«13TH INTERNATIONAL PUBLIC RELATIONS RESEARCH CONFERENCE “Ethical Issues for Public Relations Practice in a Multicultural World” Holiday Inn ...»
The event was breaking online July 17th, following the July 16th deletion of the books. Just under 2,000 people were notified the books had been deleted and their purchase price refunded. The e-mail read: “The Kindle edition books Animal Farm by George Orwell. Published by MobileReference (mobi) & Nineteen Eighty-Four (1984) by George Orwell. Published by MobileReference (mobi) were removed from the Kindle store and are no longer available for purchase. When this occurred, your purchases were automatically refunded. You can still locate the books in the Kindle store, but each has a status of not yet available. Although a rarity, publishers can decide to pull their content from the Kindle store.” People were not happy with this vague statement and wanted clarification. Customers sought clarification with posts on the Amazon.com discussion boards under the folder heading “Mysterious George Orwell refunds.” These comments helped to create the need for the crisis response.
The Kindle Crisis communication unfolded in two steps. The first step was a statement issued by Amazon.com spokesperson, Drew Herdener: "These books were added to our catalog using our self-service platform by a third party who did not have the rights to the books. When we were notified of this by the rights holder, we removed the illegal copies from our systems and from customers' devices, and refunded customers. We are changing our systems so that in the future we will not remove books from customers' devices in these circumstances." This statement was delivered to traditional and online media outlets on July 17th, shortly after the deletions. Although Herdener’s statement did not appear online at the Kindle Community site, bloggers and traditional media outlets used his statement in their discussions of the Kindle Crisis. Herdener’s statement was primarily a form of instructing information designed to explain the actions taken by Amazon.com (Sturges, 1994). Compensation and a suggestion of corrective action appear in the message as well.
The refund was offered (compensation) and Amazon.com promised not to repeat such remote deletions in the future (corrective action).
The second statement was an apology from Amazon.com CEO Jeffrey Bezos on July 23, 2009, at 12:16 PM (PDT):
“This is an apology for the way we previously handled illegally sold copies of 1984 and other novels on Kindle. Our "solution" to the problem was stupid, thoughtless, and painfully out of line with our principles. It is wholly self-inflicted, and we deserve the criticism we've received. We will use the scar tissue from this painful mistake to help make better decisions going forward, ones that match our mission.
With deep apology to our customers, Jeff Bezos Founder & CEO Amazon.com The message was posted to the Kindle Community discussion board on Amazon.com. The primary users of the discussion board are Kindle users and/or those interested in learning more about Kindles.
However, anyone who joins the Amazon.com discussion boards could read and post to that area.
Given the deletions occurred several days before, it is unclear why Bezos elected to post at that time.
However, some speculated the timing of the posting was related to earnings reports that were to appear around that time. Regardless of the reason, the response was slow in Internet time. As noted earlier, a late apology is still better than no apology at all.
At this point we have a descriptive analysis of Kindle crisis response. Based upon existing research, the Amazon.com communication could be judged effective. There was a progression from 212 instructing and adjusting information followed by reputation repair through an apology. However, such descriptive analysis is speculative in nature. We are positing that online reactions provide a source of evidence-based assessment of crisis communication effectiveness.
Examining online reactions to this online apology from Amazon.com’s CEO is an important step forward in the research on stakeholder reactions to apologies. The online environment enables us to analyze “concrete” responses/reactions that are in direct response to the crisis communication.
It also allows us to explore if people are willing to voice their support or rejection of the crisis response. In addition, it permits us to examine their reasons for supporting or rejecting the crisis response. Finally, it enables us to see if those posting report the incident and the subsequent crisis response will affect their purchase intentions and word-of-mouth communication, two potentially important outcomes of the handling of the crisis.
Research Questions The goal of this exploratory research was to examine the Kindle Community discussion board reactions to the July 23, 2009 online apology from Jeffrey Bezos. The following research
questions were posed:
RQ1: How did the Kindle Community postings indicate the apology was received by this community?
RQ2: How did the Kindle Community postings indicate the apology affected the behavioral intentions of these writers?
RQ3: How did the Kindle Community postings indicate the apology affected the word-ofmouth communication intentions of these writers?
RQ4: Did reactions to the apologies differ for Kindle owners and Kindle non-owners?
RQ5: What suggestions for additional corrective action appear in the posts?
Method Procedure The data for this investigation were 210 relevant responses that appeared on the Kindle Community discussion board in the two-day period following Bezos’s apology on July 23 at 12:16 pm PDT. Each post was treated as the unit of analysis. The Kindle Community provides an appropriate forum for the analysis of reactions to the apology because participants are Kindle users, potential Kindle users, and/or Amazon.com customers. By posting on this site, Bezos indicated this is the intended audience for the apology and therefore responses appearing on the Kindle Community are appropriate data for our analysis. The first Kindle Community post was stamped one minute after the time stamp of Bezos’s apology post, indicating some rapid reponses from the Kindle Community.
In order to be included in the data set the posting had to indicate it was a response to the apology. Examples of phrases signaling it was a response include: “That took a lot of courage,” “Thank you for posting,” “Amazing apology,” and “Actually, I don’t think an apology was needed.” Postings unrelated to the apology and postings indicating they were responses to other people’s postings also were excluded from the data set because we wanted to examine only online comments that were reactions to Bezos’s apology. Of the 210 posts, 65.7% (n = 138) followed Bezos’s apology on the the 23rd and 34.3% (n = 72) appeared on the 24th.
Coding categories Four categories of information in the responses pertained to our research questions and were coded for the analyses. First, the writers’ reactions to the apology were coded into one of three mutually exclusive response categories: (1) acceptance of the apology (e.g., the writer indicated s/he accepted the apology through comments that praised the apology or Bezos); (2) conditional acceptance of the apology (e.g., these comments took the form of “yes… but…”; posts indicate the apology was appreciated but lacking in some way. For example, the writer indicated the apology was accepted but the problem was not solved or it was nice that an apology was posted 213 but the books were still deleted); or (3) rejection of the apology (e.g., the writer indicated s/he rejected the apology outright or the apology was unacceptable because it was insufficient in some way). These reactions are important because they signal the extent to which message posters accepted the apology and are an indication crisis response effectiveness.
Second, message contents were coded for statements of behavioral intention about purchase behaviors. The purchase behavior categories were: (1) positive behavioral intention (e.g., will remain a loyal customer, will continue to buy e-books, intend to purchase Amazon.com products);
(2) negative behavioral intention (e.g., will no longer use the Kindle or purchase e-books from Amazon.com); or (3) no purchase-related behavioral intention.
Third, word-of-mouth communication intentions were coded. The word-of-mouth communication categories included: (1) positive word-of-mouth (e.g., will communicate favorable messages to others about the Kindle, will endorse the Kindle to others, will recommend e-books from Amazon.com); (2) negative word-of-mouth (e.g., will say negative things about Amazon.com and/or about the Kindle, will tell others not to buy a Kindle or Amazon.com products); and (3) no word-of-mouth communication intention.
Fourth, postings were coded according to whether the writers included a statement about Kindle ownership. Two categories were used: (1) stated they owned a Kindle or (2) did not mention they owned a Kindle.
Fifth, messages were coded for the inclusion of suggestions for additional corrective action.
The four categories were: (1) no call for corrective action; (2) a call for corrective action as a solution to this specific situation (e.g., provide free copies of the deleted Orwell books); (3) a call for corrective action to address the larger issue of the user agreement and/or terms of service that allows Amazon.com to control Kindle user’s libraries of e-books (e.g., calling for an end to Amazon.com’s ability to delete books, or to end DRM or digital rights management); and (4) a general call to replace “illegal” e-books with legal copies of the book (e.g., how “illegal e-book” problems will be handled in the future).
Two coders coded the Kindle Community postings. The coders reached 95% agreement on the first three categories, 100% agreement on the fourth category, and 88% agreement on the fifth category. Coding discrepancies were then resolved through discussion.
(1, N = 188) = 51.08, p.001), and no behavioral intentions stated and negative behavioral intentions, (1, N = 188) = 51.08, p.001). Overall, the majority of the reactions to the apologies did not include statements of behavioral intentions. But when statements of behavioral intention were included, twice as many reflected positive behavioral intentions than negative intentions.
Word-of-Mouth Communication References to word-of-mouth communication were examined for RQ3. The distribution of the word-of-mouth communication variable revealed the great majority (96.2%, n = 202) of the reactions to the apology did not mention an intention to engage in word-of-mouth communication.
Only 8 (3.8%) postings mentioned either a positive (n = 5) or negative (n = 3) word of mouth intention. Due to the small n sizes, no statistical analyses were performed.
The Janis-Fadner Coefficient of Imbalance was used to examine RQ4. The Janis-Fadner Coefficient of Imbalance was originally developed to detect bias in news media coverage but has been extended to other types of messages and has been used to evaluate organizational reputations (Deephouse & Carter, 2005). The limitation is that the data must be consistent with the tri-part structure of the formula (Hurwitz, Green, & Segal, 1976). The Janis-Fadner Coefficient of Imbalance utilizes three coding categories: (1) favorable; (2) unfavorable; and (3) neutral. When the numbers are entered into the formula, a value ranging between +1 and -1 is created. Values near zero indicate balanced treatment in the messages. The Janis-Fadner Coefficient of Imbalance is designed so that it will always: (1) increase when the frequency of favorable content increases; (2) decrease when the frequency of units of unfavorable content increases; (3) equal zero if the units of content are balanced/neutral; (4) equal zero if the numbers of units of favorable content are equal to the number of unfavorable content (Janis & Fadner, 1943). The Janis-Fadner Coefficient of Imbalance is superior to simple proportions because the resulting value: (1) provides an assessment of the strength and direction of the imbalance and (2) is standardized so that researchers can compare the scores from different data sets. For instance, the Janis-Fadner Coefficient of Imbalance for two different media channels could be compared or messages from different stakeholder groups could be compared.
For RQ4, we compared comments posted by self-identified Kindle owners (n = 61) and nonowners (n = 25). The 124 posts that did not explicitly reference Kindle ownership status were excluded. For reactions to the apology, the Janis-Fadner Coefficient of Imbalance for owners was.53 compared to a score of.09 for non-owners and.40 for the entire sample. For behavioral intention, the Janis-Fadner Coefficient of Imbalance for owners was.11 compared to -.20 for nonowners and.02 for the entire sample. Word-of-mouth communication was not analyzed given the small number of posts related to that topic. The data indicate that those who own Kindles were more accepting of the apology and indicated positive behavioral intentions. Non-Kindle owners were near the neutral point for accepting the apology and indicated they would not purchase or use a Kindle.
These findings are consistent with research that demonstrates positive feelings towards an organization and/or product affect reactions to crises (e.g., Pullig, Netemeyer, & Biswas, 2006).
One could argue that the non-Kindle users were irrelevant to the crisis response and that Amazon.com rightly chose to focus on retaining current users. However, the non-Kindle owners were still Amazon.com customers and many indicated they had considered buying Kindles but the crisis reinforced their choice not to buy or triggered a decision not to buy a Kindle. Although the non-Kindle owners are still relevant to the crisis response because they are potential owners, they would be a lower priority than retaining customers. Even though the Janis-Fadner Coefficient of Imbalance score was negative, there were non-Kindle owners who indicated they still intended to purchase a Kindle.