This project explores the different methods used to analyze the calls received by the Crisis Clinic across geography and time in order to find useful insights in terms of discovering important trends, correlations and possible causations. We analyze the call trends of 4 different lines: Crisis Line, Teen Link, Recovery Line, and 211, and specifically focus on the most common problem areas and needs, which we have analyzed with respect to geography in terms of ZIP codes and cities, and with respect to time from January 2010 to May 2014. Our exploration with the data shows that it is possible to extract useful information on the call behavior of the callers across geography and time through visual analysis. Based on these results, we explain how managerial decisions specifically relevant to funding of the Crisis Clinic can be enhanced, and also focus on the aspect of increasing public awareness through hosting the final set of visualizations as a dashboard on Tableau Public.
Role: UX Researcher, Designer, Data Wrangler, Presenter
Key Activities: Literature review, data curation, interviews, prototyping and user evaluation, data visualization, usability study, presentation
Team Members: Abhigyan Kaustubh (AK), Emily Greenberg, Lana Pledger, Rijuta Trivedi
Timeline: Apr 2014 – Jun 2014 (10 weeks)
Tools: Excel, Tableau, Powerpoint, SQL
Crisis Clinic is at the heart of the Seattle-King County safety net providing a broad array of telephone-based crisis intervention and information and referral services. For many people in emotional distress or needing community services assistance, they are their “first call for help.” Every year, the Crisis Clinic receives a huge number of phone calls from King County residents in need of emotional support and community services. It has four main programs through which it provides its services:
- The 24 Hour Crisis Line offers emotional support to those in crisis or considering suicide;
- King County 2-1-1 offers information and referrals to community services based on its database of more than 5000 services;
- WA Recovery Help Line provides a state wide service offering emotional support and linkage to substance abuse, problem gambling and mental health services to anyone in Washington State;
- Teen Link offers emotional support and assistance to teens by providing a teen-answered help line.
As a nonprofit organization, Crisis Clinic depends on the financial support of local government, United Way of King County, corporations and foundations, and the generosity of donors to keep its doors open and provide services. In addition, it also serves as a central point for crisis resources that includes training, outreach, and a bridge to other organizations that may provide specialized support.
In this project we investigate the different ways of making interactive visualizations of the callers’ dataset to gain insights into its presence in the King County area, and also explore and understand patterns and trends in the calls they receive across geography and time. We envision that these visualizations and insights will allow the staff at Crisis Clinic to better allocate resources in a targeted way, more effectively communicate the impact of Crisis Clinic to current and prospective funders, and allow the general public to better understand and appreciate its work in the King County area.
1. Interactive data visualization for global road safety data by Pulitzer Center (McCarey, 2013).
This was an example of a spatial overview with additional data visualization summaries revealed on mouse-over. In this example, by clicking on a country one can access an assortment of road safety data, ranging from trend lines on highway fatalities to statistics on the types of vehicles most likely to be involved in fatal crashes. We were impressed by the use of details-on-demand to quickly provide summary statistics for individual regions.
2. “How fast is LAFD where you live?” – A map to compare the LAFD’s performance across LA (Welsh, 2014).
In this example, the spatial map shows a block-by-block analysis of how long it takes LAFD units to reach victims after the agency picks up a 911 call. A pop up displaying further data is revealed on mouse-over with statistics on average response time, the total number of responses over five years, the breakup of medical and fire related responses, and average arrival and dispatch times. Filters on the side further enable scanning the information based on neighborhoods as well as type of need (e.g. cardiac). This analysis helps reveal “simultaneous incidents” as a vexing issue and difficult to manage in some parts of the city with current staffing levels.
3. San Francisco Crimespotting, designed by Stamen Design is an interactive map of crimes in San Francisco (San Francisco crimespotting, 2014).
This spatial visualization overlays crime events on a block by block basis. The users can filter by crime type, date and time of the day and find out individual crime event by clicking on each point. In addition to information about the happenings in neighborhoods, this visualization helps answer questions about patterns like: is there more number of crimes this week than last week, more this month than last, etc. This provided a unique way of exploring crime volume during different times of the day, a trend that we believed may be interesting to explore in the Crisis Clinic’s data set.
The Crisis Clinic logs all of its calls into a 2009 SQL server database. Mike Maloy and Terry Morgan, part of the IT Staff, generously granted us access to a server owned by the Crisis Clinic that provided a snapshot of the current database. This snapshot was used to create the interactive visualization which was presented to the Crisis Clinic staff during the user studies and evaluations, with the assumption that they could later connect Tableau to their live database if desired after project completion.
We were engaged with the process of data processing throughout the project as we were uncovering new requirements through our usability studies (described below). To incorporate these, we had to consider defining new variables that the users were interested in and manipulate and clean data accordingly.
Key Dimensions that were finally explored and visualized:
Personas and Scenarios
1. Susan, the Program Director
As the director of 211 (The Community Information Line), Susan needs to be able to quickly answer inquiries from stakeholders about Crisis Clinic’s operations; she recently received an inquiry asking how many Veterans have accessed services in Snohomish. Using the interactive dashboard, she is able to quickly use filters to scope down to this demographic and answer the question. If she is too busy, she will ask Hannah to help out.
2.Tom, the Fund Raiser
Tom is responsible for talking to a local legislature in Bellevue about getting more funding for the Crisis Clinic. He goes to the interactive dashboard and can quickly zoom in on data revolving around call volumes from Bellevue residents. He captures a visualization that will help him communicate community needs to the legislature, and convince him that the Crisis Clinic is providing his town with an invaluable service. He also shares the dashboard to the public to increase the visibility of Crisis Clinic work and provides information about community needs and trends.
3. Mary, the Curious Student
Mary heard from a friend who called Crisis Line Clinic. This peaks her interest and she would like to find out more about what type of Calls are most prominent for the area where she lives. She opens the dashboard on the public website and notices that for the city she live in – Sammamish, Mental Disorders are the largest problems category. This makes her aware of the issues she hasn’t thought before and makes her find out even more about the clinic.
Upon deciding on the topic of the visualization and the personas we would be catering to, we had a brainstorming ideation session, where we individually came up with sketches for answering our initial set of main questions based on initial interviews with the Crisis Clinic:
The main questions that we started off with were the following:
- Distribution of calls by region (ZIP code). Do certain regions have more types of calls then others?
- Are there peaks in call volume during certain times of the year or special times of the year?
- Distributions of callers by other demographics; i.e. age, ethnicity, gender, veteran-status.
Initial Iterations of Sketches
Prototyping and User Evaluation
After brainstorming through various sketches, we consolidated them into creating an early mid-fidelity PowerPoint prototype to test it with our prospective users. This prototype evolved into an interactive implementation in Tableau, which was further refined based on usability testing. We went through three rounds of iterations and usability testing. During these iterations, the users were asked to complete tasks using progressive iterations of the visualization, and the visualization in turn was improved accordingly.
- What is this visualization broadly about?
- What does the size of the bubbles indicate?
- How many types of programs are there?
- Find the top 10 problems and needs of the King County area. What about all of the problems and needs of the King County area.
- Find the details (ZIP code, city name, number of callers, calls per capita) that any random bubble represents.
- Select a random ZIP code. Can you tell once you have a selected it (is there adequate feedback)?
- Select a random city. Can you tell once you have a selected it (is there adequate feedback)?
- Find the details (number of callers, calls per capita, city name) of a ZIP code of your interest.
- Find the top 10 problems pertaining to any ZIP code.
- Find the top 10 problems pertaining to any ZIP code with reference to a specific program.
- Find all problems/ needs pertaining to any ZIP code.
- Find the calls over time pertaining to a specific program.
- Find the details (number of callers, calls per capita) of a city of your interest.
- Find the top 10 problems pertaining to a city of your interest.
- Find the top 10 problems pertaining to a city of your interest with reference to a specific program.
- Zoom in to a specific area on the map. Try to pan around
- Find the calls over time pertaining to a particular program, and the problem needs associated with them.
- Select a particular time interval to find the corresponding problems/needs and details (number of calls, calls per capita) of a ZIP code and city of your interest during that interval.
- Are the labels, titles and symbols clear and legible?
- Is the vocabulary used comprehensible?
- Is the feedback of the actions you perform adequate?
- Is there anything that you would expect to see, or would like to see, that you didn’t?
Iteration I – PowerPoint Prototype
The first round of the usability testing was done using the above PowerPoint prototype. We showed it to four people at the Crisis Clinic, who were from our initial set of personas, and collected their feedback. As the visualization was not fully interactive at this point, as the users were asked to complete tasks we asked them to describe what they were thinking, and we described what changes would occur in the visualization based on particular actions.
Our key insights from this early study were:
- Make the visualization simpler and easier to use.
- Incorporate proper labelling, which was not present on the prototype. Our rationale for not doing this initially was because we had crafted the slides as a low-fidelity prototype. The users, though, interpreted it as a high-fidelity prototype and felt that that is how it would be actually incorporated. Also, as we had abbreviated a few things because of it low-fidelity nature, the users didn’t find all of it comprehensible, which caused some distraction from the main point.
- People loved the map feature and wanted it to be “pretty”, “colorful”, and legible.
- The test users indicated their preference to be able to browse by ZIP code and by city.
Iteration II –Tableau
Usability Testing II
- Great excitement about the map view and desire to find trends over time.
- Managers wished to see top 10 problems or needs for a particular area/city or ZIP code.
- The Tree map view of the problem need wasn’t as intuitive to the people as we had suspected. Particularly asked question regarding its relation to the map – some didn’t think that they were related at all, and one person asked “why is this better than a bar chart?”
- Although a novelty, the animation wasn’t successful in showing trends as the older data points were replaced by the newer ones.
- Suggestions to correlate call and problems/needs with population density (calls per capita) of a city/ZIP code.
- The Program Director was excited to show this on the website and share it with the public to create more awareness.
- The people in leadership were appreciative of using such a visualization during funding events to deliver relevant pertinent information effectively to prospective funders.
- The people comprising of the IT Task persona – Hannah, were “meh” about it, and described the way they work as described in the Persona section.
Usability Study III
The 3rd round of the testing was done individually by the project team members using the above reformatted dashboard, which consisted of informal usability sessions with four Crisis Clinic employees, coworkers (who were UX Designers) and friends, the latter to gauge the needs and perspective of the new persona – Mary, the Curious Student. Here we incorporated feedback from previous tests and added third persona – Mary. In case when people were not familiar with the Call Center we had brief introduction and explanation of what is about.
Summarizing across participants, the key challenges that the users faced were:
- Users were consistently frustrated around de-selection in Tableau – people consistently had problems figuring out how to do it, and how to get back to a “baseline” view.
- Many users were frustrated with the map controls – two people double-clicked on everything and kept zooming in (Tableau issue).
- People did not realize they could select or filter by dates.
- People did not immediately realize that the programs were clickable.
- People wanted a search field to search by ZIP code and city on the map.
- Users were confused about bubbles over the cities. Although most people eventually figured out that bubbles represented calls per ZIP code, it was not immediately obvious, and took some prompting for them to explore and understand.
- Although some users thought that being able to select cities was a good feature, it was difficult to deselect a city to get to another ZIP code and the interaction felt awkward.
For users unfamiliar with the data set, there was a common desire to know more about each program. Many questions were around what 211 does.
Evaluation & Conclusion
We believe our visualization is successful in meeting the needs in terms of answering the questions important to our three key personas. Furthermore, our visualization affords easy exploration of patterns and stories of different consumer groups of special interests (Active Military, Dependents of V/AM, Disabled, ESL, Immigrant/ Refugee and Veteran) in the King County and the WA area.
We evaluated our final visualization against some of the seminal thoughts on data visualization from books and papers from experts:
Overview: The choropleth map provides an overview of distribution of calls per ZIP code. The auxiliary charts provide an overview of calls over time and the top problems and needs across all calls.
Zoom: Map portion of the dashboard allows panning and zooming in/out of any area of interest.
Filter: The visualization allows filtering based on the city, ZIP code, type of Crisis Clinic program, time frame, problem needs as well as special interest groups.
Details on Demand: Hovering/clicking on ZIP code areas for example shows details that are pertinent to that chart, such as calls per capita, total number of calls, city name and ZIP code. This is true for all three sections of the visualization.
Principle of Effective Navigation
The visualization also allows the progressive technique of step by step exploration, fine tuning as we go. For example, we can start with an overview and decide to check for call trends for a specific city, say, Seattle. Then if we want to further see the call trends for the city of Seattle during a particular time frame, we can fine tune the filters to show that. A process like this allows certain interesting trends and facts to emerge, which might explain some of the anomalies seen in the overview. The visualization thus satisfies the principle of successfully showing from pattern to investigation of cause.
Tuftes principles of Graphical Excellence and Integrity
The visualization follows the flowing principles of graphical excellence and integrity:
- It is easily understandable.
- Visualizing this data reveals trends and patterns that might have been hard to grasp through visually examining textual forms of the data.
- It presents data in context. For example, the distribution of calls by ZIP code is shown in the form of a choropleth, i.e. geographical representation. When it highlights information, it does so in context, changing other related views to reflect data specific to that selection.
- It uses visual components appropriately. We use lines for trends over time, length of bars for number of calls in different categories, gradation of color for volume of calls, and color for program. These fall along the ideals of effective encodings for intuitive data visualization. Color is also used consistently through the visualization, allowing users to make connections across the different parts of the visualization.
- It uses labels and legends where necessary, keeping the user in context through the exploration.
Storytelling and Final Thoughts
One of the things that came up during our evaluations with Crisis Clinic staff was the idea that this would be an interesting addition to their website for anyone in the general public to explore. It would be a good way for them to look for patterns in their area of interest, or in general just explore what Crisis Clinic does. However, people that are unfamiliar with the call data may benefit from more guidance and a “walkthrough” of some key data points in the visualization. For this particular purpose, we believe it may be beneficial to incorporate a story format with the data to convey the information to the public. We also think that it would be beneficial to provide more context information such as a description of each of the programs does. Although we chose not to include this in the visualization because its primary users are familiar with the programs, we believe that contextual information could be provided by the website where it is embedded; ideally, this would be hosted by Crisis Clinic.
We do believe that this data set has many stories to tell, both to the Crisis Clinic staff, potential funders, and well as the general public. During one of our usability tests, while exploring the data one of the Crisis Clinic staff members quoted:
“I’m having an emotional reaction… it really makes our work real and affirms what we do… I think a lot of the staff and volunteers would like to see this.”
Ultimately, this visualization provided some useful insights to the Crisis Clinic, as well as exposed them to the power and potential of interactive data visualization.
For a more detailed report, please visit here.