1
Committing to physical exercise: how nudges facilitate positive
changes in behavior
By Sjaak Hoogstraaten Student number: 10466886
Student for the Master is Business Economics – specialization Managerial Economics and Strategy.
ECTS: 15
2 Statement of Originality
This document is written by Student Sjaak Hoogstraaten who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Contents of this thesis
1. Introduction & Literature review ... 4
1.1 Nudges ... 5
1.2 Monetary incentives as nudges ... 6
1.3 The power of commitment as a nudge ... 7
1.4 Research set-up ... 9
2. Methodology: ... 9
2.1 T-tests ... 11
2.2 Historical data ... 11
2.3 Survey ... 12
3. Hypothesis and research question ... 13
4. Data and Results ... 13
4.1 Data ... 13 4.2 T-tests ... 15 4.3 Historical data: ... 16 4.4 Survey results ... 17 5. Interpretation of results ... 19 6. Discussion ... 22 7. Conclusion ... 23 8. Appendix ... 24 9. References ... 25
4 Abstract
This field experiment investigates the effectiveness of a nudge in the form of a commitment to exercise. The participants in this experiment are people that signed up for weekly running classes. First it is determined that these people are willing to make a commitment to exercise. Secondly, this commitment increases their attendance rates for training sessions but only makes a
significant impact after the first three weeks. Thirdly, the participants in the treatment group that received the nudge are more likely to reach their self-set goals which are related to exercising. Exercising can reduce overweight and obesity. Therefore, the increase in exercising caused by the nudge can provide a potential solution to problems related to overweight and obesity that many countries face.
1. Introduction & Literature review
Many countries face high rates of obesity. For 2016 The World Health Organization (WHO) estimated that 1,9 billion adults were overweight of which 650 million people were obese which is 39% and 13% of the total population respectfully. The WHO and others link obesity with an increased risk of diseases such as cardiovascular diseases and diabetes (Mokdad et al., 2003; Withrow & Alter, 2011 ; WHO 2016). Withrow & Alter estimate that obesity accounts for up to 2,8% of a countries healthcare expenditure. Furthermore, they find that obese patients have higher medical costs than their normal weight peers. Exercising is a well-known method to reduce obesity (Williams, Bezner, Chesbro, & Leavitt, 2005) and thereby reduce the healthcare costs of a country. In this field experiment participants in running groups will be asked whether they are willing to make a commitment of being present the consecutive training session. The aim of this commitment is to induce them to exercise more frequently.
One potential way of increasing the exercising levels of people is by using nudges, A nudge is an economic concept that entails the approach to change the behavior of people in certain desirable directions, while maintaining their freedom of choice (see for example Sunstein, 2014, 2016). Trainers inviting runners to explicitly commit to upcoming training sessions is an example of a nudge. This nudge might be a cost effective way to increase the frequency of exercising, which may in future applications be extrapolated to reducing the problem of obesity. In this thesis I investigate through a field experiment whether a by the trainer documented previous commitment to upcoming training sessions is an effective nudge to increase exercising. The remainder of this introduction is divided into four parts. First the concept of nudges is further
5 described. Secondly the concept of monetary incentives as a nudge is discussed. Thirdly the power of commitment as a nudge is evaluated and finally the research setup is described.
1.1 Nudges
Many traditional economic theories build on the assumption that people are agents that act perfectly rational (Becker, 1962). This would mean that, when making choices, they take every possible option into account and can adequately foresee the consequences of their
choices/actions. Furthermore, these perfectly rational people, or Econs, as they are referred to in Thaler & Sunsteins’ 2009 book, in theory always choose what is best for them (Thaler &
Sunstein, 2015). If all people would be Econs, a nudge should not sort any effect. Econs are able to make the best decision for themselves and therefore do not need help from another agent.
However, more recently it has become widely accepted under scholars that in practice non-perfectly rational behavior is often exerted by people (see for example Ariely, 2012;
Loewenstein, et al., 2012). Because people are not (always) acting perfectly rational, stimulation in the form of a nudge can influence the decisions people make, and potentially help them to make better decisions.
Setting a desired default option is an example of a nudge (Thaler & Sunstein, 2009). Perfectly rational people would not forget to make a choice when faced with an important
decision. Non-perfectly rational people however, often do not divert from the default option or do not make a choice at all. This is because people are naturally inclined towards inertia and known as the default bias (Samuelson & Zeckhauser, 1988). With an appropriate default option, people would not be worse off by leaning towards inertia and thus not making a choice. Illustratively, in countries where the donor system is opt-out instead of opt-in, the rate of people that are willing to be a donor is much higher (Abadie & Gay, 2006; Thaler & Sunstein, 2009). A higher rate of registered donors benefits the wellbeing of patients and consequently benefits the healthcare system of a country.
Nudges have also been previously related to potentially aiding in solving issues such as obesity by stimulating physical exercise. Williams, et al., (2005) found that signing a behavioral contract to exercise increases participants’ commitment to their training programs. General conclusions however cannot be drawn from their study. They only focusses on a small and
6 specific sample of 43 postmenopausal African American women that have a goal of walking more. This thesis on the other had a bigger more diverse research group and the goal of the nudge was to increase exercising in the form of running not walking.
1.2 Monetary incentives as nudges
Another way to increase exercise described in literature is with the use of financial
incentives. In a 2009 paper, Charness & Gneezy investigated the effects of financial incentives to exercise. The incentives produced significant results for a group receiving high monetary
incentives to exercise but not for the group that received low incentives. This is in line with the view posed by Gneezy & Rustichini in their 2000 paper that you should pay people enough or not at all. Acland & Levy, (2015) replicated the results of the Charness and Gneezy experiment. In both of their papers the attendance rates to the gym of the participants in a group receiving high incentives were higher than the control group and continued to be higher in the period after the incentives were removed. This suggests habit formation, which Lally et al. (2010) define as the repetition of a behavior consistently. Acland & Levy However, found that attendance rates dropped after a vacation period. The conclusion of habit formation is therefore negotiable. Other authors completely disagree on the long run effectiveness of financial incentives. In his 1993 book, Kohn reviewed multiple incentive programs aimed at improving health related behavior. The individuals in the incentives group were complying better to the required improvements in their behavior, however, in the long run they exerted worse behavior than those in the no-reward control groups.
Two reasons why (monetary) incentives might not work can be found in literature. The first can be found in signaling theory. One could say that providing monetary incentives to go to the gym is a signal that going is apparently difficult and therefore people might opt not to go (Gneezy, et al 2011). The second reason can be found in the distinction between intrinsic and extrinsic motivation. Motivation is defined as the reason underlying specific behavior. (Guay et al., 2010). Benabou & Tirole 2003 (page 490) define intrinsic motivation as ‘the individual’s desire to perform the task for its own sake’ and extrinsic motivation as performing the task contingent to a reward. Extrinsic motivation often crowds out intrinsic motivation (Benabou & Tirole, 2003; Bénabou & Tirole, 2018 ; Frey & Oberholzer-Gee, 1997). Therefore, incentives only work when they are large enough. The crowding out of intrinsic by extrinsic motivation can
7 explain why Charness and Gneezy (2009) found that financial incentives only work when they are large.
There are two more disadvantages of using monetary incentives to increase commitment to physical exercise. Firstly, the incentive could only work for a short time period and can cause addiction to rewards (Suvorov, 2003). When incentives are removed the motivation to exert effort on the task is lower than before, when the contingent rewards were in place. The intrinsic
motivation is not automatically reinstated when the incentives are removed. Secondly, ethical concerns may be raised for the use of monetary incentive for health promotion. Monetary incentive schemes might interfere with the autonomy of the person receiving the incentive (Ashcroft, 2011). In his 1993 paper Kohn refers to financial incentives as bribes that are morally wrong.
There are other methods to increase good behavior such as exercising without using monetary incentives. As previously stated, Williams, et al., (2005) found that signing a behavioral contract to exercise, similar to a commitment, is effective, although this was only determined for a small sample of women. Apesteguia, et al. (2013) , found that rule compliance increases after a simple reminding email.Other non-financial nudges, such as providing feedback and providing information about performance also work to stimulate ‘good’ behavior. (Feild, 2015; Kloss, 1994; Martinez, 2014).
1.3 The power of commitment as a nudge
People often make irrational, shortsighted choices concerning rewards. They are inclined to choose immediate gains over waiting and receiving a higher gain. In other words the short-term costs outweigh the long-short-term benefits (Banerjee et al., 2010; Gollwitzer & Oettingen, 1998). A way to solve this myopia has been investigated in a lab experiment. In their 2016 study
Pietroni & Hughes, asked half of the participants to agree to read the instructions of the following experiment carefully. During the experiment participants had to make choices concerning their rewards. They could choose for an immediate gain or they could choose to receive a larger but delayed payment. The participants that agreed on reading the instructions carefully (the treatment group), chose the short-term gains less often than the participants that were not asked to agree on
8 reading the instructions carefully. Therefore, the participants in the treatment group were less affected by the myopia of preferring short-term benefits over long-term (larger) benefits.
This myopia is also present for exercising. When engaging in exercising one first encounters the cost while the benefits are delayed. Because of this myopia people often have good intentions but they fail to act on them (Gollwitzer, 1999). In their 2006 paper, Malmendier & Della Vigna find that people are overconfident about their future attendance at the gym (their good intentions) and pay for a gym membership with unlimited access, even though they would pay less if they would pay per visit.
One way to change behavior to help people achieve their good intentions such as gym attendance is through commitment and implementation intentions (Williams et al., 2005). The effectiveness of a commitment is determined by two constructs: self-efficacy and expectancy. Self-efficacy is the belief that one can successfully execute the required behaviors. A high sense of self-efficacy entails being able to increase efforts to achieve required behavior when obstacles occur (Bandura, 1982). Expectancy refers to the believe in the extent to which one’s behavior will result in favorable or unfavorable outcomes. People are more likely to engage in changing their behavior, when they believe this change will produce favorable outcomes to them (Bandura, 1982). We can conclude that since people are overconfident about future attendance to exercise (running) classes they will be willing to commit to them. This commitment to exercise can be expected to be effective if they believe that they will be able to keep their commitment (self-efficacy) and belief that exercising will result in favorable outcomes to them (expectancy).
It has been shown that commitment works to increase attendance. Subsequently, the commitment can help people to act according to their own good intentions. The commitment works when self-efficacy is high enough and works significantly better when the commitment is accompanied by an implementation intention, Implementation intentions are plans that link anticipated critical situations to goal-directed responses (Gollwitzer, 1999). Frayne & Latham, (1987), implemented a training program to increase the attendance at work of government employees. Part of the training was to write a behavioral contract with themselves, this commitment was part of the reason attendance rose after the training was held.
In the literature, goal intentions and implementation intentions are clearly distinguished. Setting a goal is an effective technique for motivating people (Latham & Locke, 1991; Locke &
9 Latham, 1984). Combining these goals with an implementation intention works even better. Numerous authors conclude that Implementation intentions work better than goal intentions(See for example: Gollwitzer, 1999; Gollwitzer & Oettingen, 1998; Lippke et al., , 2004; Sheeran, 2002). Furthermore it doesn’t matter whether goals are self-set or set by others (Latham & Locke, 1991).
1.4 Research set-up
Building on this knowledge I decided to help runners with their implementation
intentions by having their trainers invite them to make a commitment of being present at the next training session. This is an easily obtainable proximal goal which works better than a vaguely defined goal of ‘exercising more’ (Bandura, 1982).
The trainers and I help the runners to make an implementation intention, by setting a proximal goal and making a commitment/promise of coming to the consecutive session. This constitutes the nudge. One would expect that this promise would matter, as the significant impact on behavior of promises has already been shown in lab experiments (Charness & Dufwenberg, 2010). Furthermore, the runners train in the same group each weak. When people train in the same group each week and when attendance rises friendships are more likely to occur. These friendships can also increase attendance as Charness & Gneezy (2009) have shown.
In this thesis I use the nudge of a promise to commit to potentially stimulate people to exercise. The participants of this experiment are people that have long term sport and health related goals. My thesis concern 953 participants in running groups. The main findings of this thesis are that these runners are willing to make a commitment to exercise and that this
commitment causes a significant change in their behavior; they exercise more. This behavior also helps them to attain their self-set goals. These findings suggest that the nudge of proposed
commitment to a proximal goal is indeed effective. The rest of the thesis is organized as follows: Part 2 will give an overview of the applied methodology, part 3 will discuss the hypotheses, part 4 will describe the data and results, part 5 will give an interpretation of these results, part 6 will be a discussion and part 7 concludes.
10 For my experiment I wanted to see the effects of a very simple nudge on members of a running class that have weekly training sessions. My main research question was: “to what extent is a nudge to commit to upcoming training sessions an effective way to increase people’s
frequency of physical exercise?” The question was:
English: “I commit to being present at next weeks’ training”
In Dutch: “Ik zeg toe bij de training van volgende week te zijn”
Cooperation for my research was provided by Hardlopen Amsterdam, a company that gives about 90 weekly running classes divided over 8 cities. All participants in these classes sign up for a 13-week running program. Prior to the start of these 13-week programs, I asked trainers whether they would be willing to implement my nudge in their training group(s). I told them that the implementation of the nudge could help them to increase the participation rate of their
sessions. I did not tell them that asking for the commitment was a nudge or that this method was part of my research. Only the 10 out of 45 trainers that were willing to participate incorporated my nudge at the end of each of their training sessions. Therefore, my experiment is not
randomized. The participants that were in a group where the nudge was implemented were in the treatment group. The participants in the treatment group were asked the commitment question at the end of each training session, whereas the training sessions of the control group ended the same way they did before. The participants in the treatment group wrote down their names and answered the commitment question with either yes or no. With a yes, they had
promised/committed themselves to being present at the consecutive training session. See appendix A for an example of a completed form.
The trainers kept track of their attendance with the use of an excel spreadsheet. The spreadsheets of all the trainers were shared with me. For my experiment I could only focus on the first 5 weeks of attendance due to time constraints. The trainers that were implementing the nudge sent me a picture after each training session with the answers to the nudge/commitment question. Furthermore, I obtained historical data form the company.
With these two data sets I conducted different statistical estimations. First, I compared the differences between the treatment and control group using a T-test. Secondly, I used both data
11 sets to compare the effect of the nudge after its implementation. Furthermore, I also examined the motivation of the participants with a short survey to be able to enrich my findings using multiple data sources.
2.1 T-tests
For the T-test I used two different estimators of the attendance rates. First, I compared the significance of the difference in the participants’ attendance between the treatment and control group each week. Secondly, I compared the difference in attendance of the runners that were also present the previous week. The percentage of runners that were present in a specific week and in the previous week was compared between the treatment and control group. This provides a better estimate of the impact of the nudge because only the people (in the treatment group) that were present in the previous training session received the nudge to come the consecutive week.
To be able to measure the effect of the nudge I wanted to isolate the participants that received the nudge from those that did not. Therefore, I assumed that when the answer to the question/promise of being present the consecutive week was not answered and/or recorded the participants did not receive the nudge. Also, the participants that weren’t present in one of the training sessions did not receive a nudge that particular training session. For a second set of T-tests I considered the attendance of these participants as missing values.
2.2 Historical data
To account for the non-random allocation of the treatment group, it was necessary to use a difference in difference treatment with historical data to measure the impact of the
nudge/commitment. The differences in differences method is a well-known solution to deal with a non-random treatment allocation. See for example Card & Krueger, (1994). We have to make two assumptions to be able to draw conclusions using the differences in differences method The common trend assumption, and the assumption that the composition of the treatment and control group remains the same (Angrist J. D. and J. S. Pischke, 2008).
All participants in my experiment live proximal to each other and therefore experience the same differences in weather and other surrounding factors that might influence the attendance rates. Therefore, we can assume a common time trend. We can also determine that the second
12 assumption, that the composition of the groups should remain the same, holds. All participants trained with the same trainer on the same location and at the same time prior and after the intervention. We therefore have no reason to believe that their composition might have changed. When these assumptions hold we can draw conclusions from the regressions (described below) of the differences in differences method.
Using these two assumptions we can segregate the difference between the attendance rates of the treatment and the control group caused by the implementation of the nudge and caused by a time trend difference. The differences in differences can be calculated with equation 1:
(1) Y = β0 + β1*Dpost + β2*Dt+ β3*Dpost*Dt β
The variable of interest is Y, Dpost is a dummy variable that is 1 for the post-intervention period, Dt is a dummy variable that is 1 when the participant is in the treatment group. β3 is the differences in differences estimator. This estimator gives the effect of the implementation of the nudge.
For the variable of interest, I used two different measures. The same two measures that were used for the T-test used in the post-intervention period. The first is the attendance in each week (equation 2). The second is a measure of the difference in attendance of the runners that were also present in the previous weeks’ training session (equation 3).
(2) week1present = β0 + β1*Dpost + β2*Treatment+ β3*Dpost*treatment
(3) week1_2_present = β0 + β1*Dpost + β2*Treatment+ β3*Dpost*treatment
2.3 Survey
Prior to the experiment about half of the participants completed a short survey. The participants were asked for their motivation to exercise. Knowing their motivation, I could determine whether the nudge helped to change the behavior of people in a way they themselves desire.
13
3. Hypothesis and research question
My main research question is: “to what extent is a nudge to commit to upcoming training sessions an effective way to increase people’s frequency of physical exercise?” I aimed to answer my main research question by researching the following sub-questions:
I To what extent are people willing to explicitly commit themselves to exercise? II How does making a commitment influence attendance behavior in practice?
III To what extent does this nudge of commitment help people change behavior in the way they desire?
From these three sub-questions, I extrapolated three hypotheses based on the previously described theoretical considerations. These are:
H1: People are willing to explicitly commit themselves to exercise
H2. Making a commitment to exercise will positively influence attendance behavior H3. The nudge of making a commitment to exercise, will increase the ability of people to achieve their own pre-set goals
4. Data and Results
The following section first describes the data of this study. Secondly, the significance of the difference in attendance between the treatment and control group during the time of the experiment is determined using T-tests. Thirdly a historical comparison is made using the differences in differences method and finally the survey results are discussed.
4.1 Data
The 90 training sessions provided by the company I cooperated with for my experiment are given by 44 trainers in 27 different locations across seven cities (Amsterdam, Rotterdam, Utrecht, Blaricum, Schothorst, Vathorst, Leusden and Alkmaar). A total of 953 people was subscribed for one of these training sessions. 10 out of the 44 trainers were willing to apply my nudge with all or one of their groups. The other trainers either did not reply at all or had different reasons not to carry out the nudge. These reasons fall in two categories, either they did not feel
14 comfortable with carrying out the nudge towards their runners (two trainers)or they preferred to use their own methods to try to keep their runners to keep coming to their training session (seven trainers).The rest (25 trainers) didn’t indicate why they wouldn’t participate or did not reply at all.
I will first provide descriptive statistics. Table I provides the number of observations made and the number of people that were present each week. Table II records the answer to the question whether people want to commit themselves to being present the consecutive week.
Most trainers did not use the exact question I intended them to ask. A common diversion of the commitment question was: “Will you be present next week?” instead of “I commit to being present next week”. Furthermore, one of the trainers voice recorded the commitments instead of writing it down (he did send the voice messages to me). 20 percent of the trainers used the exact same question as I intended, 80 percent deviated from this question. See table VIII (in appendix B) for an overview of the deviations.
Table I Descriptive statistics
Variable Total Control Treatment
Observations Presence (in %) Observations Presence (in %) Observations Presence (in %) Week1present 939 73,27% 769 72,69% 170 75,88% Week2present 953 72,93% 783 72,16% 170 76,47% Week3present 849 64,43% 713 63,39% 136 69,85% Week4present 793 54,22% 649 51,62% 144 65,97% Week5present 820 50,49% 650 48,15% 170 59,41%
Notes. Descriptive statistics of the number of observations and the presence each week in the treatment and control group.
Table II
Answer to question of commitment
Variable Observations % yes
Yes or no week 1 128 94,53%
Yes or no week 2 99 94,95%
Yes or no week 3 78 85,90%
Yes or no week 4 80 97,50%
Yes or no week 5 83 93,98%
Notes. Participants answered either yes or no each week to the question whether they wanted to commit themselves to being present the consecutive training session.
15
4.2 T-tests
When we look at the significance between the difference attendance rates of the treatment and control group, we see that in week 4 and 5 the attendance in the control group is significantly higher; 14 and 11 percent for week 4 and 5 respectfully. When we look at the second measure of attendance, we see the same pattern with similar differences in attendance rates. People that were present in week 3 and week 4 were also significantly more often present in the consecutive week, see table III for an overview.
Table II summarizes the number of people that received the nudge and how many of them indicated that they will be present the next weeks’ session. To be able to test the effect of the nudge in the consecutive week, I changed observations of the runners that didn’t receive the nudge in the previous week to not recorded. These are a lot of observations, for example one of the trainers only implemented the nudge the first week but not in the following weeks. However, when the observations of when the nudge wasn’t implemented are deleted, the results do not change significantly. The below table III summarizes the results of the two different measures of attendance.
Table III
Differences in attendance between the treatment and the control group
Control Treatment Difference
Observations % Observations %
Treatment-control Week1present 769 73% 170 76% 3% Week2present 783 72% 170 76% 4% Week3present 713 63% 136 70% 7% Week4present 649 52% 144 66% 14%* Week5present 650 48% 170 59% 11%** W1_to_w2_present 559 78% 129 82% 3% W2_to_w3_present 515 71% 107 78% 4% W3_to_w4_present 409 60% 79 73% 7% W4_to_w5_present 300 62% 95 74% 14%* W1_to_w2_present1 559 78% 125 82% 4% W2_to_w3_present1 515 71% 97 80% 9% W3_to_w4_present1 409 60% 71 77% 17%**
16
W4_to_w5_present1 300 62% 69 75% 13%*
Notes. 1With the value changed to not recorded) if yes/no wasn't recorded, *p<.05; ** p<.01; *** p<.001
These T-tests however, might not identify a causal effect because the distribution of the nudge was not randomized. There might be a self-selection of trainers that are willing to implement this method of commitment. Therefore, I also made a comparison using historical data.
4.3 Historical data:
Of the 90 weekly training sessions that started in April 2018, 63 also started in January 2018. Meaning that these 63 training sessions had the same trainer and the same day and time of the week in which the training session was being held. For one of the groups I implemented a pilot version of the experiment, deleting that group leaves 62 training sessions. This provides me with a good sample for a differences in differences comparison. In table IV the descriptive statistics of the four different groups are presented.
Table IV
Descriptive statistics of the
Pre-intervention control group Post-intervention control group Observations Presence (in %) Observations Presence (in %)
Week1present 618 72,98% 672 71,88%
Week2present 636 65,72% 672 73,21%
Week3present 640 67,50% 614 63,84%
Week4present 640 62,97% 558 51,61%
Week5present 635 63,15% 553 49,19%
Pre-intervention treatment group Post-intervention treatment group
Week1present 84 77,38% 103 71,84% Week2present 84 73,81% 103 79,61% Week3present 84 66,67% 69 76,81% Week4present 84 63,10% 103 68,93% Week5present 84 58,33% 103 62,14% Yes or no week 1 73 95,89%1 Yes or no week 2 57 94,74%1 Yes or no week 3 54 85,19%1 Yes or no week 4 53 100,00%1 Yes or no week 5 50 94,00%1
17 Note. Descriptive statistics of the number of observations and the presence each week in the treatment and control group. 1 Is the percentage of people that said yes to the question of commitment.
Table V summarizes the results with the attendance of week 1 till 5 as the depended variables (equation 2). The first three weeks the nudge do not seem to have a significant impact, but in week four and five the attendance is significantly higher for the group that received the nudge.
Table VI gives an overview of the regression according to equation 3. For the participants that were present in the third week the nudge starts to make a significant difference. Runners that were present in week 3 and in the post intervention period treatment group, were 26,3 % more often present the consecutive week. People that were present in week 4 and in the treatment group, were 21,5 % more often present in week 5.
Table V
regression results for measure 1
W1present W2present W3present W4present W5present Dpost_intervention -0.011 0.075** -0.037 -0.114*** -0.140***
Treatment 0.044 0.081 -0.008 0.001 -0.048
Dpost*Treatment -0.044 -0.017 0.138 0.172* 0.178*
Constant 0.730*** 0.657*** 0.675*** 0.630*** 0.631***
N 1477 1495 1407 1385 1375
Notes. W1 present refers to the presence in week 1, W2present to the presence in week 2 etc. The first row are the dependent variables for the regressions according to measure 1. * p<.05; ** p<.01; *** p<.001
Table VI
regression results for measure 2
W1 to W2 W2 to W3 W3 to W4 W4 to W5 Dpost_intervention 0.074** -0.013 -0.095** -0.088* Treatment 0.068 -0.049 -0.074 -0.102 Dpost*Treatment 0.000 0.196* 0.263** 0.215* Constant 0.717*** 0.726*** 0.699*** 0.724*** N 1060 985 897 757
Notes. W1 to W2 presence refers to being present in week 1 and the consecutive week 2. W2 to W3 to being present in week 2 and 3, etc. The first row are the dependent variables according to measure 2. * p<.05; ** p<.01; *** p<.001
18 Participants filled out a survey which was a short enquiry into their exercising history, current injuries and/or other issues. The participants were also asked for their motivation to participate in the training sessions. When their motivation was asked they could select one or more of the below options:
- I want running to become a habit - I want to be able to run for longer. - I want to be able to run faster
- I want to move more and build stamina - Something else namely, …..
When they selected the option “Something else namely, ….’ they could write down their own alternative motivation. 412 out of the 953 participants that were subscribed completed the survey. 323 of them, the big majority with 78,40 percent, indicated that they wanted to move more and/or turn running into a habit. See figure I for an overview of the survey results and appendix C for the complete survey.
19
5. Interpretation of results
In this section the extent to which my three earlier mentioned hypotheses are correct based on the results will first be discussed. Subsequently, it will be discussed to what extent the results confirm my main hypothesis that a nudge to commit to upcoming training sessions is an effective way to increase people’s frequency in exercising.
The first hypothesis whether people are willing to make a commitment can be confirmed. Between 86 and 98 percent (see Table II) of the people that received the nudge committed themselves to being present the consecutive week. This are convincing numbers, but these numbers cannot be extrapolated to the entire population. The participants in my experiment already signed up for running classes and the percentage of people that are willing to commit themselves who did not already sign up for running classes will likely be lower. The commitment made however does not automatically translate into behavior according to it. To quote (Latham & Locke, 1991, page 217): 0 50 100 150 200 250 300
Alternative Habit Longer Faster Moving
Figure 1 Survey Results
Notes.This chart summurizes the goals particpants wanted to achieve with signing up for running classes. They could select one or multiple options. 88 participants formulated their own alternative goal. 175 participants selected that they wanted running to become a habit. 151 participants indicated they wanted to be able to run for longer, 71 to be able to run faster and 266 indicated that they wanted to move more and build stamina.
20 ‘Commitment refers to the degree to which the individual is attached to the goal,
considers it significant or important, is determined to reach it, and keeps it in the face of setbacks and obstacles. It must be stressed, however, that the feeling of commitment does not
automatically lead one to act in accordance with it.’
We are therefore not able to answer hypothesis 2 without further analyzing the results.
In my experiment the commitment made, only starts to make a significant impact after the first three weeks. We see that people are significantly more often present in week four and five of the program (see table V). An explanation could be that all participants starts off with good intentions of going to running classes and all participants have enough willpower to go the first couple of times. However, after these first couple of weeks, excuses for not going may be made more easily without a clear commitment. Therefore, my second hypothesis is also (partly) confirmed. Making a commitment matters, although only after a certain amount time.
Two main reasons for the commitment to work can be distinguished. The first reason is the previously introduced implementation intention that is established when participants indicate they will be present the consecutive weeks training session. In his 1999 article Gollwitzer shows that implementation intentions work to reach desired behaviors/results. The argumentation he uses is that when people form implementation intentions, the goal directed behavior will be triggered automatically when the specific situation is encountered.
The second reason is that people want to see themselves as honest and honorable people (Ariely, 2012) that do not like to lie (Gneezy, 2005; Vanberg, 2008) and, above all, they are overconfident about future self-control (Malmendier & Della Vigna, 2006). Most people indicate that they indeed will be present to consecutive week and they do not want to renege on this promise. They might exert extra effort to be present after they have made a commitment. Furthermore, we know from the 2006 Malmendier & Della Vigna, study that people are overconfident about future efficiency and self-control. But the commitment combined with the aversion to lying and thereby disappointing themselves might push people to go to the training sessions more often. In a 2011 paper discussing a series of experiments concerning self-deception Chance et al, show that people indeed deceive themselves and fail to recognize that they do. In my experiment that would entail that people expect they will go to all the training sessions and therefore sign up but, go less often than they expected beforehand. They have higher
21 expectations about their future performance/behavior than is observed in practice. When the time comes to exercise, they might cheat themselves and come up with an excuse not to go. This can be either an explanation for why the commitment might not work or why the commitment might work. It might not work if people are expecting they will go, and therefore easily answer yes to the question of commitment but come up with an excuse not to go at the last moment. However, people might expect they will go and therefore easily commit themselves. At the last moment they might think about the commitment they made and don’t want to renege on this promise. Looking at the results of my research the latter is the stronger pull on peoples’ behavior.
If we prolong the trend that the nudge starts to have a significant impact after the first three weeks, we can expect that the participants that receive the nudge will also be present more often in the weeks following week 5. A counterargument to this could be that the power of the nudge might be decreasing when it has been used many times. However, people still would be disinclined to lie about being present and the commitment is still being made each week.
Furthermore, consistently going to run with their group each week can result in habit formation. Charness & Gneezy, (2009) show that paying people (sufficiently) to go to the gym increases their attendance rates, interestingly this higher attendance rate is also maintained when the financial incentive is removed. Possibly, a good habit of working out has been formed. An
explanation is that people see the (potential) benefits of their exercise and decide to continue with this behavior. Similarly, people will still be going more often after the first 5 weeks and after having committed themselves, since they see or experience the benefits.
My third hypothesis is that the nudge of making a commitment to exercise, will increase the ability of people to achieve their own pre-set goals. Since people sign up for a 13-week training program we can assume they have a goal of going to all these training sessions and to work on their general fitness and health. This is also corroborated by the short survey that was conducted prior to the start of the first training session. Close to 80 percent of the runners had a goal that was related to exercising (running) more often (See Figure 1). Since the commitment nudged people into going to the running classes more often it seems fair to say that asking for the commitment helps the runners to reach the goals that they set for themselves. My third hypothesis is thereby also confirmed.
To answer my main research question, we see, looking at the results that a nudge is an effective way to increase people’s physical exercise although it does not produce immediate
22 results. From week 3 onwards participants that received the nudge were 13% till 18% more often present (table V) and it seems reasonable to extent this trend further.
6. Discussion
I will first discuss the two main limitation of this experiment. Firstly, in my experiment, the main differences compared to my intended research set-up, was the specific way the
commitment was asked of all participants. Many trainers did not use the exact commitment question I had intended them to use (see appendix table VIII). They asked for the commitment in a weaker form. One would expect that this weaker form should also weaken the power of the commitment, and therefore not have such a strong effect on the attendance of participants. Even if this is the case, the results of this experiment are still significant. But they could potentially have been even more convincing.
The second limitation of my experiment was the non-randomization of the distributions of the nudges. This was solved by using the differences in differences method. The differences in differences provided significant results and I was therefore able to draw conclusion based on these results.
In this experiment a nudge was used to induce a change in behavior. A nudge might work in the short run but might do more harm than good in the long run (Kohn 1993). Additionally it could be a problem that the desired outcomes are identified by experts, these experts do not always know what is best for the people that are influenced by the nudge they designed. (Chriss, 2015). In my experiment the goal of exercising more corroborated with the goal the participants had set for themselves. The autonomy of the participants was also not affected since they were able to choose whether they wanted to commit themselves or not. However, there is no consensus on the long run effectiveness of nudges. More research into the long-run effectiveness of nudges would be useful. Building on this it would also be interesting to determine which types of nudges are effective in the long run and which types of nudges aren’t.
It would also be an interesting topic for future research to determine whether a habit formation occurs with the commitment nudge used in this experiment The results of the 2009 Charness & Gneezy paper show that monetary incentives effectively stimulate people to go to the gym. This increase in attendance was observed the first 8 weeks that the participants received the
23 monetary incentives, the attendance rates were also higher when the incentives were removed for the group that was in the initial incentive group. This suggests the formation of a good habit. If asking for a commitment also results in habit formation, the nudge would be a very cost-effective way to stimulate people to form good habits. It would also be interesting to see which method works best to urge people to change their behavior and form good habits. For example, the difference in effectiveness of the commitment and a financial incentive can be investigated by exposing one group to the commitment-nudge, another group could earn financial incentives for ‘good-behavior’ and a third would be the control group.
7. Conclusion
The purpose of this thesis was to investigate whether a nudge in the form of a commitment to exercise in an effective way to increase the attendance rates of people in training sessions. This was the main research question of my thesis. Worldwide countries face problems with
overweight and obesity (WHO, 2016). A nudge could increase the exercising behavior of people and thereby provide a potential solution to this problem. To answer my main research questions, I drafted three sub-questions namely:
I To what extent are people willing to explicitly commit themselves to exercise? II How does making a commitment influence attendance behavior in practice?
III To what extent does this nudge of commitment help people change behavior in the way they desire?
Based on these three research questions I drafted three hypotheses that result in three main conclusions. The first confirmed hypothesis is that people are, once signed up for an exercise program, willing to commit themselves to upcoming training sessions. Secondly. this
commitment causes them to attend the training significantly more often after the first three
weeks, which partly confirms my second hypothesis. Participants start off with good intentions to exercise and the commitment is a significant push to help them maintain these good intentions after the first couple of weeks when their good intention are becoming harder to pursue. The third main conclusion is that the commitment helps people to attain their self-set goals. These goals are either in the direction of exercising more or a performance related goal; to run faster of longer. A combination of exercising more and a performance related goal is also possible. Going to the running classes more often helps to attain all of these goals, confirming my third hypothesis.
24 We can therefore conclude that a nudge in the form of a commitment to upcoming training sessions is an effective way to increase people’s frequency in physical exercise. It is useful for economists to know that similar nudges could also prove effective and applicable in different scenario’s when they are implemented correctly. The increase in exercising caused by my nudge could help to solve overweight and obesity problems and reduce the healthcare costs of a country.
8. Appendix
Appendix A:
Form for the documentation of the nudge.
Notes. The forms were accompanied by a description of which day the training session was held and which group it was.
Appendix B
The below table VIII summarized the deviation from the commitment question that were being used.
Table VIII
The different questions that were used by trainers
Question Percentage of runners in the control group that
25
As intended1 20.00%
Will you be present2 44.71%
Reported on sheet3 30.00%
Voice recorded4 5.29%
Notes, 1 the question I intended was used. 2 the question used was: ‘Will you be present’ 3 The results of the questions were reported on the answer sheet. 4. The answers to the commitment question were voice recorded by the trainer.
The participants of the running groups were asked to fill out a short survey. One of the questions concerned their motivation to run. The entire survey is shown below:
Appendix C:
Survey:
Heb je gezondheidsklachten waarvan wij op de hoogte moeten zijn? Zo ja, welke?
Gebruik je medicijnen waarvan wij op de hoogte moeten zijn? Zo ja, welke?
Heb je fysieke beperkingen of last gehad van blessures? Zo ja, welke?
In het afgelopen jaar hoe vaak heb je gesport per week?
Wat is je trainigsdoel?
o Ik wil dat hardlopen een gewoonte wordt o Ik wil langer kunnen hardlopen
o Ik wil sneller kunnen hardlopen o Meer bewegen/ conditie opbouwen o Iets anders namelijk, …
Als je loopervaring hebt. Hoeveel km kun je aaneen lopen?
Hoe 'ben je bij Hardlopenamsterdam gekomen?
9. References
26 cross-country study. Journal of Health Economics, 25(4), 599–620.
Acland, D., & Levy, M. R. (2015). Naiveté, Projection Bias, and Habit Formation in Gym Attendance. Management Science, 61(1), 146–160.
Angrist J. D. and J. S. Pischke. (2008). Mostly Harmless Econometrics : An Empiricist ’ s Companion. Massachusettts I nstitute of Technology and The London school of Economics.
Apesteguia, J., Funk, P., & Iriberri, N. (2013). Promoting rule compliance in daily-life: Evidence from a randomized field experiment in the public libraries of Barcelona. European Economic Review, 64, 266–284.
Ariely, D. (2012). Why We Lie. The Wall Street Journal, 1–6.
Ashcroft, R. E. (2011). Personal financial incentives in health promotion: Where do they fit in an ethic of autonomy? Health Expectations, 14(2), 191–200.
Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37(2), 122–147. Banerjee, A. V., Duflo, E., Glennerster, R., & Kothari, D. (2010). Improving immunisation coverage in rural
India: Clustered randomised controlled evaluation of immunisation campaigns with and without incentives. BMJ (Online), 340(7759), 1291.
Becker, G. S. (1962). Irrational Behavior and Economic Theory. Journal of Political Economy, 70(1), 1–13. Benabou, R., & Tirole, J. (2003). Intrinsic and Extrinsic Motivation, 70(April), 489–520.
Bénabou, R., & Tirole, J. (2018). Self-Confidence and Personal Motivation Author ( s ): Roland Bénabou and Jean Tirole Published by : Oxford University Press Stable Quaterly Journal of Economics, 117(3), 871–915.
Card, D., & Krueger, A. B. (1994). Minimum Wages and Employment : A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania Author ( s ): David C. The American Economic Review, 84(4), 772–793.
Chance, Z., Norton, M. I., Gino, F., & Ariely, D. (2011). Temporal view of the costs and benefits of self-deception. Proceedings of the National Academy of Sciences, 108(Supplement_3), 15655–15659. Charness, G., & Dufwenberg, M. (2010). Bare promises: An experiment. Economics Letters, 107(2), 281–
283.
Charness, G., & Gneezy, U. (2009). Incentives to Exercise. Econometrica, 77(3), 909–931. Chriss, J. J. (2015). Nudging and Social Marketing. Society, 52(1), 54–61.
Feild, J. (2015). Improving Student Performance Using Nudge Analytics. Proceeding of the 8th International Conference on Educational Data Mining, 464–467.
Frayne, C. A., & Latham, G. P. (1987). Application of social learning theory to employee self-management of attendance. Journal of Applied Psychology, 72(3), 387–392.
Frey, B. S., & Oberholzer-Gee, F. (1997). American Economic Association The Cost of Price Incentives : An Empirical Analysis of Motivation Crowding- Out. The American Economic Review, 87(4), 746–755. Gneezy, U. (2005). Deception: The Role of Consequences. The American Economic Review, 95(1), 384–
27 Gneezy, U., Meier, S., & Rey-Biel, P. (2011). When and Why Incentives (Don’t) Work to Modify Behavior.
Journal of Economic Perspectives, 25(4), 191–210.
Gneezy, U., & Rustichini, A. (2000). Pay Enough or Don’t Pay at All*. Quarterly Journal of Economics, 115(3), 791–810.
Gollwitzer, P. M. (1999). Implementation Intentions; Strong Effects of Simple Plans. American Psychologist, 54(7), 493–503.
Gollwitzer, P. M., & Oettingen, G. (1998). The emergence and implementation of health goals. Psychology and Health, 13(4), 687–715.
Guay, F., Chanal, J., Ratelle, C. F., Marsh, H., Larose, S., & Boivin, M. (2010). Intrinsic, identified, and controlled types of motivation for school subjects in young elementary school children. British Journal of Educational Psychology, 80(4), 711–735.
Kloss, R. J. (1994). A Nudge is Best: Helping Students through the Perry Scheme of Intellectual Development. College Teaching, 42(4), 151–158.
Lally, P., Jaarsveld va, C. H. M., Potts, H. W. W., & Wardle, J. (2010). How are habits formed: Modelling habit formation in the real world. European Journal of Social Psychology Eur., 40(June 2009), 998– 1009.
Kohn, A. (1993). Punished by Rewards: The Trouble with Gold Stars, Incentive Plans, A’s, Praise, and Other Bribes. Boston: Houghton Mifflin.
Latham, G. P., & Locke, E. A. (1991). Self-regulation through Goal Setting. Organizational Behavior And Human Decision Processes, 50(1), 212–227.
Lippke, S., Ziegelmann, J. P., & Schwarzer, R. (2004). Behavioral Intentions and Action Plans Promote Physical Exercise: A Longitudinal Study with Orthopedic Rehabilitation Patients. Journal of Sport and Exercise Psychology, 26(3), 470–483.
Locke, E. A., & Latham, G. P. (1984). Goal Setting: A Motivational Technique that Works!
Loewenstein, G., Asch, D. A., Friedman, J. Y., Melichar, L. A., & Volpp, K. G. (2012). Can behavioural economics make us healthier? BMJ (Online), 344(7863), 1–3.
Malmendier, U., & Della Vigna, S. (2006). Paying Not to Go to the Gym. The American Economic Review, 96(3), 694–719.
Martinez, I. (2014). The Effects of Nudges on Students’ Effort and Performance: Lessons from a MOOC. EdPolicyWorks, (19), 1–33.
Mokdad, A. H., Ford, E. S., Bowman, B. A., Dietz, W. H., Vinicor, F., Bales, V. S., & Marks, J. S. (2003). Prevalence of obesity, diabetes, and obesity-related health risk factors, 2001. Jama, 289(1), 76–79. Pietroni, D., & Hughes, S. V. (2016). Nudge to the future: capitalizing on illusory superiority bias to
mitigate temporal discounting. Mind and Society, 15(2), 247–264.
Samuelson, W., & Zeckhauser, R. (1988). Status quo bias in decision making. Journal of Risk and Uncertainty, 1(1), 7–59.
Sheeran, P. (2002). Intention—Behavior Relations: A Conceptual and Empirical Review. European Review of Social Psychology, 12(1), 1–36.
28 Sunstein, C. R. (2014). Nudging: A Very Short Guide. Journal of Consumer Policy, 37(4), 583–588.
Sunstein, C. R. (2016). Do Poeple Like Nudges? SSRN Electronic Journal. Suvorov, A. (2003). Addiction to Rewards. MIMEO, 1–51.
Thaler, R.H., & Sunstein, C.R. (2009). Nudge: Improving Decisions about Health, Wealth and Happiness, London: Penguin.
Vanberg, C. (2008). Why Do People Keep Their Promises? An Experimental Test of Two Explanations. Econometrica, 76(6), 1467–1480.
World Health Organazation website fact sheet. http://www.who.int/en/news-room/fact-sheets/detail/obesity-and-overweight
Williams, B. R., Bezner, J., Chesbro, S. B., & Leavitt, R. L. (2005). The Effect of a Walking Program on Perceived Benefits and Barriers to Exercise in Post-Menopausal African American Women. Journal of Geriatric Physical Therapy December, 21(4), 332–342.
Withrow, D., & Alter, D. A. (2011). The economic burden of obesity worldwide: A systematic review of the direct costs of obesity. Obesity Reviews, 12(2), 131–141.
