Project Title

Effects of Model Facial Expression & BMI in Health Advertising on Consumer Intent to Achieve Health Goals: An Eye-Tracking Study

Names of Principal Applicant and Co-Applicants (please also include the names of the institutions and email addresses)
Dr Kerrie Bertele, University of Hertfordshire

Dr Ariadne Kapetanaki, University of Hertfordshire

Dr Paul Connell, Stoneybrook University

Introduction

Globally, an estimated 1.46 billion adults are overweight and 502 million are obese (Swinburn et al. 2011). People with excess weight are prone to getting type 2 diabetes, heart disease and certain cancers (Flegal at al. 2007). Obesity can also affect individuals’ self-esteem and mental health (Krishen and Worthen 2011). Social marketing campaigns that include mass media advertising to promote healthy eating are among the initiatives proposed by the World Health Organisation in their portfolio of actions to tackle obesity (WHO/Europe, 2008). In this vein, developed countries have allocated significant budgets to promote healthy eating behaviours through advertising. Despite the huge public spending on programs and campaigns to enhance healthy eating behaviors, obesity rates reflect citizen’s persistence in making unhealthy choices. The prevalence of obesity is continually increasing and despite public expenditure to tackle the problem “no national success stories have been reported in the past 33 years” (Ng et al. 2014).

Limited research examines the role of advertising promoting healthy eating. Effective antismoking advertising has received substantial attention, focusing on the message valence and intensity (Reardon et al. 2006), the use of graphic images (Andrews et al. 2014) and the effectiveness of message themes (Pechmann et al. 2003). The relationship between advertising and obesity has also been heavily studied, but the focus has been on the impact of food advertising on rising obesity levels (Beales and Kulick 2013), particularly of advertising targeted at children (Harker, Harker and Burns 2007). Some notable exceptions examining effective health campaigns include Kees, Burton and Tangari’s (2010) investigation of the role of temporal framing on consumer’s health goals. The aim of this research is to identify the optimal portrayal of models in health advertising in order to enhance health advertising effectiveness. Attentional avoidance is an effective defense against emotive, distressing or threatening stimuli, and reduces accessibility of the message (Hansen, Hansen and Shantz, 1992) and ad persuasiveness (Keller and Block 1997), therefore capturing consumer attention is an important goal of health advertisers.

This project involves two eye-tracking experiments: Study 1 considers the direct impact of model facial expression (Serious x Fake Smile x Duchenne Smile) on the consumer attention and intent to diet, and questions whether this is mediated by visual attention to the advertisement. The second study investigates the interaction of facial expression x BMI. Participant hunger, gender, their BMI and self-esteem are controlled. Self-esteem is often linked with body image satisfaction; lower self-esteem is related to higher levels of body image dissatisfaction (Krishen and Worthen 2011). Those with higher body image dissatisfaction are more likely to want to change their body, and may be more influenced by the experiment stimuli. The findings are aimed at informing effective advertisements to enhance consumer intent to eat healthily. If every adult reduces their BMI by just 1%, about 100,000 new cases of cancer would be prevented (National Cancer Institute, 2012). Small steps taken to address obesity can therefore have an important impact.

Report of Activities

Empirical Research completed:

To date two eye-tracking experiments have been carried out.

Study 1

The first study was a three-way between-subjects design investigating model facial expression (serious x fake smile x Duchenne smile) on consumer intent to achieve health goals and visual attention. This experiment was conducted with students from a UK university (Brown and Richardson 2012), with a minimum of 29 participants per condition (n = 89). This sample size was consistent with studies adopting a similar eye-tracking experimental design (Hollitt et al. 2010). Participants were given £5 compensation for taking part in the study.

This study involved a four step process:

First a photo-shoot was carried out with a model recruited to capture the required facial expressions. This was completed in the film studies department at the University of Hertfordshire by a professional photographer. Next the facial expressions were pre-tested via an online questionnaire (n=68). The facial expression manipulation was tested for valence of expression and genuineness of expression. Results are detailed in the method and results sections below. The public service advertisement stimuli were then created using photoshop. Finally the main eye-tracking experimental study was carried out.

Study 2

Study 2 was a 3 x 2 between-subjects factorial design investigating the impact of facial expression (serious x fake smile x Duchenne smile) x model BMI (low x high) on visual attention and intent to diet. This study was conducted with a combination of students at a UK university as well as participants at a local shopping mall, to enhance the heterogeneity of the sample (n = 151). There were a minimum of 25 participants in each of the six conditions, consistent with similar eye-tracking research studies (Hoover et al. 2010). Participants were compensated with either a £5 voucher to spend in the shopping mall (mall consumers) or £5 cash (University students).

This study involved two stages. First the same stimuli as in Study 2 were manipulated using photoshop to develop high model BMI/low model BMI versions of the advertisements. Next the main eye-tracking experimental study was carried out.

Data Analysis & Write up

The questionnaire and eye-tracking data was cleaned, analysed and written up. It was prepared for a journal article submission and conference presentation.

Journal Submission

This research was written up and submitted to the Journal of Advertising Special Issue: Effective Health Messages in Advertising. It was sent to reviewers and we are waiting on feedback.

Planned Activities for the Future

Data Collection:

We plan to carry out a third study based on feedback from our submission to the Journal of Advertising. We will either complete a third eye-tracking experimental study investigating the impact of model appearance on consumer health goals focusing on an alternative variable to Study 1 & Study 2. Alternatively we will carry out qualitative research in the form of interviews to get a deeper insight into the impact of models in public service advertisements on consumer intent to live healthily.

Conference Attendance:

We look forward to presenting our project at the British Academy of Management Conference in September 2015, and then going on to collect further data and spread our results to the consumer psychology research area in Europe (EMAC 2016) and North America (ACR 2016).

Were the Research Aims and Objectives met?

The overarching aim of this research was:

To identify optimal portrayal of models in health advertising in order to determine effective health advertising strategies and take a step towards tackling the global obesity crisis.

To this end, specific objectives include:

  1. Investigate the impact of facial expression (serious, smiling (forced), smiling (spontaneous)) in health food advertising on consumer affective responses and intentions to adopt a healthy lifestyle.
  2. Facial expression is known to impact perceived physical attractiveness, yet no study has examined the influence of facial expression on individual response to health marketing. This represents an important contribution.
  3. Identify the extent to which the impact of facial expression on consumer behaviour intentions is mediated by visual attention to elements of the ad.
  4. This project hypothesises that the persuasiveness of health food advertising portraying models will be mediated by visual attention to the advertisement
  5. Investigate the impact of model BMI on consumer affective responses and intentions to adopt a healthy lifestyle
  6. Identify the extent to which the impact of model BMI on consumer affective and behavioural responses is mediated by visual attention to elements of the ad.
  7. Determine the interaction effect of model BMI and facial expression on consumer affective and behavioural responses.
  8. Examine whether this interaction effect is mediated by visual attention to elements of the advertisement.
  9. Identify the moderating impact of relevant individual difference variables (e.g. BMI, self-esteem, self-control) on consumer response to different health advertisements.
  10. Explore the coping strategies (e.g. attentional avoidance, selective accessibility) used by individuals sharing certain individual difference variables (e.g. High BMI, Low self-esteem) when presented with distressing or threatening health advertisements (e.g. morbidly obese/extremely thin imagery)

The primary aim and objectives 1-7 have been met by Study 1 & Study 2.

Findings from the two experiments offer an important contribution in relation to illustrating optimal model presentation strategies in health advertising. The first study indicates that, for female consumers, a Duchenne smiling model is optimal. Specifically, alongside stimulating a higher intent to diet than the serious and fake expressions, a Duchenne smiling model leads to higher attention to the verbal message than a serious expression. Emotional contagion theory supports the role of portraying a Duchenne smile to enhance audience response (Hatfield, Cacioppo and Rapson 1994).

The results provide a tale of caution for advertisers who employ distressing and graphic images when warning against the consequences of obesity. Not only can distressing imagery lead to attentional avoidance (Brown and Richardson 2012), but it can fail to garner the positive consumer emotions spurred by smiling models which are in turn attributed to positive behaviour intent (Ekman, Davidson and Friesen 1990). While intent to diet is not affected by the model BMI, evidence from the second study overwhelmingly supports the role of model BMI in capturing visual attention to the advertisement. Higher model BMI is more effective in gaining visual attention when combined with a serious or Duchenne smiling model, and is found to be most effective when the target audience is overweight. This is consistent with social comparison theory, explaining how individuals compare themselves with models that are similar to them and avoid comparisons with heavier or thinner models as a means of self-enhancement (Festinger 1954). Given that public campaigns are attempting to change behaviours and reverse obesity levels, higher BMI individuals are likely to be an important target audience.

The preliminary evidence from these experiments suggests that advertisers should consider both the model facial expression and BMI when designing public health advertisements. The facial expression has an important significant impact on consumer intent to diet, and the model BMI has a strong influence on visual attention. Combining findings from the two studies, a Duchenne smiling model with a high BMI is optimal to both capture visual attention and enhance behavioral intentions.

Objective 8 [Explore the coping strategies (e.g. attentional avoidance, selective accessibility) used by individuals sharing certain individual difference variables (e.g. High BMI, Low self-esteem) when presented with distressing or threatening health advertisements (e.g. morbidly obese/extremely thin imagery)] had not been met. This will be achieved via a third study post the duration of the grant.

Limitations or challenges encountered

Methodological limitations:

The stimuli employed in the experiments include a female model only, which could have impacted how the male participants responded to the stimuli. It also prevented an analysis of the impact of advertisement-audience gender congruence on the impact of the model presentation strategies on attention and behaviour intent, which represents an interesting avenue for future research. Second, in an attempt to increase the external validity of the study, the health advertisement stimuli were placed among four real life advertisements; for cars, washing up liquid, and perfume. Post-experiment discussions with some participants revealed that some of them believed they were going to be questioned on the content of the advertisements, which may have led to increased focus on the branded advertisements, diluting the impact of our manipulated stimuli on their responses. Finally, intent to diet was measured with a two-item scale (Brown and Richardson 2012), alternative scales measuring intentions to behave in a manner consistent with the ad message (e.g. Chandran and Menon’s (2004) five item scale) may offer additional insight.

Sample recruitment challenge:

We proposed recruiting 120 students for Study 1, we were only able to recruit 90 participants (89 valid responses). Given the time required to complete this offline study it is a challenge to recruit participants. In Study 2 we proposed recruiting 120 participants. For this study we increased the number of experimental conditions from three to six conditions. Ideally we would have recruited a minimum of 180 participants; however we were only successful in recruiting 150 participants. We managed to meet the minimum acceptable sample size for eye-tracking experimental research (Hollitt et al. 2010). In order to increase our sample size we went on to collect a further 14 participants offering £10.

Personal challenges:

Two of the researchers (PI Dr Kerrie Bertele and co-investigator Dr Ariadne Kapetanaki) went on maternity leave during the course of the grant duration, leaving less time to complete the project than was forseen on application of the grant. This left only 9 months as opposed to 12 months to complete our objectives, which proved very challenging.

Analysis of Methods

The methodology adopted was a series of eye-tracking experiments. Two studies have been completed:

Study 1

Study 1 was a three-way between-subjects design investigating model facial expression (serious x fake smile x Duchenne smile) on consumer intent to achieve health goals and visual attention. This experiment was conducted with students from a UK university (Brown and Richardson 2012), with a minimum of 29 participants per condition (n = 89). This sample size is consistent with studies adopting a similar eye-tracking experimental design (Hollitt et al. 2010). Participants were given £5 compensation for taking part in the study.

Stimuli Development.

The first step was to develop three stimuli with the required facial expressions. A model was recruited and a professional photo-shoot was conducted. The model was paid £100 for participating in the study. The model was shown images of the facial expressions by the researchers and three photographs were selected from a sample of 100 photographs. Following the photo-shoot three advertisements were created and a verbal advertisement copy was added ‘Eat healthily to avoid obesity and live better’. Both positive (‘live better’) and negative (‘avoid obesity’) framing are included in this slogan to rule out the potential confounding impact of the valence of the information framing (Levin, Schneider and Gaeth 1998).

FIGURE 1

PHOTO SHOOT IMAGES

Pre-test.

To test the validity of the facial expression manipulation a questionnaire was conducted using an online consumer panel (61.8% male, 96% from the US, 88.2% third level educated). Participants were paid for their participation (n = 68). Each participant was randomly exposed to one of the three facial expressions followed by an identical questionnaire. The questionnaire included Izard’s (1977) differential emotions scale (DES), which proposes a set of ten basic emotions - interest, enjoyment, surprise, fear, sadness, anger, shame, guilt, disgust, and contempt - each captured with three distinct items. We are particularly interested in whether the model displays enjoyment (items: delighted/happy/joyful) or sadness (items: downhearted/sad/discouraged). It is expected there will be no significant difference in the other emotions and that they should be relatively low. The manipulation of the fake vs. Duchenne smile was tested by asking how genuine the smile of the model was perceived. This was measured on a seven-point Likert scale anchored by extremely fake/extremely genuine (Krumhuber and Kappas 2005).

Design and Procedure.

The experiment involved calibrating the participant’s pupils to capture the eye-tracking data. They were then allowed to process through the study at their own pace. To increase external validity the manipulated advertisements were placed randomly among four other neutral advertisements (e.g. perfume, washing powder, car). All of the advertisements were black and white to avoid the confounding impact of color (Gorn et al. 1997). Each participant was randomly assigned to one of the three experimental conditions. The five advertisements were followed by an identical questionnaire. A Tobii X2-30 eye-tracker was used to collected visual attention data including gaze duration (milliseconds) and fixation density for the whole ad and specific ad elements (Hollitt et al. 2010). Both duration of gaze -representing longer gaze duration achieved by refixations of the pupil (Russo and Leclerc 1994)- and fixation density - the number of pupil fixations on a given area of the advertisement which is conceptually equated to attention to that area of the stimulus (Feiereisen, Broderick, and Wong 2008)- are used to assess visual attention in this paper. The Tobii X2-30 eye-tracker is a small piece of hardware attached to the base of a monitor. It has a frequency of 30 Hertz, meaning the screen is refreshed 30 times per second. A 24” monitor is used to present the images, consistent with real-life magazine sizes. It captures highly accurate gaze-position data in real-life conditions – allowing participants to move their head freely and accurately in varying light conditions.

Study 2

The second study introduces the role of model BMI on consumer responses. Study 2 was a 3 x 2 between-subjects factorial design investigating the impact of facial expression (serious x fake smile x Duchenne smile) x model BMI (low x high) on visual attention and intent to diet. This study was conducted with a combination of students at a UK university as well as participants at a local shopping mall, to enhance the heterogeneity of the sample (n = 151). There were a minimum of 25 participants in each of the six conditions, consistent with similar eye-tracking research studies (Hoover et al. 2010). Participants are compensated with either a £5 voucher to spend in the shopping mall (mall consumers) or £5 cash (University students).

Stimuli Development.