Communication Portfolio Assignment 7

Journal of Business Communication 50(4) 362 –382

© 2013 by the Association for Business Communication Reprints and permissions:

sagepub.com/journalsPermissions.nav DOI: 10.1177/0021943613497053

jbc.sagepub.com

Article

Effects of Directness in Bad- News E-Mails and Voice Mails

Frank Jansen1 and Daniel Janssen1,2

Abstract In this study, we explore the effects of channel choice (e-mail vs. voice mail) and message structure (direct vs. indirect) on the receiver’s perception of bad-news messages. We conducted an experiment in which bad-news e-mails and voice mails were presented to participants who evaluated their response to the messages via a questionnaire. The results indicate that e-mail is more comprehensible, while voice mail is more persuasive and effective for maintaining a personal customer relationship. Furthermore, messages with an indirect structure (explanation → bad news) are valued more highly than direct messages (bad news → explanations). We also found interaction effects of channel and structure, the most important being that the preference for the indirect structure is limited to e-mails.

Keywords medium, channel, e-mail, voice mail, direct/indirect message structure, readers’ evaluations

Introduction

In the early summer of 2010, Apple’s iTunes store received a complaint from one of their customers, Eva. Eva’s daughter had downloaded a—supposedly—free applica- tion for Eva’s iPhone. However, a couple of days later, Eva found her credit card had been billed for €109.99. She decided to send an e-mail to iTunes’ customer service:

1Utrecht University, Utrecht, Netherlands 2University of Antwerp, Antwerp, Belgium

Corresponding Author: Daniel Janssen, Communication Studies, UIL-OTS, Utrecht University, Domplein 29, 3512 JE Utrecht, Netherlands. Email: [email protected]

497053 JBC50410.1177/0021943613497053Journal of Business Communication[END SUBLIST][END SUBLIST][END SUBLIST]Jansen and Janssen research-article2013

Jansen and Janssen 363

My daughter has downloaded a free game on my iPhone (Fishies) and now my credit card will be charged for €109.99 euro. Now I see on my account that the same has happened with Blowfish (€89) which was supposed to be a free download as well. So what’s going on here?

iTunes responded within 24 hours with the following e-mail:

Hi Eva,

My name is Sarah from Apple iTunes Support. I understand that your daughter purchased “Fishies” and you were charged €109.99 when it was supposed to be free. I am sorry for any frustrations this may have caused you, but I assure you I am more than happy to look into this further for you.

I have reviewed your purchase history and confirmed the actual purchase of the app “Fishies” was free; however, it was the purchase of “Chest of 1950 Pearls” which cost €109.99. This is what is called an “in app” purchase. This means that your daughter made the purchase within the app (or within the game). Sometimes, in order to advance to the next level in these games, you need to make an app purchase, which in turn, costs money.

( . . . )

Seeing your iTunes Store purchase history and order numbers ( . . . )

I have refunded you in full for this purchase.

You will see a credit of €109.99, plus any applicable taxes, in three to five business days. ( . . . )

I appreciate your understanding. ( . . . )

Take care,

Sarah

iTunes Store Customer Support

This e-mail conversation is interesting for a number of reasons, two of which are the expansion of the e-mail channel1 and the rise of new conventions and subgenres (cf. Baron 2000). It also illustrates how business and business communication have devel- oped. People buy goods online from virtual stores, pay electronically with their credit card, complain by way of e-mails, receive electronic (and sometimes even automatic) e-mail responses, and—in this case—get their money credited back to their bank or credit card account. And all this takes place in just a matter of hours. This also applies to other interactive channels such as texting and Twitter, which use the written medium, and voice mail2 that uses the oral medium. Not too long ago, customers in a

364 Journal of Business Communication 50(4)

situation like Eva’s had to write a complaint letter and would then have to wait to receive a letter back; this correspondence would have taken weeks.

In many instances, e-mails have simply replaced letters. People apply for a job not by sending a letter of application or requesting a form, but by sending an e-mail or filling in a web form. The same people may get rejected not by a letter, but by an e-mail. The same goes for many other forms of business communication: new chan- nels replace old channels and old genres rematerialize within those new media. This is a gradual development that has been called “remediation” in media studies (see, e.g., Bolter & Grusin, 1999). E-mail may well be considered the most important new chan- nel just because of its volume. As early as 2007, the research firm IDC showed a dra- matic increase of e-mail volume worldwide, rising from 31 billion per day in 2003 to a whopping 97 billion per day in 2007. More recent data from 2008 shows that the number of e-mail users has risen to 1.3 billion, while the numbers of e-mails that were sent, reached an estimated 210 billion.

The advantages of e-mail are numerous and often mentioned. E-mail is fast and cheap and readily available for almost everyone. Organizations and their customers can communicate all over the world at the speed of light virtually for free. Seen in this light the success of e-mail is understandable. Sarah’s e-mail shows how powerful this channel can be.

But in a way Sarah’s task was not too difficult. She had to convey positive news to Eva and positive news does not put much constraint on media-choice. If you win the lottery, you do not care whether people inform you by e-mail, telephone, letter, or whatever other channel. This situation may be totally different in the case of bad-news communication (Timmerman & Harrison, 2005). Would Eva be as content with an e-mail if iTunes had not refunded her or would a more “personal” medium have been more effective in that situation? The question then is: What are the company’s other options? Of course a face-to-face meeting would be the best choice. All the empirical evidence shows that mediated channels is no match for face-to-face conversations (Baltes, Dickson, Sherman, Bauer, & LaGanke, 2002). But in a modern-day business world face-to-face meetings are not always feasible. That leaves two other possibili- ties, namely telephone and voice mail.

In this study we compared e-mail to voice mail. We opted for voice mail because voice mail and e-mail are relatively new media. Furthermore, they share an important characteristic: voice mail and e-mail are noninteractive and asynchronous channels. Apart from this, they differ in one important dimension: e-mail is a written medium, voice mail a spoken medium. In the next section, we will argue why a spoken medium may be more effective for bad-news communication.

In this study, we focus on one genre, namely bad-news messages, and on two chan- nels, namely e-mail and voice mail. Our main goal is to explore which of the two channels is more effective for delivering unwelcome news to customers.

Furthermore, we are interested in the effects of structural directness on the recep- tion of bad news. Giving an explanation for the bad news in a conversation is more effective when implemented in a prophylactic way (before the bad news) than as a remedy (after the bad news). The same applies to written bad-news messages in

Jansen and Janssen 365

e-mails. Earlier research has shown that readers prefer to read the explanation before the bad news (Jansen & Janssen, 2011). The question is, however, if listeners to bad- news messages in voice mails have the same preferences.

In the next section, we will elaborate somewhat on the results of theoretical and empirical studies of channel choice and message structure in order to give the research questions and hypotheses in the next section a solid ground. In the Method section, we address the methodological issues for this study and explain how our experiments were conducted. In the Results section, we present the outcomes; this is followed by the Conclusion and Discussion section.

Bad News, Directness, and Media Choice

In this section, we will give account for our decision to focus on bad-news messages. After that, we will compare the two channels that we used in our experiments—e-mail and voice mail—in greater detail. Finally, we will explain why we added structural directness as a second independent variable.

Bad News

One of the major challenges in business writing is to communicate a clear message while keeping the client happy and preserving the image of the organization (Bovée & Thill, 2000; De Jong et al., 2002; Janssen, 2007). It goes without saying that commu- nicating good news, news that is favorable for the receiver, is a simpler task than com- municating bad news, news that conveys information with unfavorable consequences for the recipient, for example, denying an insurance claim or rejecting a job applicant. Good communication calls for a cooperative attitude from senders and receivers. However, in bad-news communication the interests of the speaker and the senders conflict which makes cooperation difficult and communication more complex. Also, the demand for clarity in the message may easily conflict with the need to preserve the organization’s image and the relationship between the organization and its customer. The proper balance between clarity and good relations is an important theme in (lin- guistic) pragmatics, especially in politeness theory (Brown & Levinson, 1987).

Furthermore, bad-news communication induces an interesting conflict of a more psychological nature. Timmerman and Harrison (2005) observe that managers often dodge tasks that entail conveying bad news to their subordinates (the so-called MUM effect). And if not, they are inclined to distance themselves from the news and the receivers by choosing a channel that masks their lack of involvement with the subor- dinates, a strategy that is also predicted by the hyperpersonal model (Walther, 1996). It is exactly for this reason that e-mail is often the medium of choice for bad-news communication: avoiding the receiver. Although Timmerman and Harrison suggest that managers should resist these all too human inclination, research tells us that many senders prefer e-mail for delivering bad news (cf. Sproull & Kiesler, 1986). What we do not know is how medium choice in bad-news communication affects the receivers.

366 Journal of Business Communication 50(4)

Handling bad-news communication is something every communication specialist should master. For that reason, bad-news or crisis communication has become an essential part of most courses in business communication all over the world. From a research perspective, it is thus essential to study effective and less effective instances of bad-news communication and the strategies that are used in it (Jansen & Janssen, 2010).

Channel Choice: E-Mail Versus Voice Mail

In this paragraph, we will discuss the relevant differences between the channels e-mail and voice mail. This discussion will lead to two hypotheses: (a) e-mails are more com- prehensible and (b) voice mails are more personal.

If we compare voice mail and e-mail, we see similarities and differences. E-mail and voice mail are both relatively new channels, made possible by the computer and its electronic infrastructure. The most fundamental difference between the two chan- nels is that they capitalize on different media: e-mail on the written medium and voice mail on the spoken medium, which entails that receiving means reading for e-mail and listening for voice mails. The spoken medium is time-bound and the written one space- bound (Crystal, 2001).3 Listeners are, in a way, hostages of the speaker. They have no other option than to submit themselves to speakers, and to process their words one after the other. The permanence of texts gives readers more freedom of choice: They may access the parts of the message in the order in which they appear in the text, or they may not. They may peek ahead, reread (difficult) parts of the text, and so on. By consequence, it is more difficult to understand a complex message in speech than in writing (see Chaiken & Eagly, 1983; Unnava, Burnkrant, & Erevelles, 1994). The field study by El Shinnawy and Markus (1997) corroborates these findings. El Shinnawy and Markus report that their participants prefer e-mail when the messages contain, for example, numerical data that are difficult to remember correctly. Therefore, we expect the e-mails in our experiment to be more comprehensible than the voice mails.

The next difference between the two channels has to do with the possibility to transmit vocal cues. Spoken language conveys vocal cues; written language does not. Vocal cues are informative in their own right. They inform the reader about stable characteristics of the sender such as sex and age, and temporary ones such as his or her sincerity and emotional state (Scherer, 2003). It is highly probable that receivers appreciate this information because it helps them evaluate the interpersonal aspects of the message, especially in bad-news communication.

Apart from the possible direct relevance of vocal cues, they are also indirectly rel- evant: The voice contributes to a feeling of copresence and it raises the risk of leakage of unwanted emotions.

Written and oral media also differ to the extent that can convey a feeling of copres- ence. “Co-presence” and “social presence” are the terms for the receiver’s perception of the sender’s presence in the communicative situation (Keil & Johnson, 2002; Rice, 1993). When receivers hear the sender’s voice, they will experience him or her as more present and thus more involved in the interaction than when they read a text.

Jansen and Janssen 367

There is not much empirical evidence for these claims, but research does show that e-mailers are more self-absorbed and less interested in or empathic with the receiver. For example, Watts Sussman and Sproull (1999) found less sugarcoating in bad news transmitted by e-mails than in face-to-face and telephone conversations (see also Sproull & Kiesler, 1986, and Kruger, Epley, Parker, & Ng, 2005). We may thus hypothesize that receivers appreciate a copresence affording channel (voice mail) more than a channel preventing it (e-mail).

Vocal cues are not only a chance for senders to communicate more effectively; they form a potential risk as well. In the previous section, we mentioned Timmerman and Harrison (2005) who concluded that the most suitable channel for the receiver (in terms of “interactional justice”) should be the one that is most transparent in convey- ing potentially ambiguous information about the sender’s sincerity and involvement. Receivers will wonder: Is the sender of this bad-news message sincere for instance when he expresses his sympathy? When the listener senses those “emotional leaks” he or she will be sympathetic toward the sender and more inclined to accept the bad news.

Finally, the production process of the message in the two channels differs. The production of even a short message presumes that the sender takes a series of steps: generate, formulate, and articulate the message (Levelt, 1989). For a literate person this process is easier in the written medium than in the oral medium. The permanency of writing enables him to divide the composition process in parts: making a draft first, rethink the text, and revise it when necessary. This process results in a polished pre- sentation where all production problems have become invisible in the final product. Producing an immaculate voice mail is more difficult than writing a flawless e-mail (Dingwall 1992). In spite of all the possible preparations (planning, making a draft on paper), during the final act of speaking much can go wrong with hesitations, pauses, restarts, grammatical errors, and so on as a result.

On the other hand, in everyday conversation we are very tolerant toward errors and mistakes in oral communication. And the fact that receivers know how difficult it is to leave a decent voice mail may influence their perception. They may evaluate voice mail messages more positively than comparable e-mails.

At the end of this paragraph we summarize the differences between e-mail and voice mail that may be relevant for the receiver in Table 1.

Table 1. Differences Between E-Mail and Voice Mail.

E-mail Voice mail

Communication mode Writing/reading Speaking/listening Access Divers Only linear Vocal cues Absent Present Copresence Low High Emotional transparency Low High Message production Easy Difficult

368 Journal of Business Communication 50(4)

Direct and Indirect Structure

As mentioned earlier, we focus on “directness” as an element of structure in bad-news messages. A classic distinction in business communication textbooks on bad-news messages is that between direct and indirect structures (Bovée & Thill, 2000; Janssen, 2007; Jansen & Janssen, 2011). In the indirect approach, the writer buffers the news by presenting reasons or explanations before the bad news. In the direct approach, the writer presents the bad news first and the reason later.

The direct structure may have a positive effect on the readers’ comprehension of the message. Britt and Larson (2003) experiments showed that complex sentences (consisting of an embedded clause with a claim and a causative clause introduced by the connector because), in which the claim was presented first were read faster than those in which the reason was presented first. Furthermore, they found that claim-first arguments were recalled more accurately and in the exact order in which they had been presented than reason-first arguments.

Jansen and Janssen (2011) found the indirect structure to be more effective when it comes to persuasion and effects on image. They found no differences in comprehen- sion. Indirectness seems to make a direct appeal to the rationality of both the sender and the receiver by giving the reader the opportunity to closely follow the reasoning of the writer.

Jansen and Janssen (2011) came to their findings by experimenting with letters and e-mails, written media. It is still unclear whether an experiment with an oral medium like voice mail would lead to the same results. Unnava et al. (1994) found an effect of presentation order in their experiments with oral (radio) and written adver- tisements, a primacy effect of the first argument in the radio condition. They explain this effect by stating that that the listeners process and remember the first argument and use it as a frame for the remainder of the arguments. In the written condition the readers did not have to capitalize on order because they could reread the text, skip parts, and so on.

Although there are substantial differences between our experiment and Unnava et al.’s we expect that directness has more effect in voice mail than in the e-mail.

Research Questions and Hypotheses

Based on our review of relevant theories and research, the following research ques- tions and hypotheses can be formulated:

Question 1: Which channel, e-mail or voice mail, do receivers of bad-news mes- sages prefer? Hypothesis 1: E-mail (messages) will be evaluated as more comprehensible

than voice mail (messages). Hypothesis 2: Voice mail (messages) will be evaluated as more personal than

e-mail (messages).

Jansen and Janssen 369

Question 2: Which message structure do the receivers prefer: direct or indirect? Hypothesis 3: The direct structure (bad-news → explanation) will be evaluated

as more comprehensible than the indirect structure (explanation → bad news).

Hypothesis 4: The indirect structure will be evaluated as more persuasive and more personal.

Question 3: Is the preference for a structure independent from the media? In other words, is there an interaction between structure and media? Hypothesis 5: The positive effect of the indirect structure will be more promi-

nent in voice mail than in e-mail.

Method

Material

We created seven topics for bad-news messages in a business context. The gravity of the bad news was in all cases intermediate, from a cancelled trip via an appliance that turns out to be a total loss to the announcement that a student having a part-time job as a waiter will be given a different task. For each topic we composed a primary message and a secondary message that functioned as a replication and manipulation check. For instance, we wrote a message about the high repair costs of cell phones and of a televi- sion set, and about a cancelled flight and a cancelled railway trip (see Table 2).

All e-mail messages followed the same format. This format contained a salutation, a neutral introductory sentence, and either the bad news followed by three explana- tions or three explanations followed by the bad news. Then the messages ended with a neutral statement, a greeting, and the name of the sender.

Table 2. Context of Bad-News Messages and the Specific Topics.

Context of bad news Specific topics

1 Electronic gadget cannot be repaired Broken iPod Broken watch 2 Holiday trip is cancelled Flight Rail journey 3 Internal application is turned down For participation in an

(expensive) company course For a higher position 4 A medical test failed and has to be redone IMR-scan Taking blood sample 5 The repair costs of an electronic appliance are much

higher than first expected Broken telephone Broken television set

6 The application for an external job is turned down

Manager Management trainee

7 A student who has a secondary job as junior waiter is degraded to the function of dishwasher

In a pancake restaurant In a gourmet restaurant

370 Journal of Business Communication 50(4)

Example of a Bad-News E-Mail

This e-mail message regards/concerns your reservation of a business class ticket flight number CA 1773 for 15/2/2009 from Amsterdam to Toronto (Canada).

Regrettably, we have to cancel this flight for the following reasons: the airport of Toronto is on strike that week. Besides, we expect extreme bad weather conditions in Toronto where the airport will be closed for all flights for four days. At the moment, we are also encountering some unforeseen technical problems with the airplane, which we are trying to solve as soon as possible.

We apologize for the inconveniences and would like to propose an arrangement with you. You are requested to contact Can Airlines for further information, telephone number 020–340 74 24.

With kind regards,

Can Airlines

Jan-Peter de Vries, Client Manager

We presented the e-mails to our participants on a laptop computer, in a realistic e-mail-makeup. The mean length of the letters was 186 words (minimum 180, max- imum 191).

As far as the voice mails are concerned, their structure was almost identical with the e-mails:

Example of a Bad News Voice Mail

Good day, this is Remon Spaas of Telfort Repair service speaking.

We have received your mobile telephone and examined it on defects.

The battery of your phone is broken. Furthermore the display has to be replaced because of water damage. Besides this, the entire interior works has to be cleaned.

For these three reasons, the costs of repair of your mobile phone will amount to €315.

If you accept this estimate, please let us know by contacting our client service.

The phone number is 0900–9596.

Greetings

The only difference with the e-mail is the position of the sender’s self-identification. As was mentioned in the introduction, the position of this “internal” self-identification in e-mails is conventionally at the end of the message, namely in the signature. The conventional position for self-identification in voice mails, however, is at the

Jansen and Janssen 371

beginning, directly after the address to the recipient “Hi/hello, It’s NAME of the X company” (Goutsos 2001, Lange 1999, Knoblach and Alvarez-Cáccamo 1992). For reasons of ecological validity, we decided to follow this convention, by inserting “This is [NAME] speaking” directly after the salutation. After the main body of the voice mail text, the closing move consisted of closing and goodbye formulas, and again, the name of the sender.

Participants

Of the 1,133 participants in this second experiment 533 (53%) were men and 600 (47%) were women. Given that fact that we used a 2 × 2 × 14 design (e-mail/voice mail, direct/indirect, 14 different messages) this implies that we had ±18 participants in every condition. The participants were between 15 and 77 years of age: M = 27.0 (SD = 12.5). Randomization checks revealed that participants were distributed equally in respect to their gender and age across all the conditions. Nearly all participants were students of secondary vocational education schools, or had graduated from these types of school. They had a Dutch cultural background and were all native speakers and flu- ent readers of Dutch. Nobody was paid for his or her contribution. As in the previous experiment, we asked many participants to volunteer for the experiments during a train ride as they commuted to work; others were recruited by our students in libraries, on campus, at work, or at home.

Independent Variables

The first independent variable was channel. We presented our participants with bad- news messages in an e-mail or voice mail format. The second independent variable was the presentation order of the company’s decision (the bad news for the customer) and the explanation for it. The explanation included three independent reasons. We had two main reasons for inserting an elaborate argumentation. By doing so, the structural differences between the two conditions were as large as would be accept- able in a realistic setting. Furthermore, we wanted our participants to process the entire fragment instead of peeking ahead or scanning the paragraphs until they found the decision. To present realistic argumentation, we did a pretest in which readers other than the participants in the experiment evaluated the plausibility of the explana- tion. The text in both conditions was identical, with one exception: in the decision- first condition, the phrase announcing the upcoming reasons was at the end of the sentence; whereas, the phrase referring back to the reasons in the explanation-first condition was at the beginning of the sentence, as in the following examples:

Example of the Decision-First Condition

The cost of repair of your mobile phone will amount to € 315 for three reasons.

The battery of your phone is broken. Furthermore, the display has to be replaced because of water damage. Besides this, the entire interior work has to be cleaned.

372 Journal of Business Communication 50(4)

Example of the Explanation-First Condition

The battery of your phone is broken. Furthermore, the display has to be replaced because of water damage. Besides this, the entire interior work has to be cleaned.

For these three reasons, the costs of repair of your mobile phone will amount to € 315.

Dependent Variables

The effects of the independent variables were measured with a questionnaire by which the participants could evaluate the messages (see the appendix). First, the participants gave a general assessment of the overall quality of the message in the form of a report mark (ranging from 1 to 10 as is the convention in the Netherlands) for the entire mes- sage. Then, they used the 7-point Likert-type scales and semantic differential scales (1-7) to evaluate the messages on

• comprehension; e.g., “I understand the message completely” (Items 2, 14, 16, 17, and 19)

• agreement; e.g., “I can imagine this kind of decision” (Items 3, 4, 10, and 12) • positive sender traits, e.g., “The sender impresses me as honest” (Items 11, 15, 22) • attitude of the sender toward receiver (Items 8, 18, 22) • image of the company, e.g., “My view of the organization is positive/negative”

(Items 24, 25, 26)

In the last part of the questionnaire, we included a proposition that we used as a manip- ulation check for bad news (Item 27) and an open question to test whether the partici- pant had comprehended the message (Item 28): “Three reasons for the decision are mentioned in the message. What was the second reason in your own wording?” The answers to this open question were coded as 0 for no answer or for reasons other than mentioned in the message, and 1 for one or more reasons mentioned in the text. This question was placed toward the end of the list to prevent it from causing the partici- pants to attach more value to the arguments. Demographic questions about the partici- pants’ gender and age together with two background questions about the estimated importance of these kind of bad-news letters for the participants and their experience with this type of messages completed the questionnaire.

Design and Procedure

We opted for a between-participant design to avoid contamination effects. Thus, all participants evaluated only one message in one condition. The procedure was as fol- lows. After greeting and welcoming the participant, the experiment leader presented a written instruction:

Suppose you are a client of Telfort (a Dutch mobile phone company) and you bought a telephone with a subscription a year ago. As this telephone has not been functioning well recently, you have sent it to Telfort. A few days later you receive the following message:

Jansen and Janssen 373

Next the participant had the opportunity to ask questions and then the message was presented to them onscreen. After the participant had indicated that he or she had fin- ished reading, the text was made invisible, and they received the questionnaire. Although there was no limit set to the reading time, most participants spent only 3 minutes on the reading task (M = 195 s, SD = 152).

Participants in the voice mail condition were allowed to listen to the message as many times as they wanted, but almost no one made use of this opportunity. It took the participants less time to listen to the message than to read it. The messages lasted between 24 and 70 seconds (M = 43 seconds, SD = 16).

Manipulation Checks

Statistical analysis revealed that we could distinguish four reliable clusters of proposi- tions (the dependent variables) in the data: comprehension, agreement, sender traits, attitude of sender, and image of the company (see Table 3).

All have acceptable to good Cronbach’s alphas, with .61 being a generally excepted minimum for clustering (Field, 2009). Also, we checked whether the par- ticipants’ evaluation of the messages for each topic in the table was essentially the same, which they were for all clusters (p > .05). Subsequently, we checked the manipulation of bad news by analyzing the mean score for the proposition, “The message entails bad news indeed.” The mean score was 5.6 (SD = 1.4) on a 7-point scale, so we can safely assume that the participants considered the message bad news. Furthermore, the participants had sufficient prior knowledge about the topics in the messages (Item 32). The mean score for experience was 4.0 (SD = 2.1). The participants did not consider the messages themselves to be important (Item 31) considering a mean of 2.5 (SD = 1.6). We found no effects of channel and structure on these three propositions (p > .05).

Results

As expected, the messages were not too difficult for the participants. The results show that 790 of the 1,133 participants (= 70%) correctly answered Question 28 (see the appendix). We also found no statistically significant effects of channel or message structure (p > .05) on the number of correct answers on this question. So all messages

Table 3. Reliability of the Clusters of Dependent Variables.

Cluster Propositionsa Cronbach’s α

Comprehension 2, 14, 16 .64 Agreement 3, 10, 12 .78 Positive sender traits 11, 15, 22 .74 Positive attitude sender toward receiver 18, 21, 23 .68 Image organization 24, 25, 26 .86

a. The other propositions (5, 6, 7, 9, 13, and 20) could not reliably be attributed to any cluster.

374 Journal of Business Communication 50(4)

were relatively easy to understand, a conclusion that is in line with the rather low result for the proposition that measured the perceived level of difficulty most directly, namely Item 19 “the message is difficult”: M = 2.6 (SD = 1.1). The short reading times of the e-mails and the absence of requests to replay the voice mails reported in the procedure section point in the same direction.

The next question we need to answer is whether the channel, in general, has an effect on the evaluation of the messages (see Table 4).

The results are unambiguous. We see two distinct effects of channel. First, the par- ticipants find—as expected—e-mails more comprehensible than voice mails. Second, voice mail leads to more agreement, a more positive image of the sender, a more posi- tively evaluated relationship and a better image.

Is there an effect of message structure as well? Does a direct structure lead to dif- ferent evaluations than does the indirect structure? In Table 5, we present the mean scores of direct and indirect structures on all dependent variables.

The effects of structure are less prominent than were the effects of channel. First, the scope of the message structure’s influence is narrower, because structure only affects agreement and sender image. Second, the effects we did find are considerably weaker, as the lower eta squares indicate. The direction of the differences is identical

Table 4. Means Scores and Standard Deviations (SDs) for Channel of the Clusters of Dependent Variables (1 = Negative Evaluation, 7 = Positive Evaluation) and Report Mark (1 = Low Evaluation, 10 = High Evaluation).

E-mail Voice mail F p η2

Report mark (1) 6.2 (1.5) 6.4 (1.3) 5.9 .015 .005 Comprehension 5.7 (1.0) 5.1 (1.2) 64.4 .001 .054 Agreement 4.6 (1.5) 4.9 (1.3) 10.7 .001 .009 Positive sender traits 4.6 (1.2) 5.0 (1.0) 50.2 .001 .042 Positive attitude sender 3.9 (1.3) 4.4 (1.2) 33.7 .001 .029 Image organization 4.0 (1.4) 4.4 (1.4) 29.3 .001 .025

Table 5. Means Scores and Standard Deviations (SDs) for Structure of the Clusters of Dependent Variables (1 = Negative Evaluation, 7 = Positive Evaluation) and Report Mark (1 = Low Evaluation, 10 = High Evaluation).

Direct Indirect F p η2

Report mark 6.3 (1.4) 6.3 (1.5) .011 ns Comprehension 5.4 (1.1) 5.4 (1.2) .245 ns Agreement 4.7 (1.4) 4.8 (1.4) 4.448 .035 .004 Positive sender traits 4.7 (1.2) 4.9 (1.1) 7.476 .006 .007 Positive attitude sender 4.1 (1.3) 4.2 (1.2) 2.160 ns Image organization 4.1 (1.5) 4.5 (1.4) 2.123 ns

Jansen and Janssen 375

for all evaluative dimensions: the indirect structure (decision last) is clearly valued more highly than the direct (decision first) structure.

Finally, we tested whether we could find an interaction effect between channel and message structure. By way of multivariate analyses of variance we evaluated if direct- ness had a different effect in e-mails than it had in voice mails (see Table 6 for the results).

The results in Table 6 show no interaction effect on the image of the organization, but significant interaction effects on report mark, comprehension, agreement, sender traits, and attitude toward the receiver.

The results in Table 6 show that direct voice mails get a significantly higher overall evaluation than direct e-mails, indicated by the report mark. In other words, the par- ticipants appreciated to direct voice mail more than the direct e-mail.

Statistical analysis revealed no significant difference between the direct and the indirect structure within the channels. E-mail is considered more comprehensible than voice mail and this finding is even more prominent in the indirect messages where the decision followed the explanation.

The other three clusters with significant interaction effects give a slightly different picture. As for agreement, directness has no effect in voice mails (p > .05) but it has in e-mails. Readers of a direct e-mail are considerably less inclined to agree with the decision than the receivers of indirect e-mail, F(1, 572) = 7.8, p = .005, η2 = .014. Furthermore, it turns out that a direct voice mail elicits more agreement than does a direct e-mail, F(1, 568) = 14.6, p < .001, η2 = .025.

As for the positive sender traits, the pattern is roughly the same as for agreement cluster. We see no effects of directness in voice mail. In the e-mails, however, the direct structure leads to a less positive evaluation of the sender than does the indirect structure, F(1, 572) = 16.7, p < .001, η2 = .028).

The effects of channel and structure on the attitude of the sender toward the receiver are completely in line with the results on two previous dependent variables. Within the voice mail condition structure has no effect (p > .05), but in e-mail it does. The sender

Table 6. Means Scores and Standard Deviations (SDs) for Channel and Structure of the Clusters of Dependent Variables (1 = Negative Evaluation, 7 = Positive Evaluation) and Report Mark (1 = Low Evaluation, 10 = High Evaluation).

E-mails Voice mails

F p Direct Indirect Direct Indirect

Report mark 6.1 (1.5) 6.2 (1.6) 6.4 (1.3) 6.3 (1.3) 5.9 .015 Comprehension 5.6 (1.0) 5.7 (1.0) 5.2 (1.2) 5.0 (1.3) 4.1 .043 Agreement 4.4 (1.5) 4.8 (1.5) 4.9 (1.3) 4.9 (1.3) 4.5 .034 Positive sender traits 4.4 (1.2) 4.8 (1.2) 5.1 (1.0) 5.0 (1.0) 11.7 .001 Attitude sender 3.8 (1.3) 4.1 (1.1) 4.4 (1.1) 4.2 (1.1) 7.7 .006 Image organization 3.8 (1.4) 4.1 (1.4) 4.4 (1.4) 4.4 (1.3) 2.1 NS

376 Journal of Business Communication 50(4)

of a decision last e-mail is considered more sympathetic than the sender of a decision first e-mail, F(1, 572) = 8.3, p = .004, η2 = .014). The superiority of voice mail over e-mail is more apparent in messages with direct structures, F(1, 588) = 36.1, p < .001, η2 = .06, than in messages with indirect structures, F(1, 581) = 4.7, p = .03, η2 = .008.

No other interaction effects were found; nor did we find a significant interaction of any of the dependent variables in relation to age or gender of the participant.

Conclusion and Discussion

The outcomes of the experiment permit us to answer the first research question about the receivers’ preference for e-mail or voice mail? Hypothesis 1 stated that e-mail (messages) would be evaluated as more comprehensible than voice mail (messages). Hypothesis 2 stated that voice mail (messages) would be evaluated as more personal than e-mail (messages). Both hypotheses are supported by the data (see Tables 4 and 5). Our participants considered e-mail to be more comprehensible than voice mail, and valued the persuasive and interpersonal characteristics of voice mail higher than those of e-mail.

The second research question addressed the issue of message structure. Which mes- sage structure do the receivers prefer: direct or indirect? Hypothesis 3 states that direct structure (bad-news → explanation) would be evaluated as more comprehensible than the indirect structure (explanation → bad news). Hypothesis 4 reads—the indirect structure will be evaluated as more persuasive and more personal. Although the effects of structure were more limited and weaker than those for channel, our results support the hypotheses.

Finally, the research question about the interaction of channel and structure: Is the positive effect of the indirect order more outspoken in voice mail condition as Hypothesis 5 stated? No, it is the other way around. Positive effects of indirectness are limited to the e-mail condition.

Before we discuss the theoretical relevance of our findings we have to consider some of the limitations of our study.

First, in our experiment we compared two channels and excluded face-to-face. We had good reasons for this. First, incorporating of an unmediated condition in the design would have been superfluous because we know beforehand what the results would be. All the empirical evidence shows that mediated channels are no match for unmediated communication, such as face-to-face conversations (Baltes et al., 2002). Furthermore, the instantaneous feedback possibilities of face-to-face interaction would have resulted in completely different and thus incomparable conversations. To study the effect of voice the vocal cues had to be held constant and in perfect relation to the written cues in the e-mails.

The messages in our experiments were—although divers—also small and limited to bad news of moderate gravity. This choice made our experiment realistic on our view, because no sender would leave a voice mail for more serious bad news (e.g., a dismissal or the death of a beloved one). However, we would encourage experiments on other types of bad news (e.g., turning down an invitation or a marriage proposal or

Jansen and Janssen 377

refusing to lend a person a certain amount of money). Furthermore, we would like to encourage experiments with larger messages. Our voice mails and e-mails consisted of no more than an introduction, the bad news, the three explanations, and a closing line. This mini format—although realistic—may have had an impact on the comprehension results (see also discussion).

Also, we used only two channels for the bad news. Extending the experiments to other channels, in particular to other written media (e.g., texting, letters on paper, chat- ting) would be interesting, especially to answer the question whether the differential effects we found should be ascribed to differences in channels or to differences in media (spoken vs. written).

The most important limitation in our study may well be the following. All voice mail messages were recorded by one single male speaker who as an experienced, authoritative speaker spoke fluently without false starts and made use of pitch and intonation patterns. It would be interesting to test a variety of speakers (male-female, high pitch-low pitch, slow-fast) to see if their voice mail messages lead to different results. Furthermore, the fact that the voice mails were “optimal” (nice sympathetic voice, no hesitations or errors) is defendable because it enabled us to compare them in a fair way with the e-mail texts, which were also flawless in style and spelling; but the absence of speech characteristics like hesitations and false starts made the voice mails perhaps less realistic. It would be interesting to repeat this experiment with voice mails with filled pauses and other speech characteristics.

And finally, the specific linguistic and cultural background of our participants who are speakers of Dutch in the Netherlands, may threaten the external validity of the results. We see no a priori reasons why the evaluation of the two channels would be influenced by cultural aspects, but the evaluation of message structure may be culture- specific. In Jansen and Janssen (2011) we explain why we think the Dutch preference for indirect structure is a fortiori valid for other cultures.

But if we assume that our conclusions are valid, how can we explain the consistent superiority of the voice mail message when it comes down to establishing good cus- tomer relations? In the first place we have to realize that the eventual usability draw- backs of voice mail were of little concern for our participants in the experiments because the messages, in their written and oral varieties, were presented to them directly. So the participants were not hindered by practical shortcomings of voice mail such as in-boxes that are to find or open on some telephones, voice mail messages often cannot be selected, voice mails may be “empty” because the caller did not leave a complete message, and so on. So the differential effects of the channels in our exper- iment can only be attributed to the difference in medium: oral versus written.

An obvious explanation for the positive effects of voice mail can be derived from medium richness theory (Daft & Lengel, 1984): The possibility of the voice mail chan- nel to relay emotions by way of the human voice makes this channel more apt for the sender to show empathy with the reader. This is in line with findings in Shapiro, Buttner, and Barry (1994) that face-to-face and telephone contact promote interac- tional, interpersonal justice (Timmerman & Harrison, 2005).

Many other experiments point out the extraordinary quality of voice. Apart from the classic experiment by Chaiken and Eagly (1983), we may refer to Keil and Johnson

378 Journal of Business Communication 50(4)

(2002) for the greater impact of voice mail (compared with e-mail) on the social pres- ence of the sender and Gelinas-Chebat and Chebat (1992) for the effect of intonation and speech intensity on the evaluation of commercial messages.

The results of a recently conducted additional experiment on a smaller scale (160 participants) point in the same direction. In this experiment we compared normal voice mail messages with computer synthesized voice mail messages (with a typical flat intonation contour). The idea behind this experiment was that normal voice mail mes- sages would signal empathy while computer-synthesized speech would reveal no information about the sender. The results support the idea that the presence of “voice” in bad-news message contributes to the receiver’s sense of interpersonal justice. The participants evaluated the synthesized voice mail message significantly lower than the messages with normal intonation (Visser, 2011).

We can interpret our results within the context of the hyperpersonal model as well: The higher evaluation of voice mail may be considered a sign of the receiver’s appre- ciation of the sender’s courage to use a channel that might expose him or her as insin- cere, or the receiver’s appreciation that the sender initially planned to call him up (instead of using the MUM strategy).

Evidently, media richness and hyperpersonal are not mutually exclusive. So it is possible that they apply both in this case: An oral message is more empathical and elicits more respect from the receiver, than a comparable written message.

On the basis of the experimental evidence to date we are not yet in a position to exclude another explanation of a more mundane nature. Voice mail may be considered more attractive, because of its infrequent use. The participants were perhaps merely pleasantly surprised by the fact that someone left a voice mail message in a situation (communicating bad news) where e-mail is the medium of preference (El-Shinnawy & Markus, 1997).

This leaves us with the final question about the effect of the direct and indirect structure in e-mails and voice mails. Is there an explanation for the fact that the evalu- ation of voice mails turns out to be totally structure insensitive, while the indirect structure in e-mails elicits a higher evaluation on the relational variables? Unfortunately, we have only a speculative answer. It is possible that our participants, knowing that an eventual replay of the voice message would be cumbersome, paid more attention to the details of the message than readers did. If this had been the case the listeners may have interpreted the explanations in the indirect structure immediately as signals for the upcoming bad news, thus, blurring the difference between the direct and indirect struc- ture. The listeners could have predicted the bad news correctly. We do not know yet if this is a general effect of the indirect structure; but for now we consider it a hypothesis that can and needs to be tested in future experiments.

For now we may conclude that in situations in which the relationship between send- ers and receivers is fragile, as is the case in bad-news communication, senders may well consider voice mail as a medium. Although senders often choose e-mail to con- vey bad news to avoid direct confrontation, this study shows that presenting the same information orally is beneficial for the acceptance of the message and the relationship between senders and receivers.

Jansen and Janssen 379

Appendix

Questionnaire

1. Which report mark do you give to the message? 1 2 3 4 5 6 7 8 9 10 2. I understand the message completely 3. The decision was sufficiently well-founded 4. I can imagine this kind of decision 5. I prefer to receive a message like this as a letter 6. The message could be shorter 7. I would react immediately to this message 8. The sender is self-confident 9. The style of the message is official 10. The reasons for the decision are acceptable 11. The sender impresses me as honest 12. I agree with the decision of the organization 13. I prefer to receive a message like this as a voice mail 14. I could catch on without difficulties 15. The sender is friendly 16. The goal of the message was clear right away 17. The message was complete 18. The sender adopts an arrogant attitude 19. The message is difficult 20. I prefer to receive a message like this as an e-mail

The sender is

(21) standoffish/involved (22) Insincere /sincere (23) Understanding/not understanding (24) My view of the organization is positive/negative

I think the organization is

(25) Unprofessional/professional (26) Reliable/unreliable (27) The message entails bad news totally disagree/totally agree (28) Three reasons for the decision are mentioned in the message. What was the

second reason in your own wording? (29) Sex (30) Year of birth (31) Is this type of message important for you? Not important/important (32) Do you have any experience with receiving these kind of messages? No

experience/much experience

380 Journal of Business Communication 50(4)

Authors’ Notes

The authors are mentioned in alphabetical order. The authors have contributed to this article (and the research) equally and both take full responsibility for its content.

Acknowledgments

We would like to thank Rosalie Brem, Sharon Brilman, Joy Hofman, Ilona Lawnik, Tomas de Smet, and Annewil van Wijlen who helped in carrying the experiment out and Ted Sanders, Mike Huiskes, and two anonymous reviewers for their helpful comments on an earlier version.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

1. We follow Lyons’s (1977, pp. 67-70) terminology by using “medium” for the more general distinction between oral versus written communication, and “channel” for the different mes- sage transportation and display systems within the oral and the written medium. As we have incorporated for each medium just one channel in our experiment, the terminological distinc- tion is of little relevance.

2. Rice and Danowsky (1993) suggest a terminological dissociation between “voice messag- ing” and “voice answering.” By “voice answering,” they mean the use of a spoken message by which an absent receiver of a call invites the caller to leave a message. “Voice messaging” is reserved for the manipulation of recorded messages, for example abbreviate, or revise and resend them. When we speak about voice mail we mean the latter.

3. Baron (1998) doubts the relevance of dichotomies in this field. She presents the uses of each old and each new channel in so-called spectra, which blur the clear distinctions present in the dichotomy of Crystal (2001).

References

Baltes, B. B., Dickson, M. W., Sherman, M. P., Bauer, C. C., & LaGanke, J. S. (2002). Computer- mediated communication and group decision making: A meta-analysis. Organizational Behavior and Human Decision Processes, 87, 156-179.

Baron, N. (1998). Letters by phone or speech by other means: The linguistics of e-mail. Language & Communication, 18, 133-170.

Baron, N. (2000). Alphabet to e-mail: How written English evolved and where it’s heading. New York, NY: Routledge.

Bolter, J. D., & Grusin, R. (1999). Remediation: Understanding new media. Cambridge: MIT Press.

Bovée, C. L., & Thill, J. V. (2000). Business communication today. Englewood Cliffs, NJ: Prentice Hall.

Jansen and Janssen 381

Britt, M. A., & Larson, A. A. (2003). Constructing representations of arguments. Journal of Memory and Language, 48, 794-810.

Brown, P., & Levinson, S. C. (1987). Universals in language use: Politeness phenomena. Cambridge, England: Cambridge University Press.

Chaiken, S., & Eagly, A. H. (1983). Communication modality as a determinant of persuasion: The role of communicator salience. Journal of Personality and Social Psychology, 45, 241-256.

Crystal, D. (2001). Language and the Internet. Cambridge, England: Cambridge University Press.

Daft, R. L., & Lengel, R. H. (1984). Information richness: A new approach to managerial behav- ior and organizational design. Research in Organizational Behavior, 6, 191-233.

De Jong, Jaap (ed.) et al (2002). Handboek bedrijfscorrespondentie [Handbook for Business Correspondence]. Groningen: Martinus Nijhof.

Dingwall, S. (1992). Leaving telephone answering machine messages: Who’s afraid of speaking to machines? Text & Talk, 12, 81-101.

El-Shinnawy, M., & Markus, M. L. (1997). The poverty of media richness theory: Explaining people’s choice of electronic mail versus voicemail. International Journal of Human- Computer Studies, 46, 443-467.

Field, A. (2009). Discovering statistics using SPSS. London, England: Sage. Gelinas-Chebat, C., & Chebat, J.-C. (1992). Effects of two voice characteristics on the attitudes

toward advertising messages. Journal of Social Psychology, 132, 447-459. Goutsos, D. (2001) Sequential and interpersonal aspects of English and Greek answering

machine messages. Pragmatics, 11, 357-377. Jansen, F., & Janssen, D. (2010). Effects of positive politeness strategies in business letters.

Journal of Pragmatics, 42, 2531-2548. Jansen, F., & Janssen, D. (2011). Explanations first: A case for presenting explanations before the

decision in Dutch bad-news messages. Journal of Business and Technical Communication, 25, 36-67.

Janssen, D. (Ed.). (2007). Zakelijke communicatie [Business communication]. Groningen, Netherlands: Noordhoff Uitgevers.

Keil, M., & Johnson, R. D. (2002). Feedback channels: Using social presence theory to compare voicemail to e-mail. Journal of Information Systems Education, 13, 295-302.

Knoblauch, H., & Alvarez-Cáccamo, C. (1992). “I was calling you”. Communicative patterns in leaving a message to answering machines. Text & Talk, 12, 473-505.

Kruger, J., Epley, N., Parker, J., & Ng, Z.-W. (2005). Egocentrism over e-mail: Can we com- municate as well as we think? Journal of Personality and Social Behavior, 89, 925-936.

Lange, I. (1999). Die sequenzielle Struktur von Anrufbeantworter-kommunikation [The sequen- tial structure of voice mail communication]. InList (Issue No. 14). Retrieved from http:// www.inlist.uni-bayreuth.de/issues/14/inlist14.pdf

Levelt, W. J. (1989), Speaking. From intention to articulation. Cambridge: MIT Press. Lyons, J. (1977). Semantics I. Cambridge, England: Cambridge University Press. Rice, R. E. (1993). Media appropriateness: Using social presence theory to compare traditional

and new organizational media. Human Communication Research, 19, 451-601. Rice, R. E., & Danowsky, J. A. (1993). Is it really just like a fancy answering machine?

Comparing semantic networks of different types of voicemail users. Journal of Business Communication, 30, 369-397.

Scherer, K. R. (2003). Vocal communication of emotion: A review of research paradigms. Speech Communication, 40, 227-256.

382 Journal of Business Communication 50(4)

Shapiro, D. L., Buttner, E. H., & Barry, B. (1994). Explanations: What factors enhance their per- ceived adequacy? Organizational Behavior and Human Decision Processes, 58, 346-368.

Sproull, L., & Kiesler, S. (1986). Reducing social context cues: Electronic mail in organiza- tional communication. Management Science, 32, 1492-1512.

Timmerman, P. D., & Harrison, W. (2005). The discretionary use of electronic media: Four considerations for bad news bearers. Journal of Business Communication, 42, 379-389.

Unnava, H. R., Burnkrant, R. E., & Erevelles, S. (1994). Effects of presentation order and com- munication modality on recall and attitude. Journal of Consumer Research, 21, 481-490.

Visser, R. (2011). Het gebruik van voicemailberichten in slachtnieuwscommunicatie [The use of voicemail messages in bad news communication] (Unpublished master’s thesis). Utrecht University, Utrecht, Netherlands.

Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal, and hyperpersonal interaction. Communication Research, 23, 3-44.

Watts Sussman, S., & Sproull, L. (1999). Straight talk: Delivering bad news through electronic communication. Information Systems Research, 10, 150-165.

Author Biographies

Frank Jansen is an assistant professor in the Communication Studies Unit of Utrecht University. He has lectured in grammar, dialectology, and rhetoric at several universities. His current research interests include the style and structure of electronic texts and politeness phenomena in business communication.

Daniel Janssen is a senior researcher and lecturer in the Department of Dutch at the Utrecht University and an international guest professor at the University of Antwerp. His main research interests include cognitive and social aspects of writing processes and (improving) professional communication. He has published several articles and textbooks on professional writing pro- cesses and on written and oral business communication.

Copyright of Journal of Business Communication is the property of Association for Business Communication and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.