The Iterative Design Process in Social Science Research: Scientific Productivity of the Health Retirement Study

Background

The product of the University of Michigan Health Retirement Study (HRS) is survey data. Visualizing the number of publications in the form of journal articles, books and book chapters, dissertations, and working papers using HRS data is the best metric to measure this product's scientific productivity. No funding or grants will be available without the visualization of this metric. This visualization will be updated using an iterative design process.



Static HRS Data Visualization

The Scientific Productivity of HRS visualization before the redesign: (1) This file was exported from Microsoft Office as a giant GIF (2) A summary table supplemented this graphic.

Design Goals

The online bibliography was also being updated alongside the visualization. This online bibliography update utilized the Parallel Design Process to evaluate 3 possible solutions instead of the iterative design process. Although this article covers the process behind the data visualization, the bibliography will also factor into the user personas below.

The design goals were to improve functionality of the online bibliography and portability of the data visualization for the target audience based on user research:

"I am the director of the National Institute on Aging and I want to show the director of the National Institutes of Health this study's influence on the scientific world."

Funders

"I work for the New York Times. I need to write an article about published research by a Health Retirement Study investigator. "

Members of the Press

"I am a researcher, faculty or student and need to cite an HRS publication in my own paper."

Academics

"I am helping one of our investigators prepare a presentation for a conference and need to create a bibliography."

HRS Staff

First Iteration

The first iteration was a hand drawn sketch. It allowed incorporation of feedback into design changes very quickly.



Static HRS Data Visualization

The first hand drawn iteration: (1) The orientation of the graphic stayed vertical (2) Mistakes are not a big deal with quick drawings

Second Iteration

The second iteration was in d3, a library created by the former New York times data editor Michael Bostock. Although an ambitious choice, the department did not have the resources of a New York Times Data Team.



The second iteration in d3: (1) The d3 library requires extensive resources. There were issues with portability on different devices (2) What looks like just a blank space is actually the absence of the summary table from the original visualization - it was left out of the hand drawn sketch by accident and the team liked it

Third Iteration

The third Iteration and final iteration was with the Google Charts library.

The third iteration in Google Charts: (1) This allowed portability between devices (2) The orientation of the final version changed from vertical to horizontal due to client preference.

Key Takeaways From Our Process

This process was based on approval from the director of the study in an intensely political environment. Nevertheless, incorporating the iterative design process and user personas into such a personality driven product environment yielded great results.


Health Retirement Study Redesign