A Study on Graphical Representation

How to cite

Pei, Zhuan; Matsudaira, Jordan; Shen, Yi; Lieberman, Carl; Korting, Christina. 2019. "A Study on Graphical Representation." AEA RCT Registry. AEARCTR-0004331. https://doi.org/10.1257/rct.4331-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Zhuan Pei* , Jordan Matsudaira , Yi Shen , Carl Lieberman , Christina Korting

* Primary investigator

Registry ID
AEARCTR-0004331
DOI
https://doi.org/10.1257/rct.4331-1.0
First registered
Last updated
Status
On going
Trial dates
2018-11-14 – 2019-12-31
Countries
Private
JEL codes
C1
Registry record
http://www.socialscienceregistry.org/trials/4331

Abstract

Regression discontinuity design (RDD) has gained traction across social science fields as a leading quasi-experimental strategy to assess policy impacts. Graphical representation is to a large extent responsible for this growing popularity, and it is now an integral part of any well-executed RDD study. While there is statistically-based guidance on RDD graphical representation, the success of a graphical technique should ultimately lie in the correct perception of policy impacts by the reader of the graph. To date, no research has been conducted to study the role of different RDD data visualization techniques on reader perception. This project fills this gap: it aims to 1) analyze how data visualization techniques affect perception by varying exposure to an array of graphical parameters in a randomized experiment, and 2) use findings from the experiment to establish a set of graphical guidelines for practitioners. Visual representation in the related regression kink design (RKD) will be similarly studied. The findings of this project will contribute to a broader literature on data visualization and serve to create a framework for
evaluating graphical practices in the social sciences.

Intervention

Our treatments correspond to graphical representation choices.

Experimental design

Subjects are randomized into different graphing treatments, shown a series of regression discontinuity or regression kink graphs, and are asked to classify whether a discontinuity or kink exists.

Design details

Randomization method
Between subject randomization by a computer for regression discontinuity study Mixed-design (within and between subject randomization) by a computer for regression kink study
Randomization unit
Individual
Clusters
we do not cluster sample
Observations
Nonexpert sample: 5 rounds of experiments; overall 1,632 unique surveys initiated (a subset of the subjects participated in more than one round of the experiment) Expert sample: Will send study link via email to up to 2,500 individuals. Eye tracking study: up to 50 subjects
Arms
Nonexpert sample: 88 individuals per treatment arm for regression discontinuity studies 72 individuals per treatment arm for regression kink studies Expert sample: Not subject to fixed sample size: depending on the response rate
Primary outcomes
Binary variable indicating whether a graph was correctly classified by the participant.
Secondary outcomes
Binary variable indicating the participant's stated level of confidence (from bonus choice). Time spent per question. Stated magnitude of discontinuity (for expert sample only).

Cited by

No citation data available for this registration yet.

Keywords