Treatment Delivery: Improving Psychotherapeutic Results

TREATMENT DELIVERY: IMPROVING PSYCHOTHERAPEUTIC RESULTS

BY INTEGRATING AFFECTIVE FORECASTING PRINCIPLES

EMILIA RUTH BROWN

A dissertation submitted in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

WASHINGTON STATE UNIVERSITY Department of Psychology

JULY 2016

To the Faculty of Washington State University:

The members of the Committee appointed to examine the dissertation of EMILIA RUTH

BROWN find it satisfactory and recommend that it be accepted.

______Paul Kwon, Ph.D., Chair

______Bruce R. Wright, M.D.

______Sarah L. Tragesser, Ph.D.

ACKNOWLEDGMENT

I would like to thank my advisor, Paul Kwon, Ph.D., for his mentorship and help on this project.

In addition, I would like to thank my committee that included Bruce R. Wright, M.D., and Sarah

L. Tragesser, Ph.D. for their time and energy. A special thanks goes out to Melissa Falkenstern

for her partnership in creating the overall research study in which my data was collected.

iii

TREATMENT DELIVERY: IMPROVING PSYCHOTHERAPEUTIC RESULTS

BY INTEGRATING AFFECTIVE FORECASTING PRINCIPLES

Abstract

by Emilia Ruth Brown, Ph.D. Washington State University July 2016

Chair: Paul Kwon

Research suggests that factors within the therapeutic environment such as the therapeutic alliance and client motivation are strongly tied to treatment success, yet the majority of psychotherapy research focuses on overall treatment modules and/or intervention tasks rather than the specific ways therapists might influence their clients through the way treatments are delivered. This research experiment integrates affective forecasting (AF) research to study whether the delivery of a psychotherapeutic intervention affects participants’ level of impact bias, treatment participation, and outcome. Analyses were conducted on data collected from a longitudinal online experiment completed Fall 2013 – Spring 2014 that randomly assigned participants to one of three conditions: delivery, delivery AF, or control. It was predicted that delivery condition and AF levels would affect participation and outcome variables. Analyses revealed that the manipulation was unsuccessful at creating a significant difference in AF levels across the intervention groups. Levels of AF were associated with length of time spent on the task at T-1 and college adjustment. Results had small effect sizes, suggesting that significance was related to the large sample size. Limitations of the study and future directions were discussed.

TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS ...... iii

ABSTRACT ...... iv-v

LIST OF TABLES ...... vi

CHAPTER

1. INTRODUCTION ...... 1-15

Affective Forecasting: Impact Bias ...... 4-8

Delivery of a Task: Adjusting for Impact Bias ...... 8-12

Intervention Task: Pennebaker’s Expressive Writing Task ...... 12-14

Present Study ...... 14-15

2. METHOD ...... 15-17

3. MEASURES ...... 17-19

4. RESULTS ...... 19-23

5. DISCUSSION ...... 23-26

REFERENCES ...... 27-35

APPENDIX

A. APPENDIX A: THREE EXPERIMENTAL CONDITIONS ...... 36-39

B. APPENDIX B: INTERVENTION AND CONTROL TASKS ...... 40-41

C. TABLES ...... 42-49

LIST OF TABLES

1. Table 1. Means and Standard Deviations of Variables ...... 42

2. Table 2a. Summary of Multiple Regression Analysis for Effects of AF and Delivery Group on Time Spent on Task at Time-1 ...... 43

3. Table 2b. Summary of Multiple Regression Analysis for Effects of AF and Delivery Group on Time Spent on Task at Time-2 ...... 44

4. Table 3. Correlations Between Outcome Variables at T-1 ...... 45

5. Table 4a. Summary of Multiple Regression Analysis for Effects of AF and Delivery Group on Change in Anxiety Symptoms ...... 46

6. Table 4b. Summary of Multiple Regression Analysis for Effects of AF and Delivery Group on Change in Depression Symptoms ...... 47

7. Table 4c. Summary of Multiple Regression Analysis for Effects of AF and Delivery Group on Change in Satisfaction with Life ...... 48

8. Table 4d. Summary of Multiple Regression Analysis for Effects of AF and Delivery Group on Change in College Adjustment ...... 49

Introduction

Use of stereotypically female and racial minority names in an email request to meet with professors resulted in biased responding from professors in more lucrative academic departments across the United States (Milkman, Akinola, & Chugh, 2012). When providing one’s name, including a middle initial resulted in being rated higher in status and receiving higher scores on writing evaluations (Van Tilburg & Igou, 2014). These headlining articles and others have made social cognitive research a hot topic in recent years due to the emphasis on how seemingly miniscule differences in the delivery of information can affect perception. This research highlights the malleability of our thoughts, emotions, and behaviors based on subconscious responses to the manner in which information is provided. Despite the variety of areas in which this information could be used, as of yet most of this research has remained in the general domains of social psychology and business.

Evidence has been accumulating that suggests the need for integrating this information into the domain of clinical psychology in order to ascertain the role of social cognitive factors in the psychotherapeutic process. Specifically, researchers have found little difference between the overall efficacy and effectiveness of the evidence-based treatments used today, often finding variability between therapists rather than treatment type (Luborsky et al., 2002; Kim, Wampold,

Bolt, 2006; Serlin, Wampold, & Levin, 2003). When differences have been found, many of these findings have been variable and appear to largely result from therapist allegiance, as the psychotherapy found to be statistically superior is typically the one most used and researched by those conducting the experiments (Elkin, 1999; Falkenström, Markowitz, Jonker, Philips, &

Holmqvist, 2013; Luborsky et al., 1999). However, evidence has been found for psychotherapeutic variability among therapists, despite trends to increase standardization (Elkin,

1999; Kim et al., 2006). In fact, superior therapeutic outcomes have even been found for a subset of therapists that is independent of theoretical orientation and treatment style (Okiishi,

Lambert, Nielsen, & Ogles, 2003).

Given this evidence, researchers have been exploring alternative components of therapy.

Despite the tendency to ignore the variability in therapists within intervention studies, research has revealed that therapists often should be considered a random factor in statistical analysis due to their variability (Elkin, 1999; Serlin et al., 2003). The magnitude of the therapist effect varies between studies; however, the average therapist effect in treatment outcome studies is believed to be between 5-10% (Crits-Christoph & Mintz, 1991; Wampold & Brown, 2005). One of the primary causes of this therapist effect may be between-therapist differences in ability to implement Rogerian techniques (Zuroff, Kelly, Leybman, Blatt, & Wampold, 2010) and develop the therapeutic alliance (Baldwin, Wampold, & Imel, 2007). However, in terms of explaining the actual differences in this ability, researchers of the therapist effect note that “a wide variety of characteristics could be pertinent, including personality traits, interpersonal styles, interpersonal skills, and preferred therapeutic strategies and interventions” (Zuroff et al., 2010, p.

693). In fact, for decades researchers and clinicians have noted the importance of social influence within the context of psychotherapy (Goldfried & Davila, 2005; Goldstein, 1966;

Strong, 1968).

In this paper, I argue for the further incorporation of social cognition concepts into research on psychotherapy in order to identify and better understand the variability between therapists. Specifically, I am suggesting that psychotherapy be viewed as a multi-component treatment that can be separated into the therapeutic intervention and the delivery of that intervention. Furthermore, I theorize that social cognitive concepts will be particularly important

in the delivery of interventions given the unique social environment that occurs between a therapist and a client. This way, researchers can begin to develop techniques for the general implementation of interventions that may reduce the variability in treatment outcomes across individual therapists and increase the overall efficacy of these treatments.

An example of this division between intervention and delivery can be seen with the completion of exposure therapy for a phobia. When introducing the idea of systematic desensitization, therapists can deliver information about this intervention to the client in a variety of ways. For example, one therapist may provide research findings on the efficacy of the intervention whereas another might focus on personal narratives of clients who have completed the intervention successfully or on identifying the client-specific advantages of completing the treatment in order to enhance motivation. These differences in the delivery of the exposure treatment could result in variations in client understanding of the treatment, motivation to participate in treatment, and treatment outcome. Identifying social cognitive concepts that impact the treatment process could help identify what components of the delivery of a treatment might be important in maximizing treatment success.

Researchers have provided evidence for the importance of the implementation of social cognition within psychotherapy. For example, an experiment by Alden, Mellings, and Laposa

(2004) revealed differences in the response of clients with generalized social phobia to a social exposure session depending upon how feedback about the exposure was delivered. Specifically, negatively framed feedback (e.g., “voice did not tremble”) resulted in lower anxiety about a future, similar session than positively framed feedback (e.g., “voice remained steady”). This experiment among others has shown the importance of Tversky and Kahneman’s framing postulate (1981) for medical and psychotherapeutic settings (Gallagher & Updegraff, 2012;

Meyerowitz & Chaiken, 1987; Trupp, Corwin, Ahijevych, & Nygren, 2011). In addition, multiple experiments have been completed focusing more specifically on cognitive dissonance and effort justification, the desire to explain the effort necessary to complete a task, within intervention settings. These experiments revealed the importance of high effort and high choice on treatment outcome for both weight management and phobia interventions (Axsom & Cooper,

1985; Cooper, 1980).

Despite this evidence, the transition to a more standardized look at the delivery of interventions has been slow to develop. Perhaps the closest that researchers and clinicians have come to this integration are microskills training, a training basis for developing skills such as empathy (Truax & Carkhuff, 1967), and Motivational Interviewing (MI), a guidebook of therapeutic techniques to enhance motivation (Miller & Rollnick, 2012). However, microskills training has limited research focusing on clinical populations and treatment outcome (Ridley,

Kelly, & Mollen, 2011) and MI focuses solely on motivation and has been criticized by some as being no more effective than treatment-as-usual (Butler et al., 2013; Smedslund et al., 2011).

Thus, further research is still necessary in order to help identify how to deliver therapeutic techniques in the most beneficial manner. In this paper, I explore how the delivery of an intervention may affect treatment motivation and outcome, particularly within the context of affective forecasting, a social cognitive concept related to anticipated emotional response.

Affective Forecasting: Impact Bias

Affective Forecasting (AF) is a person’s prediction of the emotional experience that will occur following a future event. Interestingly, there are certain components of AF that are more easily predicted than others. Participants in AF research often are able to accurately predict the valence and types of emotional experiences they will endure. In contrast, participants often are

unable to predict the intensity and duration of these emotions accurately. The overestimation of the intensity and duration of future emotions, the most common pattern of error in AF predictions, is referred to as impact bias and will be the focus of our experiment (Wilson &

Gilbert, 2003).

A common example of an AF condition used to measure impact bias would be the end of a romantic relationship (Eastwick, Finkel, Krishnamurti, & Loewenstein, 2008; Gilbert, Pinel,

Wilson, Blumberg, & Wheatley, 1998; Tomlinson, Carmichael, Reis, & Aron, 2010). When a person predicts how a break-up will feel, he or she is able to predict the types of future emotions accurately (e.g., positive versus negative, angry versus sad versus happy). Impact bias is seen when predicting how intense those emotions will be and how long the emotions will last.

Specifically, a person will predict that the emotions will be more intense and last longer than what is actually experienced, particularly when predicting negative future events (Gilbert et al.,

1998).

One of the reasons for the interest in AF is to be able to understand how predictions about a future event can affect emotions and behaviors. Woodzicka and LaFrance’s (2001) study on gender harassment revealed differences between actual and predicted emotions and behaviors when comparing a hypothetical versus an actual sexual harassing interviewer. 68% of the women in the hypothetical scenario predicted that they would refuse to respond to at least one of the sexually harassing questions and 28% predicted that they would take further action. In contrast, 100% of the women in the actual scenario answered the questions and 52% ignored the harassment altogether. One of the primary theories presented for this difference in predicted and actual behaviors were the differences in predicted emotions, as many women predicted that they

would feel angry whereas more women in the actual scenario felt afraid. This study emphasizes the importance of emotions in accurate prediction of future behaviors.

There is also more direct evidence for the link between AF and future mood and behavior. When participants in a study were asked to rank the likelihood that they would take a course based on the course description, their rankings of the level of predicted enjoyment were shown to be statistically important in the decision-making process (Falk, Dunn, & Norenzayan,

2010). A study by Totterdell and colleagues (1997) monitored mood every day for two weeks, asking participants to provide their current mood as well as a predicted mood for the upcoming day. This study found that future mood states were related to predicted mood. This relationship was independent of daily “hassles” experienced by the participants. This research provides evidence for how predicted emotions may relate to future behaviors and mold future emotional responses.

There also appears to be an interactive relationship between emotional states, predicted emotional responses, and future behaviors. When manipulating empathy gaps, participants in an experiment by Van Boven et al. (2012) had greater inaccuracy and were more likely to over- predict their willingness to participate in future embarrassing situations if they were in a neutral

(or “cold”) emotional state. In contrast, participants who were manipulated into feeling anger or fear (“hot” emotions) prior to their predictions were much better at estimating their willingness to participate in an embarrassing situation. Interestingly, these predictions also resulted in different levels of future willingness to participate in embarrassing situations. Although the neutral emotional state resulted in over-predictions about willingness to participate, this group also had a higher overall level of willingness to participate than the hot emotional state (Van

Boven, Loewenstein, Welch, & Dunning, 2012).

AF research has also begun to look at the effects of impact bias on interventions. For example, an experiment attempting to reduce impact bias in order to increase participant willingness to complete a colorectal examination showed that lower levels of predicted discomfort was associated with increased behavioral intention to receive the examination in the future (Dillard, Fagerlin, Dal Cin, Zikmund-Fisher, & Ubel, 2010). Additionally, researchers looking at the effects of AF on exercise found that finding ways to increase positive emotion predictions resulted in increased intent to exercise in the upcoming week (Ruby, Dunn, Perrino,

Gillis, & Viel, 2011). Although there is limited information on how AF would relate specifically to clinical psychology, research on factors such as anticipatory anxiety have also explored the realm of predicted emotions. Research in this area has shown that the avoidance seen in panic disorder with agoraphobia is predicted by the anticipated number of future panic attacks rather than the actual frequency that occurs (Cox, Swinson, & Kuch, 1991). Given this information, it is likely that impact bias can result in avoidant behaviors and affect the psychotherapeutic process.

In fact, there is evidence to suggest that impact bias may have a greater impact on those with symptoms of depression and anxiety, the main categories of disorders seen within psychotherapy. Specifically, higher levels of depressive and anxiety symptoms result in greater impact bias when predicting negative affect. These effects remain when controlling for the higher level of negative affect experienced by those with these symptoms. Interestingly, depressive symptoms are also associated with more realistic predictions of positive affect. In other words, in addition to over-predicting the negative intensity of future negative events, those with greater depressive symptoms are also less likely to over-predict the positive intensity of future positive events (Wenze, Gunthert, & German, 2012). Another study found that dysphoria,

lifetime hypomanic symptoms, and anxiety were all associated with this pattern of predicting greater negative affect and less positive affect; however, statistical analyses revealed that dysphoria was the primary underlying cause of this pattern (Hoerger, Quirk, Chapman, &

Duberstein, 2010). There is also evidence to suggest that particularly low levels of predicted positive affect is a distinguishing symptom of dysphoria in individuals with a past history of at least one suicide attempt (Marroquín, Nolen-Hoeksema, & Miranda, 2013). These studies show the importance of integrating AF research into theories of anxiety and depression, as varying levels of impact bias appear to be related to symptoms of psychopathology and may be a contributing factor to symptoms of anxiety and, particularly, depression (Marroquín et al., 2013;

Wenze et al., 2012).

Delivery of a Task: Adjusting for Impact Bias

In order to correct for impact bias, it is important to understand how these faulty predictions occur. Two components of AF have been identified as major factors in the development of impact bias, focalism and immune neglect. Focalism has been defined as the tendency to make predictions based solely on the event being questioned. Overestimations of the influence of the event, therefore, result from predictions that do not account for other events that will be co-occurring in the person’s life (Wilson, Wheatley, Meyers, Gilbert, & Axson, 2000).

In addition, there is evidence to suggest that this focalism is primarily a focus on the initial emotional response that results in a disregard for how the emotions will dissipate over time

(Eastwick et al., 2008). Immune neglect is a person’s lack of accounting for his or her subconscious ways of coping with events. There is a greater likelihood of overestimating the effects of a solitary event on a person’s life when he or she is unaware of the coping strategies and defense mechanisms available that make adaptation possible (Wilson & Gilbert, 2005). This

immune neglect also appears to relate to limited recollection of previous, similar experiences in order to predict future outcomes (Buehler & McFarland, 2001).

Wilson et al. (2000) completed an experiment that showed the effect of focalism on AF.

Undergraduate students were asked to complete a measure of AF related to a future football game, estimating how they would feel the week following the game if their team won or lost.

Prior to completing this measure, the students were randomly assigned to either a control group or a diary group. The diary group was asked to write about their daily activities during the upcoming week. Participants in both groups were also asked to estimate how often they believed they would think about the football game during the following week. After the football game, participants provided information on their actual level of happiness. When comparing predicted and actual happiness, results showed that participants in the control group had greater impact bias than participants in the diary group. In addition, this difference was mediated by the estimation of how often participants would think about, or “focus” on, the football game in the upcoming week. These results show the effects of focalism and how reducing the focus on the one event can improve AF accuracy.

An experiment by Gilbert and colleagues (1998) provides an example of the process of immune neglect. In this experiment, participants were presented with the possibility of obtaining a job. The participants were randomly assigned to one of two groups that varied on the level of fairness. In the low fairness group, participants were rejected by an individual who was poorly informed about the participants’ job capabilities. In contrast, the high fairness group participants were rejected by a group of people who were highly informed about the participants’ job capabilities. Prior to finding out if they had obtained the job, participants predicted how they would feel if they did not receive the job. The researchers found that participants predicted

statistically identical future emotions across the two groups; however, those in the low fairness group reported statistically higher levels of happiness than those in the high fairness group 10 minutes after not obtaining the job. The researchers explain that, when faced with a negative outcome, participants attempted to justify the outcome based on internal and external factors.

Those in the low fairness group were able to identify clear external factors that could have affected their ability to obtain the job. However, the participants in this experiment did not account for this coping strategy when making affective predictions. As a result, participants were unable to appropriately adjust their AF, resulting in greater impact bias for those in the low fairness group.

Based on these two components, researchers have identified ways of reducing impact bias. As noted above, Wilson et al. (2000) revealed how asking participants to list other activities that will be completed following the event in question can reduce focalism and improve

AF. Research has also revealed that asking participants to focus on past, similar experiences prior to making AF predictions results in lower levels of impact bias (Buehler & McFarland,

2001). Perhaps the most compelling evidence for how and why to reduce impact bias is the previously mentioned research completed by Dillard and colleagues (2010). Their experiment randomly assigned participants to one of two groups. Both groups were provided with educational information on colorectal examinations, an infamously uncomfortable preventative medical procedure. One group was also provided with a narrative written by someone who had gone through the experience. This narrative emphasized the person’s ability to cope with the examination and noted other experiences in the person’s life aside from the medical appointment as a way to reduce immune neglect and focalism, respectively. Results showed that participants

in the narrative group, after reading the informational material, reported fewer concerns about potential barriers to getting the procedure and higher levels of behavioral intent.

The information provided above suggests that integrating research on impact bias could help to improve the delivery of psychotherapeutic interventions. In fact, multiple psychotherapeutic techniques incorporate portions of this strategy. In cognitive and behavioral treatments for depression and anxiety, therapists often use monitoring in order to ascertain the frequency, intensity, and triggers of certain psychological symptoms (Beck & Beck, 2011).

Many researchers and clinicians have noted the tendency for those with depression and anxiety to focus on negative stimuli and perceive neutral stimuli in a negative way (Murphy, Sahakian,

Rubinsztein, Michael, Rogers et al., 1999; Schultz & Heimberg, 2008). This suggests that an important component of monitoring would be an emphasis on identifying and providing feedback on discrepancies between believed versus actual emotional responses to daily symptoms and treatment completion. In addition, including an emphasis on strategies used within research on impact bias could help reduce impact bias and avoidant behaviors and, in turn, improve overall treatment participation and response.

The present experiment will focus on how impact bias might affect the beginning stages of delivering a psychotherapeutic intervention. Specifically, I am interested in how the initial delivery of an intervention task can be manipulated to reduce focalism and immune neglect related to treatment concerns. I theorize that AF is particularly important in the beginning stages of an intervention, as impact bias related to beginning the treatment (e.g., fears of increased symptoms, fear of exposure to distressing stimuli) could result in avoidance and lowered participation. It is theorized that, like previous research in other settings has shown, reminding participants of their ability to cope with negative events and emphasizing the impact of other

events on future emotion will reduce impact bias related to beginning a psychotherapeutic intervention. This, in turn, will result in greater participation in treatment and better therapeutic outcomes, as decreased predictions of negative affect related to participation in the intervention should result in reduced avoidance of the intervention and reduced future negative affect.

Intervention Task: Pennebaker’s Expressive Writing Task

In order to test this theory, it is necessary to manipulate the delivery of a well-established treatment intervention in order to measure the variability in impact bias and how this variability might affect treatment success. Pennebaker’s expressive writing task was chosen due to its history of working as a short-term intervention for college student participants (Hoyt & Yeater,

2011; Pennebaker & Beall, 1986; Pennebaker, Colder, & Sharp, 1990; Pennebaker, Kiecolt-

Glaser, & Glaser, 1988). The expressive writing task asks participants to write about a past traumatic event with a focus on emotions and cognitions (see Appendix B). This task is usually completed over a period of 15-20 minutes and repeated up to 3 additional times (Pennebaker &

Beall, 1986); however, these effects can also occur after a single completion of the writing task

(Greenberg, Stone, & Wortman, 1996).

A plethora of research has been gathered on the expressive writing task and multiple articles are available that explore its overall effects (Frattaroli, 2006; Smyth & Pennebaker,

2008). Despite the short-term nature of the intervention, it has an average unweighted effect size of .075 (Cohen’s d = .151), with higher effects occurring for participants with a trauma history

(Frattaroli, 2006). The advantages of this intervention cover a wide range of areas, showing increased physiological and psychological health. Examples of specific effects of the intervention on health include reduced future doctor visits (Pennebaker & Beall, 1986), improvements in disease symptoms such as rheumatoid arthritis and breast cancer (Creswell et

al., 2007; Danoff-Burg, Agee, Romanoff, Kremer, & Strosberg, 2006), improvements in mood and management of trauma-related stressors (Smyth, Hockemeyer, & Tulloch, 2008), and reduced distress and depression (Frattaroli, 2006).

Although the theoretical understanding of this intervention remains at least partially unclear, there have been multiple experiments exploring why this intervention works. An analysis of the writing of participants in early expressive writing experiments revealed evidence for the importance of developing meaning and insight about the event (Pennebaker, 1993). In addition, a recent experiment with women diagnosed with breast cancer revealed that self- affirmative statements mediated the effects of emotional expression during the task on future physical symptoms (Creswell et al., 2007). Research has also shown the importance of enhancing the meaningfulness and integrative processing of the writing task, as these factors increase positive affect following the task (Schutte, Searle, Meade, & Dark, 2012). Although more information is necessary to pinpoint the theoretical underpinnings of the intervention, what is important for our experiment is its background of research in being an effective, short-term intervention.

The basic narratives used in this experiment for the treatment groups and the control group were adapted from Pennebaker (1997) and Pennebaker (n.d.). The two treatment groups had the same overall narrative with additional delivery components meant to manipulate AF levels that were adapted from Dillard et al. (2010). The treatment group that focused on reducing impact bias emphasized participants’ previous ability to cope with difficult events and the participants’ ability to quickly return to everyday life. This narrative also included an analogy from the Cognitive Processing Therapy manual (Resick, Monson, & Chard, 2008) in order to further emphasize the temporary nature of emotional release. Later in this paper, this

group is referred to as Delivery AF. In contrast, the other treatment narrative included an emphasis on the negative feelings that participants tend to feel after completing the task and suggested that participants could discontinue the task if they felt emotionally overwhelmed. This was done to minimize participant focus on their ability to cope with the treatment intervention while also attempting to remain consistent with a narrative that may be feasibly delivered by clinicians within a psychotherapeutic context. This group is referred to as the Delivery group throughout this paper. These narratives can be found in Appendix A.

Present Study

The present study will examine this theoretical relationship between treatment delivery and treatment participation and outcome over a one week time period. More specifically, it will test the theory that treatment delivery with a focus on reducing impact bias will result in greater treatment success. Data was collected from an experiment that randomly assigned participants to one of three conditions: one of two intervention groups or a control group. The intervention groups only varied in the initial delivery of the intervention (i.e., the description of the intervention provided prior to the instructions) as discussed in the previous section on

Pennebaker’s expressive writing task. This variability was meant to manipulate focalism and immune neglect in order to affect levels of impact bias (see Appendix A). The experiment measured multiple components of psychopathology, participation, and actual versus predicted affect in order to measure AF and psychotherapeutic success.

It is believed that lower levels of impact bias will result in spending more time on the intervention task and greater likelihood that participants will complete the task on two different occasions. It is also hypothesized that this difference will result in greater improvements in psychotherapeutic outcomes. Additionally, it is hypothesized that the manipulations in the

delivery of the two treatment groups will result in statistically significant differences in the level of AF for each group. I predict that the interaction between intervention group and level of AF will relate to changes in psychotherapeutic outcomes. Specifically, I hypothesize that being assigned to the intervention group emphasizing a reduction in impact bias in addition to reporting lower levels of AF will result in better psychotherapeutic outcomes. In contrast, it is hypothesized that the combination of being assigned to the intervention group minimizing coping strategies in addition to reporting higher levels of AF will result in worse psychotherapeutic outcomes.

Method

Participants

Participants consisted of 718 introductory psychology undergraduate students at a large, public university in the Pacific Northwest that completed an online experiment in Fall 2013-

Spring 2014. Students received course credit for their participation. 74.0% of the sample identified as female, 24.5% as male, and .1% as in transition. 1.4% chose not to respond. The age of participants ranged from 18 to 55 with a mean age of 20. The majority of participants identified as Caucasian (60.6%) with the most common other racial/ethnic identifications being

Multiracial (8.2%), Hispanic (7.7%), Asian/Asian American (6.4%), and African/African

American (3.9%). 1.1% identified as Pacific Islander, 0.7% as “Other,” 0.6% as Native, and

0.1% as Middle Eastern. 0.7% of participants chose not to respond to this question. All participants considered themselves proficient in the English language. 271 participants completed Time 2 of the study. An ANOVA found significant differences in the T-1 levels of T-

1 Length, T-1 SWLS, and T-1 BDI between those who completed T-2 and those who did not.

Specifically, those who completed T-2 of the study tended to spend longer on the T-1 writing

task and had lower levels of T-1 SWLS and T-1 BDI in comparison to those who did not complete T-2. No other variables explored in the study were significantly different between the two groups.

This data was part of a larger project consisting of multiple online studies. The participants used in our analyses self-selected into the experiment by either electing to do a

“cognitive task” or a “writing intervention.” Both groups of students completed the same measures and writing intervention. 14.5% of the sample selected into the study through the

“writing intervention” and 85.5% through the “cognitive task.” No differences in outcomes were found when only analyzing participants who self-selected into the writing intervention.

Therefore, both groups of participants were used in my analyses in order to improve power.

Procedure

Time 1 (T-1): Participants completed a series of questionnaires including measures of psychological functioning (current affect, depressive and anxiety symptoms, life satisfaction, adjustment). Participants were then randomly assigned to one of three delivery conditions:

Delivery, Delivery AF, and Control. All three conditions introduced the upcoming task

(Appendix A) and then measured predicted future affect (immediate and one week later).

Participants completed the task (Appendix B) and affect was measured a second time. At the end of the experiment, participants were provided with information about the importance of repeating the task and were then asked to complete the task multiple times over the next week.

Participants then provided an email address in order to receive the second portion of the experiment.

Time 2 (T-2): One week after T-1, participants were emailed a link to T-2 of the experiment. They were asked to follow the link and complete the experiment within 48 hours.

Participants were asked about the frequency of times they completed the task over the past week and the average time spent on the task. Next they completed a series of questionnaires (current affect, depressive and anxiety symptoms, life satisfaction, adjustment) and then completed their randomly assigned task condition a second time.

Measures

Affective Forecasting:

Affect: Current, Predicted. Items used in this measurement were obtained from Buehler

& McFarland (2001) and Wenze et al. (2012), articles that adapted the Positive and Negative

Affect Schedule-Expanded Form (PANAS-X; Watson & Clark, 1994). Participants were asked to rate their current and predicted level of the following positive and negative emotions: sad, happy, angry, nervous, enthusiastic, jittery, hostile, excited, lonely, and content. (The specific wording was as follows: “Please rate how you [currently feel / expect to feel directly after completing this writing task / expect to feel one week after completing this writing task] in terms of the following emotions…”) It was rated using a 7-point Likert scale from “Not at All” (1) to

“A Lot” (7). Consistent with past research, the positive and negative affect scores were moderately to highly intercorrelated with Cronbach’s alphas ranging from .66 to .83. Thus, composite affect variables were calculated using an average of the positive affect scores and the reverse-coded negative affect scores. Affect was calculated for baseline affect (AF-Initial), predicted affect immediately following the task (AF-PredImmed) and one week after the task

(AF-Pred1Wk), affect following the task (AF-PostTask), and affect one week after the task (AF-

1Wk).

Participation in Treatment:

Participation was measured in multiple ways. A simplistic measurement of participation was created by coding participants for whether they completed the prompt at T-2 of the study or not (T-2 Completion). In addition, participation was measured by the length of time participants took to complete the prompt both at T-1 and T-2 (T-1 Length and T-2 Length). At T-2, participants were asked how many times they completed the task (WkFrequency) and the average length of time that was spent completing the task over the past week. (Participants were notified that their responses to the T-2 questions would not affect their participation credits.) The

WkFrequency and the weekly average duration spent on the task were multiplied in order to calculate the WkDuration.

Treatment Outcome Variables:

Anxiety. The Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer, 1988). The

BAI is a 21-item self-report measure that asks participants to rate symptoms from “Not at all” (0) to “Severely – I could barely stand it” (3). A total score was calculated (range of 0 to 63). The participant scores ranged from 0 to 55. The BAI had an internal consistency of .91 and .93 for T-

1 and T-2, respectively. The test-retest reliability was .62 across a one week period.

Depressive Symptoms. The Beck Depression Inventory-II (BDI-II; Hawley et al., 2013;

Osman, Downs, Barrios, Kopper, Guiterrez, & Chiros, 1997). The BDI-II is a self-report measure of depressive symptoms with responses on a scale ranging from 0 to 3. The BDI-II correlation with other depressive inventories has been found to be satisfactory. 20 of the 21 items were included with the exclusion of the question asking about suicidal ideation (a possible range of 0 to 60. Scores in this sample ranged from 0 to 51. The BDI-II had an internal consistency of .91 and .90 for T-1 and T-2, respectively. The test-retest reliability was .78 across a one week period.

Life Satisfaction. Satisfaction with Life Scale (SWLS; Diener, Emmons, Larsen, &

Griffin, 1985). The SWLS is a 5-item self-report measure that is rated from 1-7 with a total score that ranges from 5 to 35. Evidence of divergent and convergent validity has been found.

The internal consistency was .89 at T-1 and .90 at T-2, and the test-retest reliability was .73 over a one week period.

Adjustment. College Adjustment Test (CAT; Pennebaker, 2013). The CAT is a 19-item questionnaire measuring adjustment to college that is rated by self-report from 1 to 7. This measure had an internal consistency of .75 at T-1 and .78 at T-2 and a test-retest reliability of .79 over a one week period.

Results

First, I tested the assumptions of the data. See Table 1 for a review of the means and standard deviations of the variables. Normality was tested using histograms and normal P-P plots. Based on these graphs, T-1 BAI, T-2 BAI, T-1 BDI, T-2 BDI, and T-2 SWLS were transformed using a square root function. In addition, I also checked the skewness and kurtosis of the measures used. Prior to the transformations, the skewness of the variables mentioned above ranged from 1.21 to 1.76 and the kurtosis of the variables ranged from 1.56 to 3.43.

Following the transformation, skewness ranged from -.07 to .40 and kurtosis ranged from -.64 to

-.10. I also reviewed the skewness and kurtosis of all other measures used in this paper. Given the large size of the sample, significance tests were not reviewed, as even minimal differences from normal can result in significance. All values were within the acceptable range of +2 and -2 with the exception of T-1 Length that had a positive skew and kurtosis, 3.69 and 29.05, respectively. Based on this information, T-1 Length was transformed using a square root function. This resulted in a skewness of .57 and a kurtosis of 3.94.

I conducted a manipulation check to look at the differences between the delivery groups on AF-PredImmed and AF-Pred1Wk. This was done using a MANOVA. The one-way

MANOVA revealed a significant effect of delivery on the AF variables, Wilk’s λ = .714, F (4,

1428.00) = 65.63, p < .001. Power to detect significant effects was 1.00. Using a Bonferroni adjustment, I looked for significance at or below an alpha level of 0.03. Both AF-PredImmed, F

(2, 715) = 132.62, and AF-Pred1Wk, F (2, 715) = 10.64, were significant at p < 0.001. Upon further inspection, the mean scores revealed that only the control group showed a significant difference in levels of AF. This difference was expected given the less emotional nature of the control task. The AF manipulation was ineffective at creating a significant difference in AF scores across the intervention groups and, therefore, was ineffective.

Participation

Multiple analyses were conducted to look at treatment participation. A stepwise discriminant analysis was conducted to predict which participants completed the T-2 intervention task. The predictor variables explored were T-1 BAI, T-1 BDI, T-1 Length, AF-PredImmed,

AF-Pred1Wk, and delivery group. As previously mentioned, participants chose to participate in the study by either selecting a description of a cognitive task or an intervention task. Therefore, self-selection into the study was also a predictor in order to ascertain if this self-selection affected the results. The stepwise discriminant analysis for completion of the T-2 intervention found a significant relationship between completion of the written task and two of the predictors,

T-1 Length (0.828) and T-1 BDI (-0.48). Specifically, the longer the T-1 Length, the more likely a person was to complete T-2. In addition, the lower the T-1 BDI score, the more likely a person was to complete T-2. No other predictors improved the significance of the analysis and were thus removed by the analysis. 63.8% of cases were correctly predicted by this model.

I used multiple regression analyses to explore the factors that may be associated with the length of time spent on the intervention task at T-1 and T-2. These analyses were conducted on the two intervention groups. For T-1 Length, step 1 controlled for the effects of baseline levels of anxiety and depression. Step 2 looked at the main effects of delivery group, AF-PredImmed, and AF-Pred1Wk. Step 3 analyzed the interactions between delivery group and the two types of

AF, AF-PredImmed and AF-Pred1Wk. For T-2 Length, step 1 controlled for T-1 Length. Step 2 controlled for WkFrequency, WkDuration, and baseline levels of anxiety and depression. As was done for T-1 Length, steps 3-4 looked at the main and interactive effects of delivery group and AF predictions.

As seen in Table 1a, there were significant main effects of AF-PredImmed and of AF-

Pred1Wk on the length of time spent on the intervention task at time one. Specifically, higher levels of AF-PredImmed (i.e., predicting more positive affect) resulted in spending less time on the task. In contrast, higher levels of AF-Pred1Wk were associated with spending more time on the intervention task. There were no significant interactions between delivery group and AF-

PredImmed.

Table 2b reviews the variables contributing to T-2 Length, the time spent on the task at the follow-up one week later. The only significant main effect found for time two was T-1

Length, the length of time participants spent on the task at time one. Specifically, the longer the

T-1 Length, the longer the participant spent on the T-2 task. In contrast to T-1, there were no significant effects of delivery group, AF, or the interactions.

Outcomes

A MANOVA was conducted to explore the effects of the randomly assigned experimental group on the change in outcome variables across time. The one-way MANOVA

revealed a nonsignificant effect of experimental group on the outcome variables, Wilk’s λ = .98,

F (8, 438.00) = .63, p = ns. No significant difference in treatment outcomes was found between the three groups. Thus, the intervention groups did not show a significant improvement in treatment outcomes when compared to the control group.

Analyses were also completed to explore the effects of treatment group and AF on treatment outcomes. Given the moderate correlations among the outcome variables, analyses were completed separately for depression, anxiety, life satisfaction, and adjustment (see Table 3).

Multiple regression analyses examined the roles of treatment group, AF, and treatment participation on T-2 mental health outcome variables. These variables were T-2 BAI, T-2 BDI,

T-2 SWLS, and T-2 CAT. Step 1 controlled for baseline level of the outcome variable being analyzed. Step 2 looked at T-1 Length, WkFrequency, and WkDuration. Step 3 looked at the main effects of intervention group and the two types of AF. Step 4 analyzed the interactions between intervention group and the two types of AF.

Tables 4a-4d show the results of these four analyses. T-1 level of the outcome variable had a significant effect on all four analyses. As expected, higher baseline rates of depression, anxiety, satisfaction with life, and college adjustment increase resulted in significantly higher T-

2 rates of these outcomes. In addition, WkFrequency had a significant effect on all of the T-2 outcome variables when keeping all other variables constant. Specifically, the more times a participant completed the task during the week, the higher their levels of the outcome variables.

In addition, a significant main effect of WkDurations was found across the outcome variables.

The longer amount of total time the participant spent on the task during the week, the lower their levels of the outcome variables. In addition, some associations were found for some of the individual outcome measures. There was a negative association between T-1 Length and levels

of SWLS, revealing that less time spent on the task at T-1 resulted in higher levels of satisfaction with life. For CAT, a statistically significant effect was found for AF-Pred1Wk. Higher levels of AF-Pred1Wk resulted in significantly higher levels of CAT. Delivery group and the interactions did not have significant effects on the outcome variables.

Discussion

The results obtained must be reviewed carefully. Many of the hypothesized trends did not occur or conflict with expected results. For example, the results would suggest that what causes increases in psychopathological symptoms also causes increases in resistance factors.

This does not make theoretical sense. This is most likely the result of the large sample size, as it increases the likelihood of finding erroneous significance. This likelihood is supported by the small effect sizes found for each of the results. After my review of the results, I carefully consider sources of error for this study.

Unfortunately, manipulations checks revealed limited success for the AF manipulation and the intervention task. Specifically, the differences in delivery across the treatment groups did not result in statistically different AF levels. As such, it is not surprising that delivery group was not a significant predictor in the analyses conducted. In addition, a MANOVA revealed that the intervention groups did not result in improved outcomes in comparison to the placebo group.

This is inconsistent with numerous past experiments. The effects of delivery and AF on intervention outcomes would likely be minimal in this sample given the limited efficacy.

When looking at participants’ likelihood of completing the intervention task one week later, greater initial participation and lower levels of depressive symptoms result in greater likelihood of continuing with future tasks. However, there is over 35% of the difference between those who completed T-2 and those who did not. This means that there are additional variables

affecting T-2 completion. Contrary to my hypotheses, AF and delivery group were not significant predictors of completing the task a week later. It should be noted that self-selection was not a significant predictor of T-2 completion. In other words, there was no difference between participants who chose to participate in an intervention task and those who believed they would complete a cognitive task. No other participation factors or outcome variables were significantly related to AF or delivery group.

Possible Sources of Error

Perhaps the largest limitation of this study is the use of an online subject pool.

Participants in the subject pool have an additional external motivation of earning class credits in exchange for completing the study. This motivator could have drastically impacted participation and, perhaps, attention to the intervention task. Given the longitudinal nature of the study, this motivator may have drastically impacted who returned to complete T-2. For example, the strongest predictor variable for who chose to complete the T-2 intervention task might be if participants still needed psychology research credits. Perhaps this variable could account for the

35% of unknown variance that helps to predict which participants would complete T-2 and which would not.

As previously stated, it is also possible that the online nature of the study increased variability across participants. For example, having participants attend an in-person session may help to regulate the time spent on the task and reduce distractability. There is much greater control over a participant’s environment during an in-person study. For example, instructions could have been read out loud to increase the likelihood that participants were actively engaged during the delivery portion of the task. Also, there is greater control over distractors such as the internet and socializing with others. To our knowledge, few articles have been published that

uses this task online (Bond & Pennebaker, 2012) which may suggest that the intervention is more successful in in-person settings. In fact, there is some evidence to suggest that completing narratives by writing is superior to typing. For example, researchers have found that writing helps to process information for future recall (Mueller & Oppenheimer, 2014). More importantly, those who write an emotional narrative have greater negative affect and disclose more about the event than those who type the narrative (Brewin & Lennard, 1999).

Additionally, the study focused on a nonclinical population. This likely limited the variance in outcome variables and may have minimized motivation to complete the task. As seen in Table 1, variability in most of the outcome variables was relatively low and the BAI and

BDI-II levels were primarily in the subclinical range. This may suggest that the limited improvement may be due to low T-1 levels of mental health symptoms. However, it should be noted that multiple analyses looked at self-selection and found no differences between those who chose to participate in an intervention versus those who chose to participate in a cognitive task.

It is possible that higher levels of symptom distress and lower levels of functioning are necessary to perceive differences in participation and outcomes. Despite these limitations, a large sample size was collected in order to be able to detect small changes in distress and functioning consistent with a subclinical population.

Specific adaptations to our methodology may be helpful in maximizing future research success. Specifically, it may be beneficial to complete studies in an in-person setting or using auditory instructions in order to minimize environmental distractions and increase attention. In addition, it would likely be advantageous to focus on groups that are more directly motivated by the intervention itself. For example, it could be helpful to explore differences in delivery in a mental health clinic or a hospital. Perhaps the most helpful addition would be a manipulation

check to determine the level of attention given to the delivery narratives. For example, one could ask participants multiple choice questions about the delivery narrative at the end of T-1. This would help rule out inattention as a reason for any failures of the manipulation task. In addition, there may also be ways to improve the manipulations used to affect levels of AF. As discussed earlier, the narrative used to manipulate AF primarily focused on immune neglect. Future narratives could also attempt to increase focalism so that both components of impact bias are equally explored when delivering information. Perhaps what might be most helpful is the use of a clinical population in order to increase internal motivation and reduce the likelihood that participation is due to external variables such as course credit.

Although the results of this study were minimal, there is evidence to suggest that AF and delivery are factors that affect participation and some outcomes in an intervention. Given the limitations of this study that were discussed above, it would be beneficial to further explore these concepts, particularly within more clinical settings. Developing a better understanding of methods of delivering information could improve client motivation and treatment outcomes, independent of the treatment interventions.

References

Alden, L. E., Mellings, T. M., & Laposa, J. M. (2004). Framing social information and

generalized social phobia. Behaviour Research and Therapy, 42(5), 585-600.

Axsom, D. & Cooper, J. (1985). Cognitive dissonance and psychotherapy: The role of effort

justification in inducing weight loss. Journal of Experimental Social Psychology, 21,

149-160.

Baldwin, S.A., Wampold, B.E., & Imel, Z.E. (2007). Untangling the alliance-outcome

correlation: Exploring the relative importance of therapist and patient variability in the

alliance. Journal of Consulting and Clinical Psychology, 75(6), 842-852.

Beck, J. S., & Beck, A. T. (2011). Cognitive behavior therapy, basics and beyond. New York,

NY: The Guilford Press.

Beck, A.T., Epstein, N., Brown, G., Steer, R.A. (1988). An inventory for measuring clinical

anxiety: Psychometric properties. Journal of Consulting and Clinical Psychology, 56(6),

893-897.

Bond, M., & Pennebaker, J. W. (2012). Automated computer-based feedback in expressive

writing. Computers in Human Behavior, 28, 1014-1018.

Brewin, C. R., & Lennard, H. Effects of mode of writing on emotional narratives. Journal of

Traumatic Stress, 12(2), 355-361.

Buehler, R., & McFarland, C. (2001). Intensity bias in affective forecasting: The role of temporal

focus. Personality and Social Psychology Bulletin, 27, 1480-1493.

Butler, C. C., Simpson, S. A., Hood, K., Cohen, D., Pickles, T., Spanou, C., Mccambridge, J.,

Moore, L., Randell, E., Alam, M. F., Kinnersley, P., Edwards, A., Smith, C., & Rollnick,

S. (2013). Training practitioners to deliver opportunistic multiple behavior change in

counselling in primary care: A cluster randomized trial. British Medical Journal, 346, 10.

Cooper, J. (1980). Reducing fears and increasing assertiveness: The role of dissonance reduction.

Journal of Experimental Social Psychology, 16, 199-213.

Cox, B.J., Swinson, R.P., Norton, G.R., & Kuch, K. (1991). Anticipatory anxiety and avoidance

in panic disorder with agoraphobia. Behavior Research and Therapy, 29(4), 363-365.

Creswell, J. D., Lam, S., Stanton, A. L., Taylor, S. E., Bower, J. E., & Sherman, D. K. (2007).

Does self-affirmation, cognitive processing, or discovery of meaning explain cancer-

related health benefits of expressive writing? Personality and Social Psychology Bulletin,

33(2), 238-250.

Crits-Christoph, P., & Mintz, J. (1991). Implications of therapist effects for the design and

analysis of comparative studies of psychotherapies. Journal of Consulting and Clinical

Psychology, 59(1), 20-26.

Danoff-Burg, S., Agee, J. D., Romanoff, N. R., Kremer, J. M., & Strosberg, J. M. (2006). Benefit

finding and expressive writing in adults with lupus or rheumatoid arthritis. Psychology

and Health, 21(5), 651-665.

Diener, E., Emmons, R.E., Larsen, R.J., & Griffin, S. (1985). The Satisfaction with Life Scale.

Journal of Personality Assessment, 49(1), 71-75.

Dillard, A.J., Fagerlin, A., Dal Cin, S., Zikmund-Fisher, B.J., & Ubel, P.A. (2010). Narratives

that address affective forecasting errors reduce perceived barriers to colorectal cancer

screening. Social Science & Medicine, 71, 45-52.

Eastwick, P. W., Finkel, E. J., Krishnamurti, T., Loewenstein, G. (2008). Mispredicting distress

following romantic breakup: Revealing the time course of the affective forecasting error.

Journal of Experimental Social Psychology, 44(3), 800-807.

Elkin, I. (1999). A major dilemma in psychotherapy outcome research: Disentangling therapists

from therapies. Clinical Psychology: Science and Practice, 6(1), 13-32.

Falk, C. F., Dunn, E. W., & Norenzayan, A. (2010). Cultural variation in the importance of

expected enjoyment for decision making. Social Cognition, 28(5), 609-629.

Falkenström, F., Markowitz, J. C., Jonker, H., Philips, B., Holmqvist, R. (2013). Can

psychotherapists function as their own controls? Meta-analysis of the crossed therapist

design in comparative psychotherapy trials. Journal of Clinical Psychiatry, 74(5), 482-

491.

Frattaroli, J. (2006). Experimental disclosure and its moderators: A meta-analysis. Psychological

Bulletin, 132(6), 823-865.

Gallagher, K. M. & Updegraff, J. A. (2012). Health message framing effects on attitudes,

intentions, and behavior: A meta-analytic review. Annual Behavioral Medicine, 43, 101-

116. http://dx.doi.org/10.1007/s12160-011-9308-7

Gilbert, D. T., Pinel, E. C., Wilson, T. D., Blumberg, S. J., & Wheatley, T. P. (1998). Immune

neglect: A source of durability bias in affective forecasting. Journal of Personality and

Social Psychology, 75(3), 617-638.

Goldfried, M. R., & Davila, J. (2005). The role of relationship and technique in therapeutic

change. Psychotherapy: Theory, Research, Practice, Training, 42(4), 421-430.

Goldstein, A. P. (1966). Psychotherapy research by extrapolation of social psychology. Journal

of Counseling Psychology, 13(1), 38-45.

Greenberg, M. A., Stone, A. A., & Wortman, C. B. (1996). Health and psychological effects of

emotional disclosure: A test of the inhibition-confrontation approach. Journal of

Personality and Social Psychology, 71, 558-602.

Hawley, C. J., Gale, T. M., Smith, P. St. J., Jain, S., Farag, A., et al. (2013). Equations for

converting scores between depression scales (MADRS, SRS, PHQ-9 and BDI-II): Good

statistical, but weak idiographic, validity. Human Psychopharmacology: Clinical and

Experimental, 28, 544-551.

Hoerger, M., Quirk, S.W., Chapman, B.P., & Duberstein, P.R. (2012). Affective forecasting and

self-rated symptoms of depression, anxiety, and hypomania: Evidence for a dysphoric

forecasting bias. Cognition and Emotion, 26(6), 1098-1106.

Hoyt, T., & Yeater, E. A. (2011). The effects of negative emotion and expressive writing on

posttraumatic stress symptoms. Journal of Social and Clinical Psychology, 30(6), 549-

569.

Kim, D.-M., Wampold, B. E., & Bolt, D. M. (2006). Therapist effects in psychotherapy: A

random-effects modeling of the National Institute of Mental Health Treatment of

Depression Collaborative Research Program data. Psychotherapy Research, 16(2), 161-

172.

Luborsky, L., Diguer, L., Seligman, D. A., Rosenthal, R., Krause, E. D., Johnson, S., Halperin,

G., Bishop, M., Berman, J.S., & Schweizer, E. (1999). The researcher’s own therapy

allegiances: A “wild card” in comparisons of treatment efficacy. Clinical Psychology:

Science and Practice, 6(1), 95-106.

Luborsky, L., Rosenthal, R., Diguer, L., Andrusyna, T.P., Berman, J.S., Levitt, J.T., et al. (2002).

The dodo bird verdict is alive and well – mostly. Clinical Psychology, Science and

Practice, 9(1), 2-12.

Marroquín, B., Nolen-Hoeksema, S., & Miranda, R. (2013). Escaping the future: Affective

forecasting in escapist fantasy and attempted suicide. Journal of Social and Clinical

Psychology, 32(4), 446-463.

Meyerowitz, B. E. & Chaiken, S. (1987). The effect of message framing on breast self-

examination attitudes, intentions, and behavior. Journal of Personality and Social

Psychology, 52, 500-510. http://dx.doi.org/10.1037/0022-3514.52.3.500

Milkman, K.L., Akinola, M., & Chugh, D. (2012). Temporal distance and discrimination: An

audit study in academia. Psychological Science, 23(7), 710-717.

Miller, W. M., & Rollnick, S. (2012). Motivational interviewing: Helping people change. (3rd

ed.). New York, NY: The Guilford Press.

Mueller, P. A., & Oppenheimer, D. M. (2014). The pen is mightier than the keyboard:

Advantages of longhand over laptop note taking. Psychological Science, 25(6), 1159-

1168.

Murphy, F. C., Sahakian, B. J., Rubinsztein, J. S., Michael, A., Rogers, R. D., Robbins, T. W., &

Paykel, E. S. (1999). Emotional bias and inhibitory control processes in mania and

depression. Psychological Medicine, 29(6), 1307-1321.

Okiishi, J., Lambert, M.J., Nielsen, S.L., & Ogles, B.M. (2003). Waiting for supershrink: An

empirical analysis of therapist effects. Clinical Psychology and Psychotherapy, 10, 361-

373.

Osman, A., Downs, W.R., Barrios, F.X., Kopper, B.A., Gutierrez, P.M., & Chiros, C.E. (1997).

Factor structure and psychometric characteristics of the Beck Depression Inventory-II.

Journal of Psychopathology and Behavioral Assessment, 19(4), 359-376.

Pennebaker, J. W. (1993). Putting stress into words: Health, linguistic & therapeutic

implications. Behavior Research and Therapy, 31, 539-548.

Pennebaker, J. W. (1997). Writing about emotional experiences as a therapeutic process.

Psychological Science, 8(3), 162-166.

Pennebaker, J.W. (2013). The College Adjustment Test (CAT). Measurement Instrument

Database for the Social Science. Retrieved from www.midss.ie.

Pennebaker, J. W. (n.d.). Writing and Health. James W. Pennebaker. Retrieved June 3, 2013,

from http://homepage.psy.utexas.edu/HomePage/Faculty/Pennebaker/Home2000/Writing

andHealth.html

Pennebaker, J. W., & Beall, S. K. (1986). Confronting a traumatic event: Toward an

understanding of inhibition and disease. Journal of Abnormal Psychology, 95(3), 274-

281.

Pennebaker, J.W., Colder, M., & Sharp, L.K. (1990). Accelerating the coping process. Journal of

Personality and Social Psychology, 58(3), 528-537.

Pennebaker, J. W., Kiecolt-Glaser, J. K., & Glaser, R. (1988). Disclosure of traumas and immune

function: Health implications for psychotherapy. Journal of Consulting and Clinical

Psychology, 56(2), 239-245.

Resick, P.A., Monson, C.M., & Chard, K.M. (2008). Cognitive processing therapy:

Veteran/military version. Washington, DC: Department of Veterans’ Affairs.

Ridley, C.R., Kelly, S.M., & Mollen, D. (2011). Microskills training: Evolution, reexamination,

and call for reform. The Counseling Psychologist, 39(6), 800-824.

Ruby, M. B., Dunn, E. W., Perrino, A., Gillis, R., & Viel, S. (2011). The invisible benefits of

exercise. Health Psychology, 30(1), 67-74.

Schultz, L. T., & Heimberg, R. G. (2008). Attentional focus in social anxiety disorder: Potential

for interactive processes. Clinical Psychology Review, 28(7), 1206-1221.

Schutte, N. S., Searle, T., Meade, S., Dark, N. A. (2012). The effect of meaningfulness and

integrative processing in expressive writing on positive and negative affect and life

satisfaction. Cognition and Emotion, 26(1), 144-152.

Serlin, R. C., Wampold, B. E., & Levin, J. R. (2003). Should providers of treatment be regarded

as a random factor? If it ain’t broke, don’t “fix” it: A comment on Siemer and Joorman

(2003). Psychological Methods, 8, 524-534.

Smedslund, G., Berg, R. C., Hammerstrom, K. T., Steiro, A., Leiknes, K. A., Dahl, H. M., &

Karlsen, K. (2011). Motivational interviewing for substance abuse. Campbell Systematic

Reviews, 6. 10.4073/csr.2011.6

Smyth, J. M., Hockemeyer, J. R., & Tulloch, H. (2008). Expressive writing and post-traumatic

stress disorder: Effects on trauma symptoms, mood states, and cortisol reactivity. British

Journal of Health Psychology, 13, 85-93.

Smyth, J. M., & Pennebaker, J. W. (2008). Exploring the boundary conditions of expressive

writing: In search of the right recipe. British Journal of Health Psychology, 13, 1-7.

Strong, S. (1968). Counseling: An interpersonal influence process. Journal of Counseling

Psychology, 15(3), 215-224.

Tomlinson, J. M., Carmichael, C. L., Reis, H. T., Aron, A. (2010). Affective forecasting and

individual differences: Accuracy for relational events and anxious attachment. Emotion,

10(3), 447-453.

Totterdell, P., Parkinson, B., Briner, R. B., & Reynolds, S. (1997). Forecasting feelings: The

accuracy and effects of self-predictions of mood. Journal of Social Behavior and

Personality, 12(3), 631-650.

Truax, C. B., & Carkhuff, R. R. (1967). Toward effective counseling and psychotherapy:

Training and practice. Chicago, IL: Aldine.

Trupp, R. J., Corwin, E. J., Ahijevych, K. L., & Nygren, T. (2011). The impact of educational

message framing on adherence to continuous positive airway pressure therapy.

Behavioral Sleep Medicine, 9, 38-52. http://dx.doi.org/10.1080/15402002.2011.533993

Tversky, A. & Kahneman, D. (1981). The framing of decisions and the psychology of choice.

Science, 211, 453-458. http://dx.doi.org/10.1126/science.7455683

Van Boven, L., Loewenstein, G., Welch, E., & Dunning, D. (2012). The illusion of courage in

self predictions: Mispredicting one’s own behavior in embarrassing situations. Journal of

Behavioral Decision Making, 25, 1-12.

Van Tilburg, W.A.P., & Igou, E.R. (2014). The impact of middle names: Middle name initials

enhance evaluations of intellectual performance. European Journal of Social Psychology.

Doi: 10.1002/ejsp.2026.

Wampold, B. E., & Brown, G. S. (2005). Estimating variability in outcomes attributable to

therapists: A naturalistic study of outcomes in managed care. Journal of Consulting and

Clinical Psychology, 73(5), 914-923.

Watson, D., & Clark, L. A. (1994). The PANAS-X: Manual for the Positive and Negative Affect

Schedule–Expanded Form. Unpublished Manuscript, University of Iowa, Iowa City.

Wenze, S.J., Gunthert, K.C., & German, R.E. (2012). Biases in affective forecasting and recall in

individuals with depression and anxiety symptoms. Personality and Social Psychology

Bulletin, 38(7), 895-906.

Wilson, T.D., & Gilbert, D.T. (2003). Affective forecasting. Advances in Experimental Social

Psychology, 35, 345-411.

Wilson, T.D., & Gilbert, D.T. (2005). Affective forecasting: Knowing what to want. Current

Directions in Psychological Science, 14(3), 131-134.

Wilson, T.D., Wheatley, T., Meyers, J.M., Gilbert, D.T., & Axsom, D. (2000). Focalism: A

source of durability bias in affective forecasting. Journal of Personality and Social

Psychology, 78, 821-836.

Woodzicka, J. A., & LaFrance, M. (2001). Real versus imagines gender harassment. Journal of

Social Issues, 57(1), 15-30.

Zuroff, D. C., Kelly, A. C., Leybman, M. J., Blatt, S. J., & Wampold, B. E. (2010). Between-

therapist and within-therapist differences in the quality of the therapeutic relationship:

Effects on maladjustment and self-critical perfectionism. Journal of Clinical Psychology,

66(7), 681-697.

Appendix A. (Adapted from Pennebaker, 1997; Pennebaker, n.d. AF conditions developed using

Dillard et al., 2010. Delivery AF uses an analogy adapted from Resick, Monson, & Chard, 2008)

Below are the three experimental conditions. Differences between the two intervention conditions are in italics.

1. Delivery

In the following task, you will be asked to write about a distressing and/or traumatic

emotional experience in your life. I will request that you write about this experience in

great detail, as this helps you to remember the experience vividly and allows you to feel

the exact emotions that you felt when the experience occurred.

This task is often used to help reduce the impact of these negative experiences on a

person’s daily life, as negative experiences can often result in worry, low mood, and

avoidance of similar events. This task will allow you to express and organize your

thoughts and feelings related to the experience which, in turn, will make the feelings

become more tolerable over time. The experience can then be stored in your memory

with less intense feelings, resulting in less distressing memories of the experience. In

fact, researchers and therapists have used this task with great success. The task has been

linked to improved academic achievement, sleep, physiological health, and psychological

well-being.

Do not be surprised if you feel your emotions almost as strongly as you did at the time of

the incident. Many people also report feeling sad or depressed after writing; however,

this feeling usually goes away within a few hours. Over time, if you continue to allow

yourself to focus on your thoughts and emotions related to the experience, your feelings

will become less intense and less overwhelming.

Many people report that after writing, they sometimes feel somewhat sad or depressed.

Like seeing a sad movie, this typically goes away in a couple of hours. If you find that you

are getting extremely upset about a writing topic, simply stop writing or change topics.

2. Delivery AF