• No results found

With a little help from my friends? Analyzing the potential of Social proof to overcome resistance and increase intention to pay for online news

N/A
N/A
Protected

Academic year: 2021

Share "With a little help from my friends? Analyzing the potential of Social proof to overcome resistance and increase intention to pay for online news"

Copied!
31
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

to overcome resistance and increase intention to pay for online news

Manuel Valiño student number 11181532

Master's Thesis

Graduate School of Communication, University of Amsterdam Master’s programme Communication Science

Supervisor: dr. Marieke L. Fransen July 21, 2017

(2)

Abstract

Media organizations struggle to get enough paid circulation revenue to sustain their businesses. They use messages that appear on the screen while consumers are reading articles to try and

persuade them to subscribe. However, these messsages have not been successful so far in attracting a considerable number of new paying readers. Therefore, it appears vital for the future of media organizations to test techniques that can increase the persuasiveness of those subscription messages. In previous studies Social proof has been found an effective way to enhance persuasion. But its efficacy in the context of (paid) online news consumption has not been tested, neither has been its potential to overcome resistance to the above mentioned subscription messages. A model was suggested where different levels of social proof (weak vs. strong) would increase paying intention by means of a lower resistance to subscription messages. Nevertheless, the results showed no significant effect of provision of social proof on readers' intention to pay for online news, or resistance to subscription messages. More surprisingly, resistance did not significantly predict paying intention, although a qualitative measurement included in the online questionnaire reflected the “annoyance” and “frustration” experienced by most respondents. Alternative explanations, such as the relevance of deeply rooted attitudes against paying for online news, and design limitations to take into account in future research on the topic are discussed.

(3)

In the early days of online journalism, Robert D. Ingle, then editor of the San Jose Mercury News, made an unfortunate forecast about the rise of electronic newspapers. “I don't think anyone in their right mind would use this as a substitute for the newspaper” (Kurtz, 1993), said Ingle in 1993. Back then only three million Americans were online (Salwen, Garrison, & Driscoll, 2004). Today more than 1.3 billion people worldwide, or about 40 per cent of global Internet users, read newspapers online (Henriksson, 2016). Television and the Web are, across countries, the most frequently accessed sources for news, while readership of printed newspapers has declined significantly, especially in the west (Newman et al., 2017; Henriksson, 2016).

Despite the rise in online news usage, the gold rush that newspaper owners and managers expected in the decade of 1990 never came and media companies are still struggling to find a sustainable economic model (Salwen, Garrison, & Driscoll, 2004). Digital advertising revenue grew 51 per cent worldwide between 2010 and 2015, up to US$10 billion, but is still far from becoming as important a source of revenue as in print (US$69 billion in 2015). A similar pattern is observed in the case of circulation revenue. Paid digital circulation revenues increased 30 per cent in 2015 –and 547 per cent over the previous five years–, but they represent less than four per cent of print

circulation revenue, with only about one in five users paying for online news (Henriksson, 2016). Regarding digital circulation, significant differences exist among countries. Payment figures tend to be higher in Nordic countries and, recently, also in the United States (Newman et al., 2017; Lichterman, 2016a). Seventy-eight per cent of American newspapers with circulations over 50,000 were using a digital subscription model in 2015 (Williams, 2016). The most common strategy has been implementing a metered paywall, where readers have to pay 10 to 15 euros per month after reading a certain number of articles (Williams, 2016). Although there are some successful examples such as The New York Times reaching one million digital-only subscribers (Williams, 2016) and getting more than 50 per cent of their revenue directly from readers (Doctor, 2017), the vast

(4)

The situation is even less promising in most Southern and Central European countries, where “most online content remains free as there isn’t the subscription culture to build on” (Newman et al., 2017, p. 24). More importantly, among those who do not pay for online news the likelihood that they will do it in the future is extremely low: only seven per cent in Germany, nine per cent in the United Kingdom or 11 per cent in Spain (Newman et al., 2017). As Williams (2106) points out, “going forward each new subscriber must actively be won, converted, persuaded”.

For some the reason for not paying is quite straightforward: news is not worth paying for. Apparently, more readers could be persuaded of paying if the content would be sufficiently

valuable, convenient, and relevant (Newman et al., 2017). Consumers “expect the improved product to be offered first, and then to be asked to pay” (Doctor, 2017). Others argue that they can not afford paying for news (Newman et al., 2017). However, an overwhelming majority says that so much online news is freely available that they simply do not mind going somewhere else (Newman et al., 2017; Cook & Attari, 2012), even if what they end up reading is not their preferred option (Chyi & Lee, 2013). As Goyanes (2014) indicates, “online news has become a commodity” (p. 743).

A culture of free dominates online media consumption. People are used to getting news for free online and they expect to keep doing so (Outing, 2009). In 2001 almost eight out of ten

“strongly disagreed that newspapers should charge for site access” (Donatello, 2002, p. 36). Twelve years later, a third of those who said they would not subscribe to an online newspaper argued that they did not approve of user-financed news websites (Kammer et al., 2015). This principled position towards (not) paying for online news seems to influence people's intention regardless of the price and creates a resistance to the fact of paying for online news (Kammer et al., 2015).

Testing strategies to overcome this resistance and attract new subscribers is, therefore, much needed. Provision of social proof is a technique frequently used to increase the effectiveness of persuasive messages. People often rely on similar others to infer a course of action (Cialdini, 2001, Amblee & Bui, 2012). This must be especially true for a context as social as that of online news,

(5)

where almost one third of consumers use social media and news aggregators as their preferred way of getting to content (Newman et al., 2017). This research, thus, aims to demonstrate that provision of social proof can be an effective way of reducing resistance and increasing persuasiveness.

When people are confronted with opposing attitudes of others they identify with –i.e., people who are the same age or have similar interests–, they may reassess their views and even change them so that they can align with the group norm (Festinger, 1954). In the specific case of paying for online news, then, it is expected that providing non-subcribers with social proof that people similar to them pay for online news will reduce their resistance to subscription messages, increasing their intention to pay. This study tries to answer the following research question:

RQ: Does providing non-subscribers with social proof –that others like them are already subscribers– increase their intention to pay for online news? And is resistance to persuasion the

underlying mechanism that mediates the effect of social proof on paying intention?

The current study builds on Festinger's theory of social comparison processes (Festinger, 1954) and on the subsequent work on social influence and social proof (Reingen, 1982; Li & Pavlou, 2014; Koch & Benlian, 2015; Amblee & Bui, 2012). It also seeks to complement the existing literature on persuasion to resistance (Fransen et al., 2013; 2015a; 2015b) by offering empirical evidence of the potential effectiveness of social proof as a technique to reduce resistance to persuasion. Finally, this research ambitions to provide media professionals with further tools to get new subscribers and make their businesses sustainable. This is especially relevant in a period when “online advertising gets gobbled up by Facebook and Google” (Lichterman, 2016b) and the rise of ad-blocking software poses still another threat for advertising revenue (Newman et al., 2016, 2017).

(6)

Theoretical background Resistance to paying for online news

As early as 2001 research was already being conducted about people's attitudes toward paying for online news (Donatello, 2002). One of the respondents of that study explained that “there are so many free sources of news, it doesn't make sense to pay” (p. 37). The argument is similar to the one used nowadays by the majority of online news readers (Newman et al., 2017), emphasizing that media companies have not succeeded in persuading consumers to pay for their products. Most respondents in Newman et al.'s (2017) study pointed to the widespread availability of free content as the main reason for not paying for online news. This reluctance seeems to be sustained over time: in many of the 36 countries analysed by Newman et al.'s (2017) around 80 per cent of those who had not paid for news the previous year said it was unlikely that the would pay in the future.

Kammer et al. (2015) further investigated the reasons for not paying, concluding that “the disinclination reflects a general resistance to the very idea of paying for news” (p. 113); 25.1 per cent of the respondents in their study about the free-to-fee transition said they would not pay for online news in the first place and 36.4 per cent would not pay for news at all in any medium.

The very nature of the Internet contributes to this opposition to pay for online content. The Web was created primarily to encourage the free exchange of information (Kahin & Varian, 2000) and this idea causes a disapproval of user-financed news websites (Kammer et al., 2015). Moreover, competition among media companies for online traffic discourages any strategy to charge for content, as the adoption of paywalls has resulted in significant readership loss in the past (Salwen, Garrison, & Driscoll, 2004; Newman et al., 2016; Cook & Attari, 2012).

This aversion to paying for online news consumption arguably manifests itself as a resistance to the subscription messages used by media companies to attract new paying readers. When consumers are asked to pay in order to continue reading they may feel that their freedom is threatened or eliminated (Fransen et al., 2015b). Online users will then feel motivated to restore that

(7)

freedom, according to the theory of psychological reactance (Brehm, 1966; Brehm & Brehm, 1981). Therefore, they will try to access the same or similar news without having to pay for it. Cook and Attari's research (2012) illustrates this idea. They found that The New York Times' readers reacted negatively to paying for previously free content, decreasing their visits, devaluing the newspaper or even using loopholes to avoid paying. Therefore, my first hypothesis states that:

H1: Resistance to the subscription messages used by online newspapers to attract new paying readers is negatively related to the readers' intention to pay for online news

Social proof as a persuasion-enhancing strategy

A number of strategies can be used to resist persuasive attempts, such as messages asking to pay for online news. Fransen et al. (2015b) studied the different motivations that lead to adopt one tactic or another. That choice, they concluded, depends on each person's specific motive for resisting the message. In the case of subscription messages, as discussed earlier, consumers may experience psychological reactance, which occurs when people feel that their freedom is threatened or

eliminated (Brehm, 1966; Brehm & Brehm, 1981). But it can also be the case that they are reluctant to change because they are satisfied with the current situation –consuming online news for free– or they believe that the proposed change –paying for reading news online– does not make sense (Fransen et al., 2015b). Threats to freedom usually trigger contesting or empowerment resistance strategies, while reluctance to change is related to empowerment tactics (Fransen et al., 2015b).

Non-subscribers are likely to put into action one or several of these resistance strategies to avoid being persuaded to pay. Thinking of other people who share one's views has proved to be a somewhat effective way to resist persuasive messages (Jacks & Cameron, 2003). This

empowerment strategy is known as social validation and takes place when consumers “actively look for (significant) others who share their existing beliefs” (Fransen et al., 2015a, p. 9), thus validating

(8)

their attitudes and behaviours. The fact that most people are not paying for online news (Newman et al., 2017) makes it easy for readers to find evidence that not paying is the right thing to do.

The concepts of social proof and social validation are rooted in Festinger's theory of social comparison processes (Festinger, 1954), which states that “there is a drive to evaluate accurately one's opinions and abilities” and that “this evaluation is frequently only possible by comparison with others” (p. 124). Therefore, when people are confronted with messages asking them to subscribe to a certain news website they will rely on other people's attitudes as a reference to decide what they should do. As mentioned above, this social element is likely to be used as a resistance strategy, especially considering the extended view that news should be for free.

However, subscription messages can also include certain elements that are likely to enhance persuasion. Cook and Attari (2012) proved that minimizing psychological reactance –by framing the paywall in terms of financial necessity– led to increased willingness to pay among readers of The New York Times. Similarly, Kammer et al.'s qualitative study (2015) discovered that

acceptance –as opposed to resistance– of the very premise of news organizations made respondents more likely to pay for accessing Danish newspaper politiken.dk’s news. The researchers conclude, “if the audiences acknowledge that journalism is a costly activity and experience that the fee supports quality journalism, they say they will be more likely to pay” (p. 117).

Providing social influence may neutralize the readers' resistance to subscription messages, as “those who discover that most others in the group disagree with them become relatively less

confident that their opinion is correct and a goodly proportion change their opinion” (Festinger, 1954, p. 122). Making available the comparison with others whose opinions are different should thus result in a reassessment of one's views, which can lead to different intention and behaviour.

Empirical evidence for the effectiveness of social proof to enhance persuasion is found in Reingen (1982). He demonstrated that showing a list of fictitious donors to a person just before asking for a donation would increase the likelihood of that person making a donation. In a series of

(9)

experiments, Reingen proved the effectiveness of informative social influence (“As you can see, other students have given a donation already. Would you be willing to help also by giving a single donation?"), but found no effect of providing normative social influence (“We need your financial support so that you can help us in preventing heart attacks for people you might even know”), compared to the control condition (no normative neither informational social influence).

Amblee and Bui (2012) found that the daily deal website Groupon is able to generate sales by providing social proof to potential customers. In this case social proof was the number of coupons sold a few hours after the launch of the deal and the authors proved that it is strongly related to total coupon sales when the offer is closed, predicting total sales even better than price.

In the case of paying for online news, I found other indicators of the probable positive effect of providing social proof to potential paying readers. The Wall Street Journal recently tested a strategy so that when a new reader accesses a story shared by a Journal subscriber or reporter, he/she is offered a 24-hour free pass –in exchange for entering his/her email address (Wang, 2016a). More than 50% of the newspaper's revenue comes nowadays directly from readers, up from the industry average of 25% a decade ago (Doctor, 2017), and half of all sales are a result of something that has been shared (Wang, 2016b). This social referrals strategy shows that people seem to be more likely to pay for content after this content has been shared by other users from their network.

Another example is the increase in subscriptions and donations seen by some American news organizations shortly after an appeal made by television host John Oliver (Lichterman, 2016a). Oliver “made subscribing to the Times and Post and supporting ProPublica hip” (Doctor, 2017). Although using this type of endorsements is not strictly the same as providing readers with social proof –as here is the source that is relevant–, it shows that previously uncontested views may be affected when contrasting them with those of (significant) others (Fransen et al., 2015b).

Additionally, there is evidence that points to a different effect of social proof depending on whether identification with the referent group (i.e., that group the social proof makes reference to) is

(10)

weak or strong. This idea relies on the literature about social norms and identity theory. When describing intervention strategies to reduce alcohol consumption on campus, Prentice (2008)

explains that those “that target more proximal groups are likely to have more impact” (p. 173). Stok et al. (2012) also based their hypotheses about how social norms affect fruit intake on the

empirically demonstrated assumption that “when identification with a certain group is strong, this group’s behaviours will influence behaviour more than when identification is weak” (p. 958).

Referring to one of Reingen's experiments (1982), Cialdini (2001) suggests that social evidence would have not been as effective to generate persuasion had the friends' and neighbours' names on the list of donors shown to residents of Columbia been “those of random strangers” (p. 75). Taking all the above described evidence into account, it is expected that:

H2a: Subscription messages that include social proof that other people are paying for online news will lead to a higher intention to pay for online news than those messages not including social proof

H2b: Subscription messages that include strong-tie social proof –from people that share several characteristics with the receiver– will lead to a higher intention to pay for online news than those messages including weak-tie social proof –from people the receiver feels less identified with–

The role of resistance as mediator

While the use of social validation as a way to resist persuasion is widely accepted, it is not that clear whether provision of social proof also enhances persuasion by means of a decrease in resistance.

Social proof would be, in this case, an Omega persuasion strategy, according to the Knowles and Linn's (2004) distinction between Alpha and Omega tactics. The former seek to achieve

persuasion by enhancing the persuasiveness of the message itself; the latter try to remove or reduce resistance to it. Though the authors do not mention social proof explicitly, they discuss the potential

(11)

for reducing resistance of providing consumers with numerical anchors that can make an offer look more attractive (“Eighty thousand customers have purchased this thirty dollar service”, p. 126).

Although in Reingen's research (1982) the threat to freedom is not as evident as in the case of paywalls –one can choose whether to make a donation or not, but one can not continue reading news unless he/she pays for it–, it is likely that people will experience resistance when requested to make a donation. This resistance decreased when providing evidence that others had complied.

The tendency of people to remain skeptical and adopt a defensive attitude towards

advertising (Fransen et al., 2015b) means that participants exposed to Groupon deals in Amblee and Bui's study (2012) did probably experience certain resistance towards the offers shown and that social proof had some effect over it, increasing the attractiveness of those deals with more sales. To the best of my knowledge social proof has not been explicitly suggested or tested as a resistance-neutralizing strategy. However, the literature reviewed points in this direction. I expect that providing social proof to readers will reduce their resistance to the subscription messages:

H3a: Subscription messages that include social proof that other people are paying for online news will lead to a lower resistance to those messages than messages not including social proof

H3b: Subscription messages that include strong-tie social proof –from people that share several characteristics with the receiver– will lead to a lower resistance to those messages than messages including weak-tie social proof –from people the receiver feels less identified with–

The final hypothesis this study aims to test refers to the suggested mediation role of resistance in the relationship between social proof and paying intention. Provided that resistance to subscription messages is a most relevant factor in people's intention to pay, and that paying intention increases when social proof is included in subscription messages, it is reasonable to expect that a reduced

(12)

resistance due to provision social proof is the underlying mechanism increasing paying intention:

H4: The effect of social proof on readers' intention to pay for online news is mediated by the readers' resistance to the subscription messages

Figure 1. Conceptual model of the current study.

Method Participants and Design

A convenience sample of 143 individuals1 (82 female, 61 male; Mage = 30.97 years, range = 18-65,

SD= 9.76) voluntarily participated in this study, which was administered online between June 26

and July 10, 2017. Fifty of the respondents said they were living in Spain at the time, while other 43 declared to reside in The Netherlands. Most of them (18.9%) had a Bachelor's degree as highest level of education completed (1 = Less than high school degree; 8 = Professional degree; M = 5.23,

SD = 1.37), and said they were strongly interested in news (0 = not at all interested; 100 =

extremely interested; M = 78.30, SD = 18.40). On average, they were paying for one online

entertainment or news services (e.g., Netflix, Spotify, Blendle, The Economist) when the study took 1Three hundred and twenty-eight people started the online questionnaire, but only 43.6% finished it. Most of the partial

respondents quit the survey when they were warned not to use mobile devices to answer it (n=67) or when they were instructed to take a couple of minutes to navigate through the news website before the survey would continue automatically (n=73). One hundred and forty-three respondents got to the end of the questionnaire (N=143).

(13)

place (0 = none; 1 = one; 2 = two; 3 = three or more; M = .97, SD = .94). To test the hypotheses a between-subjects experimental design was adopted with Social proof (absent vs. weak tie vs. strong tie) as independent variable, paying intention as dependent variable and resistance as mediator.

Procedure

The link to the study was distributed on social media –i.e., Facebook and LinkedIn– and

participants were randomly assigned to one of the three experimental conditions. Respondents were told that the aim of the online survey was to examine “the effects of certain message characteristics on the usage of online news”. Therefore, they were asked to navigate through an online news website specifically designed for the purpose of this study, The Journal2.

Before being exposed to the newspaper, they were asked a series of questions in order to create a reader profile. These questions included some demographics, how interested in news they were, which their preferences in terms of news topics were, etc. Some of the answers were used to introduce the manipulation: after being instructed to take some time to navigate and read through The Journal's front page and articles, respondents were abruptly exposed to a subscription message asking them to pay to continue reading. Depending on the condition, this message would include no shared characteristics with other subscribers, one shared interest or several traits in common.

Social proof

The subscription message informed participants that they could not continue reading content unless

2Only the most prominent headlines and picture on the front page were accessible in order to avoid overcomplicating the design of the stimuli. Note that respondents were exposed to The Journal by embedding HTML code from the prototyping tool invision (www.invisionapp.com) into the Qualtrics online survey. Before the front page would appear on the screen the prototype had to load all the images (each of the articles) it contained. The inclusion of more articles would have made the loading time even larger, which might have caused some participants to quit. The Journal's front page was adapted from Reuters World News front page (www.reuters.com/news/world) and the seven articles available were adapted from Reuters investigates (www.reuters.com/investigates), a section of Reuters that publishes in-depth, award winning reports (IRE, 2014; SABEW, 2015; Reuters, 2017; European Press Prize, 2017;). These items included, for instance, a story on Putin's plan to build a health clinic for elite officials or another on how lead poisoning affects areas across Los Angeles. See Appendix or visit https://projects.invisionapp.com/share/TPC81QA3N for an overview.

(14)

they paid for it. This message addressed directly each participant by the username previously chosen by him/her, but was different depending on the experimental condition –see figure 2–. In the no social proof condition, the message only included a possible justification for paying –that producing high quality journalism costs money–, and the payment request itself (“The Journal depends on readers like you to bring you high quality, independent journalism. Subscribe now to join our community and continue enjoying our content”). In the weak tie social proof condition, the message was identical, but it also informed the reader that he/she had a shared interest with most subscribers (“More than 60% of our subscribers share with you an interest in 'topic'”). In the strong tie social proof condition more common characteristics were added. Thus, the reader was told that he/she shared with most subscribers a certain age, interest (as in the previous condition) and social media use (“Our subscribers share with you a similar age ('years old' on average). More than 60% of them are also interested in 'topic' and, like you, spend 'hours' per day on social media”).

Figure 2. Subscription messages shown in the weak (left) and strong Social proof conditions.

The above described manipulation was decided so that respondents would identify more strongly (strong tie) with other subscribers that share with them age, interest and social media use (strong

(15)

social proof), and less strongly with those having just a similar interest in common (weak tie; weak social proof). Support for this manipulation choice was inferred from previous literature. For instance, Stok et al. (2012) conducted a pilot test to conclude that participants identified more with Dutch students than with the Dutch population. Similarly, it was expected in the current study that respondents would feel more close to other subscribers when they share three characteristics (age, topic interest and social media use) with them than when they share only one (topic interest).

Paying intention

After exposure to the subscription message, paying intention was measured using four items, each one presented as a particular payment scheme: “How likely it is that you would pay a monthly subscription fee of less than 10 euros / between 10 and 15 euros / a single article fee of less than 0.50 euros / between 0.50 and 1 euros” (0 = extremely unlikely; 100 = extremely likely; Cronbach's

α = .79, M = 21.45, SD = 20.79). These amounts include the average online subscription prices that

newspapers in the United States and most European countries charge (Owen, 2016, 2017).

Resistance: qualitative measure

Respondents were provided with an open-response field and instructed to describe briefly their thoughts when they saw the subscription message (“What did you think when you saw the message in the red box asking you to subscribe to The Journal? Please state your first thoughts in one or two sentences”). This thought-listing task was incorporated to the questionnaire as a qualitative measure of resistance in order to complement the quantitative measure included in the model as mediator.

Resistance: quantitative measure

Resistance to subscription messages was assessed using 10 items from a questionnaire developed by Fransen et al. (2013) to measure resistance strategies. These items were adapted to measure

(16)

validation (“I thought about people who do not approve of paying for online news”), attitude bolstering (“I thought about the arguments I have for my opinion about paying for online news”) and source derogation (I had negative thoughts about the news website The Journal as a

brand/product). They were used to measure resistance as a whole (1 = strongly disagree; 7 = strongly agree; Cronbach's α = .80, M = 3.74, SD = 1.03). Similarly, variables for each of the resistance strategies (two items each ) were also assessed and subsequently computed: social validation (r = .71, p < .001; M = 3.36, SD = 1.60), counterarguing (r = .38, p < .001), attitude bolstering (r = .71, p < .001; M = 4.12, SD = 1.57), assertions of confidence (r = .61, p < .001; M = 3.38, SD = 1.56) and source derogation (r = .76, p < .001; M = 3.85, SD = 1.61).

Manipulation check

To check whether respondents perceived the manipulation as expected a single item was included asking them about the content of the subscription message (1 = the message did not include any shared characteristics with other subscribers; 2 = the message did mention a common interest; 3 = it mentioned several common characteristics such as age, interest and social media use).

Furthermore, in the absence of a pilot test it was necessary to include two more questions in the questionnaire to check whether identification with the groups worked as expected. This allowed me to test whether the inferences from previous literature were correct (i.e., that respondents would feel more close to other subscribers when they share three characteristics with them than when they share only one; see above Social proof). Respondents were asked to what extent they identified with a group of people having just one interest in common with them (0 = Not at all identify; 100 = Strongly identify; M = 69.71, SD = 24.06) or with people sharing with them age, interest and social media use (0 = Not at all identify; 100 = Strongly identify; M = 65.65, SD = 29.34).

(17)

Debriefing

After completing the questionnaire, respondents were thanked and debriefed (“The news website The Journal and the figures provided in the subscription messages, such as 'More than 60% of our subscribers are about the same age as you', were created for their use in this study and are not real”).

Results

To check if the manipulation worked across the three conditions, a crosstabulation analysis was conducted with the three conditions, and the question regarding which message respondents recalled to have seen as variables. There was a statistically significant difference between the three groups, X2 (4, N = 143) = 84.46, p <.01. A z-test with Bonferroni-adjusted p values confirmed significance in the expected way: 69.4 per cent of respondents in the no social proof condition said they were actually exposed to the message including no social proof, while the percentages of respondents in the other two conditions that selected that option were 21.3 per cent and 10.6 per cent (see Table 1). The manipulation check did show however that as many as 44 respondents did not select the expected answer –the one in accordance with their condition– and should have, therefore, been excluded from further analyses. Their inclusion in the study was justified based on the fact that social proof might have an at least partially unconscious effect on receivers.

Nevertheless, it was decided to run the analyses using this smaller subsample as well (n = 99). I also checked whether respondents did actually identify more strongly with the message in the strong social proof condition than in the weak social proof condition. When comparing their answers, only 63 participants said that they identified with the groups in the expected way, the difference between the identification with the strong tie referent group and the weak tie referent group being higher than zero. Though it was not the purpose of this study to check what constitutes strong and weak social proof, these results point to a limitation which will be discussed further later.

(18)

Table 1

Manipulation check

The subscription message mentioned... Condition … no commoncharacteristics

with subscribers … one common intererest with most subscribers … several common characteristics with most subscribers No Social proof (n = 49) 34 (69.4%)* 10 (21.3%) 5 (10.6%) Weak Social proof (n = 46) 9 (18.4%) 30 (63.8%)* 7 (14.9%) Strong Social proof (n = 48) 6 (12.2%) 7 (14.9%) 35 (74.5%)*

Note. * indicates a percentage of respondents significantly different than the other two in each condition (p<.05)

Randomization

Randomization across the three conditions (no social proof, n = 49; weak social proof, n = 46; strong social proof, n = 48) was successful for age, as shown by the results of a one way ANOVA conducted with experimental condition as independent variable and age as dependent variable, F (2, 140) = .19, p = .824. There was no significant difference in terms of age. No significant difference was found either in terms of gender, X2 (2, N = 143) = .85, p = .656. Another one way ANOVA with education as dependent variable showed that randomization was also successful, F (2, 140) = .18, p = .840, with no statistically significant differences in terms of education across the three conditions. Two more ANOVAs were conducted with online news use (2 items, 1 = less than once a week; 8 = more than ten times a day; r = .57, p < .001; M = 5.13, SD = 1.78), F (2, 140) = .49, p = .615, and interest in news, F (2, 140) = .001, p = .999, showing a successful randomization regarding these variables as well. Randomization also worked for purchase of online services, X2 (6, N = 143) = .47,

p = .998, and regarding the rating of The Journal (“Please rate The Journal as journalistic product”;

(19)

Paying intention

To test H1 a linear regression was conducted with paying intention as dependent variable and resistance as independent variable. The regression showed that the model was not significant, F(1, 141) = 1.41, p = .237. Consequently, resistance did not predict paying intention in the current sample, b = 2.01, b* = .10, t = 1.19, p = .237, and H1 had to be rejected. Though not significant, the result pointed however to an unexpected increase in paying intention as resistance increased as well.

Table 2

Several linear regressions conducted with paying intention as dependent variable (N = 143)

Predictor b SE b* t Resistance 2.01 1.69 .10 1.19 Social validation 1.55 1.09 .12 1.43 Attitude bolstering 2.51* 1.09 .19 2.30 Assertions of confidence 3.41* 1.08 .26 3.15 Source derogation -2.21* 1.07 -.17 -2.06

Note. Results of several linear regressions, not of a regression with several independent variables. * indicates p<.05

Taking each component of resistance independently, attitude bolstering, F(1, 141) = 5.30, p < .05, asertions of confidence, F(1, 141) = 9.93, p < .05, and source derogation, F(1, 141) = 4.26, p < .05 also predicted significantly paying intention, while social validation did not, F(1, 141) = 2.04, p = . 155 (See Table 2). Attitude bolstering, b = 2.51, b* = .19, t = 2.30, p < .05, and asertions of

confidence, b = 3.41, b* = .26, t = 3.15, p < .05, predicted 2.9% (adj. R2 = .029) and 5.9% (adj. R2 = .059) of the variance in paying intention. In both cases intention to pay for online news rose as did

(20)

attitude bolstering or asertions of confidence. Different was the effect of source derogation, b = -2.21, b* = -.17, t = -2.06, p < .05. Paying intention decreased by 2.21 with each additional unit of source derogation, which predicted 2.2% of the variance in paying intention (adj. R2 = .022).

Subscription messages, “what a nuisance”

Although no statistical analyses were performed on the basis of the qualitative measure of

resistance, the respondents' thoughts were informally coded to have a more accurate view of what their attitudes were towards subscription messages and paying for online news. This informal results were meant to complement somehow the quantitative measurement of resistance.

Most respondents (n = 29) said to be “annoyed” and “frustrated” when confronted with the subscription messages. “What a nuisance. I can't be bothered paying for this service”, declared one of them. At least 14 participants went even further and manifested their intention to “close” and “exit” the website as a reaction to the payment request, a possible avoidance resistance strategy: “I would be off it straight away and never return, as I have done in the past”. Another respondent referred to the subscription request as “the kind of message that you ignore”. Some even admitted to be willing to use loopholes to read The Journal for free (“How can I bypass the paywall?”).

The culture of free that dominates online news consumption was also very present in the thoughts of the respondents, as observed in messages such as “I'm not going to pay for usual news” or “I don't think common knowledge information should cost money”. The widespread availability of free content was one of the reasons most respondents offered for not willing to pay, as they did not seem to perceive a particularly high quality in the contents of The Journal: “There are many options for free and I will only subscribe if it is a great journal”, “I am not going to pay for the very same content I can access online for free", “I can read about the same things for free elsewhere”.

Several participants showed concerns about the payment procedure (“I was worried if I clicked next I would get an automatic charge”), and some reacted negatively to the level of

(21)

customization provided in the subscription message (“Where did they get this information from? It felt kind of weird to have this personalization call to subscribe”), even questioning the accuracy of the figures presented (“I thought the message was fake because it was too personalized”).

Few respondents showed support for the fact of paying for online news and said they “understand” the reason for the request. “I thought that I should help journalism and subscribe”, said one of them. While the previous groups of respondents do seemed to experience resistance to subscription messages, these were probably the ones feeling the lowest levels of resistance.

Table 3

Intention to pay for online news

How likely is it that you would pay for reading The Journal?

Condition M SD

No Social proof (n = 49) 22.58 21.50

Weak Social proof (n = 46) 18.21 19.72

Strong Social proof (n = 48) 23.42 21.12

Note. Measurement for paying intention goes from 0 = extremely unlikely, to 100 = extremely likely

Social proof

A one way ANOVA with paying intention as dependent variable and social proof (no vs. weak vs. strong) as independent variable was conducted to test H2a and H2b. The assumption of

homogenous variances between the groups was met, Levene's F (2, 140) = .44, p = .642. Results of this ANOVA showed no significant effect of Social proof on paying intention, F (2, 140) = .84, p = . 433. Being provided with no social proof (M = 22.58, SD = 21.50), weak social proof (M = 18.21,

SD = 19.72) or strong social proof (M = 23.42, SD = 21.12) did not lead to a statistically significant

(22)

To see if similar results would be obtained among respondents that correctly recalled the

subscription message, the same analysis was run taking into account only those participants (n = 99). Therefore, another ANOVA showed no significant difference between absence of social proof (M = 20.23, SD = 21.70), provision of weak social proof (M = 17.14, SD = 17.80) or provision of strong social proof (M = 20.22, SD = 20.77) on paying intention, F (2, 96) = .24, p = .785.

Resistance

Hypotheses H3a and H3b were tested using another one way ANOVA with resistance as dependent variable and Social proof as independent variable. Homogeneity of variances was met, Levene's F (2, 140) = .77, p = .465. Results showed no significant effect of Social proof on resistance, F (2, 140) = .93, p = .397. Being provided with no social proof (M = 3.74, SD = 1.05), weak social proof (M = 3.59, SD = 1.08) or strong social proof (M = 3.88, SD = .96) did not lead to a significantly different resistance to subscription messages. Therefore, H3a and H3b must be rejected.

Again I also checked whether taking into account only those respondents that selected the right option in the manipulation check would make a difference in terms of these results. Another one way ANOVA showed no significant difference between absence of social proof (M = 3.69, SD = 1.06), provision of weak social proof (M = 3.52, SD = 1.04) or provision of strong social proof (M = 3.89, SD = 1.04) on resistance to subscription messages, F (2, 96) = 1.01, p = .370.

In accordance with the results presented so far, a relationship between Social proof and paying intention mediated by resistance could never have occured, since no significant relationship was observed between social proof and paying intention, or social proof and resistance or resistance and paying intention. Therefore, H4 also had to be rejected for the sample of this study.

Discussion

(23)

resistance-neutralizing strategy, improving the persuasiveness of subscription messages used by newspapers to attract new paying readers. The results showed however that provision of social proof did not affect significantly the level of resistance displayed towards those messages, neither did it cause a

significant change in respondents' intention to pay for online news. It could not be proved either that resistance is significantly related to paying intention. More surprisingly, some resistance strategies did significantly predict paying intention, but in the opposite way as expected. Only source

derogation was proved to be negatively related to paying intention, with respondents showing significantly lower intention to pay for online news as they displayed a higher source derogation. Finally, the role of resistance as a mediator of the relationship between social proof and paying intention could not be demonstrated as, in the current sample, that relationship was not significant.

As expected from the literature, paying intention was extremely unpopular among the respondents of this study (M = 21.45, SD = 20.79, Mode = .00) and this highly skewed distribution might explain, to some extent, the somehow surprising results obtained. This is particularly so for the non-significant positive relationship observed between resistance and paying intention.

Moreover, the moderate and relatively stable level of resistance to subscription messages (M = 3.74, SD = 1.03) across respondents might as well indicate that other factors apart from resistance to subscription messages are better predictors of intention to pay for online news. It might be the case, for example, that deeply rooted attitudes towards (not) paying for online news affect paying intention regardless of the subscription messages consumers are exposed to. If this is true, which should be further investigated, social proof would fail to predict resistance as well as paying intention, as the results of the current study show. Provision of social proof appears insufficient to change those well-established attitudes about (not) paying for online news and efforts sustained over time may be needed to change those perceptions, in a similar way as changing societal approval of smoking took almost 30 years and enormous resources (Mermelstein & Riesenberg, 1992). This may be particularly relevant for Southern and Central European countries where there is

(24)

not a subscription culture to build on (Newman et al., 2017).

In line with the previous argument is the fact that people are usually quite certain about their opinion regarding paying for online news. This can also help explain the results observed in the current study, since social proof is more likely to enhance persuasion in situations of uncertainty where people rely on others' opinions and behaviours (Cialdini, 2001; Koch & Benlian, 2015).

Moreover, Reingen (1982) refers to a length-of-list effect observed in his experiments about inducing compliance. The expected effect –donating money– was obtained “only when the number of other compliers was sufficiently large” (p. 114). In the current study, the figures displayed in the subscription messages (“More than 60% of our subscribers...”) showed to respondents may have been perceived as insufficient to comply with the payment request.

Finally, despite the high interest in news declared by the respondents (M = 78.30, SD = 18.40, Mode = 100), they may have not given much importance to the fact of not being able to read The Journal. According to reactance theory (Brehm, 1966; Brehm & Brehm, 1981), the greater the importance of the freedom threatened, the greater is the amount of reactance aroused. Participants did not perceive The Journal as a particularly good journalistic product and, aware of the numeous alternatives to get news for free, they may not have experienced a great deal of resistance .

Limitations

There are some important limitations to take into account when assessing the validity of this study. First of all, a pilot test would have allowed to check in advance whether the different values of social proof displayed (i.e., weak vs. strong) work in the intended way. Unfortunately, this was not possible due to time constraints and the identification check included in the questionnaire revealed that 80 respondents did not identify as expected, not perceiving a closer tie with the group of subscribers referred to in the strong social proof condition. Although it was not a goal of this study to test what are the defining elements of weak and strong social proof, the fact that a considerable

(25)

number of respondents did not confirm the assumption behind the distribution of conditions compelles to interpret the results cautiously and makes it necessary to replicate this study.

The decision to display the subscription request in the form of a pop up message was intented to lead to a higher resistance. However, it may as well have affected the respondents' paying intention, as seems evident from the qualitative measure of resistance, where several participants shared their annoyance towards this advertising technique despite not having a strong attitude against paying for online news: “I thought the pop-up was disruptive and annoying. I understand that newspapers are struggling and they need to make money but they could do the pop-up when you enter the site, not in the middle of me reading an article”. This represents another limitation for the interpretation of the results and should be taken into account for future reseach.

References

Amblee, N.C. & Bui, T.X. (2012). Value proposition and social proof in online deals: An exploratory study of Groupon.com. Proceedings of the 14th Annual International

Conference on Electronic Commerce, 294-300

Brehm, J. W. (1966). A theory of psychological reactance. New York: Academic Press.

Brehm, J.W. & Brehm, S.S. (1981). Psychological reactance: A theory of freedom and control. San Diego, CA: Academic Press.

Chyi, H.I. & Lee, A.M. (2013). Online news consumption. A structural model linking preference, use, and paying intent, Digital Journalism, 1(2), 194-211

Cialdini, R. (2001). Harnessing the science of persuasion, Harvard Business Review, 79 (9), 72-79 Conner, M. & Sparks, P. (2005). Theory of planned behaviour and health behaviour. In M. Conner

& P. Norman (Eds.), Predicting health behaviour: Research and practice with social

cognition models (2nd ed., pp. 170-222). Maidenhead, UK: Open University Press

(26)

paywall, Cyberpsychology, behaviour, and social networking, 15(12)

Doctor, K. (2017). Newsonomics: Trump may be the news industry’s greatest opportunity to build a sustainable model. NiemanLab. Retrieved from

http://www.niemanlab.org/2017/01/newsonomics-trump-may-be-the-news-industrys-greatest-opportunity-to-build-a-sustainable-model/

Donatello, M. (2002). What consumers tell us about paying for news online. Econtent, May 2002, 36-40

European Press Prize (2017). Comrade capitalism. The Investigative Reporting Award 2015. Retrieved from http://www.europeanpressprize.com/article/comrade-capitalism/ Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117–140 Fransen, M.L., Ter Hoeven C.L., & Verlegh P.W.J. (2013). Strategies to resist advertising. In Botti

S, Labroo AA (eds). Advances in Consumer Research 41. Association for Consumer Research: Duluth, MN; 792.

Fransen, M. L., Smit, E.G., & Verlegh, P.W.J. (2015a). Strategies and motives for resistance to persuasion: an integrative framework. Frontiers in Psychology, 6, 1201

Fransen, M.L., Verleghb, P.W.J., Kirmanic, A., & Smit, E.G. (2015b). A typology of consumer strategies for resisting advertising, and a review of mechanisms for countering them.

International Journal of Advertising, 34 (1), 6-16

Goyanes, M. (2014). An empirical study of factors that influence the willingness to pay for online news. Journalism Practice, 8(6), 742-757

Henriksson, T. (2016). Full highlights of World Press Trends 2016 survey. World Association of

Newspapers and News Publishers. Retrieved from

http://www.wan-ifra.org/articles/2016/06/12/full-highlights-of-world-press-trends-2016-survey IRE (2014). 2013 IRE Award winners. Investigative Reporters & Editors. Retrieved from

(27)

Jacks, J.Z. & Cameron, K.A. (2003). Strategies for resisting persuasion. Basic and Applied Social

Psychology 25(2), 145-161.

Kahin, B. & Varian, H.R. (2000). Internet publishing and beyond: The economics of digital information and intellectual property. Cambridge, MA: MIT Press

Kammer, A., Boeck, M., Hansen, J.V. & Hauschildtd, L.J.H. (2015). The free-to-fee transition: audiences’ attitudes toward paying for online news. Journal of Media Business Studies, 12 (2), 107–120

Knowles, E.S. & Linn, J.A. (2004). Resistance and Persuasion. Mahwah, NJ: Lawrence Erlbaum Associates

Koch, O.F. & Benlian, A. (2015). Designing viral promotional campaigns: how scarcity and social proof affect online referrals, 36 International Conference on Information Systems

Kurtz, H. (1993) Click! On-line newspaper is a mixed read. The Washington Post. Retrieved from https://www.washingtonpost.com/archive/politics/1993/10/23/click-on-line-newspaper-is-a-mixed-read/c9b05b61-3732-437c-9a65-2e5779158e92/?utm_term=.1022884fa302

Li, T. & Pavlou, P.A. (2014). How does social influence really affect consumer decisions? Insights from an eye tracking study. Fox School of Business Research Paper, 14-033

Lichterman, J. (2016a). After Trump’s win, news organizations see a bump in subscriptions and donations. NiemanLab. Retrieved from http://www.niemanlab.org/2016/11/after-trumps-election-news-organizations-see-a-bump-in-subscriptions-and-donations/

Lichterman, J. (2016b). The Winnipeg Free Press’ bet on micropayments will generate about $100,000 in revenue this year. NiemanLab. Retrieved from

http://www.niemanlab.org/2016/04/the-winnipeg-free-press-bet-on-micropayments-will-generate-about-100000-in-revenue-this-year/

Lichterman, J. (2016a). After Trump’s win, news organizations see a bump in subscriptions and donations. NiemanLab. Retrieved from

(28)

http://www.niemanlab.org/2016/11/after-trumps-election-news-organizations-see-a-bump-in-subscriptions-and-donations/

Mermelstein, R. J., & Riesenberg, L. A. (1992). Changing Knowledge and Attitudes About Skin Cancer Risk Factors in Adolescents; Health Psychology, 12(6), 371-376

Newman, N., Fletcher, R., Levy, D.A.L., & Nielsen, R.K. (2016). Digital news report 2016, Reuters Institute for the Study of Journalism. Retrieved from

http://reutersinstitute.politics.ox.ac.uk/sites/default/files/Digital-News-Report-2016.pdf Newman, N., Fletcher, R., Kalogeropoulos, A., Levy, D.A.L., & Nielsen, R.K. (2017). Digital news

report 2017, Reuters Institute for the Study of Journalism. Retrieved from

https://reutersinstitute.politics.ox.ac.uk/sites/default/files/Digital%20News%20Report %202017%20web_0.pdf

Outing, S. (2009). What a persuasive-technology psychologist can tell us about paying for news online, Editor & Publisher. Rertieved from

http://www.editorandpublisher.com/columns/what-a-persuasive-technology-psychologist-can-tell-us-about-paying-for-news-online/

Reuters (2017). Awards. Reuters News Agency. Retrieved from https://agency.reuters.com/en/about-us/awards.html

SABEW (2015). 2014 BIB Winner’s list. Society of American Business Editors and Writers. Retrieved from https://sabew.org/2015/04/2014-bib-finalists-list/

Sehl, A., Nielsen, R.K., Cornia, A., & Simon, F. (2017). Pay models in European news. Reuters Institute for the study of journalism. Retrieved from

http://www.digitalnewsreport.org/publications/2017/pay-models-european-news/ Prentice, D. (2008). Mobilizing and weakening peer influence as mechanisms for changing

behavior. Implications for alcohol intervention programs. In M. J. Prinstein & K. A. Dodge [Eds.], Understanding peer influence in children and adolescents (pp. 161–180). New York, NY: Guildford Press

(29)

Reingen, P.H. (1982). Test of a list procedure for inducing compliance with a request to donate money, Journal of Applied Psychology, 67(1), 110-118

Salwen, M. B., Garrison, B., & Driscoll, P. D. (2004). Online news and the public. Routledge Stok, F.M., de Ridder, D.T.D., de Vet, E. & de Wit, J.B.F. (2012). Minority talks: the influence of

descriptive social norms on fruit intake. Psychology & Health, 27(8), 956–70

Wang, S. (2016a). The Wall Street Journal is changing up its paywall, offering guest passes and expanded link-sharing on social. NiemanLab. Retrieved from

http://www.niemanlab.org/2016/08/the-wall-street-journal-is-changing-up-its-paywall-offering-guest-passes-and-expanded-link-sharing-on-social/

Wang, S. (2016b). The Wall Street Journal is confident its “bendier” paywall will draw the paying readers it needs to survive. NiemanLab. Retrieved from

http://www.niemanlab.org/2016/12/the-wall-street-journal-is-confident-a-loosened-paywall-will-draw-the-paying-readers-it-needs-to-survive/

Williams, A.T. (2016). Paying for digital news: The rapid adoption and current landscape of digital subscriptions at U.S. newspapers, American Press Institute. Retrieved from

https://www.americanpressinstitute.org/publications/reports/digital-subscriptions/single-page/

(30)

Appendix

(31)

Referenties

GERELATEERDE DOCUMENTEN

The study produces four findings: (1) the presence of online reviews has a positive effect on consumers’ purchase intention, (2) product perception does not mediate this

At this point, recommender systems shift from using explicit feedback willingly provided by users to implicit feedback that is inferred from user behaviour.. Matrix factorization is

Thus, for a consumer that is highly involved with a product category, their involvement level has a moderating impact on the influence the online- and offline social

a) Die komponente van die Stanislavski-sisteem kan nie sonder die konteks van die totale omvang van sy ideologie ondersoek word nie. b) Die afhanklikheid van die komponente van

Modellen zijn zeker niet foutloos, maar het onderzoek naar conversies heeft mijns inziens geen baat bij het alternatief, te weten meer

5.18: Different phase patterns that constitute the total phase pattern imprinted onto the SLM for the generation of overlapped spot patterns by two laser beams of different

In episode three, the editor/author utilises bodies and spaces such as the king, the Babylonians, Daniel, the lions’ den, the prophet Habakkuk and food to demonstrate the

In the first phase of digital divide research (1995-2005) the focus was also on the two first phases of appropriation of digital technology: motivation and physical access..