Master’s Thesis
Activity-based Lifestyle Changes through Text Message Inducement - An Observation Study -
Graduate School of Communication University of Amsterdam
Author: Maximiliane Roettger Student ID: 1060243
Master’s Programme Communication Science
Supervisor: Dr. Gert-Jan de Bruijn Word Count: 8665 Faculty of Social and Behavioural Sciences Pages: 30
TABLE OF CONTENTS
Introduction 1
Theoretical Framework 2
Promoting physical activity with communication means 2
The theory of planned behaviour (TPB) 4
Physical activity and the intention-behaviour gap 5
Motivational nudges to overcome the intention-behaviour gap 5
Action planning versus affective attitude 8
Possible other influences on successful PA 10
Methods 11 Sample 11 Experimental Design 12 Procedure 13 Measures 14 Results 18 Randomization check 18
Total sample comparison in PA 18
Increase in step-count Friday1 (no message) vs. Friday3 (five messages) 21
Increase in step-count between week2 and week3 22
Impact of the PMG message on PA 23
Impact of AMG message on PA 24
Conclusion & Discussion 25
Implications for future studies 29
References V
Appendix A – Original Messages XVII
Appendix B – Instructions and tools for participants XVII
Pedometer XVII
Diary XVIII
Information Sheet XIX
Appendix C – Questionnaires XXIII
Baseline XXIII
Follow up XXXII
FIGURES AND TABLES
Figure 1. Theory of planned behavior ... 4
Figure 2. Study course of 5 weeks ... 14
Figure 3. Comparison – increase in average step-count ... 19
Table 1. Overview of mean differences between AMG and PMG (Randomization) ... 18
Table 2. Overview average step-count AMG and PMG (daily basis) N=29 ... 20
Table 3. Overview average step-count AMG and PMG (daily basis) N=29 ... 21
Table 4. Mean differences, SD comparisons PMG and AMG in PA (N=29) ... 21
Table 5. Overview average step-count AMG and PMG (weekly basis) N=29 ... 22
Symbols and Abbreviations Symbols d Cohen’s d F Observed F value M Mean Max Maximum Min Minimum N Number of Respondents p Significance level SD Standard Deviation t observed t value α Cronbach’s alpha η2 eta squared Abbreviations
ANOVA Analysis of variance
AJ Affective Judgments
AMG Affective Message Group
AMGyes AMG message received = 1; not received = 0
PA Physical Activity
PAdaycomparison Physical activity change between Friday1 and Friday3 PAweekcomparison Physical activity change between week2 and week3
PMG Planning Message Group
Abstract
Objectives: A Five-week study was conducted to determine the effect of persuasive messages sent via mobile phones on two different post-intentional factors (action planning and affective attitude) and subsequently on physical activity (PA) of German adults (N=29).
Methods: Pedometers were distributed to the participants (16 females; 13 males; 49.5±17.2 years). The baseline questionnaire included items of the physical enjoyment scale (PACE), items assessing action planning behaviour, and items of the neighbourhood environmental walkability scale (NEWS). Participants received text messages twice a week for four weeks. One group received messages framed with a planning-supportive appeal (PMG; n=14); the other group received messages with an affective appeal (AMG; n=15). Physical activity (PA) was assessed via step-count measured with a pedometer. Participants
completed a follow-up questionnaire at the end of the study.
Results: Considering effect sizes, one could conclude that AMG worked better than PMG on the first two Saturdays. However, this difference vanished and changed into PMG being more effective than AMG on Saturday3. Overall, one could observe an increase in step-count in a weekly comparison (week2 and week3) with PMG having a greater increase than AMG. T-test results, though, revealed that PMG messages did not work better than AMG messages, as the observed difference was not significantly big. Results of the linear-regression analyses revealed no significant impact of messages (PMG or AMG) on post-intentional factors (action planning or affective attitudes) and subsequently not on a change in step-count.
Conclusion: None of the hypotheses could be confirmed since the sample size was too small to yield significant results. However, one could observe a tendency in the
effectiveness of the intervention method, with a planning-supportive message being more effective than a message with an affective appeal. This research provides useful
information for future studies to examine how the PA level can be increased. Key words: physical activity, text messages, theory of planned behaviour, intention-behaviour gap, post intentional factors, action planning, affective attitude.
Introduction
The modern lifestyle in Western world is a crucial problem in its population’s health. A mostly sedentary behaviour in life and work together with an unhealthy diet in containing large quantities of fast food and processed foods high in calories and carbohydrates are the underlying reasons for an increasing number of severe health problems (Carrera-Bastos, Fontes, O'Keefe, Lindeberg & Cordain, 2011).
According to the World Health Organization (WHO), inactivity as a result of lifestyle changes is considered one of the leading causes of obesity. Both, inactivity and obesity, rank among the five main risk factors for mortality (WHO 2009a). Obesity (measured by the Body Mass Index, BMI) together with high blood pressure, dyslipidaemia and insulin-resistance is part of the so-called ‘Deadly Quartet – the Metabolic Syndrome’
(International WHO 2009b). A BMI > 25 means obesity, a BMI > 30 means Adiposity. In Germany, for example, one in two adults is overweight (destatis.de, 2014), 20% suffer from adiposity (Szarek, 2014). The German Association for Adiposity (Adipositas-Gesellschaft.de, 2015) forecasts € 25.7 bn. of direct costs to treat obese people in 2020, twice as much as in 2003 (Szarek, 2014). This amount does not include indirect costs such as a loss in productivity.
An increase of physical activity (PA), however, has been proven to contribute to an improvement of many of the above mentioned conditions – even type 2 diabetes (Bravata et al., 2007). Still, the majority of adults and children is simply not sufficiently physically active to maintain a healthy lifestyle (Hallal et al., 2012). Therefore, the increase of PA must have priority for public health. The costs for treatment and rehabilitation and the impact on the work capacity and efficiency of younger patients are expected to become an almost unbearable burden for health insurances and the entire socio-economic system. Effective strategies and interventions to increase PA are rare (Dishman, Oldenburg, O’Neal & Shephard, 1998; Baumann et al., 2012). There is still a need to find and to
develop supportive interventions. Programmes that motivate a large number of affected people can be provided at low cost considering public interest.
This pilot study aims to examine such an intervention with the focus on communicative support, which reaches people during their daily routine and helps to bridge the gap between the intention of being physically active and the implementation of the resolution. The purpose of this pilot study is to examine, within a small group of Germans, a trend in the effectiveness of such a communicative support sent by text messages on mobile phones: Does a support via text messages with regard to PA impact people’s PA positively?
At the beginning, the paper presents a theoretical framework, followed by the method section and the description of the study. Results will be analysed and presented. In the final summary, the conclusion will examine strengths and weaknesses of the study and provide an outlook for potential follow-up studies.
Theoretical Framework Promoting physical activity with communication means
Physical activity (PA) is defined as ‘any bodily movement produced by skeletal muscles that requires energy expenditure’ (WHO.int, 2014) and includes all movements during work and leisure in everyday life (WHO.int, 2014). PA is done for various reasons and therefore, it has different meanings. In this study, PA is defined as a regular movement such as normal walking on daily basis. In his so-called SLOTH model, Cawley (2004) defines five domains of PA concerning ‘Sleep, Leisure, Occupation, Transportation and Home’. People choose to be either active or rather inactive in one of these five domains depending on various factors, which can be individual/psychological, social or
environmental (Pratt, Macera, Sallis, O'Donnell & Frank, 2004; Baumann et al., 2012). These factors may be barriers that can lead to a reduced PA in daily life, even though most
people intend to be physically active (Rhodes, de Bruijn & Matheson, 2010; Baumann et al., 2012; Rhodes & de Bruijn, 2013a).
Over the last decades, researchers have developed a range of intervention strategies
designed to improve PA, yielding various effects (Dishman & Buckworth, 1996; Marcus et al., 2006; Parrott, Tennant, Olejnik & Poudevigne, 2008). Media based interventions, e.g., print-mailings, telecommunication and e-mailings, still belong to the most successful interventions (Parrot et al., 2008). On top of that, several reviews about text messages on health behaviour change have shown that text messages are a promising intervention method (e.g., Lau, Lau, Wong & Ransdell, 2011; Plotnikoff, McCargar, Wilson & Loucaides, 2005). Plotnikoff, McCargar, Wilson and Loucaides (2005) concluded that participants receiving weekly mail messages were more successful in changing their behaviour and exercised more than those of a control group.
Mobile phones are more and more becoming tools to organise and coordinate daily life (Consolvo, Everitt, Smith & Landay, 2006; Patrick, Griswold, Raab & Intille, 2008). Consequently, the next step is to investigate whether text messaging directly to an
individual’s mobile phone is a suitable intervention technique. Almost 80 % of the German population own a mobile phone and 40 % of the population use the short message services (SMS) more than frequently (Statista, 2014). Additionally, short messaging offers a lot of advantages (wide acceptance, speed, relatively low costs and good accessibility, to mention only a few) compared to other communication devices. All this suggests the use of this type of message transmission for a PA intervention technique (Klasnja & Pratt, 2012). Until today, there are a few studies on the differences in message framing to improve exercise behaviour (e.g., Jones, Sinclair, Rhodes & Courneya, 2004). Their focus is based on the assessment of differences effected by positively versus negatively framed messages. The result shows that people respond more favourably to the positive ones. Additionally, Chatzisarantis and Hagger (2005) found that messages targeting salient beliefs were
effective in improving attitudes and thus also intentions (Parrott et al., 2008). This
emphasizes that messages can have an impact on factors of socio-cognitive models, such as the theory of planned behaviour (TPB) (Ajzen 1991, Conner & Sparks, 2005).
The theory of planned behaviour (TPB)
The TPB states that behavioural achievement depends on both motivation (intention) and ability (behavioural control/self-control) (Ajzen, 1991; Conner & Sparks, 2005). The model tries to predict the intention of an individual to engage in a specific behaviour at a specific time and place. It suggests that attitude, subjective norm and perceived
behavioural control (PBC) with their underlying beliefs are the three key predictors of intention. Intention in turn is the key predictor of behaviour (see Figure 1) (Rhodes & Nigg, 2011).
Attitude is defined as an overall evaluation of an individual’s relation to certain behaviour. A distinction is drawn between affective attitudes (positive or negative evaluation of engaging in a behaviour) and instrumental attitudes (belief about outcomes) (Russell, 2003). Subjective norms are assumed to assess the social pressures people feel when they perform a specific behaviour. Additionally, PBC is defined as the individual’s perception of the ability to execute a behaviour (Conner & Sparks, 2005; Rhodes & Nigg, 2011). The TPB has been used extensively in health behaviour studies (Armitage & Conner 2001; Hagger, Chatzisarantis & Biddle, 2002) and will be used as base model in this study.
External variables - demographics - personality - environmental influences behavioural beliefs normative beliefs control beliefs behaviour intention attitude subjective norm PBC perceived behavioural control
Physical activity and the intention-behaviour gap
Traditional models like the TPB (Ajzen, 1991; Conner & Sparks, 2005) put their focus on how to change and increase intention to engage in a specific behaviour. Although many people have positive intentions to change their behaviour through a higher degree of physical activity, they often do not behave accordingly. In scientific literature, this discrepancy between intention and behaviour is mostly referred to as the ‘intention-behaviour gap’ – it describes the invisible process that does or does not convert intention into actual behaviour and execution (Sheeran 2002, Sniehotta, Scholz & Schwarzer, 2005, Schwarzer, 2008; Rhodes, de Bruijn, Matheson 2010; de Bruijn, Rhodes & van Osch, 2011; Rhodes & Dickau, 2012).
In fact, intentions only explain 23 % - 30 % of variances in physical activity (Sheeran, 2002; Blanchard et al., 2005; Webb & Sheeran, 2006). Rhodes and de Bruijn (2013a) concluded in their meta-analysis that there are more people who actually fail to transform their intention into PA than people who do not have any intention at all from the outset. Therefore, the focus should be on post-intentional factors, either cognitive ones or affective ones (Rhodes & de Bruijn, 2013a; Sniehotta, Scholz & Schwarzer, 2005). A solid body of research suggests that action planning and affective attitude are two of those factors (Ajzen, 2001; Rhodes, Blanchard, Matheson & Coble, 2006; Rhodes, de Bruijn & Osch, 2011; Rhodes & de Bruijn, 2013a).
Motivational nudges to overcome the intention-behaviour gap Physical activity and action planning
Action planning, the detailed planning of when, where and what people will do, should facilitate the implementation of intentions. Gollwitzer (1999) calls the framing of
intentions for having a plan ‘implementation intentions’. In research on the topic this term is often used interchangeably with ‘planning’. By formulating an implementation intention, a person specifies when and where to be active and therefore has the increased opportunity
of recognising where and when to act. A lot of research has proven that formulating such implementation intentions does increase the likelihood of performing health behaviours, like PA (Oettingen & Gollwitzer, 2000; Milne, Orbell & Sheeran, 2002; Gollwitzer & Sheeran, 2006).
The concept of action planning has been tested in various studies as moderator, mediator or even moderated-mediator (for an overview, see Mistry, Sweet, Latimer-Cheung & Rhodes, 2014). It is established as an effective strategy for closing the intention-behaviour gap. Latimer, Ginis and Arbour (2006) found evidence of the efficacy of implementation intentions for promoting physical activity among participants. In fact, participants who formulated those specific intentions, engaged in more physical activity than participants in the control group. Through the formulation of implementation intentions, the mental representation for a situation becomes more precise, and therefore easier to accomplish for people (Gollwitzer & Sheeran, 2006). Although, the benefits of action
planning/implementation intentions are clear, the compliance to action planning interventions is poor (Sniehotta, 2009, Mistry, 2013).
A recent study by Sweet, Brawley, Hatchell, Gainforth and Latimer-Cheung (2014) examined the effects of promoting action planning, using different persuasive messages as means for increasing PA. Individuals who read messages about the benefits of action planning made more plans than people who read about the costs of not creating any plan. There is still a great challenge to get people to make their own detailed plan so that they actually act in the right way (Mistry et al., 2014). Offering an external nudge, so that people feel more motivated to think about a concrete plan as to when and where to be active, could encourage them to establish one and to implement PA in their daily life (Hypothesis 1).
Physical activity and affective attitude
In addition to cognitive constructs such as planning, there are other post-intentional factors that are assumed to be predictors of the intention-behaviour discordance. One of the most important ones is affective attitude (i.e. enjoyment and pleasure from the PA) (Rhodes, Fiala & Conner, 2009; Conner, Rhodes, Morris, McEachan & Lawton, 2011; Rhodes & de Bruijn, 2013a). A growing body of research indicated that affect plays a large role in health behaviour change and decision making, be it a predictor of intentions (Ajzen, 2001; Lowe, Eves & Carrol, 2002; Keer, van den Putte & Neijens, 2010) or be it a direct predictor of behaviour (Kiviniemi, Voss-Humke & Seifert, 2007; Lawton, Conner & Parker, 2007). Russell (2003) explained in his study that affective qualities emerge from experiencing an emotion while being engaged in behaviour. Therefore, these affective qualities will be automatically attributed to this certain behaviour, e.g., physical activity (PA) (Russell 2003) and seem to lead to a higher motivation for further execution of PA. Lawton, Conner and McEachan (2009) built on this assumption in their study. They concluded that
affective attitudes were strong predictors for almost all behaviours, be it health risk behaviours (e.g., smoking and drinking) or health promoting ones (e.g., flossing teeth and PA). Additionally, they proved that affective attitudes predict directly behaviour.
Therefore, this effect is not being mediated by intention (Conner et al., 2011, Lawton, Conner & McEachan, 2009). Furthermore, they concluded that affect is more important than cognition in predicting intention and health-related behaviours. This assumption was tested by some research that assessed differences in behaviour when using affective or cognitive persuasive messages (Keer, Conner, van den Putte & Neijens, 2013; Keer, van den Putte, Neijens & de Wit, 2013). Conner et al. (2011) compared e-mail messages that informed participants about potential risks of being inactive and about benefits of being active. They used either an affective or cognitive appeal, with the affective one being more successful. Further studies examined the effects of affective messages in contrast to
cognitive messages related to the change of attitudes towards a specific behaviour and demonstrated that affect-based messages can be effective in producing behaviour change (Haddock, Maio, Arnold & Huskinson, 2008, Conner et al., 2011). Sirriyeh, Lawton and Ward (2010) found that PA levels increased significantly more among inactive adolescents who received messages targeting affective beliefs than in the control group and in the group with messages targeting instrumental beliefs.
Keer, van den Putte, Neijens and de Wit (2013) found that affect-based arguments are essential for changing people’s intentions in their relationship to health behaviour. Therefore, targeting affective benefits (having fun while being active) is an effective intervention (Lawton, Conner & McEachan, 2009). It guides attention and behaviour to a specific object (Monahan, 1995, Russell 2003) and is likely to produce strong effects as positive moods and to foster attitude changes (Monahan, 1995; Petty & Brinol, 2010). With an external nudge to encourage people to have fun while being active, their affective beliefs about PA could increase, and therefore could lead to higher PA in their daily life (Hypothesis 2).
Action planning versus affective attitude
Both planning and affective attitude are seen as post-intentional factors that can lead to behaviour adoption. The potential role of planning in intent translation is widely discussed in research. Affective attitude on the other hand has not received as much attention until recently (Rhodes & de Bruijn, 2013a).
Rhodes, Fiala and Conner (2009) analysed the effects of affective judgements (AJ) on physical activity. AJ include all affective factors such as the overall pleasure/displeasure, enjoyment and general feeling generated by PA (Keer, Putte & Neijens, 2010). In their meta-analysis Rhodes et al. (2009) found 83 out of 85 correlation-studies with significant positive correlations between AJ and PA. Additionally, there are twenty experimental-studies, which assessed impacts on AJ either as dependent or mediating variable in the PA
domain. However, very hardly any of them yielded significant results. Among these studies, there are two studies on PA via persuasion, which included AJ factors and showed some positive findings (Parrot et al., 2008; Hardeman, Kinmonth, Michie & Sutton, 2009). This could lead to the assumption that AJ and therefore affective attitude may have an impact on persuasion attempts for PA, although the two studies have little overall evidence.
For planning, there is strong empirical evidence that it mediates the intention – PA relationship and it is seen as an important factor in intervention methods (Bartholomew, Parcel, Kok, Gottlieb & Fernández, 2011; Luszczynska et al., 2010; Hagger &
Luszczynska, 2013). Systematic reviews have analysed the impact of action
planning/implementation intentions on health-related behaviours (Gollwitzer & Sheeran, 2006; Bélanger-Gravel, Godin & Amireault, 2013). Although implementation intentions, and therefore planning were seen as a powerful strategy to overcome the
intention-behaviour gap, Bélanger-Gravel, Godin and Amireault (2013) found only small to medium effect sizes of -0.21 up to 0.74 (Cohen’s d) between planning and PA. These effects of planning on PA were still bigger than the ones that were found of AJ on PA.
Considering these findings in effect sizes in the affect-PA and planning-PA relationships, it is expected that the planning-supportive text messages will yield more positive effects than the affective-based text message (Hypothesis 3).
The main expectations of this study are summarised in the following hypotheses. H1: A supportive text message that encourages German adults to generate a
concrete plan when to be active (a) leads to an increase in physical activity (b) and this relationship is mediated by action planning.
H2: A supportive text message that encourages German adults to have fun while being active (a) leads to an increase in physical activity (b) and this
relationship is mediated by affective attitude.
H3: A supportive text message that encourages German adults to generate a concrete plan when to be active leads to a higher increase in physical activity than the text message that encourages German adults to have fun while being active.
Possible other influences on successful PA
As mediation does not apply to everyone in the same way the assumptions suggest other factors to have an impact on the motivation – PA relationship. They may be the reason why people succeed in PA or not (Pratt et al., 2004; Baumann et al. 2012). The factors might depend on psychological or environmental conditions or barriers (Koring et al., 2011).
One potential psychological barrier is identified repeatedly as one of the most important determinants: self-efficacy (Hagger, Chatzisarantis & Biddle, 2002). Self-efficacy reflects an individual’s confidence in the ability to be successful in a specific behaviour such as PA (Bandura, 1998; Colberg et al., 2010, Olander et al., 2013). It is crucial for individuals to regulate their thought processes while dealing with a new or a difficult situation
(Luszczynska, Schwarzer, Lippke & Mazurkiewicz, 2011). When thinking about their future behaviour people reflect their perceived personal competence to perform the behaviour internally (‘I am confident that I can be physical at least once a week’)
(Sniehotta, Scholz & Schwarzer, 2005). People who hesitate with their confidence might fail to act on their plans, but might be motivated more through the encouragement on an affective appeal. Therefore, self-efficacy can be expected to have an impact on the
message-PA relationship (Bélanger-Gravel, Godin & Amireault, 2013; Rhodes & de Bruijn, 2013b).
Next to psychological factors as perceived self-efficacy, there are others that influence the intention-behaviour relationship and might be the reason for variations in effect sizes. Studies from health literature and urban planning (Sallis, King, Sirard & Albright, 2007; Duncan, Spence & Mummery, 2005; Transportation Research Board & Institute of Medicine, 2005) proved that people are more active when they live in walkable neighbourhoods, meaning a well-established land-use mix (proximity of homes and destinations such as shops), with access to facilities like parks and sidewalks or good connected streets. Women aged 50+ were found to be more physically active in an attractive neighbourhood scenery than women of the same age who do not live in an attractive scenery (Sallis et al., 2007). Researchers also observed better results among younger population samples (university students) or clinical populations compared to results in the general population (Bélanger-Gravel, Godin & Amireault, 2013). Van Stralen, de Vries, Mudde, Bolman and Lechner (2009) assessed the main determinants of physical activity among older adults. Besides psychological factors such as self-efficacy, physical environmental determinants including land-use mix and traffic safety as well as socio demographic determinants such as age and gender were found to be crucial
determinant for being physical active (van Stralen, De Vries, Mudde, Bolman & Lechner, 2009; Sallis et al.; 2007; King et al., 2003; Morris, McAuley & Motl, 2007).
Methods Sample
A convenience sample of 31 German respondents aged between 25 and 72 participated in the experiment. The sample consisted of 54.8% females (n=17). Half of the sample
(53.1%) was living in a big city with more than 100 000 inhabitants and the majority (71.9%) was highly educated.
Participants had to have access to a computer with internet connection, as the baseline and follow up questionnaire were distributed online. Additionally, subjects received a package including a pedometer (sponsored by the Beurer GmbH, Beurer AS 50 Elektronischer Aktivitätssensor), information and instructions about the study and about general benefits of physical activity together with a step-count diary.
Two of the participants dropped out because they lost their pedometer in the first week; therefore, only 29 subjects participated in the main analysis.
Experimental Design
The study employed a single-factor between-subjects design, which included two groups to which participants were randomly assigned (conditions: emotional-motivational framed and planning-support framed messages). It consisted of a baseline questionnaire, an intervention phase of 4 weeks and a follow up questionnaire on the last day of the study (December 21st).
To assess whether the independent variable ‘message-based support’ affected their step-count (dependent variable), the manipulations provided an affective appeal in the text message for 15 subjects and a planning-supportive text message for the other 14 subjects. To examine whether self-efficacy, land-use mix or perceived traffic-safety had a
significant impact on that relationship, the baseline provided items to ask for demographic and environmental variables as well as for potential barriers such as self-efficacy.
Manipulation check
To check whether the manipulation was successful, at the beginning of the follow-up, all participants had to answer the question whether they received the PMG message, ‘yes’ or ‘no’.
Those who indicated ‘yes’ were led to a second item asking how often they had received the PMG message within one week: ‘never’ up to ‘twice per week’.
Those who answered ‘no’ were led to a corresponding item asking how often they had received the AMG message: ‘never’ up to ‘twice per week’.
A frequency analysis of PMGyes proved that nPMGyyes=14. Those 14 subjects indicated that they had received their message twice per week. The remaining subjects (nAMGyes=15)
indicated that they had received the AMG message twice per week as well. The manipulation was successful.
Procedure
The whole study (baseline, intervention and follow up) was run in a time frame of five weeks. The intervention itself started on November 24th. The baseline and instructions were distributed by e-mail. The subjects had one week (November 16th-November 23rd) time to agree to participate and to fill in the questionnaire. They received brief information on the background of the entire study and were informed that their contribution was entirely voluntary and that they had the right to opt out of the study at any point of time. Participants were offered the chance to keep the pedometer as an incentive for their involvement. The precondition was that only those subjects, who provided their mobile phone numbers, could be included in the study. Upon agreeing to proceed with the experiment participants were then asked for their mobile phone number, some basic demographic information (such as age and gender), information about their attitude towards PA and planning-behaviour as well as information about their environment. Subjects were instructed to proceed with their normal daily life when wearing the pedometer. They were also asked to write down their daily step-count in the diary every evening around the same time and to transmit the step-count via e-mail to the researcher every Sunday.
The participants were then randomly assigned to one of the two groups: PMG (Planning Message Group) (n=14, nmale= 7; nfemale= 7; Mage= 51.50); AMG (Affective Message Group) (n=15, nmale= 6; nfemale= 9; Mage= 47.53). This design did not meet guidelines for a randomized control trial because the researcher was aware of group allocation, while the participants were not. All participants had to complete the self-report baseline measures. Afterwards, the PMG received the planning-supportive text message. The AMG received the text message that contained an affective text. All participants received their messages twice a week (Wednesdays and Saturdays) in the morning (7.a.m, CET).
Messages were sent in the morning in order to reach everybody at the same point of time and provide the same conditions for everyone (working and retired people). At the end of week4, participants of both groups completed a follow-up measure to check the
manipulation and its effectiveness.
Measures
Each group received a message containing a text to motivate them to be active. However, according to the group participants were in, those texts were framed differently. The message for the PMG (Planning Message Group) was aimed to target the planning ability of participants through the encouragement to plan an activity in their daily life: ‘Being active is easier if you think about when and where you will be active. Check your schedule and see how you can be active today!’
The message for the AMG (Affective Message Group) was created around the notion of improving affective attitude for PA and ran as follows: ‘Having fun while being active is good for your health! Pick an activity that you enjoy and just do it!’
Original messages were stated in German as the sample consisted of Germans only. The variable ‘group’ was coded into PMG = 0 and AMG = 1.
Two additional variables were computed out of the follow-up questionnaire, according to whether people received the PMG or AMG message. Variables were named as ‘PMGyes’, coded as 1=yes, I have received PMG message and 0=no, I did not and ‘AMGyes’, coded as 1=yes, I have received AMG message and 0=no, I did not.
Outcome measures
The bases for the primary outcome measures were changes in PA between weeks and the differences in PA in a daily comparison. The bases for the secondary outcome measures were change in affective attitude and change in action planning.
PA was assessed by using pedometers. Data about the daily step-count were collected by the electronic pedometer ‘Beurer AS 50 Elektronischer Aktivitätssensor’ (Beurer GmbH, Ulm, Germany). It measured total steps on a daily basis. The participants transmitted the daily results to the researcher once a week.
The first comparison assessed the difference between the average numbers of steps of the entire group on Friday of week1 to the average number of steps from Friday of week3. Researchers had chosen Friday1 for the comparison because the first four days had to be indicated as familiarization phase. All participants had to get used to the Beurer pedometer and the study design. Thus, the first text message on Wednesday1 could not be included in the evaluation, since almost half of the subjects, especially the older participants, did not set up their pedometer right. So, the first valid data was transmitted for the Friday of week1.
PAdaycomparison was computed by subtracting the average step-count of the Friday of week1 (Friday1) from average step-count of the Friday of week3 (Friday3).
For a direct comparison of the PA increase between two single weeks with respect to the frequency of messages, the focus was on week2 and week3. As mentioned above, the first week was not taken into account, because of this so-called familiarization phase. In
addition, two participants had given an incorrect phone number, so they did not receive any message on Wednesday1. This problem was solved and they got the text messages from the second message onwards. Also week4 was not included in the weekly comparison because a lot of participants went on holiday. It meant a daily routine different from the previous weeks. The focus was on week2 and week3 to get a solid comparison, as both weeks provided the same daily routines and conditions.
PAweekcomparison was computed subtracting step-count week2 from average-step-count week3.
Affective attitude towards PA was measured by an adoption of the physical activity enjoyment scale (PACE) (Motl et al., 2001; Kendzierski & DeCarlo, 1991). Participants were questioned to rate the degree to which they agreed with statements like ‘When I am physically active, I enjoy it’ or ‘When I am physically active, I feel bored” with ‘1’ indicating 'totally disagree ‘and ‘5’ indicating ‘totally agree’ with 15 items using a five-point Likert scale. Seven items were coded reversely, and therefore had to be recoded. After recoding, all 15 items were summed up and yielded a scale with α=.88.
Affective attitude change was measured by computing a new variable, subtracting baseline values from follow-up values.
Planning was assessed with items similar to those used by Sniehotta, Scholz and
Schwarzer (2005). Participants were questioned to indicate the truth of the statements with six items measured on a five-point Likert scale. A higher score indicated a higher
motivation for planning. The item stem, ‘I make a detailed plan every day…’ was followed by the items (a) ‘…where to be active’, (b) ‘…when to be active’, (c) ‘…how to be active’, (d) ‘…how often to be active’, (e) ‘…what to do when something interferes with my plans’ and (f) ‘…how to cope with possible drawbacks’. All six items were summed up and yielded a scale with α=.87.
Change in planning behaviour was assessed by computing a new variable, subtracting baseline values from follow-up values.
Covariates
Self-efficacy was assessed by three items measured on a five-point Likert scale, running from 1 ‘not true at all’ up to 5 ‘totally true’. The item stem, ‘I am confident that…’ was followed by statements like ‘I can integrate some physical activity into my daily life’. The scale including all three items yielded a Cronbach’s α=.69. Subsequently one item (‘I am confident that I can be physical active at least once per week’) was excluded from the scale, so that the reliability increased α=.73.
Traffic safety was assessed by the adoption of the neighbourhood environmental
walkability scale (NEWS) (Saelens, Sallis, Black & Chen, 2003). Three items measured the perceived traffic safety in the neighbourhood of each participant on a five-point Likert scale running from 1 ‘don’t agree’ up to 5 ‘totally agree’. Higher scores on the scale indicated at higher perceived traffic safety. One item was coded reversely and had to be recoded. The scale reliability analysis provided α=.70.
Land-use mix was assessed by the walking proximity from home to various types of facilities. Responses ranging from 1-5 minutes walking distance (coded as 1) up to 30+ minutes walking distances (coded as 5). Lower scores on the land-use mix scale indicated a closer average proximity. The computed scale land-use mix yielded α=.95.
Sex and age were assessed by asking participants to indicate their gender and age in years. Three age groups were built (new variable ‘age group’), 25-40 years (n=10), 41-61 years (n=11) and 61+ years (n=9). ‘Sex’ was coded 1 = female and 0=male.
Results Randomization check
To check whether the random assignment to the two different groups was successful and led to an even distribution of the 29 subjects by age, sex, self-efficacy, perceived traffic safety and land-use mix, independent t-tests were conducted. The comparison of the means of the two groups (whereby nPMG=14 and nAMG=15), with ‘age’, ‘sex’, ‘self-efficacy’, ‘traffic safety’ and ‘land-use-mix’, as dependent variables and ‘group’ (PMG = 0, AMG = 1) as independent variable, verified whether the randomization had worked. Table 1 gives an overview of the means and standard-deviations.
Table 1. Overview of mean differences between AMG and PMG in age, sex, self-efficacy, traffic-safety and land-use mix (with nAMG=15; nPMG=14)
MAMG SDAMG MPMG SDPMG t(27) p age 47.53 16.59 51.50 17.21 .631 .545 sex .60 .51 .50 .52 -.525 .468 self-efficacy 3.56 1.01 4.25 .83 1.97 .781 traffic-safety 3.33 1.09 3.59 .79 .733 .440 land-use-mix 2.91 1.03 3.14 .94 .623 .589
As the independent t-tests yielded no significant differences between the two groups in the dependent variable, the randomization was successful. Both groups were similar in their distribution of age, sex, self-efficacy, perceived traffic safety and land-use mix.
Total sample comparison in PA
On the first Friday of the study, participants of both groups (N=29) had a mean step-count of 7004.38 (Min=1323, Max=15031). Exactly two weeks later, on the third Friday, when
all 29 subjects had received four messages, the average step-count was at 7620
(Min=3040, Max=12670). This indicated a small tendency of increased steps of the total sample, but this increase was not significant, with Mdaycomparison= 615.62,
SDdaycomparison=3668.67, t(28)=.90, p=.374.
A small increase in average step-count of the total sample was also observed in the
comparison of two weeks (N=29) from the average step-count of week2 (Mweek2=6573.30,
Min=1875.71, Max=13050.29) to week3 (Mweek3=7206.67, Min=3034.86, Max=14983.71).
This increase in step-count of week3 to week2 was almost significant with, Mweekcomparison=633.36, SDweekcomparison=1787.40, t(28)=1.91, p=.067.
To summarize, one could observe a slight increase in step-count be it between week2 and week3 or be it between the first Friday of the study period and the third Friday (after everybody had received four messages). The total development of step-count during the four study weeks is also shown in Figure3. It presents the average increase in step-count, measured after the reception of the weekly messages. Starting point is the first Saturday after the familiarization phase. One can see an increase after Saturday2 (after 5 messages), but a striking drop in week4 with the Christmas season starting.
Figure 3. Comparison – increase in average step-count, after messages received (N=29) Saturday1: 1 message; Saturday2: 3 messages; Saturday3: 5 messages; Saturday4: 7 messages
Saturday 1 Satturday 2 Saturday 3 Saturday 4
Total Sample 6591 6069 8930 6750 PMG 5735 5507 9231 5900 AMG 7390 6593 8649 7544 2000 3000 4000 5000 6000 7000 8000 9000 10000 A V E R A G E S T E P -C O U N T
Comparison of PMG and AMG on a daily basis (Saturday1 vs. Saturday2 vs. Saturday3)
To test whether the increase in step-count varied significantly between the two groups (PMG and AMG), univariate ANOVAs were conducted. Table 2 shows an overview of the average step-count of PMG and AMG on a daily comparison (Saturday1 vs. Saturday 2 vs. Saturday3).
Table 2. Overview average step-count AMG and PMG (daily basis) N=29
Saturday1 Saturday2 Saturday3
M SD M SD M SD
PMG (n=14) 5735.15 3882.70 5506.50 2896.81 9230.71 3900.19 AMG (n=15) 7389.80 3811.61 6593.07 3956.81 8648.73 5288.63
Although one could observe a difference between PMG and AMG at all three Saturdays, with AMG being more effective than PMG on the first two Saturdays, conducted
ANOVAS yielded no significant results.
The first ANOVA with Saturday1 as dependent variable and group (coded PMG=0; AMG=1) as independent variable yielded no significant effect, F(1,27)=1.34, p=.257, η2=.05. However, one could observe a medium effect size of d=.43 meaning that AMG was more effective on Saturday1 than PMG.
The second ANOVA with Saturday2 as dependent variable and group as independent variable also yielded no significant effect, F(1,27)=.70, p=.409, η2=.025, d=.31. However, the small effect size indicated that AMG was still more effective than PMG. The following ANOVA using Saturday2 as dependent and group as independent variable, while
controlling for Saturday1, yielded a no significant result with F(1,26)=.126, p=.725, η2=.005. This result indicated that differences between the groups almost vanished when it was controlled for activities on the previous Saturday.
A fourth ANOVA with Saturday3 as dependent variable and group as independent variable yielded no significant result either, F(1,27)=.11, p=.740, η2=.004, d=.10. The result
showed that the PMG and AMG did not differ significantly. The following ANOVA including Saturday1 and Saturday2 as covariates, was not significant, F(1,25)=.90, p=.352, η2=.034. However, PA on Saturday3 was dependent on activities of the previous Saturdays.
Considering these findings in effect sizes, one could conclude that AMG worked better than PMG on Saturday1 after the message was received. However, this difference vanished through the following Saturdays and changed into PMG being more effective than AMG on Saturday3. This effect was also dependent on activities of previous Saturdays.
Increase in step-count Friday1 (no message) vs. Friday3 (five messages)
When focussing on the increase of step-count between Friday1 (no message) and Friday3 (five messages) one could observe a difference between the groups, with PMG being more effective than AMG. Additionally, one could even observe a decrease in step-count in AMG (an overview of means and standard deviation is given in Table 3 and mean differences are displayed in Table 4).
Table 3. Overview average step-count AMG and PMG (daily basis) N=29
Friday1 Friday3
M SD M SD
PMG (n=14) 6091 3177.55 7511.57 3189.53
AMG (n=15) 7856.87 3475.84 7721.20 2873.07
Table 4. Mean differences, SD comparisons PMG and AMG in PA (N=29)
MPMG SDPMG MAMG SDAMG MAMGPMG F(1,27) p PAdaycomparison 1420.57 4472.35 -135.66 2663.15 1556.24 1.32 .261
The ANOVA with PAdaycomparison (Friday3-Friday1) as dependent variable and group (PMG=0, AMG=1) as independent variable showed no significant difference in the effectiveness of message type on PA, F(1,27)=1.32, p=.261, η2=.047, d=.42.
The medium effect size, though, indicated that the increase in step-count from Friday1 up to Friday3 differed between the two groups, as PMG worked better than AMG.
These results, considering effect sizes of group comparison in the increase in step-count from Friday1 to Friday3 support H3, since the planning supportive appeal worked better than the affective appeal.
Increase in step-count between week2 and week3
When concentrating on the increase of step-count between week2 and week3, one could observe a difference between the groups, with PMG being more effective than AMG (an overview of means and standard deviation is given in Table 5 and mean differences are displayed in Table 6)
Table 5. Overview average step-count AMG and PMG (weekly basis) N=29
The ANOVA with PAweekcomparison (week3-week2) as dependent variable and group as independent variable, showed no significant difference in the effectiveness of message type on PA, F(1,27)=1.15, p=.293, η2=.041, d=.39. The small to medium effect size, though, indicated that the increase in step-count from week2 up to week3 differed between the two groups, as PMG worked better than AMG.
Table 6. Mean differences, SD comparisons PMG and AMG in PA (N=29)
MPMG SDPMG MAMG SDAMG MAMGPMG F(1,27) p PAweekcomparison 1000.66 2010.28 290.55 1541.50 710.11 1.49 .293
These results, considering effect sizes of group comparison in the increase in step-count from week2 to week3 support H3, since the planning supportive appeal worked better than the affective appeal.
Week2 Week3
M SD M SD
PMG (n=14) 6004.64 2104.78 7005.31 2919.92 AMG (n=15) 7104.05 7104.05 7394.60 2894.50
Impact of the PMG message on PA
To test whether step-count is dependent on the fact that one received the PMG message or not (H1), two linear regression analyses were conducted.
Comparison Friday1 with Friday3
The analysis with PAdaycomparison as dependent and PMGyes (coded 1=received, 0=not received) as independent variable was not significant, F(1,27)=1.32, p=.261. This result indicated that PMG message had no overall impact on change of step-count between Friday1 and Firday3. To test whether this relationship is mediated by action planning (H1b), a mediation analysis was conducted. It was first assessed whether the PMG message had an impact on action planning. A linear regression model with PMGyes as independent and change in planning behaviour as dependent variable yielded no significant result, F(1,27)=.29, p=.593. Therefore, the PMG message had no significant impact on motivation for action planning. The subsequent regression analysis with change in planning behaviour as independent and PAdaycomparison as dependent variable yielded no significant effect, with F(1,27)=.01, p=.925, indicating that planning behaviour had no impact on change in PA.
Comparison week2 to week3
Also the second regression analysis with PAweekcomparison as dependent and PMGyes as independent variable was not significant, F(1,27)=1.15, p=.293. This result indicated that PMG message had no overall impact on change of step-count between week2 and week3. To test whether this relationship is mediated by action planning (H1b) a mediation analysis was conducted. It was first assessed whether the PMG message had an impact on action planning. A linear regression model with PMGyes as independent and change in planning behaviour as dependent variable yielded no significant result, F(1,27)=.29, p=.593. Therefore, the PMG message had no significant impact on motivation for action planning. The subsequently regression analysis with change in planning behaviour as independent
and PAweekcomparison as dependent variable was not significant either, F(1,27)=.23, p=.634. This result indicated that planning behaviour had also no impact on change in PA between two weeks.
Inferring from these insignificant results, H1a and H1b have to be rejected, as there is no significant impact on step-count and no verifiable mediation effect of action planning. Impact of AMG message on PA
To test whether step-count is dependent on the fact that one received the AMG message or not (H2), two linear regression analyses were conducted.
Comparison Friday1 to Friday3
The analysis with PAdaycomparison as dependent and AMGyes (coded 1=received, 0=not received) as independent variable was not significant, F(1,27)=1.32, p=.261. This result indicated that AMG message had no overall impact on change of step-count between Friday1 and Firday3.
To test whether the main effect of AMG message on PA is mediated by affective attitude (H2b), a mediation analysis was conducted. It was first assessed whether the AMG message had an impact on affective attitude change. A linear regression model with AMGyes as independent and affective attitude change as dependent variable yielded no significant result, F(1,27)=.63, p=.435. Therefore, the text message had no significant impact on affective attitude change. The subsequent regression analysis with affective attitude change as independent and PAdaycomparison as dependent variable yielded no significant effect, with F(1,27)=3.13, p=.088. This result, close to significance, strengthens the assumption of affective attitude being a direct predictor of behaviour.
Comparison week2 to week3
The second regression analysis with PAweekcomparison as dependent and AMGyes as independent variable was not significant, F(1,27)=1.15, p=.293. This result indicated that AMG message had no overall impact on change of step-count between week2 and week3.
To test whether the main effect of AMG message on PA is mediated by affective attitude (H2b), a mediation analysis was conducted. It was first assessed whether the AMG message had an impact on affective attitude change. A linear regression model with AMGyes as independent and affective attitude change as dependent variable yielded no significant result, F(1,27)=.63, p=.435. Therefore, the text message had no significant impact on affective attitude change. The subsequent regression analysis with affective attitude change as independent and PAweekcomparison as dependent variable was not significant either, F(1,27)=2.23, p=.147. This result indicated that affective attitude did not have an impact on change in PA between week2 and week3.
Inferring from these insignificant results, H2a and H2b have to be rejected, as there is no significant impact on step-count and no verifiable mediation effect of affective attitude.
Although one can observe a small tendency towards an improved PA between week2 and week3 and the first Friday and the third Friday, no significant impact of message types on PA nor a significant mediation could be examined. Therefore, H1 and H2 have to be fully rejected.
Conclusion & Discussion
The aim of the present study was to examine the effectiveness of a communicative support via text messages on physical activity and to assess differences in the effectiveness of the type of the message frame. Additionally, the question was whether one could narrow or even close the ‘intention-behaviour gap’ addressing two post-intentional factors, action planning and affective attitudes. The study was innovative in its aim to distinguish between those two factors and to compare the effectiveness of them.
As stated above, action planning is seen as an effective strategy for closing the ‘intention-behaviour gap’ (Latimer, Ginis & Arbour, 2006). But it was argued that the compliance to interventions targeting action planning is still poor (Sniehotta, 2009; Mistry, 2013). This
study aimed to help people to formulate a concrete daily plan about their PA, by providing them with motivational nudges via text messages (PMG message). This should lead to higher action planning and hence to higher PA. Results indicated that the PMG message had no significant impact on action planning. Furthermore, contrary to former studies action planning had no significant impact on change in PA.
Affective attitudes were proven to be strong and direct predictors of almost all behaviours, with not being mediated by intention (Lawton, Conner & McEachan, 2009). Therefore, this study suggested that addressing affective attitudes through affective framed text messages would increase attitudes towards PA, and thus yield an increase in PA. Results could not support this assumption either. On the other hand, it has to be considered that the effect of affective attitude on the change in PA on daily basis (PAdaycomparison) was very close to significant (p=.088). Given the small number of participants, it is in line with former research, claiming that affective attitudes are direct predictors of behaviour.
Although the actual effect of text-support on action planning or affective attitudes and subsequently on PA was not significant, there was a difference between the two groups. Participants who received a text message including a planning supportive appeal (PMG) had a greater increase in PA than participants who received a message with an affective appeal (AMG): Effect sizes (ddaycomparison=.42; dweekcomparison=.39) supported Hypothesis 3
as the planning supportive appeal led to a higher increase in step-count than an affective appeal. However, the day to day comparison between the Saturdays indicated that AMG was more effective than PMG on the first two Saturdays after having received the
messages, which goes contrary to the assumption that planning-support works better than affective appeal.
Even though, not all tested hypotheses could be accepted due to insignificant results, one could observe an overall increase in step-count during the four weeks. Both groups showed an increase in step-count till week3, but also both groups decreased in week4 again
(Christmas season starting) (Figure3), with the AMG reaching in week4 even a value below the starting value. The increase of the total sample between week2 and week3 was close to significance (p=.063).One can observe a positive trend overall towards increased PA with applying the text-message intervention, be it PMG or be it AMG. Still, it is not possible to draw far-reaching conclusions about the actual effectiveness, since the current study contained a very small sample of N=29 and served as a pilot study to filter a
tendency.
Limitations of the intervention design and suggestion for follow up study
In general, one has to take a closer look at the study limitations to understand the study results. A crucial factor was the point of time and also the short time frame for the
experiment. With the experiment starting at the end of November and running almost until Christmas, the period was not optimal but predetermined by the timeline of the thesis. Over the four weeks the weather changed more than usual; particularly, towards the end of the study, streets became icy and made it difficult to walk outside, especially for older participants. This could be an explanation for the drop in week4. In addition, Christmas holidays in Germany started in week4, therefore, daily routine changed for all participants. Given these kinds of influencing factors, research could only concentrate on a change in PA during a two week period.
The fact that everybody themselves was responsible to record the step-count in their diaries one can assume that the transmitted step-count was biased. Due to the self-reporting, it is difficult to assess the correctness of the data, as no one was able to control the transmitted step-count. It is possible that some of the numbers might mirror mean values estimated on step-count of previous days. For better control, a possible solution could be to establish devices for digital transfer of the data. Participants were encouraged to write down their daily step-count, but PA activities in which they could not wear the pedometer, were not
taken into account. Participants, for example, who went swimming once a week, had been physically active, but this was not reflected in the diary.
However, not only the rather imprecise diary records might have caused inconsistencies, the pedometer itself caused some trouble. Participants’ feedback gave an insight into the problems with the pedometer. The tool was too small and too difficult to use, especially for the older participants. Additionally, because there was no individual support provided locally, due to the geographic distance of the researcher to the participants, everybody had to manage the set-up of the pedometer themselves. This might have led to differences in settings, which could have caused inconsistencies for the comparisons.
As mentioned above, this research was meant to be a pilot study for future research as it is new in its approach in comparing the effectiveness of affective and planning supportive appeals for closing the intention-behaviour gap. Future research should build on these findings, given that one can observe some tendency in a successful direction in this study. It would benefit from a bigger sample with a greater span in educational and environmental background to have more differences in the population. Additionally, a longer period of time is suggested, to assess whether an intervention would work even after a
familiarization phase and then people have forgotten about the pedometer in their pockets and the self-monitoring via pedometer won’t contribute to the intervention. Participants were encouraged to walk more during the first days, since they were more ambitious at the beginning to please the researcher. Participants’ feedback indicated that people were checking their pedometer less towards the end of the study.
A longer familiarization phase without sending any messages is also recommended, to have a solid comparison between a time period without a message and a time period with a message.
To avoid inconsistency a digital transmission of the diary records could be an advantage for every study, as self-reporting is not as accurate as an objective digital storage.
This study focused on differences between peoples’ PA within a four week period. It only assessed the differences in PA in terms of the two conditions PMG and AMG. However, the effects of messages direct on a single person’s behaviour could not be examined by the proposed design. There are other studies that made use of multi-level modelling to assess between-person but also within-person changes over time (Sniehotta, Presseau, Hobbs, & Araújo-Soares, 2012; Conroy, Maher, Elavsky, Hyde & Doerksen, 2013). Using such intra-individual modelling provides knowledge about how a single person responds to a treatment like receiving motivational messages. Conroy, Maher, Elavsky, Hyde, and Doerksen (2013) showed in their study that between-person variation only reflects the surface of behaviour change whereas it is possible with intra-individual modelling to gain insights into within-person variation.
Therefore, it is recommended for a follow-up study to use within-person analysis. It would provide more accurate and detailed insight into the direct effects of messages on an
individual’s behaviour.
Implications for future studies
Besides the suggestion for improvements on the current pilot study, future studies should focus on more sample specific messages to reach the target group. The study did not include a pre–test to measure potential barriers of the targeted sample. The messages were designed based on previous research and the content was rather general than tailored to the sample. In this context, the advice of psychologically educated motivation experts could improve the impact of the messages. Especially for samples within a clinical set up, for example, it is crucial to know what the barriers might be that prevent patients from cooperating. If text messages do not reach the patient, no effect will be achieved.
Feedback from the current sample provided insights, that additional text messages containing peer comparisons and information about their rank among the sample would have motivated them to be more physically active. This would be a good approach for future studies when designing intervention messages, but it has to be seen with caution as it could not be applied to every sample composition.
Text messages have been used as intervention methods for changing behaviour in various studies. Therefore, it would be also interesting to compare support via human voice and support via text messages. Since text messages are considered to be more neutral and impersonal, so that sympathy and antipathy to a real but unknown person is excluded (Peissner, Biesterfeldt & Heidmann, 2004), a personalised human voice might encourage participants to be more active. Participants might feel more supported by a real person and more connected to real life than through a neutral text-message.
To sum it up, the aim of this study was to assess whether communicative support via text messages could help to bridge the intention-behaviour gap by targeting post-intentional factors. The pilot study did not yield any significant impact, but a tendency towards positive change in increasing physical activity could be observed. Future studies should build on it to drive the findings forward. As soon as those kinds of interventions have been proved successful, a roll-out to a broad population is highly recommended. In either case, people at high risk due to a metabolic disorder or disease could benefit from the type of communicative support as described in this study. Another benefit would of course be the cut in Public Health spending.
References
Adipositas-Gesellschaft.de. (2015). DAG - Deutsche Adipositas Gesellschaft: Startseite. Retrieved 10 January 2015, from http://www.adipositas-gesellschaft.de
Ajzen, I. (1991). The theory of planned behaviour. Organizational Behaviour And Human Decision Processes, 50(2), 179-211. doi:10.1016/0749-5978(91)90020-t
Ajzen, I. (2001). NATURE AND OPERATION OF A TTITUDES. Annu. Rev. Psychol., 52(1), 27-58. doi:10.1146/annurev.psych.52.1.27
Armitage, C., & Conner, M. (2001). Efficacy of the Theory of Planned Behaviour: A meta-analytic review. British Journal Of Social Psychology, 40(4), 471-499.
doi:10.1348/014466601164939
Bandura, A. (1998). Health promotion from the perspective of social cognitive theory. Psychology & Health, 13(4), 623-649. doi:10.1080/08870449808407422
Bauman, A., Reis, R., Sallis, J., Wells, J., Loos, R., & Martin, B. (2012). Correlates of physical activity: why are some people physically active and others not?. The Lancet, 380(9838), 258-271. doi:10.1016/s0140-6736(12)60735-1
Bartholomew, L. K., Parcel, G. S., Kok, G., Gottlieb, N. H., & Fernandez, M. E. (2011). Planning health promotion programs: an intervention mapping approach. John Wiley & Sons.
Bélanger-Gravel, A., Godin, G., & Amireault, S. (2013). A meta-analytic review of the effect of implementation intentions on physical activity. Health Psychology Review, 7(1), 23-54. doi:10.1080/17437199.2011.560095
Beurer.com,. (2015). AS 50. Retrieved 26 January 2015, from
Blanchard, C., McGannon, K., Spence, J., Rhodes, R., Nehl, E., Baker, F., & Bostwick, J. (2005). Social ecological correlates of physical activity in normal weight, overweight, and obese individuals. Int J Obes Relat Metab Disord, 29(6), 720-726.
doi:10.1038/sj.ijo.0802927
Bravata, D., Smith-Spangler, C., Sundaram, V., Gienger, A., Lin, N., & Lewis, R. et al. (2007). Using Pedometers to Increase Physical Activity and Improve Health. JAMA, 298(19), 2296. doi:10.1001/jama.298.19.2296
Cawley, J. (2004). An economic framework for understanding physical activity and eating behaviours. American Journal of Preventive Medicine, 27(3), 117-125.
doi:10.1016/j.amepre.2004.06.012
Carrera-Bastos, P., Fontes, O'Keefe, Lindeberg, & Cordain,. (2011). The western diet and lifestyle and diseases of civilization. RRCC, 15. doi:10.2147/rrcc.s16919
Chatzisarantis, N. L., & Hagger, M. S. (2005). Effects of a brief intervention based on the theory of planned behaviour on leisure-time physical activity participation. Journal of Sport and Exercise Psychology, 27(4), 470-487
Colberg, S., Sigal, R., Fernhall, B., Regensteiner, J., Blissmer, B., & Rubin, R. et al. (2010). Exercise and Type 2 Diabetes: The American College of Sports Medicine and the American Diabetes Association: joint position statement. Diabetes Care, 33(12), e147-e167. doi:10.2337/dc10-9990
Conner, M., & Sparks, P. (2005). Theory of planned behaviour and health behaviour. In M.Conner & P. Norman (Eds.), Predicting health behaviour: Research and practice with social cognition models (2nd ed., pp. 170-222). Maidenhead, UK: Open University Press.
Conner, M., Rhodes, R., Morris, B., McEachan, R., & Lawton, R. (2011). Changing exercise through targeting affective or cognitive attitudes. Psychology & Health, 26(2), 133-149. doi:10.1080/08870446.2011.531570
Consolvo, S., Everitt, K., Smith, I., & Landay, J. (2006). Design requirements for
technologies that encourage physical activity. Proceedings Of The SIGCHI Conference On Human Factors In Computing Systems - CHI '06. doi:10.1145/1124772.1124840 Conroy, D. E., Maher, J. P., Elavsky, S., Hyde, A. L., & Doerksen, S. E. (2013). Sedentary
behavior as a daily process regulated by habits and intentions. Health Psychology. doi:10.1037/a003162
de Bruijn, G., Rhodes, R., & van Osch, L. (2011). Does action planning moderate the intention-habit interaction in the exercise domain? A three-way interaction analysis investigation. J Behav Med, 35(5), 509-519. doi:10.1007/s10865-011-9380-2 Destatis.de. (2014). Press releases - One in two adults in Germany is overweight - Federal
Statistical Office (Destatis). Retrieved 5 November 2014, from
https://www.destatis.de/EN/PressServices/Press/pr/2014/11/PE14_386_239.html Diabetesstiftung.de,. (2015). DiabetesStiftung DDS: Broschüren. Retrieved 28 January 2015,
from http://diabetesstiftung.de/broschueren.html
Dishman, R. K., & Buckworth, J. A. N. E. T. (1996). Increasing physical activity: a quantitative synthesis. Medicine and science in sports and exercise, 28(6), 706-719 Dishman, R., Oldenburg, B., O’Neal, H., & Shephard, R. (1998). Worksite physical activity
interventions. American Journal Of Preventive Medicine, 15(4), 344-361. doi:10.1016/s0749-3797(98)00077-4
Duncan, M. J., Spence, J. C., & Mummery, W. K. (2005). Perceived environment and physical activity: a meta-analysis of selected environmental characteristics. International Journal of Behavioral Nutrition and Physical Activity, 2(1), 11.
Gollwitzer, P. M. (1999). Implementation intentions: strong effects of simple plans. American Psychologist, 54(7), 493.
Gollwitzer, P., & Sheeran, P. (2006). Implementation Intentions and Goal Achievement: A Meta‐analysis of Effects and Processes. Advances In Experimental Social Psychology, 69-119. doi:10.1016/s0065-2601(06)38002-1
Haddock, G., Maio, G., Arnold, K., & Huskinson, T. (2008). Should Persuasion Be Affective or Cognitive? The Moderating Effects of Need for Affect and Need for Cognition. Personality And Social Psychology Bulletin, 34(6), 769-778.
doi:10.1177/0146167208314871
Hagger, M. S., Chatzisarantis, N. L., & Biddle, S. J. (2002). A meta-analytic review of the theories of reasoned action and planned behaviour in physical activity: Predictive validity and the contribution of additional variables. Journal of sport & exercise psychology.
Hagger, M., & Luszczynska, A. (2013). Implementation Intention and Action Planning Interventions in Health Contexts: State of the Research and Proposals for the Way Forward. Applied Psychology: Health And Well-Being, 6(1), 1-47.
doi:10.1111/aphw.12017
Hallal, P., Andersen, L., Bull, F., Guthold, R., Haskell, W., & Ekelund, U. (2012). Global physical activity levels: surveillance progress, pitfalls, and prospects. The Lancet, 380(9838), 247-257. doi:10.1016/s0140-6736(12)60646-1
Hardeman, W., Kinmonth, A., Michie, S., & Sutton, S. (2009). Impact of a physical activity intervention program on cognitive predictors of behaviour among adults at risk of Type 2 diabetes (ProActive randomised controlled trial). Int J Behav Nutr Phys Act, 6(1), 16. doi:10.1186/1479-5868-6-16
Jones, L., Sinclair, R., Rhodes, R., & Courneya, K. (2004). Promoting exercise behaviour: An integration of persuasion theories and the theory of planned behaviour. British Journal Of Health Psychology, 9(4), 505-521. doi:10.1348/1359107042304605
King, W. C., Brach, J. S., Belle, S., Killingsworth, R., Fenton, M., & Kriska, A. M. (2003). The relationship between convenience of destinations and walking levels in older women. American Journal of Health Promotion, 18(1), 74–82.
Kendzierski, D., & DeCarlo, K. J. (1991). Physical Activity Enjoyment Scale: Two validation studies. Journal of Sport & Exercise Psychology.
Keer, M., van den Putte, B., & Neijens, P. (2010). The role of affect and cognition in health decision making. British Journal Of Social Psychology, 49(1), 143-153.
doi:10.1348/014466609x425337
Keer, M., Conner, M., van den Putte, B., & Neijens, P. (2013). The temporal stability and predictive validity of affect-based and cognition-based intentions. Br. J. Soc. Psychol., 53(2), 315-327. doi:10.1111/bjso.12034
Keer, M., van den Putte, B., Neijens, P., & de Wit, J. (2013). The influence of affective and cognitive arguments on message judgement and attitude change: The moderating effects of meta-bases and structural bases. Psychology & Health, 28(8), 895-908.
doi:10.1080/08870446.2013.764428
Kiviniemi, M., Voss-Humke, A., & Seifert, A. (2007). How do i feel about the behavior? The interplay of affective associations with behaviors and cognitive beliefs as influences on
