Perception Of Congress On The Web

Perception of Congress on the Web: Data Analysis Inspired by the Red Versus Blue United States Map Kayce N. Reed

Abstract

Indiana University

In this paper we discuss a study designed to measure the perceptions of the visual design for political websites. We use as our sample population approximately 400 different websites for United States Congressional office-holders. In the analysis and presentation of our results we use the United States map divided into congressional districts, and then encode the study participants’ perceptions of the visual design of the websites according to a color scale that maps to the perceived favorability of the site. Our motivation for the visualization is, of course, the well known red-blue map that depicts election results according to support for a particular political party. Further, we intend to look at the competitiveness of the districts where congressional elected officials reside, the noticeable features on their websites (such as if they have a blog or visible contact information), as well as the results in relation to the number of years an individual has served in a congressional office. Beyond, developing an understanding of the political landscape, our study suggests a set of heuristics that tend to lead to more favorable perceived website designs.

School of Informatics 901 E. 10th St Bloomington, IN 47408 USA [email protected] Dennis P. Groth Indiana University School of Informatics 901 E. 10th St Bloomington, IN 47408 USA [email protected]

Copyright is held by the author/owner(s). CHI 2008, April 5– April 10, 2008, Florence, Italy ACM 1-xxxxxxxxxxxxxxxxxx.

2

Keywords Information visualization, information communication, political websites, design heuristics

ACM Classification Keywords H5.m. Information interfaces and presentation (e.g., HCI): Miscellaneous.

Introduction As humans, we oftentimes make split second decisions about our perception of the world around us. Our study measures this initial perception that one has when shown a website of a congressional elected official. Inspired by a number of studies conduced in recent years to determine the initial perceived trustworthiness of electronic commerce websites, we intend that our study will help in forming design heuristics for political websites and also lead to an increased insight into general politics online.[1] Recently, Ballew and Todorov research into perception and politics, where rapid judgments based solely on the facial appearance of political candidates predicted the outcomes of gubernatorial elections.[2] This initial perception, if captured, can have an enormous impact on how individuals make decisions about the persons they put into political office (whether in the form of a photograph or in the form of a website).

Hypotheses: Perception of Congress Online Individuals nearly always make an initial judgment of quality when viewing something for the first time. We believe that the time it takes an individual to do this is the same whether it is based on favorable or unfavorable judgments.

H1: The time it takes an individual to make a judgment of the visual quality of a political website is independent of the quality. H2: There is a relationship between the number of years a member of congress has served in office and the perceived quality of that individual’s website. H3: There is a relationship between the competitiveness of the congressional district from which an individual serves and the perceived quality of their website.

Conducting the Study In preparation for the study, we have captured screenshot images of nearly 400 congressional elected officials’ websites. In order to avoid a potential bias in the study, we removed all elected officials from our home state of Indiana (because many of the participants in our study will be residents of Indiana). We also omitted any website that was clearly affiliated with a political party (if there was an elephant or donkey pictured on the website, they were excluded). The study was conducted with twenty participants (ten women and ten men) by showing them twenty random screenshots of congressional elected officials’ websites. Following the viewing of an image, individuals are asked to rank how favorable or unfavorable they find the visual design of the website to be. Pictured below is an example that a participant might encounter.

3

to be used in mainstream media channels. [3] By using this popular format to explain and provide insight into large amounts of data, unique information can also be depicted using similar data analysis methods.

Figure 1: Sample screenshot image

Participants are asked to answer a single question, a simple likert scale from “strongly unfavorable” to “strongly favorable”, once for each of the twenty images shown to them. Each step in the study is timed in order to test our first hypothesis (H1). We end the study with a short survey which asks the individuals how they acquire political news and information and how much time they spend each day doing so.

Data Analysis and Use of Visualizations Mainstream media has capitalized on the effectiveness of the red versus blue United States map since the Presidential elections of 2000 and 2004 to get the attention of audiences and display a large amount of information using a familiar image. When speaking of “red states” versus “blue states”, millions of Americans instantly know what you are speaking of and an image of the United States forms in their minds. This image (whether simply of a map or possibly of a larger social and political separation between the East and West coasts versus the inland sections of the US) continues

Figure 2: a small portion of our results is pictured above

Because of the widespread use and compelling display of the red states versus blue states US map, we have chosen to use a similar presentation method for the results of our congressional study. Different views of the resulting data will be produced all within the United States map divided into congressional districts. Some of the key things that will be shown are the perceived favorability or lack thereof for the congressional elected official. Another view of the map might aim to show the competitiveness of the congressional district, which could provide more insight into the problem space. We would also like to show which congressional districts had specific features within their website that might have made them seem more favorable to the participants (such as the presence of a blog, links to

4

their Facebook account, etc). Preliminary results are shown in Figure 3.

interactive visualizations that allow user to combine datasets to derive composite insights are clearly relevant.

Design Implications Our decision to utilize the geographic representation was principally motivated by easily recognizable relationships. In particular, by relying upon the well understood geography of the United States, we built upon the red-blue map concept [4] with an aim towards discovering the impact of HCI design principles on politicians’ web sites. To the extent that these design principles mirror a larger view of the general populations’ predisposition to support certain candidates, as well as the candidates’ view of the importance of a well-designed web presence, the following insights were discovered: 1. Male study participants performed different than the female participants. Men were more likely to rate a site’s design as unfavorable much sooner than they rated a site as favorable (P<0.03). For women, there was no observed difference in time to determine favorability (P>0.94).

Conclusion and Future Research Following the completion of this study concerning the perception of congress based on the visual design of their websites, additional visualizations will be created to provide insight into the collected data. Further, design heuristics will be developed in response to the results with a goal being to create a set of best practices to follow when designing political websites. There appears to be a lot of exciting research currently going on within hci/d, social data analysis tools and software, amateur created multimedia, using visualizations to gain insight into a particular problem space, just a general welcoming atmosphere for political discourse; all of these things have been the primary motivation for this capstone dissertation within politics and human computer interaction/design.

References

2. Collectively, candidates that were women had better web site designs.

[1] Harrison McKnight, D., Choudhury, V., Kacmar, C. (2002) The Impact of Initial Consumer Trust on Intentions to Interact with a Website: A Trust Building Model. Journal of Strategic Information Systems, 2002.

3. The variation of design quality within states was relatively low, which mirrors the lack of strong competition in many states.

[2] Ballew, C., Toforov, A. (2007) Predicting Political Elections from Rapid and Unreflective Face Judgments. Proceedings of the National Academy of Sciences of the United States of America (2007).

The use of this data and representation for social data analysis is consistent with the datasets existing on the web today, such as Many Eyes.[5] We envision that this work will influence people to ask new questions of political presence on the web by using a common layout with new overlays of information. New

[3] Emanuel, R. (2006) DCCC Red to Blue Program. http://www.dccc.org/news/Red_to_Blue/index.html [4] Gastner, M., Shalizi, C., Newman, M. Maps and Cartograms of the 2004 Presidential Election Results. http://www-personal.umich.edu/~mejn/election/ [5] Many Eyes. IBM Collaborative User Experience. http://services.alphaworks.ibm.com/manyeyes/home