Celebrity Apologies

How to cite

Ho, Benjamin. 2022. "Celebrity Apologies." AEA RCT Registry. AEARCTR-0009812. https://doi.org/10.1257/rct.9812-1.0

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

Benjamin Ho*

* Primary investigator

Registry ID
AEARCTR-0009812
DOI
https://doi.org/10.1257/rct.9812-1.0
First registered
Last updated
Status
Completed
Trial dates
2022-07-28 – 2022-08-06
Countries
United States of America ()
Registry record
http://www.socialscienceregistry.org/trials/9812

Abstract

A growing literature in economics and other disciplines have studied the role of apologies in restoring the bonds of trust. Except for a few exceptions, most of these studies have been done in lab settings. In this study, we look to well publicized celebrity apologies. We collect a sample of celebrity apologies based on two online databases: a New York Times dataset of celebrities who were publicly reprimanded during the #metoo movement and a dataset compiled by the online apology blog SorryWatch who noted and analyzed a number of celebrity apologies in the recent decades. We use online raters to categorize and assess the effectiveness of these apologies. (We also will train a machine learning model based on the rating data to automate future apology classification.) We then match the categorized apology data to twitter data where we use sentiment analysis to analyze the sentiment (either positive or negative) of tweets that mention the celebrity in the time periods before the violation, after the violation but before the apology, after the apology, and in the years after the apology.
This research will examine real world impacts of apologies, and importantly tie the content of the apology to the outcomes. We classify apologies based on a game theory model from Ho (Mgmt Sci 2012). Recent related work examining real world impacts of the apology using that classification can be found in field experiments involving Uber customers (Halperin, Ho, List, Muir EJ 2022) and in stock market event studies (Fan, Ge, Ho, Mal Ssrn working paper, 2021).
Our research also explores new methods to use machine learning to classify apologies, and to assess their real world impacts.

Intervention

Using the Mturk sample we will ask workers to classify the apologies into the categories described in Ho (2012), perform an attention check, and then ask questions about the perceived warmth and competence of the apologizer at random hypothetical time intervals after the apology was tendered.

Experimental design

Using the Mturk sample we will ask workers to classify the apologies into the categories described in Ho (2012), perform an attention check, and then ask questions about the perceived warmth and competence of the apologizer at random hypothetical time intervals after the apology was tendered.

The primary purpose of this HIT is to classify the apology.. But the secondary purpose is the experimental manipulation where we randomize the hypothetical time period that has passed since the apology was tendered and ask subjects to report how their attitudes toward the apologizer has changed as a function of the time that has passed and the type of apology.

Design details

Randomization method
computer. using mturk.
Randomization unit
individual
Clusters
220 apologies, each will be shown to 12 different workers divided among 4 treatment dates
Observations
2700 HITs
Arms
660 by each date
Primary outcomes
Mturk worker assessment of apologies, and their effects.

Cited by

No citation data available for this registration yet.

Keywords