The original visualization presented as Chart 6 on Page 22 of MOM’s Report: Labour Force in Singapore 2019 is reproduced below
Figure 1: Original Visualization by MOM, Singapore
Section A: Critique of Original Viz
This section provides a critique of the original visualization and comments on both clarity and aesthetic aspects that could be further improved on. A total of 8 clarity issues and 5 aesthetic issues (as shown below) were identified.
Figure 2: Critique on Original Visualization
Clarity
| S/N | Issue | Comments |
|---|---|---|
| 1 | Title | The purpose and objective of the original visualization was not conveyed well in the title of the chart. It misleads people into thinking that the graph shows absolute number of labour force by age when it is in fact showing the share of each age group within the work force. |
| 2 | Subtitle | Subtitle that says “Per Cent” does not elaborate on what this per cent is referring to, nor explain which figure is the percentage derived from what base. Together with the title, it confuses viewers on the data being presented. |
| 3 | Axis Label | Missing explanation for the “June 2009” and “June 2019” row data – without reading the paragraph before, there is no easy way to figure out that these numbers refer to share of workers in each age group within the labour force. |
| 4 | Y-Axis | The graphs are literally floating with no Y-axis and its corresponding values (much less any gridlines) to anchor the actual value corresponding to each age group – this is as good as drawing random lines and is a critical flaw of the original viz. There is no easy way to know if the lines are corresponding to the correct Y-value even if we are provided with the data. |
| 5 | Choice of Viz Type | Since the data is segregated by discrete age bins, line graph – which are more suitable for continuous x-variables, should be avoided to prevent giving false impression that the data is continuous within the bin (e.g. a kink for the 45-49 age, it leaves room for misinterpretation that a downtrend from 45 years old changed to an uptrend at 47 years old). This is especially confusing as there are no softgrid guiding where the bins cut-off |
| 6 | Comparison Across Year | Since the objective of the visualization is to explain that there are more older workers in the labour force, the choice of two line graphs is inefficient as it requires viewers to do a mental comparison of the higher line to derive the changes over time. Bearing the objective in mind, the difference between the two years could have been calculated and directly presented instead. |
| 7 | Lead-in Paragraph | Lead-in paragraph is not supported by the visualization, in fact, it creates more confusion. While the age bins are split into 5-year bands (e.g. 15-19, with the only exception being those above 70), the paragraphs discuss the share of labour force by bigger bins, lumping together those aged 55 and over, as well as the 25-54 groups. This creates confusion as the figures quotes – “from 16% in 2009 to 25% in 2019” and “declined from 75% to 67%” - are all not interpretable from the visualization. The need for mental sums to add up the percentages across a few data points presented in the table is unintuitive and user-unfriendly. In terms of comparison between the years, author could have also enhance the readability by deriving the change in percentage points across the two years. |
| 8 | Choice of Data | As a general comment to the overall choice of data for the discussion, it might be clearer to talk about the change in Labour Force Participation Rate (LFPR) for each age group, rather than the composition of labour force – which could be more indicative of demographic change (e.g. ageing population) than shift in labour decisions. |
Aesthetic
| S/N | Issue | Comments |
|---|---|---|
| 9 | Definition | Visualization appears grainy and pixelated. |
| 10 | Data Points and Labels | No softgrid nor data point to guide visual on which year corresponds to which point |
| 11 | Font Consistency | Inconsistent font colour for same data source – blue for the line is different from the blue words on chart, which is again different for the blue in the table below |
| 12 | Choice of Font Colours | Choice of font colour does not contrast with background as well, hence not easy to read |
| 13 | Alignment | Information (Source and Note) at the bottom not aligned |
Section B: Suggested Improvement
This section provides some suggested improvements that could be implemented to resolve issues discussed in Section A.
Clarity
| S/N | Issues | Suggested Improvements |
|---|---|---|
| 1 | Title | The title of the visualization could be modified to reflect the key message that the viz is trying to convey, and should correspond accurately to the main features of the viz |
| 2 | Subtitle | Subtitle could be modified to provide key information to guide viewers in deciphering the viz (e.g. data and corresponding unit presented in the viz, year of data, location, and X-axis variable) |
| 3 | Axis Label | Provide clear axis labels with accurate units of measure to prevent confusing readers |
| 4 | Y-Axis | Inclusion of a Y-axis with suitable tick marks and guiding soft gridlines to enhance readability is important |
| 5 | Choice of Viz Type | For discrete variables (as in the bins of age range used), a bar chart would serve to provide better clarity to prevent misinterpretation |
| 6 | Comparison Across Year | To achieve the objective to explain that there are proportionately more older workers in the labour force in 2019 than 2009, the difference between the two years could have been calculated and directly presented for a more effective visualization. |
| 7 | Lead-in Paragraph | Lead-in paragraph should be supported by the viz, and any figures quoted by the paragraph should be available and obvious from the viz. Better interpretation and processing of the data (e.g. calculating and presenting the difference, if that is the objective) would also enhance clarity of the viz. |
| 8 | Choice of Data | Using data on the Labour Force Participation Rate (LFPR) for each age group could provide clearer comparison to achieve the objective of the viz |
Aesthetic
| S/N | Issues | Suggested Improvements |
|---|---|---|
| 9 | Definition | Ensure visualization is not pixelated and has suitable definition |
| 10 | Data Points and Labels | Provide clear background gridline to enhance readability and data points/labels for viewer’s ease of reference |
| 11 | Font Consistency | Ensure consistent font type and colour are used to represent distinct sets of data for consistency. Useful colors will also provide additional dimensions to enhance effectiveness of viz |
| 12 | Choice of Font Colours | To select contrasting colours for ease of reading |
| 13 | Alignment | As much as possible, to provide proper alignment within viz to reduce distraction caused by messiness |
Makeover Concept
Taking in the points discussed above, a redesigned visualization could look like the following:
Figure 3: Makeover Concept
Section C: Redesigned Visualization
Using Tableau, the redesigned visualization based on the discussion points and concept presented in Section B above is created as follows:-
The redesigned visualization can also be accessed via this link
Figure 4: Redeisgned Visualization
Section D: Step-by-Step for Viz Makeover
This section provides a step-by-step guide to recreate the redesgined visualization.
Step 1 - Download and Import Data 
First, download the Resident Labour Force Participation Rate by Age and Sex, 2009 - 2019 (Table 5) file via this link. Then, open Tableau Desktop to connect with the Excel file downloaded.
Step 2 - Data Preparation 
Check on the “Cleaned with Data Interpreter” box for Tableau to assist with data cleaning (if desire, may click on underlined “Review the results” to see data tables created). Select the “T5_T A5:Y18” to extract relevant data. Hide the first column and switch Connection on the top right to “Extract” (as we will be saving onto Tableau Public, which does not support Live connections). Save the Data Extract.
Step 3 - Data Processing 
Once the data are loaded, right-click on blank space under Data tab to select “Create Calculated Field”. Create a new field named “LFPR Change (%-pt)” with the formula “[2019]-[2009]” to create a field that tells us the change in percetage point of the LFPR for each age group from 2009 to 2019.
Step 4 - Simple Viz Set-up 
To set up the simple visualization, drag-and-drop the Age (Year) variable into the Column field, and the (aggregated) LFPR Change (%-pt) that we have just created into the Rows field, as well as Label. and Color marks. A simple visualization resembling the final viz is created.
Step 5a - Editing Viz Color 
As commented in Section B above, choosing appropriate color is important for a clear visualization - in this case, since we know there are positive and negative changes, we would want red to denote negative, and blue for positive to highlight the contrast. Double click on the Color legend tab, choose the Red-Blue Diverging palette, and check on the “Use Full Color Range” box.
Step 5b - Adding Informative Title and Subtitle 
Similarly, informative title that informs viewer of the viz’s main message, as well as accurate subtitle to provide additional information about the data is critical for a good Viz. Double click on the title to edit as shown above (selecting font size 22 for title and size 12 for subtitle, while keeping the default font type; color of the title may also be changed to provide contrast)
Step 5c - Hiding Extra Elements 
We then hide the field labels for columns by right-clicking on the label and selecting the hide option, and also hide the color bar card by clicking on the triangle at the top-right-hand corner of the card and selecting “Hide Card” (as shown).
Step 5d - Formatting Labels 
Next, we right-click on the blank space on the viz and select “Format” to trigger the Format pane - on the Fields dropdown bar, select “SUM(LFPR Change (%-pt))”. Under Default, click on drop-down for Numbers, select “Number (Custom)” and change decimal places to “1” - this will change the percentage point changes to 1 decimal place. Under the same Default category, change the font size to 10, Similarly, switch the tab from “Pane” to “Axis” on the Format panel, and switch the default font to size 10 and Bold the font.
Step 5e - Alias for X-Axis 
We may notice that the X-axis label has large spacing between the two fringe of age range - return to the Data Source Sheet, right-click on the age column, select “Edit Aliases” and proceed to remove spacing for all age bins. For the last bin, change to “70&Older”.
Step 5f - Remove Total 
Lastly, since Total does not provide meaningful information here, we right-click on it and select Exclude to exclude this data.
Step 6 - Basic Viz 
That completes the set-up of the basic visualization. Remember to give the Worksheet a name and save the Tableau Workbook. Next, we will create a table that shows supporting data on LFRP for each age group in 2009 and 2019. This will provide more information for the visualization that will be useful when making observations.
Step 7a - Supporting Tables Set-Up 
Create a new Tableau Worksheet. Drag-and-drop Age (Years) into the Columns field, aggregated 2009 and 2019 data points into the Text Mark (as shown above), click on “Show Me” on the top-right hand corner of the viz and select the Data Table Viz. Next, click on the “Swap Rows and Columns” button on the panel above (or press Ctrl+W)
Step 7b - Supporting Tables 
As a final step for the supporting data table, exclude the Total column as shown above.
Step 8a - Dashboard Setup 
To incorporate various layout elements, create a new Dashboard on Tableau. Drag-and-drop the first Worksheet and then stack the Table worksheet below. Hide the Field Labels for Columns and the Axis label for the table by right-clicking on them individually and selecting the respective hide options. On the Dashboard panel on the left, resize the Dashboard to fit automatically be selecting the “Automatic” option under Size dropdown menu.
Step 8b - Lead-In Paragraph 
Add a Tiled Text object (from the bottom left corner as shown above) by drag-and-dropping a text object above the main Viz worksheet (i.e. top of the dashboard viz). Include the write-up as shown to explains the visualization and enhance the readability by highlighting some key/obvious findings corresponding to the viz.
Step 8c - Axis Label for Supporting Table 
A good visualization should be clearly labeled - add in another text object to the left of the supporting data table at the bottom to inform viewer of the data being presented in the table, along with the unit of measure. Resize the text object and format it accordingly as shown.
Step 8d - Appropriate Spacing 
Appropriate spacing to indent different section of a viz enhances its overall readability - include Blank objects at appropriate space and readjust elements to create well-aligned viz.
Step 9a - Drawing Viewer’s Attention 
To accentuate the key message found in the lead-in paragraph, it might even help to conscientiously draw viewer’s attention to a particular segment of the viz. To achieve this, we first create a floating Vertical object and place it above the bars showing where the Senior age groups are.
Step 9b - Define Terms 
Making use of the space in the viz, we enhance the clarity further by providing the definition of “Senior”. In this case, we anchor a Tiled Text object with the definition “Senior (Aged 60 & Older)” (which is in alignment with the Singapore’s Legal definition) onto the floating Vertical object by drag-and-dropping a Text object onto the floating object while holding down the Shift key.
Step 9c - Cosmetic Changes 
We then select the floating vertical object and under the triangle on the top right corner, select “Distribute Contents Evenly” and format the layout of the object as shown above (select colour, and change opacity to 5%)
Step 9d - Cosmetic Changes II 
We further add on another line to create a visual divide for “Seniors vs the Rest”. This is done by creating another vertical object but with a width of 1px - formatting is as shown above.
Step 10 - Final Touch 
And finally, as we have floating objects which may be displaced when size of visualization changes, we revert the size of the visualization to a custom size fitting the resolution when the viz is created - in this case, 1,575px x 764px. This will also ensure the layout formatting will not have too drastic a change when the viz is uploaded onto Tableau Public.
Step 11 - Upload 
And for the very last step, we save the visualization onto Tableau Public.
Final Product available on Tableau Public 
Section E: Observations
The first key observation based on the redesigned visualization is that the three Senior age groups has the largest increase in Labour Force Participation Rate (LFPR) from 2009 to 2019 (Note: according to senior-citizen-singapore.com, the Senior Citizen Act of Singapore defines a “Senior” as someone who has attained 60 years or higher). This is immediately brought to focus both by the segmented region with shaded background and captioned as “Seniors (Aged 60 & Older)”, as well as the title, which directly informs the reader of this most obvious and key takeaway from the visualization. The largest single age-group increase in LFPR belongs to those aged 65-69 year-old - increasing by 16.2 percentage point, from 2009’s 29.9% participation rate to 2019’s 46.1% participation rate. One of the reasons for this could be the increase in the legal re-employment age from 65 to 67 from 1 July 2017. This would mean that employers must offer re-employment to eligible employees up to age of 67 instead of 65 - that is, more elderly would be protected by law to be able to continue working past 65 years old if eligible. The next two age-groups with highest increase in LFPR are the 60-64 year-old and those 70 & older - as Singapore’s average life expectancy increased from 81.4 years in 2009 to 83.6 years in 2019, it will be a new norm for more seniors to be at work for an ageing Singapore. The reasons are twofold: (i) active working age will need to be lengthen to save up more for longer retirement years; and (ii) older workers may choose to stay in the workforce longer to keep them physically and mentally active, keeping demantia and loneliness at bay. All in all, while the change in Senior’s LFPR is obvious from the visualization, its reason is manyfolds.
The second key observation from the chart is the other elephant in the room: the only age group that saw a decline in its LFPR - those aged 25-29 year-old. One plausible reason for this could be an increase in average schooling years of residents aged 25 and over, from 2009’s 9.7 years to 11,2 years in 2019. This is further supported by the increase in proportion of residents who had attained University degree as their highest qualification - from 22.5% in 2009 to 32.4% in 2019. Both these statistics point to the fact that people are increasingly spending more time schooling than working, and hence a reduction in the LFPR.
Lastly, let’s discuss the flat plateau for those aged between 30 to 59 which saw the LFPR within the age group increase modestly, ranging from 0.8 percentage point to 6.6 percentage point. From the main viz, it told the factual story that the LFPR for this group of residents did not increase much for the 10 year from 2009 to 2019. However, this is where it is crucial to drill deeper into details and understand the importance of the supporting data table included in the visualization - the actual LFPR (%) for each age group. Based on the table, it is easy to realise that the participation rate for this main bunch of residents were already very high in 2009 to begin with (from 68.4% to 89.3%) - that is, majority of the subpopulation within these age groups were already actively participating in the labour force (i.e. either employed or actively looking for job). As there are bound to be discouraged workers and some form of frictional unemployment that triggers individual to leave the labour force, it is much more difficult to increase the LFPR for age-groups where the rate was already high, especially those nearer to the 100%. In summary, it is not unexpected to see low LFPR percentage change for these age groups with higher LFPR and there is no cause for worry. It would, however, be interesting to see how this figure has changed for the year of 2020 when the unanticipated adverse event in the form of global pandemic hit.
As a rejoiner, it is thus important to provide clear and informative visualization that are aesthetically pleasing - such observations mentioned in this section would have been difficult to arrive at with the original visualization that lacked clarity in the information presented and key message it is trying to drive at.