positive responses in high numbers. So, change all the score values in reverse order, using “created calculated field” function

4.4. Create gantt chart to show the survey results

We need to make a chart so that we can easily recognize the results of the survey at a glance. In order to compare the proportion of respondents easily, we can divide axes and compare them at the same time.

1. Create “Negative Score Values” field, using “created calculated field” function. Note that you should use CNT to calculate the whole number of respondents.

2. Create “Total Negative Scores” field using “created calculated field” function. You should click “Default Table Calculation” menu and change “Comute using” from Automatic to Answer.

3. Create “Total Score” field using “created calculated field” function and click “Default Table Calculation” then change Compute using to “Answer”.

4. Create “Gantt Start Point” field using “created calculated field” function.

5. Click “Gantt Start Point” field and click Compute Using to Answer

6. Create “Percentage” field using “created calculated field” function to visualise the size of the gantt chart

7. Create “Gantt Percentage” field using “created calculated field” function and click “Default Table Calculation” then change Compute using to “Answer”.

8. Drag and drop “Gantt Percetage” field into Columns and “Country” into Rows. Also Drag and drop “Answer”.

9. Drag and drop “Question” field into “Filter” pane and change the Marks to “Gantt Bar”. Lastly, Drag and drop “Percentage” field onto “Size” button to show the percentage of respondents.

10. Click “Answer” field and “Sort” menu. Check “Manually” in “Sort by” menu and change the order as below

11. Create “Avg Score” field using “created calculated field” function. Note that you should use “CNT” field to calculate sum of respondents.

12. Change the Gantt chart color

13. Chage Marks to “Circle” and drag and drop “Avg Score” onto “Label” button. Then click and change “Compute using” to cell

14. Click lable and change Font format and alignment as below

15. Click “Avg Score” in Columns and check “Dual Axis”

16. Right click above axis and click “Edit axis”. Check “Fixed” range for changing the range of axis from 1 to 5

17. After changing Size of Circle chart and color to “White”

19. Add reference to show “0%” more clearly

20. Add reference to show “Average line”. Type label “ Score = ”

18. The final Gantt chart!

4.5. Create Confidence Interval chart

The number of respondents are different among countries. Therefore, we cannot confidently say that the average scores of answers we have created above reflect correctly the opinion of whole country. For this reason, we can show the “confidence interval range” to give the better information to users. Based on the formula, when the number of sample increases, the range of confidence interval shorten.

1. First of all, we need to create 7 calculated field to visualise confidence level. The formula for the fields are below.

1.. (CI) No of Strongly agree

IF [Answer]=“Strongly agree” THEN [CNT] ELSE 0 END

2.. (CI) % of Strongly agree

SUM([(CI) No of Strongly agree])/SUM([CNT])

3.. (CI) Z_95%

1.959964

4.. (CI) Prop_SE

SQRT(([(CI) % of Strongly agree]*(1-[(CI) % of Strongly agree]))/SUM([CNT]))

5.. (CI) Prop_margin of Error 95%

[(CI) Z_95%]*[(CI) Prop_SE]

6.. (CI) Prop_Upper Limit 95%

[(CI) % of Strongly agree] + [(CI) Prop_margin of Error 95%]

7.. (CI) Prop_Lower Limit 95%

[(CI) % of Strongly agree] - [(CI) Prop_margin of Error 95%]

2. Drag and drop “(CI) % of Strongly agree” field into Columns and “Country” into Rows

3. Drag and Drop “(CI) Prop_Lower Limit 95%” and “(CI) Prop_Upper Limit 95%” fields onto the column axis.

4. Change Measure Values graph to “Line” and click left button and change to “Path”

5. Right click on the upper axis and check “Synchronize Axis”

6. Click “Contry” in Rows and check “Sort” to sort Descending order by “(CI) % of Strongly agree”.

7. Add reference line to show average of “(CI) % of Strongly agree” over the countries.

8. Add tooltip to show the detail figures

9. Create a new worksheet for showing total number of respondents by country. Drag and drop “Country” field into Rows and “CNT” onto “Label” button. Also change the color shading to make clearer.

10. Click “Country” field to sort by “(CI) % of Strongly agree”

4.6. Create tooltip charts for showing detail of answers

1. Changes of answers through weeks

• Drag and drop “Qweek” into Columns and “Avg Score” into Rows. In order to show the figures, put “Avg Score” onto “Color” and “Label” button in Marks
• Change chart format to “Line”
• Add reference line to show the average value of “Avg Score” through week

2. Answers by Gender

• Drag and drop “Gender” into Columns and “Avg Score” into Rows. In order to show the figures, put “Avg Score” onto “Color” and “Label” button in Marks.
• Change chart format to “Bar”
• Add reference line to show the average value of “Avg Score” through gender

3. Answers by Age

• Drag and drop “Age (Group)” into Columns and “Avg Score” into Rows. In order to show the figures, put “Avg Score” onto “Color” and “Label” button in Marks.
• Change chart format to “Bar”
• Add reference line to show the average value of “Avg Score” through age group

4. Answers by Employ Status

• Drag and drop “Employ Status” into Columns and “Avg Score” into Rows. In order to show the figures, put “Avg Score” onto “Color” and “Label” button in Marks.
• Change chart format to “Bar”
• Add reference line to show the average value of “Avg Score” through Employ Status

5. Answers by Household Children

• Drag and drop “Household Children” into Columns and “Avg Score” into Rows. In order to show the figures, put “Avg Score” onto “Color” and “Label” button in Marks.
• Change chart format to “Bar”
• Add reference line to show the average value of “Avg Score” through Household Children

6. Detailed figures for Country_gantt chart

• Duplicate “Country_gantt chart” worksheet
• Remove “AGG(Avg Score)” field in Columns
• Drag and drop “Percentage” field on “Label” button

7. show all tooltip charts when hover over to the “Country_gantt chart”

• Go back to “Country_gantt chart” worksheet we made ealier, and click Tooltip button in “Marks” > “All”
• In order to show all the detailed charts, type as below

4.7. Create Dashboard

1. Click Dashboard icon and change Size to “Automatic”

2. Drag and drop “A Text” object to the pane

3. Drag and drop “CI_Chart” and “Country_gantt” on the pane side by side

4. Move “Select question” filter and “Answer” legend above the two charts

5. Click triangle and check “Single Value (dropdown)”

6. Drag and drop “A Text” object below the “Answer” legend. then type below

7. Adjust sizes of boxes we put

8. Drag and drop more “A Text” objects between “Answer” label box and chart. Then type titles repectively.

9. Drag and drop “CI_chart_respondents” next to “CI_chart”.

10. The final Dashboard

5. Major observations

5.1. People in EU are more likely to get vaccination comapred to that in Asia

• If we see below dashboard we made earlier, average scores of the survey answer to “Q1. if a Covid-19 vaccine were made available to me this week, I would definitely get it” are varied by country. If you see the charts closely, people in EU countries are more likely to get vaccine, almost above average socre value of 3.39
• Especially, the average scores of answer in United Kingdom is higher than that in Japan by 0.9.
• Also, UK and Denmark people are more likely think that they strongly want to get vaccine if they are available.

5.2. The older, the more agree to vaccination

• As seen below, seniors (age group of 60-69) are more likely to get vaccine than relatively young group (age group of 30-39).
• Also, this result can be seen if we see the chart of “Answers by Employ status”. The retired are likely to get vaccine than other groups.

“Q1. if a Covid-19 vaccine were made available to me this week, I would definitely get it:”

5.3. As times go by, people think more positively toward vaccine and government

• As seen below, worries about potential side effects of vaccine are getting smaller.
• Also, people believe their government health authorities more than before.

Q3. I am worried about potential side effects of a COVID19 vaccine

Q4. I believe government health authorities in my country will provide me with an effective COVID19 vaccine

The END

---