Even if you are not a big fan of MOOCs, you probably have heard of the Harvard Happiness Course. As the most popular course in the history of Harvard University¹, “Positive Psychology” (what the course is actually named) teaches “how to get happy”.
It feels like the course became a hit overnight², like people suddenly realized that they were not happy enough yet, if at all, like me. We have all the reasons to be unhappy, but what allows us to claim that we are, actually and already, happy?
“Happiness is an allegory, unhappiness a story” — Leo Tolstoy
If it sounds to you that this article is probing into all aspects of this broad topic, it is not. It is zooming in on only one commonly-asked but never-fully-answered question, “is it wealth that makes you happier”. A similar question is asked and visualized as a section in the following article.
Among the N reasons that make you happy, wealth (usually in the form of money) is almost inevitable. For the majority of people, we spend almost 1/3 of our adult lives doing a job³ (and another 1/3 sleeping). We may like or dislike it, but that it pays (to meet our minimum expectations) is usually the prerequisite for us to take it.
What’s attempted in this article, is to provide further insights into the wealth-happiness relationship, by visualizing both what has been and not directly revealed in one of the visualizations in the above-mentioned article, using data provided by MakeoverMonday through data.world as below.
Link Between Happiness and Income
A quick preview of the wealth-happiness relationship revealed in the referenced visualization is, 1) countries or regions with higher income levels tended to have higher average happiness levels (static); 2) economic growth, in most countries or regions, came together with increasing happiness levels (animated).
Let’s take a closer look at this visualization to see how it says what is said.
Level of granularity: individual country or region
Dimensions:
- GDP per capita — X axis: how “rich” a country or region is
- Life satification — Y axis: happiness level of a country or region
- Population — Size Encoding: population size of a country or region
- Continent Each Country or Region Belong to — Color Encoding: geographic grouping of countries or regions
Time dimension: Snapshot of 2017 (static); 2005–2017 (animated)
Takeaways (what I can easily tell from the visualization):
- GDP is positively correlated with life satisfaction, so “richer” countries or regions generally have higher life satisfaction.
- Most high-GDP, high-satisfaction countries or regions are in Europe and North America (and Oceania if you hover over the continent to see details), and most low-GDP, low-satisfaction countries or regions are in Africa, with Asia and South America in the middle.
- I can’t help but noticing the two eye-catching large, red circles, which happen to be in the center of the visualization, with moderate GDP and moderate life satisfaction.
- From the animated visualization, I am able to see countries and regions moving acorss years along GDP and life satisfaction, with their start and end points connected with arrows.
The first and fourth takeaways are, I guess, exactly what the author would like the readers to notice. The other two, though not directly related to the question asked (wealth-happiness), adds to the variation of the visualization which makes it more attractive.
More About Happiness and Income
After seeing this visualization, I asked myself, is the question answered? Is it wealth that makes people happier? On first thought, the answer seems to be yes. More health, much happier. On a second thought, what if the time dimension is brought into play, in a way that life satisfaction of countries and regions with similar income levels even at different time snapshots can be compared with each other?
In other words, is life with similar income level happier today than in the past?
This is not enabled in the original visualization, as time dimension is only visible through the animation, and our eyes are unable to catch and compare the movement of each countries and regions simultaneously. The end view only links the first and last points for each country and region, but omits the dynamic changes in between.
Thus, in order to compare data across years for all countries and regions, we would need the time dimension to be added to the static visualization directly.
Combining Two Story Types
By adding the time dimension, the story to tell is both a “Change over time” and a “Contrasting values” one⁴. It is interested in, instead of the changes of life satisfaction and income level of the same country or region over time, how is the life satisfaction corresponding to the same income level in general changing over time, by comparing average life satisfaction of a given income range in different years.
Check the Tableau dashboard via the link below for my finalized dashboard.
How to Use the Dashboard
There are three ways to use this dashboard: 1) View data for certain countries or regions across years 2) View data for a certain GDP range (income level) across years 3) View all data for a specific year(s)
- To view data for certain countries or regions, either use the filters on the upper right corner, or hover over a dot on either of the two largest line charts to select a single country or region.
- To view data for a certain GDP range, (rectangular) select the dots within the targeted GDP range in the upper right line chart.
- To view data for specific year(s), refer to the lower right scatter plot and use the year filter that only applies to this plot.
Higher GDP with Higher Life Satisfaction
The first goal of the dashboard is to recreate the wealth-happiness relationship in the original visualization. Countries and regions with higher GDP have higher life satisfaction scores.
The line chart on the left serves this purpose, and more (by including the time dimension).
Level of granularity: individual country or region (each circle / dot)
Dimensions:
- Rank of GDP per capita— X axis
- Year — Y axis
- Life Satisfaction— Color Encoding: Orange-White-Blue diverging color palette is used. Orange indicates below average and blue above average. White indicates NULL values. More are explained in last section “Pain Points”.
- Lines connecting each country and region throughout the 13 years
To Take Away (what I wish can be told by the visualization):
- The higher the GDP rank, the stronger the blue and the weaker the orange (the higher the life satisfaction). This is in general true for all years.
- There are outliers though, the orange points and lines in high-GDP-rank countries and regions, and blue ones in low-GDP ranks.
- Besides, the logarithmic, instead of linear, correlation between GDP per capita and life satisfaction in a given year is displayed in the scatter plot (also mentioned in the referenced article).
Economic Growth with Increasing Life Satisfaction
The relationship between economic growth and changes in life satisfaction of countries and regions is harder to identify than I imagined, because the changes across years are not significant enough to be distinct in color. Besides, the time range of 2005 to 2017 may not be long enough to display meaningful economic growth for most countries and regions.
However, we are able to see clear country-specific trends. For example, Laos and Singapore both experienced general economic growth between 2005 and 2017, along with decreasing life satisfaction, which is best to be seen via the two line charts on the right, showing average life satisfaction and GDP across years.
These are two specific cases that contradict the general conclusions in the referenced article, which may be worth further inspection if you are interested.
Life Satisfaction of Similar GDP Levels in Different Years
Finally here we arrive at our last question, which is not directly visible in the original referenced visualization, the trend of life satisfaction changes of same GDP ranges across years.
By (rectangular) selecting a certain GDP range, e.g. around $50K in the screenshot above, we are able to see that, despite the increase of average GDP from $43710 to $47413 from 2005 to 2017, the life satsifaction shows a decreasing trend, from 7.48 to 7.06.
It is also displayed via the downward trend line in scatter plot, as well as in the selected range of upper right corner line chart that more orange is detected in later years.
Although there is not enough evidence yet to draw any conclusion about how much/less happier lives with same level of income are now than in the past. It will be an interesting topic to inspect.
Relative Deprivation
One possible theory, if this trend is real, is relative deprivation. The concept was first introduced by Professor S. A. Stouffer, and later developed by Professor R. K. Merton⁵. It entails the negative feelings generated by an unfavorable comparison to others, and basically says that feelings of deprivation are relative, rather than absolute. I will not go into details here, but it could be a direction to go if you would like to dig deeper - is life happier today with more wealth.
Happiness can also become a story to tell if we look closer. Be Curious and be happy.
Pain Points
There are a few pain points I encountered when making this dashboard, and I will go through the process how I tackled them just in case it can be helpful.
Missing Values
There are extensive missing values of both GDP and life satisfaction in the dataset. I narrowed the time range to 2005–2017, both to follow the practice in the original referenced visualization, and because of a relative lack of missingness in this range.
GDP is the foundational dimension that is used to build the visualization structure (on X/Y axis, rather than other encoding). Besides, to shed light on the questions we are interested in, life satisfaction scores without correponding GDP are somewhat meaningless. Thus, I excluded countries and regions with any missing values on GDP in 2005–2017.
The most debates I had were on missing values in life satisfaction. In most visualizations, I just filtered out single records with NULL life satisfaction, e.g. you can see certain countries and regions missing a dot in certain years in the upper right line chart.
However, since the GDP rank line chart (one on the left) provides relative data on GDP rather than obsolute values, excluding a record for certain countries or regions may distort the interpretation of GDP performance of other countries, e.g. a country or region with very low GDP may even be ranked №1 if other better-performing countries do not have life satisfaction scores on record for that year.
The missing values kept in the visualization are by default recognized by Tableau as 0, and color encoded accordingly, resulting in the scary graph below with many dark red dots. At first glance, it seems that there are many records with extremely low life satisfaction, but many are actually just NULLs (as the real bottom life satisfaction is 2.66).
In order not to distort the visualization, I’d need a non-distracting separate color for NULLs. The final solution is a three-color diverging palette with 0 in the center (colored White) as below. I have also normalized the life satisfaction values to make the mean 0.
The drawback of this methodology is that life satisfaction scores close to average are almost white, which makes them relatively undetectable. However, this does not interfere too much with our goal to visualize the trend.
Tableau community forum post that inspired my solutions:
Color Palette Difference
Life satisfaction is quantified on a range of 0–10. Although it varies extensively between wealthiest and poorest countries and regions, the difference among countries and regions in a certain GDP range is not distinct enough to be seen on a full color palatte made out of the entire dataset. Can you tell the difference between those blue points below?
In order to view the nuances, I’d need the dots to be re-colored based on the marks seleted only. This is enabled through set action, which allows changing the value of a set and corresponding measures through certain action performed on the sheet. You can see the color legend is re-callibrated based on the values of the selected marks only.
Tableau free training on set action:
Other Tableau Resources Referenced
Additional Notes
- Despite lack of rigorousness but considering readability, the following concepts are used interchangeably (without differentiation) in this article: wealth, money, and income; happiness and life satisfaction.
- The raw data used are subject to certain biases and limitations: 1) Life satisfaction is subject to selection bias and response bias depending on how the survey questions were asked and answered. 2) The article focuses on 2005–2017 due to data availability, and the trend may be different beyond this time range. 3) The article leaves out certain records with missing data, which affects its generalizability.
- Operationalization, validity, and reliability of variables referenced in this article are not further discussed, mainly because the article is aimed to raise questions rather than to make conclusions. Happy to discuss more and eager to learn more.
[1] Seph Fontane Pennock. (February 18, 2020). Positive Psychology 1504: Harvard’s Groundbreaking Course. https://positivepsychology.com/harvard-positive-psychology-course-1504/
[2] Dr. Tal Ben-Shahar started to teach this course in 2002, while the first Positive Psychology class was taught at Harvard in 1999 by Prof. Phillip Stone. The course, according to Harvard Crimson, later became the most popular course in the spring semester of 2006, and was feature in Boston Globe.
[3] A common full-time job is 8 hours / day and 40 hours / week, and 1/3 is just an approximate without regard to holidays and other variations. https://en.wikipedia.org/wiki/Full-time_job
[4] YouTube (uploaded by Udacity). (April 11, 2019). DSTND Course4 Lesson1 A2 DataStoryTypes. https://www.youtube.com/watch?v=GK5XQnNVr4A
[5] Wikipedia. Relative deprivation. https://en.wikipedia.org/wiki/Relative_deprivation