Graduation Year

2022

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Business Administration

Major Professor

Dr. Uday Murthy, Ph.D.

Committee Member

Thomas Smith, Ph.D.

Committee Member

Kristina Demek, Ph.D.

Committee Member

Clinton Daniel, DBA

Keywords

cognitive fit, cognitive load, data visualizations, gestalt principles

Abstract

The usefulness of accounting data is directly dependent on the presentation and interpretation of the data (DeSanctis and Jarvenpaa 1989; Brown-Liburd et al. 2015; Cao et al. 2015). While both corporate accounting and public accounting have a large focus on data analytics and visualizations, accountants in both fields have reported to have minimal experience creating visualizations and observe a growing use of visualizations within their role (Krumwiede 2019; Buchheit et al. 2020; Gibson et al. 2020). Vessey (1991) finds that the appropriateness of a visualization can impact a user’s decision-making performance. Although prior studies have provided suggestions of when to use a specific graph type, there is no empirical evidence regarding the relation between the use of specific graph types and analytical performance.

An analysis of Fortune 100 annual reports revealed that a variety of different graph types are used to display accounting data, with the most common graph types being line graphs, pie graphs, and stacked column graphs. Bar graphs and line graphs are predominantly used to display comparison data over time while pie graphs and stacked column graphs are predominantly used to display compositional data. Little is known, however, about the relative efficacy of different graph types to display different kinds of accounting data. There are suggestions in the practitioner literature that some graph types are more optimal than others for certain kinds of data, however these suggestions have not been subjected to rigorous empirical testing. In this study, I examine if there is a difference in decision efficiency or effectiveness when given either a bar or line graph for a comparison task and when given either a pie or stacked column graph for a compositional task. I examine both tasks in a setting of high component complexity and low component complexity to investigate if the performance differences are present in both instances. I also investigate if a user’s cognitive load when completing a task mediates the relation between graph type and decision-making performance.

From the results of this study, I find that line graph users have significantly greater decision sensitivity (ability to identify true positives) than bar graph users when completing comparison tasks of either high or low component complexity. I also find that accounting professionals have significantly greater decision sensitivity than graduate students. Interestingly, while I find that line graph users have greater decision sensitivity, I also find that bar graph users have significantly greater decision specificity (ability to identify true negatives) and overall accuracy. For the comparison task, I find no difference in decision efficiency between line graph users and bar graph users. For the compositional task, I find no differences in decision efficiency or effectiveness when comparing the performance of pie graph and stacked column graph users. Lastly, I do not find evidence to suggest that cognitive load mediates the relation between graph type and performance on accounting data analytic tasks.

Included in

Accounting Commons

Share

COinS