Nudging people towards desired
behavior, by using a socially proof
framed message.
Master’s Thesis’
T.P. Vermeiren (10003368)
Thesis supervisor
Dr. P. Verhoeven
Graduate School of Communication
Corporate Communication
Abstract
This study examines in a field experiment the nudging power of a socially proof framed message in corporate communication. By presenting customers in a restaurant with a framed message, the motivation to participate in the desired behavior of paying by card is studied. The message, both in English and Dutch, included a descriptive norm (“Join the 73% of our guests that pay by card”). The experiment was held in three restaurants, all three part of a restaurant chain. The payment method was monitored over 14 days, by measuring the number of transactions and the revenue per day per restaurant. The results of the descriptive statistics and the ANOVA test indicated that there was no proven nudging power of the message. The experimental groups had a lower percentage of card payments and a lower percentage of pin revenue than the control group. But a notable result was found at the mean revenue of the first experimental group. This group had the highest mean pin revenue, which may indicate that the message works better when customers have to pay a higher amount. Therefore, in the discussion it is suggested that future research should take this finding and other limitations in consideration and should investigate the usage of this frame more thorough. Because the nudging power of frames should not be underestimated.
Introduction
If we go to a restaurant, people tend not to go to an empty restaurant. Most preferably we want to go to a restaurant with a big cue. But why? Is the restaurant with the big cue a lot better, or do we think the restaurant is better because of all the people waiting in the cue? As American philosopher Eric Hoffer once said: “When people are free to do as they please, they usually imitate each other”
(Hoffer, 1955). People are just like herd animals, and we can’t stop the subconscious impulse to imitate each other. This mimicry can derive from imitating the facial expressions, mannerisms and postures of the persons which we are interacting with.
And all for the reason that one's behavior matches that of others in the current social environment (Chartrand & Bargh, 1999). Moreover, people don’t only have the tendency to copycat the behavior of our fellow others, but also copy the thoughts of others. Imitating the thoughts of others and thinking as a group is referred in the literature as swarm intelligence or wisdom of the crowd (Krause, Ruxton, & Krause, 2010).
Thus, it is in the nature of humans to go along with the group. Marketers are well known with this natural instinct, and are using this information in their marketing by using social proof (Tanford & Montgomery, 2015). Similar to the idea of wisdom of the crowd, social proof is using the idea that people tend to do what most people do, and tend to think that what most of the people do is right (Aronson, Wilson, & Akert, 2005). Sites like ‘booking.com’ or ‘amazon.com’ are using the principle of social proof to nudge people to, for instance, rent a room. With ques like ‘more than 500 times rented’ or ‘268 people rate this hotel with a 9.2’ and, ‘just rented by
somebody 50 minutes ago’ marketers convince customers to rent a room and nudge the customers towards purchasing their product. Research has taught that this kind of techniques where social proof is used to nudge people towards purchasing a product is a successful method (Tanford & Montgomery, 2015). But the use of social proof to convince people to do something doesn’t only work as a marketing tool. It can also be well adapted in corporate messages. Goldstein, Cialdini, and
Griskevicius (2008) used a social proof framed message to motivate hotel guests to reuse their bath towels. By presenting the guests with a card where a social norm was stated on, the researchers successfully nudged people towards the desired behavior of reusing the hotel towels.
To conclude the previous mentioned theories, it can be assumed that people could be nudged towards desired behavior like buying a product. This nudging can be resurrected via the use of framing. By presenting individuals with a framed message, people could be motivated by the message and eventually carry out the
desired behavior. As mentioned before people have the subconscious urgent to copy others, and don’t want to be different than others. Therefore, it could be
suggested that if people are given a certain social norm, people want to act towards the given norm. Based on these theories it could be concluded that a social norm, or social proof, could be a good way to nudge people towards desired behavior. As Goldstein et al. (2008) explained the presence of the nudging power of social proof in their towel study, it could be expected that social proof can be applied in more situations. In the article social proof is used to actively stimulate people to reuse their towels. But, does this also works in situations where people are motivated to do a small behavior like paying their groceries? Therefore, this thesis studies, via a field experiment, if individuals could be successfully nudged toward small behavior like paying something using a social proof framed message. This results in the following research question:
RQ: Can a corporate message, where a social proof frame is presented in, nudge individuals towards desired behavior?
This research can be a contribution to corporate communication research, where framing is not often studied in experimental research. Within communication research, frames in texts are more often studied. Therefore, this thesis could give new insights in framing research. Moreover, this research gathers new information on the corporate use of message framing via social proof. Furthermore, the concept of social proof as a frame is not thoroughly investigated within communication science. Therefore, the research on social proof frames can contribute to the scientific debate on framing.
Finally, this research could have multiple practical implications. When this research concludes that using social proof frames in corporate messages can have an impact on nudging people towards certain behavior, organizations can use this
information to optimize their corporate messages. This could increase the efficiency of the corporate use of message framing. Furthermore, the corporate messages of organizations could then, without organizations putting much effort in it, have a higher impact and better results. Therefore, this research could have high practical, as well as, scientific implications.
Theoretical Framework
Framing
Framing is one of the core subjects within communication science. Communication scientist Robert Entman explained the term framing in his famous article of 1993 as: “selecting some aspects of apperceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described” (Entman, 1993, p.52). Stating in other words that a frame can be used to give reason to a reality, or shape a reality via selection aspects in that given reality.
But communication science is not the only discipline investigating the
concept of framing. Framing is a widely investigated construct in various disciplines. The term finds its origin in the book of anthropologist Bateson (1972) where the author argued that framing can be seen as "a spatial and temporal bounding of a set of interactive messages” (p.191). Furthermore, the concept of framing finds its foundation in the research disciplines of communication science, economics, psychology, and sociology.
An important part of the foundation of framing studies lays in the sociology. Where sociological scientist Erving Goffman (1974) described framing as a tool to label schemes of interpretation. These schemes of interpretation can help
individuals or groups via rendering meaning, organizing experiences, and guiding actions. In his theory Goffman talks about ‘laminations’ and ‘keyings’. The
laminations are the frames added to a certain situation, to help individuals to give reason to that situation. Keying is when the individual gives meaning to a situation or interaction using its laminations and its primary framework. The sociological framing literature focusses on the process of constructing a reality based on the interpretation of the posed frames that the individual receives.
Snow, Rochford Jr, Worden, and Benford, (1986) elaborate on the ideas of Goffman (1974) about framing and explain the importance of frame-alignment. Since framing has to do with multiple individual interpretive orientations, individuals or organizations can have conflicts of frames with each other. Therefore, it is suggested by Snow, et. al. (1986) individuals and/or organizations strive to a linkage between these frames, such as congruent and complementary interests, values, beliefs, and goals; explained as ‘frame alignment’. But the achieved frame alignment is only temporally variable and subject to reassessment and renegotiation. Therefore, it cannot be taken for granted. But once the temporally frame alignment is achieved, the alignment can result in an adherent and constituent mobilization (Snow, et. al., 1986).
Concluding the thoughts of the sociologists, a frame is a tool to shape a reality for an individual on a certain matter or situation to give reason to the given situation. Moreover, individuals can have different frames and therefore could strive to an alignment of these frames, and this alignment can result in an adherent and constituent mobilization.
Framing and nudging
If we elaborate on this idea of frame alignment and mobilization, it can be suggested that the alignment of frames could be used as a tool to enhance
participation. This is because of the earlier explained nature of individuals to strive for frame alignment. The idea of enhancing participation can be explained via the construct of nudging.
Nudging has become an umbrella term for an approach that can enhance behavioral change (Ölander & Thøgersen, 2014). The authors explain that this behavioral influence can derive from architecting the choice. This emphasizes the idea that a choice that someone makes mostly depends on how this choice is
presented. Behavioral scientists investigated this influence of choice architecture, by using framing. Psychologists Kahneman and Tversky (1991), investigated the
influence of framing on the outcome of the decision in decisions making processes. In their article the authors conclude the presence of a framing effect in decision making when the outcome is framed differently, resulting in their prospect theory. Tverksy and Kahneman (1979) used a gain and loss frame in a choice to investigate to influence on the outcome. Their study showed that people tend to choose a positive framed message because this experienced as an aversion of risk. So, when something is positively framed people are more likely to go along with this positive frame.
The researchers Kahneman and Tversky made an important theoretical base for experimental framing research. With their prospect theory they investigated the influence of using gain or loss terms, where losses loom larger than gains
(Kahneman & Tversy, 1991). Ganzach and Karsahi (1995) have continued the
research on the use of gain or loss terms and their influence on behavior. The results of the experiment of Ganzach and Karsahi (1995) indicated that a loss-framed
message had a stronger impact than the gain-framed impact. A higher percentage of the subjects were executing the desired behavior when the loss frame was used, indicating that the use of framing can be a tool to nudge people towards the desired behavior.
Further research on the use of loss and gain frames also explained the
influence of framing on behavior. Hong, Hossain, and List (2015) analyzed the use of framing manipulations in contests. The researchers conclude that gain or loss frames can influence the productivity of work teams by using rewards or punishments. The
team that was in the loss treatment was at least 35% more likely to win the contest than the team in the gain treatment, indicating that people in the loss treatment were changing the behavior more than in the gain treatment.
As literature explains, language is an important part in framing. Gain frame words, that indicate a reward or something positive, can nudge people towards a behavior. Looking at the power of words in messages Liebrecht, Hustinx, Mulken, and Schellens (2016) explain that the language intensity can strengthen the convincing power of messages, which shows positive effects in effect studies.
Framing in corporate communication
Research in the influence of frames in texts and messages is one of the main research areas within communication science. Framing studies in communication science literature focuses on the idea of the media frame, where the media frame is seen as “the central organizing idea for news content that supplies a context and suggests what the issue is through the use of selection, emphasis, exclusion, and elaboration” (Tankard, Hendrickson, Silberman, Bliss, & Ghanem, 1999, p.3). Framing theory is based on the general idea of message shaping. When some elements of a message are left out or put in, it creates a context and shapes a perception for someone. Within the society there are continuous conflicts and negotiations between frames to shape the perceived reality for individuals. This could be between various media, between the media and the public, or between organizations and media. The use of frames in the media, or by organizations, can be illustrated as the use of frames in a corporate setting.
The use of frames by media, or organizations in the media, can be explained in a model. This model is the integrated process model of framing (De Vreese, 2005). As explained before framing can be used as an influential tool to influence outcomes of choices or nudge people towards behavior. But companies or media can also use framing as a tool to shape the publics’ opinion about a certain matter.
In the article of De Vreese (2005) the author explains that the process of framing can be divided in frame building and frame setting. Where frame building refers to “factors that influence the structural qualities of news frames” (De Vreese, 2005, p.52). The process of frame building is a continuous interaction between, for instance, social movements and journalists. Finally, the outcome of the frame building process are the frames in the text.
The case of BP’s corporate crisis clarifies this process using frames to shape the publics’ opinion. The study of Schultz, Kleinnijenuis, Oegema, Utz, and Van Atteveldt (2012) explored the use of frames in the communication from BP to the media and the public, versus the frame of the media. During the crisis BP used a frame to dissociate itself from their responsibility for the cause and furthermore presenting themselves as being solving during the crisis. The frame BP presented was partly been taken over in the US media. This article explains that the use of frames in communication is a good way to influence the public opinion and create a new reality about a certain matter for actors like the public, politics or the media. The case explains the fact that within a corporate setting the use of frames is often a ‘game’ of frame negotiation between multiple actors who are involved with the matter.
However, framing of texts or messages is not only linked to framing theory were media frames are competing with the frames of organizations. The article of Hallahan (1999) explains the application of frames in PR. In the article the author clarified the potential use of framing for strategically creating PR messages and audience responses. In the article the seven types of framing that are applicable to PR are presented. One of these seven types of framing is framing of choices. Framing choices is when decisions are posed alternative in positive or negative terms, and where these terms influence choices in uncertain situations. This refers to the article of Tversky and Kahneman (1991) where loss and gain frames are
to avoid losses than obtaining gains because the loss looms larger than the gain. However, according to the prospect theory, this does not apply when literal gains and losses are framed (Tversky & Kahneman, 1991).
The fear of missing out, via framing an opt-out, can also be conceived by individuals as a loss. Framing an opt-out, where participation is framed as the default, a default is presented to individuals. This is a very effective way to enhance participation of individuals (Johnson & Goldstein, 2003). The use of a default to enhance behavior of individuals in situations is the basis for social proof theories.
Social proof
American psychologist Robert Cialdini (2001) investigated the science of persuasion using, among other things, framing. In the article Cialdini presents six principles can be used in persuasion. One of these principles is social proof. Cialdini (1987) explains the phenomena social proof as the tendency of individuals to see actions as more appropriate when others are also doing this. People generally belief that when a lot of people are doing something, this is the right to do, a social proof to do something. This social proof can be related to the idea of using the general social norm. These social norms, or descriptive norms, can motivate people their actions by informing the individuals what is likely to be an effective or adaptive behavior in that specific situation (Cialdini, Kallgren, & Reno, 1991). People are informed about the descriptive norm by their peers. Based on this information the individuals determine how to behave in that situation (Cialdini et al., 1991).
The motivation of people to adjust their behavior to social norms is
something which derives from the animal instinct (Krause, Ruxton, & Krause, 2010). Just like animals, people have the tendency to copy others in their current social environment (Chartrand & Bargh, 1999). Individuals even shape their interpretations of situations and responses to situations, based on the behavior of others in the social environment of these individuals (Bearden & Etzel, 1982). This effect even
looms larger in novel or uncertain situations (Griskevicius, Goldstein, Mortensen, & Kendrick, 2006).
Research has taught that when social proof is used in message framing to increase desired behavior, this desired behavior is more likely to be reflected when people are presented to a social norm (Goldstein et al., 2008). This indicates the possibility of nudging people towards behavior using social proof in message frames. Research of Ölander and Thøgersen (2014) confirms this suggestion.
If a people tend to go along with the social norm, it can be expected that when people are presented with a certain social norm, individuals will go along with this social norm. If you want to nudge people towards certain behavior, a social norm could be a right motivation to get people to do the behavior. Therefore, it is expected that the desired behavior presented to individuals via a social proof framed message is more likely to be carried out by the individuals, resulting in the first hypothesis:
H1: A social proof framed message will increase the performance of the desired behavior.
Furthermore, it can be assumed that the performance of the undesired behavior will decrease because of the presented social norm. Resulting in the second hypothesis:
H2: A social proof framed message will decrease the performance of the undesired behavior.
Figure 1. Conceptual model
Methods
Research design
To the study the research question and the hypotheses, a field experiment with a 2x3 factorial design is chosen. The experimental design consists of two experimental groups and one control group, where in the experimental groups the participants are exposed to an experimental treatment in the form of a message. The control group won’t be exposed to this treatment.
A field experiment is characterized by a scientific intervention in the real world. The independent variable, the message, is being manipulated in the real life setting but the extraneous variables cannot be controlled in the experiment. The motivation to choose a field experiment is because of the fact that the behavior in a field experiment, because of its natural setting, is more likely to reflect real life. This is because the experiment is not being held in a laboratory, but in a natural
everyday situation. Because the experiment is in the real life, it can be assumed the likelihood of reflecting real life will be high. Furthermore, the natural setting of the field experiment enables it that the participants of the experiment do not know that they are participating in an experiment. This is positive because of its increased likelihood to exclude response biases like the social desirability bias. Furthermore, laboratory experiments are lacking external or ecological validity, therefore the field experiment can increase these validities.
The goal of the experiment is to identify possible behavioral changes which could occur after the exposure to a framed message. In the field experiment the message is aiming to nudge customers towards a desired payment method in a restaurant.
Experimental environment
The experiment will be held in a number of restaurants. These restaurants are part of a restaurant chain, consisting of four restaurants. The four restaurants are divided over the city center of Amsterdam. Three of the four restaurant are based on the concept of self-ordering at the register. The customers can eat their order at the restaurant or can take away. In these restaurants the customer is asked to pay
immediately at the register in cash or by card. At the other restaurant the service is based on table service, where the guests can have a seat if they want to eat at the restaurant and pay afterwards at their table.
Considering the fact of the table service at this restaurant there is chosen to not include this restaurant in the experiment. This is because of the possibility of distorted measurements during the experiment. The measurements during the experiment could be disrupted in multiple ways. Because of its table service the employee otherwise needs to present the guest at the table with the message. This could have a big impact on the results of the study and therefore there is chosen to not rely all these responsibilities on the employees. Taken this possibility of
uncertain and disruptive outcomes in consideration, the restaurant with table service will be excluded from the experiment.
The other three, all included in the experiments, are as restaurants and in their service fully comparable. But, the restaurants are not completely comparable in their customer composition. Two of the three restaurants are situated near touristic hotspots. One restaurant is on the opposite of the Amsterdam zoo ‘Artis Magistra’ and near NEMO. The other restaurant is situated in the popular ‘jordaan’ near the
Anne Frank house and the ‘nine streets’. On Google’s heat map of tourism, it can be seen that near and at the locations of these two restaurants the amount of tourists is higher than at the other location (Sightsmap.com, n.d.). Also, as an employee
working at the restaurants, it seems to be that these ‘touristic locations’ have a higher amount of tourists coming to the restaurants. Therefore, it can be assumed that these two restaurants with a higher amount visiting tourists are more or less comparable and it chosen to assign one of these restaurants to the control group and the other restaurant to the experimental group. The third restaurant, with less visiting tourist, will also be assigned to the experimental treatment.
Participants
The experiment will collect data of the three restaurant over a 14-day span. The restaurants are open for 11 hours, from 12 am until 11 pm. The message will be presented to the customers for 154 hours per restaurant. The customers in the restaurants are not aware that they are participating in the experiment.
Materials
During the experiment, the participants in the experimental groups will be
presented to an intervention. This intervention will consist of a social proof framed message presented on a plastic card next to the register. The message will be both in English, for the tourists, and in Dutch (see Figure 2). The messages presented on the card are:
• “Join the 73% of our guests that pay by card”
• “Doe net als 73% van onze gasten en betaal met pin”
In the experiment the effect of the message on nudging people towards the desired behavior is studied. To study the payment behavior of the customers, there will be looked at the total revenue in cash and in pin per restaurant per day. Also at the end of the day, when the restaurant closes, the total number of transactions will
be counted and the number of card transactions will be counted. With these data the analyses can be executed to investigate the possible existence of the impact of the message on people their behavior.
Figure 2. Example of the Framed Message Training
Important is that the employees in the restaurant are instructed to keep doing their work as they used to, and do not emphasize the message to the customers as this could influence the results of the experiment. This influence can be explained in an observer-expectancy effect were the experimenter, in this case the employee, may subtly (un)consciously communicate their expectations for the outcome of the experiment. This could result to shift the behavior of the participants toward the expected outcome of the experiment. Therefore, there is chosen not to inform the employees about the experiment. Only the restaurant managers and assistant restaurant managers will be informed that the message must be presented next to the register and must be visible at all the time during the experiment.
Intervention
The two restaurants assigned to the experimental group will have the message displayed next to the register. At the restaurant assigned to the control group, no message will be displayed. The assisting restaurant managers or the restaurant
part of their own closing procedure, on their own forms. These forms will be faxed to the office after closing hours. At the end of the experiment, all the collected data is being emailed to the researcher.
Selection of research units
As explained before the possible nudging power of social framed messages will be investigated via a field experiment at the restaurants. Because this research is being conducted via a field experiment, it can be assumed that all people entering the restaurant do this in free will. The people are not recruited to participate in a research, so it could be considered that these people walk randomly into the restaurant. But, taken in consideration that people that walk in the restaurant are motivated to go to this restaurant and actively chose to eat there, it can be assumed that the participants participating in the experiment will not be completely selected at random. Therefore, there is chosen to collect the population of the experiment via cluster sampling. The restaurants are the clusters and all the people in the
restaurant will be included in the experiment during the time of the experiment. This type of sampling is chosen because it is more convenient than other types of
sampling methods. Reasons to motivate this convenience sample are time and money. However, if the populations of the control group and the experimental groups are too much different from each other it can lead to biases or non-representativeness
Characteristics of research units
The selection of the participants in the field experiment will be executed via cluster sampling where the whole population within the cluster is being examined during the two weeks of the experiment. The restaurant chain doesn’t have a specific
at the restaurant chain, it seems to be that the customer composition of the restaurant consists of tourists and local’s off all ages and ethnics.
But because of the concept of the restaurant chain, which can be described as a hip and high end snack bar, it seems to be that the customer composition is slightly skewed when compared with a normal snack bar. Therefore, it is assumed that the composition of the local customers will lean more towards the younger and wealthy people. But basically anyone could walk in in the restaurant because of its affordable prices.
Observed variables
As mentioned before this study will explore the possible effect of nudging individuals, via a socially proofed message frame, towards desired behavior. The message is a message motivating the customer to pay their order at the register with credit or debit card instead of paying with cash. The variables measured in this experiment are ‘the total revenue in cash per day’, ‘the total revenue in card
payments per day’, and ‘the total revenue per day’. Furthermore, ‘the number of pin transactions’, ‘the number of cash transactions’ and ‘the total number of
transactions’ are counted per day. With these variables the payment behavior of the customers is being observed. It is expected that these variables are good predictors of the payment behavior of the customers, because of the nature of these variables. Furthermore, the means of the groups can be compared with each other, which could say something about differences in payment method. With these possible differences in payment method analyzed via statistical methods, ultimately a
difference in behavior could be studied. Because of privacy laws in the Netherlands, unfortunately if the customer is Dutch or tourist can’t be measured based on their payment information.
Method of analysis
To investigate the hypotheses and ultimately the research question, analyze
methods are applied via the analyzing tool SPSS. During the experiment the totals of the payments will be counted at the end of the day. This means that data recording consists of the totals of payment transactions and the totals of revenues each day. Before any testing methods are performed, a descriptive table will be plotted. This table will give an insight in the total revenue, and total pin revenue per day per restaurant. Also the mean revenue in pin and total mean revenue per restaurant will be included in this table. Moreover, via a histogram these total revenues per
restaurant and total revenues in pin will be visualized to give the reader more visual insight in the differences. Also the total pin transactions and transactions will be visualized in another histogram. Further, the revenues in pin and the pin transactions are being transformed into percentages. This is because of the problem of
comparability. When one restaurant has for instance a higher pin revenue, this revenue can be higher because the general revenue of the restaurant is higher. This could lead to skewed results. Therefore, there is chosen to also make a variable where these pin revenues are converted into a percentage and a variable where these number of transactions are converted into percentages. This percentage is based on the percentage of pin revenue compared to the total revenue. The data of the variable ‘percentage of pin revenue’ will also be visualized in a histogram.
Because, the experiment investigates the differences in the means of the number of transactions and the means of the revenue the data of the variables will be measured at an interval level. Therefore, taking the level of measurement and the hypotheses into account, it can be concluded that an ANOVA test must be used as testing method. Furthermore, post hoc tests will be performed to investigate if the group means are significantly higher or lower. As explained earlier, because of the possible differences in revenues between the three different restaurants, a
When a significant positive difference is found in the ANOVA test and within the post hoc tests between the experimental groups and the control group, it could be concluded that the message in this given population is nudging people towards the desired behavior.
Results
During the experiment data was collected in three restaurants, over a time span of 14 days. After inserting the data in SPSS, the dataset was being checked with basic statistics for missing data. Furthermore, all data were being labeled. The final sample consisted of 42 days of data (N = 42), where N is 14 days of results times 3 restaurants. During the 14 days a total of 4238 transactions (N = 4238) were
registered (M = 100.91, SD = 34.12).
Store 1 (Control) Store 2 (Exp.1 Group) Store 3 (Exp.2 Group) Total
N transactions N Pin N Total N Pin N Total N Pin N Total N Pin N Total
Day 1 54.00 85.00 42.00 78.00 45.00 75.00 141.00 238.00 Day 2 59.00 95.00 42.00 65.00 42.00 59.00 143.00 219.00 Day 3 63.00 75.00 80.00 132.00 58.00 84.00 201.00 291.00 Day 4 77.00 121.00 84.00 149.00 43.00 68.00 204.00 338.00 Day 5 108.00 147.00 140.00 209.00 74.00 94.00 322.00 450.00 Day 6 85.00 113.00 87.00 151.00 46.00 72.00 218.00 336.00 Day 7 72.00 99.00 72.00 117.00 57.00 84.00 201.00 300.00 Day 8 49.00 78.00 53.00 92.00 35.00 63.00 137.00 233.00 Day 9 76.00 118.00 65.00 102.00 26.00 49.00 167.00 269.00 Day 10 70.00 93.00 56.00 93.00 35.00 56.00 161.00 242.00 Day 11 65.00 99.00 84.00 134.00 48.00 70.00 197.00 303.00 Day 12 77.00 139.00 97.00 135.00 45.00 72.00 219.00 346.00 Day 13 96.00 139.00 90.00 138.00 42.00 73.00 228.00 350.00 Day 14 103.00 139.00 88.00 121.00 47.00 63.00 238.00 323.00 Total 1054.00 (M = 75.29, SD = 17.70) 1540.00 (M = 110.00, SD =24.36) 1080.00 (M = 77.14, SD = 25.57) 1716.00 (M = 122.57, SD = 36.33) 643.00 (M = 45.93, SD = 11.57) 982.00 (M = 70.14, SD = 11.99) 2777.00 (M = 66.12, SD = 23.63) 4238.00 (M = 100.91, SD = 34.12)
Table 1. Descriptive analysis of the number of transactions
As shown in Table 1 the number of pin transactions was the highest at
experimental group 1, with a total of 1080 pin transactions (M = 77.14, SD = 25.57). The number of pin transactions in the control group and the second experimental group were lower where the control group had a total of 1054 pin transactions (M = 75.29, SD = 17.70) and the second experimental group had a total of 643 pin
transactions (M = 45.93, SD = 11.57). Also the total number of transactions was the highest at experimental group 1 with a total of 1716 transactions (M = 122.57, SD = 36.33). This can also be seen in the Figure 3.
Figure 3. Number of transactions per restaurant.
But, as shown in Figure 4, when there is looked at the percentages of pin transactions it can be concluded that the percentages of pin transactions were the highest at the control group with an overall percentage of 68.44% (M = 68.68, SD = 7.46). The percentage of pin transactions at experimental group 1 was 62.94% (M = 62.54, SD = 5.51) and at experimental group 2 65.48% (M = 64.88, SD = 7.26).
Figure 4. Percentage of pin transactions per restaurant 0 200 400 600 800 1000 1200 1400 1600 1800 2000
Control Exp. 1 Exp. 2
Number of Transactions
Pin transactions Total Transactions 60 62 64 66 68 70 Percentage Pin transactionsPercentage Pin transactions
Furthermore, the revenues per restaurant were also descriptively analyzed as shown in Table 2.
Store 1(Control) Store 2 (Exp.1 Group) Store 3 (Exp.2 Group) Total
Revenue in € Pin € Total € Pin € Total € Pin € Total € Pin € Total €
Day 1 869.00 1098.50 830.45 1255.60 766.10 1006.60 2465.55 3360.70 Day 2 930.30 1250.10 551.45 927.45 748.10 944.10 2229.85 3121.65 Day 3 967.65 1158.95 1444.60 2240.00 890.15 1316.70 3302.40 4715.65 Day 4 1416.35 1973.35 1341.55 2648.95 781.00 1121.85 3538.90 5744.15 Day 5 1968.80 2627.85 2346.75 3099.47 1098.16 1369.66 5413.71 7096.98 Day 6 1469.90 1863.15 1536.85 2351.50 846.25 1181.58 3853.00 5396.23 Day 7 1562.70 1941.20 1459.00 2198.65 977.91 1296.41 3999.61 5436.26 Day 8 700.86 988.61 789.20 1225.20 513.70 926.25 2003.76 3140.06 Day 9 1028.75 1647.90 1000.50 1571.30 334.45 607.15 2363.70 3826.35 Day 10 1185.70 1357.60 743.35 1053.60 481.30 740.85 2410.35 3152.05 Day 11 1090.45 1451.75 1260.10 2091.60 746.71 1079.95 3097.26 4623.30 Day 12 1179.65 1911.65 1515.05 2190.05 787.35 1321.35 3482.05 5423.05 Day 13 1647.91 2211.56 1721.15 2475.65 737.65 1318.65 4106.71 6005.86 Day 14 1616.70 2286.00 1552.00 2210.10 607.60 823.86 3776.30 5319.96 Total 17634.72 (M = 1259.62, SD = 360.99) 23768.17 (M = 1679.73, SD = 497.29) 18092.00 (M = 1292.29, SD = 474.65) 27539.12 (M = 1967.08, SD = 652.48) 10316.43 (M = 736.89, SD = 200.66) 15054.96 (M = 1075.35, SD = 242.65) 46043.15 (M = 1580.05, SD = 612.31) 66362.39 (M = 1096.27, SD = 438.01) Table 2. Descriptive analysis of the revenue per restaurant.
As well as in the descriptive analysis of the number of transactions per restaurant, the analysis explains that the revenue in pin was the highest at experimental group 1 with a total revenue of €18092,00 (M = 1292.29, SD =
474.65), where at the control group the revenue in pin was €17634,72 (M = 1259.62,
SD = 360.99), and at experimental group 2 €10316,43 (M = 736.89, SD = 200.66).
This is showed in figure 5.
Figure 5. Total revenue per restaurant 0 5.000 10.000 15.000 20.000 25.000 30.000
Control Exp. 1 Exp. 2
Total Revenue
But when the percentages of the pin revenues were analyzed, as shown in Figure 6, the percentage of pin revenue was the highest at the control group with 74.19% (M = 74.70, SD = 7.16). At the experimental group 1 this percentage was 65.70% (M = 65.43, SD = 6.03) and at the second experimental group this
percentage was 68.53% (M = 68.05, SD = 8.75).
Figure 6. Percentage of pin revenue per restaurant.
After analyzing the data via descriptive analyses, ANOVA analyses were being executed to investigate if the mean differences between the groups were
significantly higher or lower. A non-significant result was found with the ANOVA analysis of the percentages of pin transactions, F (2, 39) = 2.90, p = .067. This means that there was no significant effect of the treatment on the percentages of counted pin transactions.
However, a significant result was found on the percentage of the pin revenue,
F (2, 39) = 5.85, p =.006). The test of homogeneity of variances, also showed a
significant outcome, Levene’s Statistic = 1.443, p = .25. This indicates that the null hypothesis, all equal variances between the three groups assumed, can be rejected. The control group without the experimental treatment, had the highest percentage of pin transactions revenue (M = 74.70, SD = 7.16). This was followed by
Experimental Group 2 (M = 68.05, SD = 8.75). The lowest percentage of pin revenue was found in Experimental Group 1 (M = 65.43, SD = 6.03). From the
60 65 70 75 Percentage Pin Revenue
Percentage Pin Revenue
results of the LSD post-hoc tests can be concluded that there was a significant difference found between the control group and experimental group 1 (Mdifference =
9.28, p = .002) and between the control group and experimental group 2 (Mdifference =
6.65, p = .022). There was no significant difference found between experimental group 1 and experimental group 2. These differences can also be found in the model of the ANOVA analysis (see figure 7).
Figure 7. ANOVA analysis of percentage of pin revenue.
These results show that percentages of the pin revenues were significant lower at the groups with the experimental treatment than the percentage of the pin revenue of the control group.
Conclusion
After analyzing the data in SPSS via multiple analysis methods, it is possible to answer the hypotheses and the research question with the results of the data analyses. Firstly, the data was analyzed via a descriptive method. Based on these descriptive findings it can be concluded that the mean number of pin transactions was higher in the first experimental group, than the mean number of transactions of the second experimental group and the control group. But when converting these means into percentages, something else can be concluded. Based on the
group was higher than the means of the experimental groups. Based on the results of the descriptive statistics it can be assumed that the desired behavior, motivating the customers to pay by card, was not reflected. However, when these percentages were analyzed with a ANOVA test, a non-significant result was found.
But, not only the number of pin transactions were analyzed in the experiment, also the revenues were analyzed. After analyzing the descriptive statistics of the revenues per restaurant, it can be concluded that the mean pin revenue in the first experimental group was higher than the pin revenues of the control group and the second experimental group. However, as well as with the pin transactions, when the absolute data was converted into percentages something else can be concluded. Based on the descriptive findings of these means of the percentages it can be concluded that the means of these percentages were lower in the experimental groups compared to the control group. Further, a significant difference was found in the ANOVA analysis of the percentage of pin revenue, were via the post hoc
analysis it can be concluded that the means of the experimental groups where significantly lower than the mean of the control group. However, there was no significant difference found between the two experimental groups in the post-hoc analyses.
Concluding the findings of the analyses, the hypotheses and the research question can be answered. Beforehand, based on the literature, it was expected that the stimulus would lead towards an increase of the desired behavior of card
payments. This assumption was included in the first hypothesis: “A social proof
framed message will increase the performance of the desired behavior”. Based on
the findings of the means of the pin percentage and the means of the pin revenue percentage, where the means of the control group were higher than the
experimental conditions, it can be concluded that the first hypothesis can be rejected. Further, the results of the ANOVA test on the pin revenue percentage emphasizes this finding. The mean percentage of the experimental groups was
significantly lower than the mean percentage of the control group. This conclusion is also based on the condition that in this experiment paying with card was the desired behavior.
Concluding the results of the descriptive analyses, where the means of the pin percentage and the pin revenue percentage were lower in the experimental groups than in the control group, and the ANOVA test indicating that these means were significantly lower; the second hypothesis “A social proof framed message will
decrease the performance of the undesired behavior” can be rejected. Beforehand
it was expected that the percentages of the card payments and pin revenues would increase. Given the fact this is a percentage, logically the cash payments and
revenues would then decrease. But the findings show an opposite result. Therefore, under these circumstances the second hypothesis must be rejected.
But, apart from the conditions of desired and undesired behavior, it can be concluded that in this experiment differences are found in the means of the groups. As mentioned before this conclusion is based on the outcome of the ANOVA
analysis, were significant differences were found between the control group and the experimental groups. Given the fact that the results indicate an opposite effect than expected, the research question “Can a corporate message, where a social proof
frame is presented in, nudge individuals towards desired behavior?” must be
rejected. In this study, with the given population during the time of experiment, it cannot be assumed that a corporate message in which a social proof frame is being presented could nudge towards behavior.
However, it can be assumed that these results must be viewed in the lights of some limitations. This can be assumed because of the unexpected contrary results. In the somehow similar experiment of Goldstein et al. (2008), where social proof was used to motivate hotel guests to reuse their towels, a positive result was found. The social proof framed message resulted in a higher amount of towel reuse. But a number of things were executed differently in their experiment. Firstly, in their
experiment a pretest was executed to set a good null measurement, and no control group was being used. Therefore, the impact of the research design on the results can be questioned. It can be assumed that the created groups are not fully
comparable. This could lead to biases and even non-representativeness. Therefore, it is suggested for further research that every group first has a pre-test to register a good null-measure. Thereafter, every restaurant will have the experimental
treatment. Then the two measurements can be compared to each other, and could give a purer measurement of the stimulus.
Also when this design, including a pre-test and a post-test, is being executed it must be taken into account that the measurement must be conducted during a longer period of time. This is needed to correct any fluctuations in time. In the towel study of Goldstein et al. (2008) a period 80-days was used to investigate the
possible nudging power of the message. But because of time limitations this time span was not possible in this experiment. It can be assumed that an experiment held over longer periods of time will in the end have purer results. This can be assumed because the impact of a non-representative measurement, for instance an unusually high amount of cash payments one day at a restaurant, will be higher in a shorter time period. Therefore, it is strongly suggested to execute the experiment over a longer period of time if this is possible.
This could also filter out any influence of the staff, where it could be suggested that the staff of the restaurant could be consciously or unconsciously influencing the experiment by sending the customer towards a certain behavior at the register. Although the staff was not informed of the experiment, and the staff was trained to do their work as they were used to do, staff members could push the customer towards payment behavior thus this is not fully controllable. Based on my own experience as a staff member at the restaurant chain some staff members (un)intentionally motivate customers to pay by card more than others. The suggested experimental design without any control groups, and where the
experiment is executed over a longer period could diminish the impact of this confounding variable on the results.
Finally, it can be questioned if the visibility and location of the message could be an influence in the experiment. Before the experiment was conducted, the best location of to display message was not tested. This was due to time limitations. Not testing the location of the message could result in bad visibility of the message. This could result in bad readability of the message. Therefore, it is suggested to test the best location for the message in order to exclude this uncertain factor.
Apart from the limitations which must be taken into account for future
research, a notable result was found that could be interesting to investigate in future research. When looking at the means of the pin revenues the highest mean was found at the first experimental group. Therefore, it can be questioned if the amount of the payment has an influence on the results. This could be because the mean revenue of experimental group one was in percentage lower than the control group, but the mean of the absolute revenue was still higher. Concluding these results, it can be questioned if the message has more impact on higher payment amounts than on lower payment amounts. Therefore, it is suggested that future research takes the amount of payment in consideration and investigates the question “Does the amount of the payment reinforces the effect of the message?”. Where it could be suggested that people are more likely to pay their order by card when it’s a big order, than when it’s only a single item order of a few euro’s.
Furthermore, for future research it could be interesting to investigate the content of the message by testing multiple versions of the message. In previous research it is common to test multiple versions of the message (Goldstein et al., 2008; Ölander & Thøgersen, 2014). It could be possible that the text of the social proof message has an influence on the result. Therefore, it is suggested for future research to test multiple versions of the social proof message must, to investigate which of these social proof messages has the highest impact or convincing power on
the customer. Therefore, it can be assumed that this study has a good scientific value and relevance.
Because of the little research which is done on the use of corporate messages for motivating individuals to do certain behavior, this experiment could give new insights for scientific research. Apart from the fact that framing is not often studied via experiments, the use of framing in corporate messages to nudge people towards desired behavior should also be investigated more thorough. Investigating the use of frames via experiments could give new insights in framing theory in the
communication science. The outcomes of frame-use could be different in experimental studies than in content analysis studies. Furthermore, this could contribute to scientific knowledge about framing in corporate communication messages and the relation between a frame and someone’s behavior. In
communication science it is common to look at frames in texts and how these frames could shape a context for an individual about a certain topic (Tankard et al., 1999). But this study suggest that framing could be used by organizations for more than only shaping a context.
Therefore, this study has a good practical implication for organizations. Insights in the use of frames to nudge people towards behavior could help
organizations to optimize their corporate messages and get more efficiency out of them. And, even though the results were contrary to expected results, previous research has proven that social proof messages could help organizations to get a desired behavior of their customers without putting much effort in it (Goldstein et al., 2008). Furthermore, with the notable result of the mean pin revenue it could be assumed the message could be influencing the behavior of the customer. Therefore, frame-use must not be underestimated. Because a small change in a corporate message, by using a social proof frame, could have a big influence on the results of the message!
Literature
Aronson, E., Wilson, T.D., & Akert, A.M. (2005). Social Psychology (5th ed.). Upper Saddle River, NJ: Prentice Hall.
Bateson, G. (1972). Steps to an ecology of mind: Collected essays in anthropology,
psychiatry, evolution, and epistemology. University of Chicago Press.
Bearden, W. O., & Etzel, M. J. (1982). Reference group influence on product and brand purchase decisions. Journal of consumer research, 9(2), 183-194. Chartrand, T. L., & Bargh, J. A. (1999). The chameleon effect: The perception–
behavior link and social interaction. Journal of personality and social
psychology, 76(6), 893.
Cialdini, R. B. (1987). Influence (Vol. 3). A. Michel
Cialdini, R. B. (2001). Harnessing the science of persuasion. Harvard Business
Review, 79(9), 72-81.
Cialdini, R. B., Kallgren, C. A., & Reno, R. R. (1991). A focus theory of normative conduct: A theoretical refinement and reevaluation of the role of norms in human behavior. Advances in experimental social psychology, 24(20), 1-243. De Vreese, C. H. (2005). News framing: Theory and typology. Information design
journal+ document design, 13(1), 51-62.
Entman, R. M. (1993). Framing: Toward clarification of a fractured paradigm. Journal
Ganzach, Y., & Karsahi, N. (1995). Message framing and buying behavior: A field experiment. Journal of Business Research, 32(1), 11-17.
Goffman, E. (1974). Frame analysis: An essay on the organization of experience. Harvard University Press.
Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A room with a viewpoint: Using social norms to motivate environmental conservation in hotels. Journal
of consumer Research, 35(3), 472-482.
Griskevicius, V., Goldstein, N. J., Mortensen, C. R., Cialdini, R. B., & Kenrick, D. T. (2006). Going along versus going alone: when fundamental motives facilitate strategic (non) conformity. Journal of personality and social psychology, 91(2), 281-294.
Hallahan, K. (1999). Seven models of framing: Implications for public relations.
Journal of public relations research, 11(3), 205-242.
Hoffer, E. (1955). The Passionate of mind and other aphorisms. New York, NY: Harper & Row.
Hong, F., Hossain, T., & List, J. A. (2015). Framing manipulations in contests: a natural field experiment. Journal of Economic Behavior & Organization, 118, 372-382. Johnson, E. J., & Goldstein, D. (2003). Do defaults save lives? Science, 302(5649),
1338-1339.
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica: Journal of the Econometric Society 47(2), 263-291.
Krause, J., Ruxton, G. D., & Krause, S. (2010). Swarm intelligence in animals and humans. Trends in ecology & evolution, 25(1), 28-34.
Liebrecht, C., Hustinx, L., van Mulken, M., & Schellens, P. J. (2016). Krachtige taal.
Tijdschrift voor Taalbeheersing, 38(1), 57-79.
Ölander, F., & Thøgersen, J. (2014). Informing versus nudging in environmental policy. Journal of Consumer Policy, 37(3), 341-356.
Schultz, F., Kleinnijenhuis, J., Oegema, D., Utz, S., & Van Atteveldt, W. (2012). Strategic framing in the BP crisis: A semantic network analysis of associative frames. Public Relations Review, 38(1), 97-107.
Sightsmap. (n.d.). [Sightsmap, tourism heat map of Amsterdam]. Retrieved May 12, 2016, from, http://www.sightsmap.com.
Snow, D. A., Rochford Jr, E. B., Worden, S. K., & Benford, R. D. (1986). Frame alignment processes, micromobilization, and movement participation.
American sociological review 51(4), 464-481.
Tanford, S., & Montgomery, R. (2015). The effects of social influence and cognitive dissonance on travel purchase decisions. Journal of Travel Research 54(5), 596 -610.
Tankard, J., Hendrickson, L., Silberman, J., Bliss, K., & Ghanem, S. (1991). Media frames: Approaches to conceptualization and measurement, Paper presented to the Association for Education in Journalism and Mass Communication.
Tversky, A., & Kahneman, D. (1991). Loss aversion in riskless choice: A reference- dependent model. The quarterly journal of economics 106(4), 1039-1061.
