PROJECT INTRO
We told the story of economic recovery in Washington State by examining how residents fared before, during and after the economic downturn. The target audience included policymakers, state governor and county officials. We explored multiple solutions to visualize indicators of economic well-being within Washington’s 39 counties. User study was conducted to decide the best solution. We used Tableau and Illustrator as visualization tool.
Time April 2016 - May 2016
My Role I designed and tested the 3rd visualization idea
The questions
The Great Recession was a blow to economies all over the nation. During the recession years of 2007 to 2009, high rates of unemployment and foreclosures destabilized communities. Six years later, the nation appears to have gradually recovered.
Washington state is unique in that it has experienced high population and economic growth since the recession. In 2014, the state gained almost 100,000 new residents, bringing the total state population over the seven million mark.
Through this project, we wanted to tell the story of economic recovery in Washington by examining how residents fared before, during and after the economic downturn. The questions we hoped to uncover include:
- Is cost of living keeping up with income in Washington state?
- How did the unemployment rate change before, during and after the recession in Washington state?
- How has housing affordability changed over the last ten years?
The process
data sources
We pulled country-level data from multiple sources because no one source contained all the variables we needed. While the data we use cannot be used to directly show the causal effects of the recession, they can help us visualize trends in the state over time. Our data points and sources include:
- U.S. Census Bureau’s Small Area Income and Poverty Estimates (SAIPE)[1]
- Washington Self-Sufficiency Standard[2]
- Washington State Median Household Income Estimates by County[3]
- Washington State Department of Licensing Regional Funding Map[4]
- Washington State’s 50th percentile rent estimates by county[6]
- Washington State Resident Civilian Labor Force & Employment Annual Unemployment Averages[8]
AUDIENCE
Our audience included Washington State's policymakers, Governor's office, county officials and leaders.
- Policymakers who have a stake in improving the economic well-being of Washington residents.
- Governor Jay Inslee’s office, as he has some authority over the state budget and how resources are allocated.
- County officials and leaders who need poignant, concise and user-friendly information with which to advocate (perhaps to the Governor) for their jurisdiction.
visualization iterations
Sketches were the first-round visualizations drafts. We created 3 different sketches to answer our three different research questions.
SKETCH 1a; 1b: Measures of Poverty & Well-Being by County in Washington State
With this visualization we attempted to show the current indicators of economic well-being through a mix of the poverty rate, self-sufficiency income, mean income and unemployment rate. Since the self-sufficiency income is not a traditionally used indicator of economic well-being, we experimented with how best to use this data to portray cost of living in each county relative to resident’s income, in a user friendly way.
Sketch 1a: This sketch used a heat map to show as the main visual indicator. The colors represent the difference between the mean income of the county, and the income that a family would need to actually support themselves without assistance in that county (the self-sufficiency wage). We thought the colors would show a striking difference between counties whose residents on average earn above the self-sufficiency wage and counties whose residents, on average, are earning an income well below the self-sufficiency wage. However, being that this is not something that users are used to seeing, we were unsure whether users would be able to easily grasp this message. In addition, we added in a more traditional measure, the poverty rate, as another benchmark for economic well-being, that users would be familiar with.
Sketch 1b: For the second iteration, we reversed the heat map. Poverty rate is shaded, and the difference between self-sufficiency and mean county income is labeled. We thought this iteration may be a more effective way of conveying tradition and an alternative measure of economic well-being.
SKETCH 2a; 2b; 2c: Unemployment Rates Before, During & After the Recession in Washington State Counties
With this visualization series, we wanted to show the variation of unemployment rates in Washington state counties from 2006-2015. Unemployment describes individuals who do not have a job as a percentage of the labor force. Low unemployment rates indicate better economic well-being and high unemployment rates indicate worse economic well-being: families lose wages and purchasing power, and the state loses the goods or services that could have been produced.
Sketch 2a: We first decided to show the range of unemployment rates between counties (Sketch 2a). A box and whisker plot is interesting because it describes data in quartiles (those in the “lowest” or “first” quartile are shown between the bottom whisker and the box of each year’s data set, the darker grey box shows the second quartile, the light grey box the third, and the fourth quartile is shown between the third quartile and top whisker). Two outliers, Ferry county in 2006 and 2014 are shown outside of the data range because they are greater than 1.5 times the interquartile range (Q3-Q1).
This visual features all 39 counties, organized by color into regions. This idea is helpful in theory, but it is impossible to identify individual counties, and there are too many dots to make meaningful conclusions at the regional level. However, it is interesting to see the outlier Ferry county—part of the Northeast region—fare so poorly. This visual might be most helpful to a Ferry county elected official than a statewide government official.
Sketch 2b: Applying our observations from the previous visual, we selected the same box and whisker method to show county data at the regional level, which helped declutter the visual (Sketch 2b). However, since the Washington state regions we chose are not common knowledge, it may be important to pair a sketch like this with a map of labeled Washington state counties. In this version, there are no outliers, but there are clearly regions that struggle with higher unemployment, like the Central, Northeast, and Southwest regions.
When considering what we learned from this sketch, we realized that we were looking for trends in regions over time, which led us to produce our final sketch (2c), a line graph.
Sketch 2c: Finally, Sketch 2c includes a line connecting each region’s unemployment averages for each year between 2006-2015. Visually, this is easier to see which region’s citizens suffer disproportionately from unemployment than other regions, and compared to the Washington state average. This graph also benefits from a Washington state map with labeled counties as well, so we recommend including it with the one included in Sketch 2b.
SKETCH 3A; 3b; 3c: Rent-to-income ratio among different regions of Washington from 2006 to 2015.
We wanted to track the differences in rent-to-income ratio among different regions of Washington, and how it changed from 2006 to 2015. The U.S. Department of Housing and Urban Development categorizes rent as affordable if a household is paying 30% of their income or less in rent. If a household pays over 30% of their incoming as rent, this could compromise their ability to afford other basic necessities, hence it is a powerful indicator of economic well-being. We decided to use a tree map, and thought of three different visualization solutions.
Sketch 3a: The first visualization intended to build one tree map in total, and divide the data by year first, and then by region. Each big block represents the rent-to-income ratio of different regions in a single year, so we can compare the region differences on rent-to-income ratio in each year.
Sketch 3b. The second visualization is a tree map that groups the data by region first, and then by year. Each big block represents the rent-to-income ratio of the same region in different years, so we can compare the yearly differences on rent-to-income ratio in each region.
Sketch 3c: After group discussion and rethinking the story we want to tell, we decided try a less cluttered iteration. We selected four years of data from the dataset instead of showing all 10 years of data to simplify the visualization. So, the third iteration shows the rent-to-income ratio in 2006, 2009, 2012 and 2016. We think this is representative enough to reveal trend before, during and after the recession, and it makes the visualization clearer to view.
User Study
We conducted 5 user studies as a chance to gauge whether our visualizations were clear and if we were telling an effective story. Our users were chosen based on their affiliation to policy, economics and/or familiarity with Washington state.
Key Takeaways from User Feedback
- Sketch 1: Users liked the color contrast of 1a, but thought that these visualizations in general had too many variables.
- Sketch 2: Users overwhelmingly liked the line graph best and suggested making color differences more apparent.
- Sketch 3: Users found that tree maps may not be an effective way to visualize this data; users were confused. This points to a need to visualize rent to income ratio in a new way.
Full version of user feedback can be found in our report.
Final Visualizations
We considered user feedback to fine-tune our sketches. Since all three concepts tell a slightly different piece of the story of economic well-being in Washington state, we finalized three final visuals.
Concept 1
Maps are natural for viewers to interpret, with intuitive and visually encode spatial information.
We chose to simplify Sketch 1 by using only the difference between median income and self-sufficiency as a variable. We kept the contrasting green and red colors that users found striking, though classmates pointed out that these colors may be difficult for those who are colorblind. Red and green colors are highly associated with monetary surplus and deficits, which is a key point we are conveying, and suggest specific areas where action is needed. Since the Governor and other politicians are bombarded with daily messages and our goal was to inspire quick takeaways (and also that the Governor is not known to be colorblind), we felt the color choice was not a liability and enhanced the ability for him or his aides to make quick decisions.
We changed the title to make the concept clearer, increased font size and enhanced the notes.
Concept 2
Most users found the box and whisker plot confusing, and we did not want the Governor’s office or county politicians to disregard this visual for the same reason. We focused on refining the line graph of unemployment trends across Washington state regions to allow users to make comparisons between regions. We chose region colors to make it easier to identify each region, and added the missing “Southeast” region description, as flagged by one of our users. The Washington state average line was also made as visible as possible, differentiated on the graph as a dotted black line to provide state-wide context, which is important context for a user considering local, regional, and statewide issues. Viewers can easily reference county locations by looking to Concept 1, but we made the visual cue more obvious by color coding the region names as well.
Concept 3
During user study, most users felt confusing interpreting tree maps. There were no huge differences on the size of each block due to the narrow range of data. Besides, the color-coding is confusing: the darker green blocks means people are struggling more in those counties. However, most people think darker green should mean better life situation.
Therefore, we decided to change tree map to scatter plot. The data we use is still the same, and we use different colors to represent 7 regions of Washington state, while aligning the points by the order of year.
We also show each year’s state Average Rent-to-Income Ratio by a dotted line so that users can easily compare certain county’s rent cost to the state average level. This allows the Governor’s office to make county-by-county comparisons on housing affordability, and for local politicians to determine how their county stacks up.
Brochure
We also designed a brochure for our final data visualization.
Thank you for reading.
Qin
Team
Talia Kahn-Kravis - Visualization Concept 1
Audrey Lawrence - Visualization Concept 2
Qin Jiang (me) - Visualization Concept 3
references
[1] U.S. Census Bureau. (2014). Time series small area income and poverty estimates: State and county. Washington, D.C. Retrieved from http://www.census.gov/did/www/saipe/
[2]University of Washington Center for Women’s Welfare. (2014). The self-sufficiency standard for Washington (2014). Seattle, WA. Retrieved from http://www.selfsufficiencystandard.org/node/4
[3] Washington State Office of Financial Management (2016). Median household income estimates by county: 1989 to 2014 and projection for 2015. Olympia, WA. Retrieved from http://www.ofm.wa.gov/economy/hhinc/
[4] Washington State Department of Licensing. (2016). Regional funding map. Olympia, WA. Retrieved from http://www.dol.wa.gov/business/motorcycle/funding-map.html
[5] Washington State Department of Natural Resources. (2016). DNR regions and districts. Olympia, WA. Retrieved from http://www.dnr.wa.gov/about/dnr-regions-and-districts
[6] HUD Office of Policy Development & Research. (2015). 50th percentile rent estimates, 2006 - 2015. Washington, D.C. Retrieved from https://www.huduser.gov/portal/datasets/50per.html
[7] U.S. Census Bureau. (2015). Washington quickfacts. Washington, D.C. Retrieved from https://www.census.gov/quickfacts/table/PST045215/53
[8]Washington State Employment Security Department. (2015). Washington state resident civilian labor force & employment annual unemployment averages. Olympia, WA. Retrieved from https://fortress.wa.gov/esd/employmentdata/reports-publications/regional-reports/local-unemployment-statistics
