Cocoa Powder Cocoa Butter Sugar Milk Powder Milk Fat
2. Exploring consumers beliefs and expectations: Materials and Methods
2.5 Labelling Study
The results of this study are described in Chapter 4.
2.5.1 Design
The study was a within subjects design: participants tasted both samples to rule out individual differences. The order in which the samples were tasted was counterbalanced for participants between the first and second session to allow for any order effects.
2.5.2 Participants
One hundred participants signed up, 91 attended the first session, and 87 completed both sessions (63 female, 24 male); all participants were untrained consumers. The mean participant age was 24.3 years (σ = 9.6), but participants ranged from 18 to 60 years. The mean BMI was 22.5 (σ = 3.9), which is classified as normal, but participants ranged from 16.6 (underweight) to 40.6 (obesity class III). Of the 87 participants, 63 were British, and 65 said that English was their native language. 64%
of participants said that they eat chocolate three times a week or more. 78% of participants said that they were very likely, or quite likely to try a lower-fat version of a food that they have tasted before.
2.5.3 Apparatus
Plain white ceramic plates (19 cm diameter) were used to present the chocolate to the participant. Labels, printed in black ink on white laminated paper, were displayed on the plate next to the chocolate. The chocolates were labelled either ‘Reduced-fat Milk Chocolate’ or ‘Milk Chocolate’. The labels also included a random 3-digit code to make it plausible that the label was for experimental use, rather than for the participant’s benefit (for example ‘Sample 276: Reduced-Fat Milk Chocolate’ and
‘Sample 530: Milk Chocolate’). As potentially there would be 100 participants, 20 different random numbers were used (10 for ‘Reduced-fat Milk Chocolate’ and 10 for
‘Milk Chocolate’), each participant received a different combination of numbers, to rule out any effects caused by the codes. The combinations were randomised between participants. The labels were printed (CMU Bright Roman, font size 28, centred, black ink) on plain white paper, cut to size (18cm x 6cm), and laminated.
On each testing occasion three squares of Somerfield own brand milk chocolate were given to each participant for expected liking, hedonic and sensory rating, and a further square for consumption questions. The squares of chocolate had a mean weight of 6.59g (σ = 0.57; the mean was taken from weight of all the squares in one 200g bar of the chocolate). Only full squares of chocolate were given; no half squares or broken pieces were presented. The chocolate was chosen as it did not have any branding on the surface. The chocolate was broken into squares prior to the session, and was served at room temperature.
2.5.4 Procedure
Prior to taking part in the study, potential participants were required to fill out a screening questionnaire (see Appendix 5) to assess their suitability for the study. It included questions on diet and food allergies, which was used to screen out those people who could not take part (for example, it could not be guaranteed that the chocolate did not contain traces of nuts or other allergens). The questionnaire also included questions related to chocolate consumption (‘How often do you eat chocolate?’, with the options ‘more than once a day’, ‘once a day’, ‘3 times a week’,
‘once a week’, ‘once a month’, ‘once every few months’ and ‘never’), and willingness to try low-fat foods (‘If given the opportunity, how likely is it that you would try a lower fat version of a food you have tasted before?’ with the options ‘very likely’,
‘quite likely’, ‘neither likely nor unlikely’, ‘not likely’ and ‘very unlikely’). These questions were presented amongst other consumption questions, in order to disguise the aims of the study. Participants were asked to attend two sessions four weeks apart (on the same day of the week and at the same time of day). The chocolates were tasted on two separate occasions, exactly four weeks apart to ensure that female participants were tested at the same point in their menstrual cycle on both occasions (assuming that the average cycle is 28 days long). Chocolate cravings occur more frequently perimenstrually, especially for women in the UK, Canada and America (Zellner, Garriga-Trillo, Centeno & Wadsworth, 2004), so it was important that female participants tasted the two chocolates at the same time in their menstrual cycle. Male participants were subjected to this too, to ensure that the method remained constant. It was also important that the participants tasted the two
chocolates on the same day of the week and at the same time of day in both sessions (all sessions took place between 2 and 5pm from Monday until Friday) (Kramer, Rock & Engell, 1992; Stroebele & De Castro, 2004). Participants were asked to refrain from eating and smoking for 30 minutes prior to each of the sessions. On both occasions participants took part in the experiment individually, in a small quiet room, with one experimenter present. Two locations were used for the experiment, although each participant attended the same location for their first and second sessions. The temperature of the room was taken at the beginning of each session; the mean temperature of the rooms during the first sessions was 23.8°C (σ = 1.23), and during the second session was 23.5°C (σ = 1.82). The temperature of the rooms was therefore well below the melting point of the chocolate, and similar on both occasions. There were slight differences between the two locations, such as the size and lighting of the room, but as participants attended sessions in the same location, effects are thought to be limited.
During the first session, the nature of the experiment was explained to participants orally (without specifying the aim or labelling system); a script was used to ensure that instructions were given in the same manner to each participant (see Appendix 6).
Written instructions were given, detailing what was expected of them, and how to fill out the questions that they would receive (see Appendix 7). Participants gave their written informed consent at the beginning of the first session. Participants were then required to eat one Matzo cracker (ingredients: wheat flour and water) and drink one cup of still, bottled spring water (Highland Spring), which was served at ambient
See Appendix 8 for the questionnaire used during the study. Expected liking (‘How much do you think you’ll like this chocolate?’) was rated on a visual analogue scale (VAS) (147mm) with the anchors ‘dislike extremely’ and ‘like extremely’.
Participants were presented with the sample of chocolate, and asked to rate the expected liking of the chocolate in front of them. Ratings were made by drawing a vertical mark along the horizontal line at the point that best represented their opinion.
Participants were asked to elaborate on their answer, and orally explain why they expected to like, or dislike, the product (answers were recorded by the experimenter).
The participant was then asked to consume one square of chocolate completely.
Participants were not instructed on how to eat the chocolate (i.e. to bite or not, suck or not), and if asked were told to eat it as they normally would. They were then asked to rate the chocolate for overall liking (‘How much do you like the chocolate overall?’), again on a VAS, with the anchors ‘Like not at all’ and ‘Like Extremely’. Again they were asked to orally explain their answer. The liking question was given before the sensory questions in order to prevent participants from becoming analytical in their rating of liking. The participant was then given a glossary of terms, which included a definition of each of the sensory terms they would be questioned on:
1. Melting in the mouth: the process of going from solid to liquid in the mouth;
2. Creamy / creaminess: having the consistency and texture of full fat cream;
3. Rich / richness: highly-flavoured; filling; may be overly sweet or heavy, possibly sickly or fatty;
4. Thick / thickness (mouthfeel): having a viscous consistency in the mouth when liquid; syrupy; the opposite of thin or watery;
5. Milky / milkiness: the taste / flavour of milk;
6. Smooth / smoothness: has a uniform, even texture; it isn’t gritty or grainy;
7. Sweet / sweetness: one of the basic tastes; associated with the taste of sugar.
Participants were asked to read all the definitions, and refer back to them if they needed to. The definitions were given to standardise the questions, and to ensure that all participants (those of whom English was their native language, and those with English as their second or third language) understood all the questions given to them.
The participants were then asked to work through the questionnaire at their own pace, and rate the chocolate according to a number of sensory attributes that are affected by the fat content of chocolate (melt-in-mouth characteristics and viscosity / thickness, (Guinard & Mazzucchelli, 1999) and those attributes described in focus groups as being important in chocolate (creaminess, milkiness, smoothness, sweetness and richness). Sensory attributes were accessed using both VAS questions (for example
‘How fast do you think the chocolate melts in the mouth?’ with the anchors “melts very slowly in the mouth” and “melts very quickly in the mouth”) and Just About Right (JAR) questions. JAR scales are used to measure the desirability of a specific attribute (the deviation from ideal), and the optimum levels of an attribute in a product (Lawless & Heymann, 1999). Scales typically consist of five or seven points ranging from too little to too much for a given attribute, with a mid-point of ‘just about right’.
The participant was required to determine the intensity of the attributes, and rate them according to its distance from the ideal level of the attribute. If the intensity is ideal, the consumer should rate the attribute as ‘just about right’, but if it is not ideal the
between the intensity and the ideal point (Rothman, 2007). Prompts were given, indicating when the next two squares of chocolate should be consumed.
The participants were then given a second plate, with one square of chocolate, labelled in the same way, and were asked firstly to indicate how many squares of this particular chocolate they think they would eat during a normal sitting, and secondly, how much they would consume if they were not going to restrict their intake. Finally, the participants were asked whether they would purchase the chocolate, and at what price (for a 50g bar). Typical prices for chocolate were given (budget milk chocolate:
14p; supermarket own brand milk chocolate: 28p; branded milk chocolate: 49p;
luxury milk chocolate: 89p), as a guide, although participants could give any price, and were not restricted to these categories. Participants were then asked to fill in a short questionnaire with details of time and date of last meal or snack time, and what they consumed, and if they are a smoker, when they last smoked (see Appendix 9).
None of the participants ate, or smoked in the 30 minutes prior to the session, as requested. The participants were reminded of the date and time of the second session (participants were also reminded via email nearer the time).
The second session had the same format, although the chocolate was labelled in the alternative way (i.e. if in the first session they received ‘Milk Chocolate’, in the second session they received ‘Reduced-fat Milk Chocolate’, and vice versa).
Following the second session, participants were given the same questionnaire containing questions about prior meals and snacks, and a questionnaire containing demographic questions (gender, current age, nationality, native language, current marital status, current occupation / area of study and educational status). Participants
were also given a chocolate consumption questionnaire (see Appendix 10). Sparks et al. (2001) developed a chocolate consumption questionnaire to assess both ambivalence and attitudes towards the consumption of chocolate, which includes measures of general attitude, semantic differential attitude, subjective norm, perceived behavioural control, health-conscious self-identity, intention, ambivalence and mixed feelings. Behavioural values were calculated using the methods described by Sparks et al. (2001). They were also asked to answer the DEBQ (van Strien et al., 1986).
Participants’ height and weight was measured, which was used to calculate BMI. This was all conducted after the experimental sessions, to avoid priming participants’
responses to foods. After the experiment participants were asked what they believed the purpose of the experiment was. Once the experiment was completed the participants were thanked for their participation, debriefed and compensated.
2.5.5 Data Analysis
A number of repeated measures ANOVAs (Analysis of Variance) were conducted to examine the main effects of labelling condition (‘Milk Chocolate’ or ‘Reduced-fat Milk Chocolate’) on expected liking, actual liking, sensory attribute ratings, consumption amount (normal and unrestricted consumption), purchase intent and price perception. Where appropriate, gender, age, BMI, normal chocolate consumption, willingness to try low-fat foods, restriction, external eating, emotional eating, and ambivalence were also used as between-subject variables in the analysis.
T-tests were used to examine where differences in the data lie. All these variables, except gender, willingness to try low-fat foods and normal chocolate consumption,
Groups varied slightly in size due to a large number of participants with the median value (for example participants with the same age as the median split value were placed in the same group, which meant that there were more participants in the below median age group than in the above median age group). Willingness to try low-fat foods was separated into two groups (likely and unlikely) from the original 5 groups.
Those who gave the answer ‘neither’ were excluded from analysis. Normal chocolate consumption was separated into two groups according to frequency of chocolate consumption (three times a week or more, and once a week or less).
T-tests compare the values on a continuous variable for two groups. ANOVA compares the mean scores of more than two groups; it also compares the variance between the different groups (assumed to be due to the independent variable) with the variability within each group (assumed to be due to chance i.e. error). The F ratio represents the variance between the groups, divided by the variance within the groups;
a large F ratio indicates more variability between the groups, than within groups.
There are a number of assumptions underlying the use of both t-tests and ANOVA’s:
• The dependent variable is measured using a continuous scales (interval or ratio data);
• The scores are obtained using a random sample of the population;
• Observations must be independent;
• The population are normally distributed (although large sample sizes can tolerate the violation of this assumption);
• Homogeneity of variance: samples are obtained from populations of equal variability.
Results were considered significant at p < .05, and trends towards significance were identified by p < .1. The results of the t-test are presented in the following way:
t(degrees of freedom) = t value, p = probability value. The results of the ANOVA are presented in the following way: F (hypothesis degrees of freedom, error degrees of freedom) = F ratio, p = probability value. Means and standard deviations are also presented.