Mood Induction Procedure

Those allocated to the negative mood condition were informed prior to taking part that during the lab session they would be asked to write about a negative memory of a recent personal event in their life in which they felt sad. They were told it may help for them to have an idea of what they would write about beforehand, following similar procedures by other researchers (e.g. Hernandez et al., 2003; Liotti, Mayberg, Brannan, McGinnis, Jerabek, & Fox, 2000; Liotti, Mayberg, McGinnis, Brannan, & Jerabek, 2002; Mayberg, Liotti, Brannan, McGinnis, Mahurin, Jerabek et al., 1999). Those allocated to the neutral mood condition were told they would be asked to write about a neutral memory (i.e. a memory in which they were in a non-emotional mood) such as making a routine journey or carrying out a daily routine. They too were asked to have an idea of something they could write about beforehand.

During the mood induction participants in the negative mood condition were provided with a booklet and pen and were given the following instructions: “Please think about one or more recent unhappy memories of a personal event in your life in which you felt sad. Please describe the event in detail and your reactions to it in the box below. Try to concentrate on your feelings at the time associated with the event/s. If you

can’t write for five minutes about one memory then please think of and write about another memory which has the same emotional significance.” Participants in the neutral condition were given the following instructions: “Please think about one or more memories of ordinary recent events in which you were in a neutral non- emotional mood e.g. taking a routine journey or carrying out a daily routine. Please describe the event/s in detail in the box below. If you can’t write for five minutes about one memory then please think of and write about another emotionally neutral memory.”

Whilst writing down their memories, participants in the negative condition were played five minutes of Barber’s Adagio for Strings via headphones. This piece of music has been found to be successful at inducing a sad mood in a number of previous studies (e.g. Eich & Metcalfe, 1989; Fox, Knight, & Zelinski 1998; Heene et al., 2007; Hernandez et al., 2003; Morrow & Nolen-Hoeksema, 1990). Participants in the neutral condition were alternatively played five minutes of music consisting of Chopin’s Waltzes numbers 11 and 12 played consecutively via headphones. These pieces of music have been used successfully in neutral mood conditions in a number of mood induction studies (e.g. Heene et al., 2007; Marzillier & Davey, 2005; Startup & Davey, 2001; Wood, Saltzberg, & Goldsamt 1990). Before and after the mood induction procedure, mood VAS and the PANAS were administered. This was to ensure mood induction was successful. Martin (1990) commented that the majority of studies employing a mood induction have taken a change of 10mm or more on a 100mm sadness VAS as indicative of successful mood change. Teasdale and Fogarty (1979) and Lenton and Martin (1991), however, used a 20 point difference as an indication of successful mood change. In the present study an increase in VAS rating of “sad” by 15mm or more and/or a decrease in VAS rating of “happy” was taken as a successful mood induction, based on the procedure followed by Richell and Anderson (2004). Eleven participants did not meet these criteria; therefore analyses were conducted with and without these participants.

In between each AB computer task, a booster mood induction was administered in which participants continued to listen to the same piece of music for three minutes. Music was also played for three minutes following the final task (a total time of 14 minutes of music). If participants in the negative mood condition were still in a sad mood in the final mood rating at the end of the lab session, they were played Delibe’s Mazurka from Coppelia (used as a positive mood inducer in a number of studies e.g. Bouhuys et al., 1994; Clark, Iversen, & Goodwin, 2001; Goldstein & Willner, 2002;

Parrott, 1991) whilst being asked to recall a recent happy memory. A final mood VAS was then administered to ensure mood had returned to baseline.

7.2.3. Procedure

Prior to the study, participants were emailed an information sheet and consent form, a health screen questionnaire (in order to check their eligibility to participate), the DEBQ and the BDI-II. Participants were allocated to either the negative mood induction or the neutral mood induction condition, and were allocated a time between 12pm and 5pm to take part in the lab session. Participants were asked to eat a normal meal (breakfast or lunch) 2-3 hours prior to the session and drink only water following this. At the start of the lab session, participants were asked to complete a food recall diary, baseline appetite and mood VAS and the PANAS. Following this, they carried out the mood induction procedure. After the mood induction further mood VAS and the PANAS were administered. After these mood ratings, participants completed the Stroop and dot probe tasks, the order of which was counterbalanced across participants. In between each task (and after the final task), a booster mood induction was also administered. Following this, mood VAS, PANAS, hunger, desire to eat and fullness ratings were taken. This was followed by measures of liking and wanting of crisps and chocolate and a ‘taste test’, as described in Chapter Eight. Finally, the experimenter administered mood and hunger ratings, followed by the EDI-2. Those in the negative mood condition who were still in a sad mood in the final mood rating underwent a positive mood induction. A final mood VAS was then administered to ensure mood had returned to baseline.

Height and weight measurements were taken (upon obtaining permission) and participants were provided with a written and verbal debrief. All participants in the negative mood condition were informed in the written debrief of the contact details of the University Counselling Service in the event of any significant distress following the negative memory recall. See Figure 7.1 for an outline of the procedure (intake- related data are presented in Chapter Eight as indicated in Figure 7.1).

7.2.4. Data Analysis

In order to check whether the data met parametric assumptions histograms, skewness and kurtosis statistics and Kolmogorov-Smirnov statistics were obtained and assessed. Some of the data were not normally distributed therefore, where

available, non-parametric equivalents of statistical tests were employed. In the Stroop and dot probe tasks, errors and outliers more than two SDs away from the mean RT for each individual were removed.

In order to test hypotheses one and two (that those with high restrained/emotional eating scores in a negative mood would display a significantly greater food AB than those in a neutral mood and with lower scores), separate analyses were conducted with the Stroop and dot probe data. A 2 (mood condition: negative/neutral) x 7 (word position) mixed measures ANOVA with restrained and emotional eating scores as covariates was conducted for each Stroop condition (food and neutral). Hypotheses were tested using the RT data, with a delayed colour-naming time for the target food word demonstrating an AB towards the word, and a delayed colour-naming time for the neutral word following the food word showing slowed disengagement from food (this is compared to RTs for later neutral words in the sequence). An α level of 0.05 was taken to be significant. The Stroop data were also analysed by conducting Spearman’s rho correlations for each group (negative/neutral), between RTs at word positions 1 and 2 in the Stroop task (representing orientation and disengagement) and restrained and emotional eating scores. This analysis on the Stroop data was repeated with 11 participants removed who were either not successfully induced into a negative mood following negative mood induction (n=7) or who were not maintained at a neutral mood following the neutral mood induction (n=4).5

Hypotheses one and two were also assessed using the dot probe data by conducting a 2 (mood condition: negative/neutral) x 2 (task condition: 200/2000ms) x 2 (congruence: incongruent/congruent trial) mixed measures ANOVA with restrained and emotional eating as covariates. Hypotheses were tested using the RT data, with quicker mean RTs for probes in the same location as food pictures (congruent trials) than mean RTs for probes in the same location as neutral pictures (incongruent trials) indicating an AB for food stimuli. An interference score was also calculated for each participant (incongruent trial minus congruent trial; positive value indicating AB towards the food stimuli and a negative value indicating attentional avoidance). The interference scores were analysed in a 2 (mood condition: negative/neutral) x 2 (task condition: 200/2000ms) ANOVA with restrained and emotional eating as covariates.

5 _{These participants were not removed at the start because of significantly reducing the sample size.}

Due to the complex analysis there was concern about losing power. However, it was also necessary to check whether these participants skewed the data at all.!

Email: informed consent, health screening, DEBQ, BDI-II

Allocated to negative condition (n=38). Asked to have sad

memory in mind

Allocated to neutral condition (n=39). Asked to have neutral

memory in mind

Lab session: food diary, baseline appetite and mood ratings

Mood ratings

Attention bias task 1 Sad mood induction: Barber’s

Adagio for Strings/ sad memory (5 minutes)

Neutral mood induction: Chopin’s Waltzes 11 &12/ neutral memory (5 minutes)

Barber’s Adagio for Strings (3 minutes)

Chopin’s Waltzes 11 &12 (3 minutes)

Attention bias task 2

Barber’s Adagio for Strings (3 minutes)

Chopin’s Waltzes 11 &12 (3 minutes)

Attention bias task 3

Barber’s Adagio for Strings (3 minutes)

Chopin’s Waltzes 11 &12 (3 minutes)

Hunger and mood ratings, liking and wanting of study foods

Taste test procedure

Hunger and mood ratings

EDI-2 Chapter 8 Chapter 8 Chapter 7 Chapter 7 Chapter 7

For each group (negative/neutral) Spearman’s rho correlations were conducted between restrained/emotional eating scores and the participant’s interference scores for each task condition (200/2000ms). This analysis on the dot probe task was again repeated with 11 participants removed who were either not successfully induced into a negative mood following negative mood induction (n=7) or who were not maintained at a neutral mood following the neutral mood induction (n=4).

The third hypothesis, that there would be significant associations between indicators of AB and other trait eating behaviours (external eating, DFT, bulimic symptoms, and BD), was analysed by conducting Spearman’s rho correlations between indicators of AB (Stroop word position 1 and 2 RTs; dot probe 200/2000ms interference scores) and eating behaviour questionnaire scores. Finally, Spearman’s rho correlations were also conducted between these indicators of AB and possible confounds (depression, age, BMI, and baseline hunger). As a priori hypotheses were not made for these analyses, α was adjusted for multiple comparisons: significant α set at .0125 to account for correlations for two groups at two Stroop word positions/two dot probe conditions.

Finally, multiple regression analyses were conducted with those eating behaviours/confounds (predictors) that were significantly correlated with indicators of AB (outcome). These were carried out in order to assess relative contribution of significant factors. Assumptions for stepwise regression were met: there was no multicollinearity as indicated by VIF (<10) and Tolerance (>.2) statistics, there was normality of errors as indicated by normal distribution in a histogram and through points falling close to the line in a normal probability plot, and there was homogeneity of variance of error terms (scatterplot of standardised residuals against standardised values were not skewed or curved: see Appendix 27).

7.3. Results

7.3.1. Restrained and Emotional Eating