Fit, Bias, and Enacted Sensemaking in Data Visualization: Frameworks for Continuous Development in Operations and Supply Chain Management Analytics

• Date: 01 March 2016
• Published date: 01 March 2016
• DOI: http://doi.org/10.1111/jbl.12113
• Author: Elliot Bendoly

Fit, Bias, and Enacted Sensemaking in Data Visualization:

Frameworks for Continuous Development in Operations and

Supply Chain Management Analytics

Elliot Bendoly

The Ohio State University

Data visualization has a critical role in the advancement of modern data analytics. Visualization lends assurances to data validity and com-

pleteness, as well as to the effectiveness of cleaning and aggregation tactics. It provides the means by which to explore and discover rela-

tionships otherwise hidden from default assumptions in statistical modeling. Strong visualization is also fundamental to end-result conveyance

and audience interpretation. But how can one ensure that strength? How can one avoid developing representations that are marginal in value, or

worse misleading? In this paper, I will discuss theory, evidence, and practical approaches to managing data visualization development, viewing

data visualization not simply as an outcome but as a continuous process and facet of organizational culture.

Keywords: data visualization; sensemaking; supply chain; framework; bias; fit

INTRODUCTION

Use a picture. It’s worth a thousand words. –Arthur

Brisbane

This quote—along with its various permutations—has been

thrown around for more than a century now. And there is good

reason: It is often true. In the realm of data analysis, we now use

the term data visualization to describe a specific genre of infor-

mational “picture”that allows for the effective depiction of data

and the relationships it can suggest. Some of these visualizations

are what we might more simply term graphs (control charts,

scatterplots, best-fit lines within such scatters, surface plots, con-

fidence intervals, population ellipsoids, network diagrams,

chloropleths, etc.). By themselves these choices of individual

graphical depictions are also often referred to as idioms. How-

ever, data visualization efforts often extend beyond the use of

any single graphic, and the results are often deliberately nonstatic

and interactive creations.

Well-integrated systems of idioms (i.e., the most effective of

dashboards used in practice) are similarly the result of deliberate

efforts by designers to convey more complex data stories.

Indeed, these systems can be just as critical to research as they

are to practice, given the often highly complex, nonmonotonic,

feedback-inclusive, and nominally constrained relationships

scholars are increasingly attempting to capture (Bendoly 2013b,

2014). In the words of Edward Tufte (2003): “There are many

true statements about complex topics that are too long to fitona

PowerPoint slide”(p. 31). Imagine, for example, the task of

describing how even a single supply chain disruption reverber-

ates across a network. Rather than attempt to describe the

dynamics in the form of bullet points, how much more powerful

and insightful would be annotations accompanying a dynamic

visual of supply chain activity? Perhaps a visual rendering with

snapshots of time controlled by individual viewers, denoting ful-

fillment failures along geographically mapped nodes and arcs?

Throwing in the additional ability to see the disruption unfold

visually from one time point to another with simulated automa-

tion?

Despite the existence of a wide range of ubiquitous tools

available for developing data visuals, the process by which a

strong visual representation of data, or particularly meaningful

systems of such visuals is selected, is far from a simple one. To

complicate matters, the demands on data and visualization are

increasingly multidisciplinary (Meyer et al. 2013). Indeed, in

many instances, these demands are best viewed as interdisci-

plinary, requiring not just simultaneous parallel use but truly

joint utilization and team-wise sensemaking for effective deci-

sions. As with verbal communication, differences in disciplinary

biases and terminology create barriers over which interdisci-

plinary discourse often cannot maneuver. Franco (2013) dis-

cusses these as syntactic, semantic, and pragmatic boundaries

(citing Bechky 2003; Carlile 2004). Breaking down these barri-

ers, or at least figuring out how to work around them, requires

leveraging standard nomenclature for describing visualization

and identifying where the greatest gulf between disciplines exist.

The process further relies on the use of meaningful frameworks

for the selection of appropriate visualization tactics for tasks at

hand and, moreover, organizational cultures in which data visual-

ization as an aspect of analytics is thought of as a continuous

development process, not simply an artifact but a representation

that is the foundation for further critical inquiry. In other words,

purchasing a large high resolution screen to install in a managerial

“war room”to show a map with long-term warehouse capacity

depicted by 3-D bars protruding out of a geographic plane or

curved rendering of the Earth... Not indicative of an effective

organizational approach to visualization (even if dynamic and/or

interactive). Why? Aside from the fact that 3-D bar charts are

Corresponding author:

Elliot Bendoly, Fisher College of Business, The Ohio State University,

2100 Neil Avenue, Columbus, OH 43210, USA; E-mail: bendoly.2@

osu.edu

Journal of Business Logistics, 2016, 37(1): 6–17 doi: 10.1111/jbl.12113

© Council of Supply Chain Management Professionals