7. Study Four Part One: Does negative mood interact with trait eating
7.2.2. Measures
7.2.2.2. State Measures of Mood and Appetite
7.2.2.2.1. The Positive and Negative Affect Scale (PANAS: Watson, Clark, & Tellegen, 1988)
The PANAS is a 20-item measure consisting of two distinct negative and positive affect scales. Positive affect refers to the extent to which a person feels enthusiastic, active and alert (with higher scores indicating greater positive affect), and negative affect refers to a state of distress which covers a range of negative mood states such as anger (with higher scores indicating greater negative affect: Watson et al., 1988). The PANAS requires participants to indicate the extent to which they have experienced specified mood states in a specified time frame. In the present study the time frame selected was how they were feeling at that precise moment, given that this questionnaire was used as a measure of change in state mood throughout the experiment. The PANAS has high internal consistency and reliability (e.g. Crawford & Henry, 2004; Watson et al., 1988). See Chapter Two Subsection 2.9.2 for further details on this measure. Cronbach’s alpha for the current sample was .84 at baseline measurement for the positive affect scale (comparable with reliability statistics obtained by Watson et al. 1988; .86-.90), and .63 for the negative affect scale (slightly lower than reported by Watson et al., 1988; .84-.97).
7.2.2.2.2. Visual Analogue Scales (VAS)
Mood was measured using sad, happy and anxious 100mm VAS with participants required to indicate how they were feeling on each mood state right now, ranging from ‘not at all’ to ‘extremely’. Mood VAS were given at various points throughout the study in order to assess the impact of the mood induction procedure, the tasks, and eating, on mood. Mood VAS have been used to assess the impact of mood induction, AB tasks and eating, on mood in a number of other research studies (e.g. Hepworth et al., 2010; Marzillier & Davey, 2005; Wallis & Hetherington, 2004). Hunger, fullness and desire to eat were also assessed using 100m VAS at three time points in the experiment (baseline, post mood induction and tasks, and post eating), as used for similar reasons in a number of other research studies (e.g. Hepworth et al., 2010; Wallis & Hetherington, 2004; Wallis & Hetherington, 2009). VAS measures of liking (subjective feelings of pleasure) and wanting (subjective feelings of intent or desire) for the study foods (chocolate and crisps) were also administered following the mood induction (as described in Chapter Eight). During the ‘taste test’, described in Chapter Eight, participants were also provided with VAS measures of the pleasantness, sweetness/saltiness of each of the study foods.
7.2.2.2.3. Food Recall Diary
Participants were instructed to have a normal meal (either breakfast or lunch depending on the time of the lab session between 12 and 5) two to three hours prior to attending the lab and drink only water following this. To assess adherence to this instruction participants completed a food recall diary at the start of the lab session. This required participants to record all food and drink that they had consumed that day, the time it was consumed and how much they had consumed. Such recall has been found to compare well with weighed records of intake (e.g. Bingham, Gill, Welch, Day, Cassidy, Khaw et al., 1994).
7.2.2.3. Stroop Task
7.2.2.3.1. Identification of Stroop Stimuli
Prior to the current study, 44 female students completed VAS ratings of the 12 food words from the Stroop task used in study two. The mean rating of the food words was 57.15 (SD=12.67) indicating that the words were rated as moderately appealing.
However, four words were rated below 50: ‘sugar’ (mean=48.89, SD=26.05), ‘pie’ (mean=41.55, SD=29.03), ‘cream’ (mean=32.77, SD=28.83) and ‘butter’ (mean=27.86, SD=25.02: see Appendix 6 for a full list of ratings). Therefore, these words (and their matched neutral words) were excluded from the current Stroop task. Ninety-seven females rated the neutral words from the task used in study two using Likert scales: 1) negative, 2) slightly negative, 3) neutral, 4) slightly positive, 5) positive (see Appendices VII and VIII for these ratings). Three of the matched neutral words in the food condition were rated as slightly positive and were therefore replaced: ‘companion’ was replaced by ‘container’; ‘bargain’ was replaced with ‘bracket’; and ‘premier’ was replaced with ‘parcels’. All household objects were rated as neutral, and only one of the matched neutral words (in the household object condition) was rated as positive (‘weekends’) and so this was replaced with the word ‘wrapping’. In order to have an equal number of stimuli in the two conditions, four household objects were excluded. Three household objects rated the furthest away from 3 (i.e. a completely neutral rating) were excluded (pillow, bath, photo) in addition to the word ‘sink’ given that this has multiple meanings and could therefore be considered negative. For a full list of words included in the present Stroop task see Appendix 5.
7.2.2.3.2. Stroop Procedure
Participants completed ‘Food’ and ‘Neutral’ (household object) conditions of the Stroop task, with the order counterbalanced across participants. The task began with 16 practice trials consisting of rows of ‘XXXXX’ in each of the four colours (red, blue, green and yellow). There were eight target and 48 neutral words in each of the conditions and each word was seen twice (224 trials in total). The pattern of words followed sequences of seven throughout, with the first in the sequence being the target word, followed by six neutral words. The words were presented individually in the centre of the screen. They were presented in a pseudo-randomised order; first the target words were randomised and then the individually matched neutral words for each target word were randomised. Participants responded via a response box with four colour buttons, as opposed to responding via keyboard as in studies one and two. Responding via a response box is recommended as the most accurate measure of reaction time in a Stroop task (e.g. Davidson & Wright, 2002).
In the previously employed Stroop tasks (studies one and two), no response stimulus interval (RSI) was used, i.e. each word was presented directly following the previous
response. This is in line with a number of previous food Stroop studies (e.g. Green, Corr & deSilva, 1999; Green, Elliman, Rogers & Welch, 1997: for a full description of the use of RSIs see Chapter Two Subsection 2.4.1). Sharma and McKenna (2001) compared the effect of different RSI durations on emotional Stroop performance, finding that emotional interference only occurred in a 32ms condition. The Stroop tasks in the present PhD are largely based on the original task developed by McKenna and Sharma (2004), which included an RSI of 32ms. Therefore, in the current Stroop task an RSI of 32ms was included, given that this is a brief RSI and will therefore allow lingering effects to be observed, but still allows participants a brief pause between words.
As in studies one and two of the present thesis, errors were removed in addition to response latencies above or below two standard deviations (SDs) away from the mean for each individual. A mean of 2.91% of responses were errors and were therefore removed (1.48% in the Food condition; 1.43% in the Neutral condition). A mean of 4.52% of responses were outliers and were therefore removed (2.5% in the Food condition; 2.02% in the Neutral condition).
7.2.2.4. Dot Probe Tasks
After study three, all food image ratings collected up to this point were combined to produce an overall rating of the 20 images used in the dot probe task in study three (see Appendix 15 for these ratings). Only two images were rated as unappealing overall: cheese on toast and cooked breakfast. However, the burger (mean=50.24) and the pancakes (mean=50.77) were rated as neither appealing nor unappealing. It was felt that in the previous use of the dot probe task (studies two and three) the task was too long and participants may have become bored or fatigued throughout. There were a large number of trials, as every image pair was seen eight times in each of the two conditions meaning that each participant completed 320 experimental trials (and 48 practice trials) taking approximately 15 minutes. Therefore, the following five images (rated the least appealing by a large sample of women) were excluded in the present task: burger, pancakes, cheese on toast, cooked breakfast and the doughnut.
As in studies two and three, two conditions of the dot probe task were presented to participants: one condition where all food/neutral picture pairs were presented for 200ms and another where all food/neutral picture pairs were presented for 2000ms.
In this task, 15 images were repeated eight times in each condition, with each picture appearing four times on the left and four times on the right, and four times in a congruent trial (the probe appearing in the same location as the food picture), and four times in an incongruent trial (the probe appearing in the same location as the neutral picture). Therefore, there were 120 experimental trials in each condition, with a total of 288 trials including practice trials. For a full description of the task procedure see Chapter Five Subsection 5.2.2.9. As in studies two and three, errors were removed in addition to responses below 200ms or above 2000ms. Additionally, response latencies above or below two SDs away from the mean for each individual were removed. In the 200ms condition a mean of 4.60% of responses were errors and were therefore removed (2.37% in congruent trials; 2.23% in incongruent trials). In the 200ms condition a mean of 4.01% of responses were outliers and were therefore removed (1.90% in congruent trials; 2.11% in incongruent trials). In the 2000ms condition a mean of 4.46% of responses were errors and were therefore removed (2.00% in congruent trials; 2.46% in incongruent trials), and a mean of 4.22% of responses were outliers and were therefore removed (2.08% in congruent trials; 2.14% in incongruent trials).