Using Content Analysis to Study Affect in Social Evaluations
Mike Pfarrer University of Georgia
Behavioral Strategy & Entrepreneurship Workshop University of Tennessee & SMS September 27, 2013
Why Use CA to Study Social Evaluations?
A behavioral approach to strategy eye-of-the-beholder research Bridge the gap between large-sample archival research (external validity) and small-sample, primary research (internal validity) Open the black box Social evaluations and affect
Tenor of media texts Reputation, celebrity, legitimacy, stigma Impression and crisis management
Who Started It?
Political Science and Communication
Lippman (1922): Public Opinion Janis and Fadner (1943): propaganda McCombs & Shaw (1972): agenda setting
Organization and Management Studies
Deephouse (2000): Media tenor Pollock and Rindova (2003): volume, legitimacy
And Now?
How Can You Measure Affect?
LIWC is a text analysis software program that measures the rate at which authors use positive or negative emotion words (Pennebaker et al. 2007).
Category
Positive affect
Negative affect
Examples
success, value
hurt, loss
Words
406
499
IRR
.97
.97
LIWCs dictionary has over 900 affective words
Who Is Studying It?
Pfarrer, Pollock, & Rindova. 2010. Academy of Management Journal.
Examine the effects of reputation and celebrity on:
1. 2. The likelihood of an earnings surprise How investors react to these surprises
Contributions:
1. 2. 3. High positive affect distinguishes celebrity from reputation Visibility alone is not sufficient The simultaneous possession of both is rare
Firm Celebrity: What Is It?
Celebrity as a Social Approval Asset
High level of public attention combined with positive emotional responses from stakeholder audiences (Rindova et al., 2006: 51)
Eye of the beholder social evaluation Derived from non-conforming (deviant) behavior Accrues benefits to the firm: its an asset So, fame/popularity still important, but theres more!
How Did We Measure It?
LIWC
Celebrity = high visibility and high positive affect Dramatic narratives in BusinessWeek 42,000+ articles Positive/total affect measure
There's a new generation of brands, including Amazon.com, eBay, and Starbucks, that have amassed huge global value with little traditional advertising.
But risks remain. Starbucks Corp. Chairman Howard Schultz is bracing for a boycott that could hurt his European expansion.
Who Else Is Studying It?
Zavyalova, Pfarrer, Reger, & Shapiro. 2012. AMJ.
What we did
1. 2. 3. Toy recalls; CPSC press releases IM: 5,500 press releases Newswires, hand-coded Tenor: 38k articles & web blogs Lexis-Nexis, LIWC
Contributions:
1. 2. 3. IM depends on direct guilt or by association Spillover yet also safety-in-numbers effect What about Janis-Fadner?
How Did We Measure Affect?
Coding Texts with LIWC
Positive = affective content was at least 66 % positive Negative = affective content was at least 66% negative Alternative measures Multiple firms: hand coding (10%); law of large numbers\ Negative diagnosticity
Janis-Fadner vs. new developments
Equal weighting of positive and negative articles High variance in coverage; loss of sample size Weighting of negative articles Overall positivity of business press
Related Research
1. Bermiss, Zajac, & King. 2014. Under construction: How commensuration and management fashion affect corporate rankings. Organization Science. 2. Bednar 2012. A behavioral view of the media as a corporate governance mechanism. AMJ. 3. Bednar, Boivie, & Prince. 2012. How media coverage influences strategic change. Organization Science.
4. Deephouse, Elsbach, Jones, Glynn, Kennedy, King, Misangyi, Phillips, Reger, Short, Weber, Westphal
Whats Next?
1. Opening the black box: measurement, validity, reality?
2. Content analysis techniques examining affectual content and social evaluations: Q score, JF coefficient
3. Continuous improvements in software
There is plenty left to do!
Content Analysis Website: http://www.terry.uga.edu/contentanalysis
