A Randomized Controlled Trial of a Self-Compassion Writing Intervention for Adults With Mental Illness

Abstract

Purpose:

Studies have found that writing with self-compassion about a stressful event helps promote mental health in college students and nonclinical populations. Using a randomized controlled trial, this study investigated whether a self-compassion writing intervention would lead to increases in self-compassion and proactive coping and reductions in mental health symptoms in a sample of individuals with mental illness.

Method:

Individuals with mental disorders were recruited and randomly assigned to a treatment condition in which participants wrote with self-compassion or a control condition where participants wrote about how they spent their time. Outcome measures were administered at pretest, after the 3-day intervention, and 1 month later.

Results:

Both the treatment and control groups showed significant improvements in self-compassion, proactive coping, mental health, and physical health.

Discussion:

Overall, the results suggest both self-compassion writing and writing about how one spends one’s time may be beneficial for individuals with mental illness.

Keywords

mental illness journal writing mental health field of practice randomized experiment outcome study

According to the U.S. Substance Abuse and Mental Health Services Administration’s (SAMHSA) 2016 National Survey on Drug Use and Health, 11.8 million American adults perceived an unmet need for mental health care in 2016, and the most significant barrier to mental health services was cost: 41.1% of adults identified with any mental illness and 46.2% of adults with a severe mental illness who stated they had an unmet need for mental health care explained they could not afford services (Park-Lee et al., 2017). Low-income individuals are more likely to develop a mental disorder, partly because they grow up and/or live in chronically, chaotic environments that are a constant source of stress (Anakwenze & Zuberi, 2013).

Recently, there has been a new, international push by the World Health Organization (WHO, 2017) to find research-supported, scalable mental health interventions to help people affected by adversity. Writing interventions offer great potential as scalable treatments for mental illness. They can be used by people who may be unable or unwilling to participate in psychotherapy because they are inexpensive and portable (Pascoe, 2017), and they may help reach ethnic groups that may be reluctant to engage in traditional Western talk therapy (Marsiglia & Kulis, 2015).

Writing interventions were first systematically studied by Pennebaker and Beall (1986) who asked college students to write continuously about their “deepest thoughts and feelings” about a traumatic event for 15–20 min per day for four consecutive days. Pennebaker and Beall found students who expressed their thoughts and feelings about a traumatic event through writing had fewer physician visits in the 6 months following the experiment than controls. These students also reported that their writing experience was beneficial to them long-term (Pennebaker & Beall, 1986). Pennebaker and Beall suggested the effect they found was not due to catharsis (the process of releasing repressed emotions, originally described by Freud) but rather due to a mechanism previously described by Jourard (1971) in which emotional expression leads to an increase in self-understanding, which relieves stress. Pennebaker and Beall coined their intervention “expressive writing.” By 2013, over 400 expressive writing studies with varying protocols were published (Niles et al., 2014). While numerous studies found expressive writing seems to improve health by helping people face and process traumatic experiences and individuals with different types of mental illness are more likely than the general population to experience trauma and stress (Kilgus et al., 2015; Rudnick & Lundberg, 2012), research on the effects of expressive writing interventions for individuals with mental illness has been mixed and not as consistent as studies on college students or individuals with physical illness (Baikie et al., 2012).

Researchers have been unable to identify a single causal mechanism that explains how expressive writing works (Pennebaker, 2004; Sloan & Marx, 2004). Self-compassion may represent an alternative mechanism that helps explain the effectiveness of all forms of therapeutic writing that have been tested to date, including expressive writing and positive writing, for individuals with mental illness.

Compassion may be defined as “a state of concern for the suffering or unmet need of another, coupled with a desire to alleviate that suffering” (Goetz & Simon-Thomas, 2017). Self-compassion is simply described as “compassion turned inward” (Neff & Germer, 2013). Neff (2003b) breaks down the concept of self-compassion into three components:

Self-kindness—extending kindness and understanding to oneself rather than harsh judgment and self-criticism,

Common humanity—seeing one’s experiences as part of the larger human experience rather than seeing them as separating and isolating, and

Mindfulness—holding one’s painful thoughts and feelings in balanced awareness rather than overidentifying with them.

Self-criticism has been found to be a risk factor for many different types of mental illness (Kannan & Levitt, 2013; Shahar & Henrich, 2013; Waite et al., 2015), and clinicians have found that activating adaptive emotions like self-compassion can help clients access the cognitive part of the brain and resolve maladaptive habits such as self-criticism (Greenberg, 2008). Self-compassion has a significant inverse relationship with self-criticism (r = −.65, p < .01; Neff, 2003b) and is associated with adaptive coping strategies and negatively associated with maladaptive avoidance-oriented strategies such as distraction or suppression of emotions (Neff et al., 2005). Self-compassion appears to help individuals make progress toward their goals by preventing an escalation of negative affect during challenging situations (Hope et al., 2014).

Self-compassion interventions have been found to be effective for severe symptoms of depression (Diendrich et al., 2014) and other kinds of mental illness (Jazaieri et al., 2014). One type of brief, accessible self-compassion intervention used in compassion therapy and training programs (Gilbert, 2009; Neff & Germer, 2013) is self-compassion writing. Several studies have found that writing to oneself about a problematic event using the elements of self-compassion is more effective than expressive writing for college students and nonclinical populations. Leary et al. (2007) conducted the first self-compassion writing study. The researchers asked participants to write about a difficult event and assigned them to three groups: a self-compassion writing condition, a self-esteem writing condition, and a control expressive writing condition. Participants in the self-esteem writing condition were asked first, to write down their positive characteristics; second, to write about how the event was not completely their fault; and third, to write how the event doesn’t indicate about how they are as a person. The researchers found that asking participants to write with self-compassion led participants to accept their role in difficult situations and experience less negative affect compared to participants in the expressive writing and self-esteem writing conditions.

Similarly, Odou and Brinker (2013) found self-compassion writing reduced Australian college students’ negative affect significantly more than expressive writing. Johnson and O’Brien (2013) found self-compassion writing is more effective than traditional expressive writing, and the added element of self-compassion seems to allow individuals to temper strong emotions that may arise when reexamining traumatic memories, resulting in lower levels of shame and negative affect. This suggests that an individual’s level of self-compassion may predict how much they benefit from expressive writing. It also suggests that individuals with mental illness, who experience lower levels of self-compassion than the general population (MacBeth & Gumley, 2012), may benefit more from self-compassion writing than expressive writing.

Only one previous study, by Helm (2016), investigated the effects of self-compassion writing on individuals with mental illness. Helm studied adults with a generalized anxiety disorder, randomizing participants to a control group that participated in relaxation training and a treatment group that used a self-compassion writing intervention. The treatment group showed more significant increases in positive expressivity than the control group, and both the treatment and control groups showed reduced emotion dysregulation and negative expressivity.

The authors aimed to fill a gap in the research regarding the effectiveness of self-compassion writing for individuals with clinical diagnoses. The current experiment aimed to determine whether writing with self-compassion increases levels of self-compassion, improves coping ability, decreases depression, and promotes physical health in adults with different types of mental illness including major depressive disorder, bipolar disorder, schizophrenia, and schizoaffective disorder. We expected that individuals in the treatment group who completed the self-compassion writing intervention would show a greater increase in self-compassion and proactive coping and a greater decrease in depression and physical symptoms compared to participants in the neutral control group who wrote about how they spent their time.

Method

Participants

The current study was posted on the Amazon MTurk platform between July 5, 2017, and August 11, 2017. Amazon MTurk is a website that allows the public to sign up for research studies and earn a small financial reward for their participation. MTurk has been a popular crowdsourcing site for clinical researchers for over a decade (Chandler & Shapiro, 2016). It is more feasible to conduct research with large sample sizes on MTurk than it is using traditional designs (Necka et al., 2016). Moreover, it has been found that MTurk recruitment and testing produce indistinguishable results from in-person testing when face-to-face interaction between researcher and participant is not required (Buhrmester et al., 2011; Casler et al., 2013).

The inclusion criteria for the study were English-speaking adults 18 years or older with a diagnosis of major depressive disorder, bipolar disorder, schizophrenia, or schizoaffective disorder who lived in one of the following 18 English-speaking countries: Antigua and Barbuda, Australia, The Bahamas, Barbados, Belize, Canada, Dominica, Grenada, Guyana, Ireland, Jamaica, New Zealand, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Trinidad and Tobago, United Kingdom, and United States. Consistent with our review of MTurk compensation for similar research projects, we paid participants US$4 for completing the entire study. Subjects provided electronic informed consent through the Qualtrics platform. After consenting, participants completed a screening question asking them if they had major depression, bipolar disorder, schizophrenia, or schizoaffective disorder. Individuals who answered “yes” were taken to the first day’s study, and individuals who answered “no” were disqualified. Participants were randomized by the Qualtrics platform to an experimental group (self-compassion writing condition) or a control group (neutral writing condition). This study was approved by an institutional review board. The study was not preregistered in a clinical trials registry as it was conducted as a thesis (Urken, 2018).

Procedures

Participants were asked to write for 15 min on three consecutive days (Wave 1, Wave 2, and Wave 3). At each wave, participants in the treatment group were asked to complete the treatment writing intervention, and participants in the control group were asked to complete the control writing task. The pretest survey was administered at Wave 1 before the first writing session, the posttest was administered at Wave 3 after the third writing session, and the 1-month follow-up was administered 28 days after participants completed Wave 3. After participants completed the first wave, they were invited to enter their email address, so that the second wave link could be emailed to them in 24 hr. They were informed that their email address would be kept in a separate database that was not associated with their data, and they would have 24 hr from the receipt of the email to complete the next wave’s survey. Participants were invited to complete the Wave 3 and the 1-month follow-up survey in a similar manner. Due to technical difficulties with the Qualtrics platform’s built-in timer, the researchers were not able to time participants as they wrote. In lieu of this, participants were asked to time themselves using a timer and were provided with a link to the internet timer on Google that they could use.

The total length and duration of the intervention are consistent with meta-analytic findings that showed that studies that had at least three writing sessions had marginally larger effect sizes than studies with less than three sessions and that studies with writing sessions of at least 15 min duration had significantly larger effect sizes than studies with writing sessions of shorter periods of time (Frattaroli, 2006).

Measures

Demographic variables, including ethnicity, gender, education, employment status, last year’s annual income, and country of residence, were assessed at pretest using a self-report questionnaire. Health status variables adapted from Holschuh (2007), which included questions about mental health diagnoses, symptoms, and service usage, and the four primary outcome measures, which assessed self-compassion, proactive coping, depression, and physical health, were assessed at pretest, posttest, and 1-month follow-up. Other outcome measures included an affect assessment that was administered immediately before and after each wave’s writing intervention.

Self-compassion

The Self-Compassion Scale, Short Form (SCS-SF) is a 12-item self-assessment that measures self-compassion (Raes et al., 2011). Possible scores range from 1 to 5, with higher scores representing higher self-compassion. In the current study, the Cronbach’s α for the SCS-SF was .85 for the pretest, .89 for the posttest, and .89 for the 1-month follow-up.

Proactive coping

The Proactive Coping Scale (PCS) is a 14-item self-assessment that measures autonomous goal setting and self-regulatory goal attainment cognition and behavior (Greenglass et al., 1999). Possible scores range from 14 to 56, with higher scores indicating more persuasive coping skills. In the current study, the Cronbach’s αfor the PCS was .88 for the pretest, .90 for the posttest, and .90 for the 1-month follow-up.

Depression

The Remission from Depression Questionnaire (RDQ; modified) is a 41-item self-assessment that measures a broad spectrum of experiences reported by patients as being indicative of remission (Zimmerman et al., 2013). For the current study, the positive mental health, depression, and anxiety subscales were selected and combined to measure depression. The Likert-type scale was changed from a 3-point scale to a 4-point scale, to avoid ceiling effects—possible scores on the modified scale range from 29 to 116, with higher scores indicating more significant depression. The Cronbach’s αfor this RDQ (modified) scale was .93 for the pretest, .94 for the posttest, and .95 for the 1-month follow-up.

Physical symptoms

The Cohen–Hoberman Inventory of Physical Symptoms (CHIPS) is a 33-item self-assessment that measures a variety of common physical symptoms such as back pain and diarrhea (Cohen & Hoberman, 1983). Possible scores range from 33 to 165, with higher scores indicating poor physical health. In the current study, the Cronbach’s αfor the CHIPS Scale was .82 for the pretest, .87 for the posttest, and .85 for the 1-month follow-up.

Mood

The Positive and Negative Affect Schedule (PANAS) Short Form is a 10-item self-assessment that measures how intensely participants currently feel positive and negative emotions (Kercher, 1992). The five positive emotion items get summed to create a Positive Affect (PA) subscale, and the five negative emotions items get summed to create a Negative Affect (NA) subscale. Possible scores for each scale range from 5 to 25, with higher scores on the Positive Affect subscale indicating higher positive affect and higher scores on the Negative Effect subscale indicating higher negative affect. In the current study, the Cronbach’s αs for the PANAS short form PA ranged from .79 to .84 for the three prewriting scores and .84 to .87 for the three postwriting scores; the Cronbach’s α for NA ranged from .88 to .89 for the three prewriting scores and .88 to .91 for the three postwriting scores.

Writing Conditions and Instructions

The self-compassion writing intervention integrated writing prompts from previous studies of self-compassion (Leary et al., 2007; Shapira & Mongrain, 2010), self-compassion journal-writing prompts, and material provided by the Compassion Institute (2017) for noncommercial purposes. The self-compassion writing intervention consisted of four, consecutive self-compassion writing prompts focused on the three aspects of self-compassion (mindfulness, common humanity, and self-kindness; Neff, 2003b). The first prompt asked participants to write about a recent difficult experience they had in a nonjudgmental, accepting way. The second prompt encouraged participants to acknowledge what they were hoping for and needing and to write about the core need underneath their stress or suffering, such as a need for health, safety, love, appreciation, connection, or achievement. The third prompt asked participants to write a brief letter to themselves acknowledging their common humanity and how everyone makes mistakes, fails, or gets angry sometimes. The final prompt asked participants to imagine a wise, compassionate person they trust or a compassionate figure from nature and to write down what kind words this figure might say to them during their time of distress. Participants in the treatment group were asked to complete all four of these consecutive prompts at each wave.

Consistent with previous expressive writing and compassion writing research (Baikie et al., 2012; Matthiesen et al., 2012; Wong & Mak, 2016), the neutral writing intervention instructed participants to describe in detail how they spend their time, leaving out emotions and providing just the facts. In order to match the structure of the experimental writing intervention as much as possible, the control writing intervention was broken up into four separate parts. The first writing prompt asked participants to write about how they spent their time this morning or yesterday morning, the second writing prompt instructed control participants to write about how they spent their time this afternoon or yesterday afternoon, the third writing prompt asked control participants to write about how they spent their time this evening or yesterday evening, and the final writing prompt asked control participants to write about how they planned to spend their time tomorrow. Participants in the control group were asked to complete all four of these consecutive prompts at each wave.

Safety Considerations

Participants were informed of the possible risk of breach by a third party when using the internet, prior to the commencement of the study. In addition, participants were notified that the work they perform on MTurk could be linked to their public Amazon.com profile page, and they were informed that they might wish to restrict what information they choose to share in their public Amazon.com profile to keep their work private. Lastly, telephone hotlines were provided in case the study triggered emotional distress, and participants were notified to call 911 in case of emergency.

Statistical Analysis

Analyses were conducted in SPSS, Versions 24 and 25. The level of significance was set at α = .05. Analyses of variance (ANOVAs) were utilized to explore the effect of treatment on primary and secondary clinical outcomes, and effect sizes were reported. Independent t tests were used to assess the length of participants’ writing and how long it took participants to complete the experiment. Outcome measure analyses only used data from participants who completed all three waves of the experiment.

Hypothesis 1: Based on recent research suggesting that self-compassion interventions (Albertson et al., 2014; Arch et al., 2014; Neff & Germer, 2013; Smeets et al., 2014), and self-compassion writing interventions in particular (Breines & Chen, 2012; Ziemer, 2014) increase levels of self-compassion, we hypothesized that subjects who completed the self-compassion writing intervention in the current study would show a greater increase in self-compassion at the Wave 3 posttest and 1-month follow-up than participants in the control group.

Hypothesis 2: Based on research by Neff et al. (2005) that found self-compassion is positively correlated with adaptive coping abilities and research that found self-compassion may help individuals make progress with their goals by controlling negative affect (Hope et al., 2014), we hypothesized that subjects who completed the self-compassion writing intervention in the current study would show a greater increase in proactive coping at the posttest and 1-month follow-up than participants in the control group.

Hypothesis 3: Based on research that demonstrated that self-compassion is negatively correlated with mental illness (MacBeth & Gumley, 2012) and that self-compassion interventions can protect against mental illness (Trompetter et al., 2017; Waite et al., 2015), and reduce mental health symptoms like depression and anxiety (Arch et al., 2014; Neff & Germer, 2013), we hypothesized that participants who completed the self-compassion writing intervention in the current study would show a greater improvement in depression at the posttest and 1-month follow-up than participants in the control group.

Hypothesis 4: Based on the original expressive writing study by Pennebaker and Beall (1986), subsequent expressive writing studies (Koschwanez et al., 2013; Petrie et al., 2004; Sloan & Marx, 2004), and Wong and Mak’s (2016) self-compassion writing study that showed that writing interventions are effective at improving physical symptoms, we hypothesized that participants in the treatment group would show a greater decrease in physical symptoms at the posttest and 1-month follow-up than the control group.

Hypothesis 5: Based on the previous self-compassion writing research that showed positive affect significantly improved for participants who wrote with self-compassion (Helm, 2016; Imrie & Troop, 2012; Ziemer, 2014) and positive writing research that resulted in immediate increases in positive affect after each wave’s intervention (Burton & Kiing, 2007), we hypothesized that individuals in the treatment group in the current study would show a greater increase in positive affect than the control group after each wave’s writing intervention. We measured positive and negative affect immediately before and after each wave’s intervention to be consistent with previous writing research.

Results

One-thousand nine individuals enrolled in the study; seven declined to sign the consent form, and 39 were found to be ineligible (see CONSORT diagram; Figure 1). A total of 211 participants completed all three waves and the follow-up assessment; eight participants who completed the entire study were excluded because, due to technical errors, they either completed both the experimental and control interventions or were reassigned to a different group in the middle of the experiment. The total number of participants who did not complete the study was 629. A large number of participants did not complete the study because it required three separate waves of participation and data collection, and this is not typical for many participants on the MTurk platform. Other reasons for dropout may have included technical problems that were encountered with some of the links to the Wave 2 and Wave 3 studies that were sent to participants.

Figure 1.

CONSORT diagram.

The majority of completers were White (n = 155, 76.4%) female (n = 148, 72.9%), had at least some college or training (n = 186, 91.6%), resided in the United States (n = 197, 97.0%; including Puerto Rico), and reported they experienced at least one type of psychotic symptom (n = 140, 69.0%) and saw a mental health professional within the last year (n = 151, 74.4%). In addition, over half of participants (n = 115, 56.7%) reported they were currently taking psychotropic medication, and close to half (n = 94, 46.3%) reported a history of at least one psychiatric emergency visit. Additional demographic and clinical characteristics of the sample are included in Tables 1 and 2.

Table 1.

Pretest Demographic Characteristics of Completers and Dropouts.

Characteristic	n (%) or M (SD)
Characteristic	Total Sample(n = 832)	Completers(n = 203)	Dropouts(n = 629)
Age	32.00 (9.35)	34.59 (9.91)	31.16 (9.01)
Gender
Male	269 (32.3%)	51 (25.1%)	218 (34.7%)
Female	539 (64.8%)	148 (72.9%)	391 (62.2%)
Male transgender	6 (0.7%)	2 (1.0%)	4 (0.6%)
Female transgender	5 (0.6%)	0 (0.0%)	5 (0.8%)
Gender variant/nonconforming	12 (1.4%)	2 (1.0%)	10 (1.6%)
Prefer not to answer	1 (0.1%)	0 (0.0%)	1 (0.2%)
Employment status
Full-time	394 (47.4%)	81 (39.9%)	313 (49.8%)
Part-time	114 (13.7%)	34 (16.7%)	80 (12.7%)
Student	105 (12.6%)	24 (11.8%)	81 (12.9%)
Homemaker	58 (7%)	16 (7.9%)	42 (6.7%)
Disabled	71 (8.5%)	22 (10.8%)	49 (7.8%)
Unemployed	71 (8.5%)	23 (11.3%)	48 (7.6%)
Retired	6 (0.7%)	1 (0.5%)	5 (0.8%)
Unknown	1 (0.1%)	0 (0.0%)	1 (0.2%)
Freelancer/indep. contractor/self-employed	12 (1.4%)	2 (1.0%)	10 (1.6%)
Education
Some high school	11 (1.3%)	3 (1.5%)	8 (1.3%)
High school/GED	93 (11.2%)	14 (6.9%)	79 (12.6%)
Some college	378 (45.4%)	79 (38.9%)	299 (47.5%)
College degree	258 (31.0%)	68 (33.5%)	190 (30.2%)
Advanced degree	92 (11.1%)	39 (19.2%)	53 (8.4%)
Annual income
Under US$5,000	153 (18.4%)	46 (22.7%)	105 (17.0%)
US $5,000–US$24,999	262 (31.5%)	64 (31.5%)	198 (31.5%)
US$25,000–US$49,999	241 (29.0%)	54 (26.6%)	187 (29.7%)
US$50,000–US$99,999	143 (17.2%)	32 (15.8%)	111 (17.6%)
US$100,000 or more	33 (4.0%)	7 (3.4%)	26 (4.1%)
Race/ethnicity
White	604 (72.6%)	155 (76.4%)	449 (71.4%)
Black or African American	96 (11.5%)	18 (8.9%)	78 (12.4%)
American Indian or Alaska Native	6 (0.7%)	3 (1.5%)	3 (0.5%)
Asian	34 (4.1%)	8 (3.9%)	26 (4.1%)
Native Hawaiian or Pacific Islander	2 (0.2%)	0 (0.0%)	2 (0.3%)
Hispanic	54 (6.5%)	9 (4.4%)	45 (7.2%)
Eastern European/Central Asian	1 (0.1%)	0 (0.0%)	1 (0.1%)
Mixed ethnicity	34 (4.1%)	10 (4.9%)	24 (3.8%)
Other	1 (0.1%)	0 (0.0%)	1 (0.1%)
Country
Canada	11 (1.3%)	2 (1.0%)	9 (1.4%)
Jamaica	2 (0.2%)	1 (0.5%)	1 (0.2%)
Puerto Rico	2 (0.2%)	1 (0.5%)	1 (0.2%)
Trinidad and Tobago	1 (0.1%)	0 (0.0%)	1 (0.2%)
United Kingdom	8 (1.0%)	2 (1.0%)	6 (1.0%)
United States	808 (97.1%)	197 (97.0%)	611 (97.1%)

Table 2.

Pretest Clinical Characteristics of Completers and Dropouts.

Characteristic	n (%) or M (SD)
Characteristic	Total Sample	Completers	Dropouts
Diagnosis	(n = 832)	(n = 203)	(n = 629)
Agoraphobia	4 (0.5%)	2 (1.0%)	2 (0.3%)
Anxiety	521 (62.6%)	125 (61.6%)	396 (63.0%)
Attention deficit hyperactivity disorder	9 (1.1%)	1 (0.5%)	8 (1.3%)
Autism spectrum disorder	7 (0.8%)	1 (0.5%)	6 (1.0%)
Bipolar disorder	217 (26.1%)	36 (17.7%)	181 (28.8%)
Depression (unspecified)	2 (0.2%)	1 (0.5%)	1 (0.2%)
Eating disorder	78 (9.4%)	19 (9.4%)	59 (9.4%)
Gender dysphoria	1 (0.1%)	1 (0.5%)	0 (0.0%)
Major depressive disorder	632 (76%)	174 (85.7%)	458 (72.8%)
Obsessive compulsive disorder	116 (13.9%)	28 (13.8%)	88 (14.0%)
Panic disorder	129 (15.5%)	32 (15.8%)	97 (15.4%)
Persistent depressive disorder	1 (0.1%)	0 (0.0%)	1 (0.2%)
Personality disorder	54 (6.5%)	11 (5.4%)	43 (6.8%)
Posttraumatic stress disorder	150 (18.0%)	45 (22.2%)	105 (16.7%)
Premenstrual dysphoric disorder	1 (0.1%)	0 (0.0%)	1 (0.2%)
Psychosis (unspecified)	1 (0.1%)	1 (0.5%)	0 (0.0%)
Schizoaffective disorder	19 (2.3%)	5 (2.5%)	14 (2.2%)
Schizophrenia	27 (3.2%)	4 (2.0%)	23 (3.7%)
Social phobia	2 (0.2%)	1 (0.5%)	1 (0.2%)
Substance use	65 (7.8%)	12 (5.9%)	53 (8.4%)
Unknown	2 (0.2%)	0 (0.0%)	2 (0.3%)
Psychotic symptoms	(n = 832)	(n = 203)	(n = 629)
At least one symptom	617 (74.2%)	140 (69.0%)	477 (75.8%)
Auditory hallucinations	157 (18.9%)	33 (16.3%)	124 (19.7%)
Persecutory delusions	436 (52.4%)	103 (50.7%)	333 (52.9%)
“Reading minds” or thought insertion/withdrawal	112 (13.5%)	21 (10.3%)	91 (14.5%)
Delusions of reference	165 (19.8%)	39 (19.2%)	126 (20.0%)
Grandiose delusions	103 (12.4%)	24 (11.8%)	79 (12.6%)
Last psychotic episode	(n = 616)	(n = 143)	(n = 477)
Within the last week	219 (35.5%)	48 (33.6%)	171 (35.8%)
Within the last month	131 (21.3%)	26 (18.2%)	106 (22.2%)
Within the last 3 months	92 (14.9%)	23 (16.1%)	71 (14.9%)
Within the last year	73 (11.9%)	15 (10.5%)	59 (12.4%)
More than 1 year ago	87 (14.1%)	31 (21.7%)	56 (11.7%)
Not sure	14 (2.3%)	0 (0.0%)	14 (2.9%)
Last seen by a mental health professional	(n = 828)	(n = 203)	(n = 629)
Within the last year	618 (74.3%)	151 (74.4%)	467 (74.2%)
More than a year ago	171 (20.6%)	43 (21.2%)	128 (20.3%)
Never	39 (4.7%)	9 (4.4%)	30 (4.8%)
Missing	4 (0.5%)	0 (0.0%)	4 (0.6%)
Currently seeing mental health professional	(n = 618)	(n = 151)	(n = 467)
Yes	451 (73.0%)	106 (70.2%)	345 (73.9%)
No	163 (26.4%)	42 (27.8%)	121 (25.9%)
Not sure	4 (0.6%)	3 (2.0%)	1 (0.2%)
Taking psychotropic medication	(n = 828)	(n = 203)	(n = 629)
Yes	480 (57.7%)	115 (56.7%)	365 (58.0%)
No	340 (40.9%)	84 (41.4%)	256 (40.7%)
Not sure	8 (1.0%)	4 (2.0%)	4 (0.6%)
Missing	4 (0.5%)	0 (0.0%)	4 (0.6%)
Number of psychiatric emergency visits	(n = 821)	(n = 203)	(n = 629)
One time	188 (22.6%)	48 (23.6%)	140 (22.3%)
Two times	107 (12.9%)	20 (9.9%)	87 (13.8%)
Three or more times	98 (11.8%)	26 (12.8%)	72 (11.4%)
Not sure	5 (0.6%)	1 (0.5%)	4 (0.6%)
Never	423 (50.8%)	108 (53.2%)	315 (50.1%)
Missing	11 (1.3%)	0 (0.0%)	11 (1.7%)
Last psychiatric emergency	(n = 397)	(n = 95)	(n = 302)
Within the last month	17 (4.2%)	3 (3.1%)	14 (4.6%)
Within the last 6 months	59 (14.9%)	11 (11.3%)	48 (15.9%)
Within the last year	61 (15.4%)	10 (10.3%)	51 (16.9%)
More than a year ago	258 (65.0%)	70 (72.2%)	188 (62.3%)
Not sure	2 (0.5%)	1 (1.0%)	1 (0.3%)
Overall physical health	(n = 822)	(n = 203)	(n = 629)
Poor	92 (11.1%)	32 (15.8%)	60 (9.5%)
Fair	319 (38.3%)	73 (36.0%)	246 (39.1%)
Good	338 (40.6%)	80 (39.4%)	258 (41.0%)
Excellent	73 (8.8%)	18 (8.9%)	55 (8.7%)
Missing	10 (1.2%)	0 (0.0%)	10 (1.6%)

The average character count for the treatment group’s writing samples (M = 688.21, SD = 304.75) did not differ significantly from the control group’s writing samples (M = 651.29, SD = 336.95); t(201) = −0.82, p = .41. In addition, participants in the treatment and control groups did not differ significantly in how long they took to complete the experiment; t(201) = 0.45, p = .66. Due to technical complications, not all participants completed the interventions within 24 hr after receiving the invitation for the next wave. Some participants took up to 17 days to complete the writing interventions which were intended to last a total of three consecutive days, and participants took between 30 and 45 days to complete the entire experiment.

Hypothesis 1: It was predicted that the treatment group’s self-compassion scores (as measured by the SCS-SF) would show a greater increase from pretest to posttest and 1-month follow-up compared to the control group. This hypothesis was not supported.

A repeated measure ANOVA comparing self-compassion scores at pretest and posttest scores yielded a significant within-group effect, F(1, 201) = 30.39, p < .01). Self-compassion increased significantly from pretest to posttest for both the treatment and control groups with a large effect size (η² = .13), indicating the increases were both substantial (Cohen, 1988). A repeated measure ANOVA comparing self-compassion scores at posttest and 1-month follow-up also yielded a significant within-group effect; F(1, 201) = 13.65, p < .01. Self-compassion increased significantly from posttest to 1-month follow-up for both the treatment and control groups, with a medium effect size (η² = .06), indicating that the increases were both substantial. Interaction effects and between-group effects were not significant.

Hypothesis 2b: It was predicted that the treatment group’s proactive coping scores would show a greater increase from pretest to posttest and 1-month follow-up compared to the control group. This hypothesis was not supported.

A repeated measures ANOVA comparing proactive coping scores at pretest and posttest scores yielded a significant within-group effect; F(1, 201) = 18.22, p < .01. Proactive coping increased significantly from pretest to posttest for both the treatment and control groups, with a medium effect size (η² = .08), indicating the increases were both substantial. A repeated measure ANOVA comparing posttest and 1-month follow-up scores also yielded a significant within-group effect; F(1, 201) = 11.74, p < .01. Proactive coping increased significantly from posttest to 1-month follow-up for both the treatment and control groups, with a medium effect size (η² = .06), indicating both increases were substantial. Interaction effects and between-group effects were not significant.

Hypothesis 3: It was predicted that the treatment group’s depression scores (as measured by the RDQ [modified]) would show a greater decrease from pretest to posttest and 1-month follow-up compared to the control group. This hypothesis was not supported.

A repeated measure ANOVA comparing RDQ (modified) scores at pretest and posttest scores yielded a significant within-group effect; F(1, 201) = 91.63, p < .01. Depression decreased significantly from pretest to posttest for both the treatment and control groups, with a large effect size (η² = .31), indicating the decreases were both substantial. A repeated measures ANOVA comparing RDQ (modified) scores at posttest and 1-month follow-up also yielded a significant within-group effect; F(1, 201) = 11.64, p < .01. Depression decreased significantly from posttest to 1-month follow-up for both the treatment and control groups, with a medium effect size (η² = .06), indicating the decreases were both substantial. Interaction effects and between-group effects were not significant.

Hypothesis 4: It was predicted that the treatment group’s physical symptom scores would show a greater decrease from pretest to posttest and 1-month follow-up compared to the control group. This hypothesis was not supported.

A repeated measures ANOVA comparing CHIPS scores at pretest and posttest scores yielded a significant within-group effect; F(1, 201) = 75.74, p < .01. Physical symptoms decreased significantly from pretest to posttest for both the treatment and control groups, with a large effect size (η² = .27), indicating both the decreases were substantial. A repeated measures ANOVA comparing CHIPS scores at posttest and 1-month follow-up scores did not yield any effects. Physical symptoms did not decrease significantly from posttest to 1-month follow-up. Interaction effects and between-group effects were not significant.

Hypothesis 5: It was predicted that the treatment group’s positive affect scores (as measured by the PANAS-PA) would show a greater increase after each wave’s writing intervention than the control group. This hypothesis was not fully supported.

Repeated measures ANOVAs were performed comparing PANAS-PA preintervention and postintervention scores for each of the three writing interventions administered at Wave 1, Wave 2, and Wave 3. A repeated measures ANOVA comparing the PANAS-PA preintervention and postintervention scores for Wave 1 yielded a significant interaction effect; F(1, 201) = 26.68, p < .01, with a large effect size (η² = .12). Post hoc testing was completed. Pairwise comparisons using the LSD (Least Significant Difference) post hoc criterion for significance showed the mean difference for the treatment group was significant (M = 2.60, p < .01), and the mean difference for the control group was not significant (M = 0.39, p = .196). Therefore, only the treatment group’s positive affect increased significantly from preintervention testing to postintervention testing during Wave 1.

A repeated measures ANOVA comparing PANAS-PA preintervention and postintervention scores for Wave 2 yielded a significant within-group effect; F(1, 201) = 15.47, p = .000. Both the treatment group and control group significantly increased in positive affect from preintervention testing to postintervention testing for Wave 2. This result had a medium effect size (η² = .07), indicating that both increases were substantial.

Similar to the Wave 1 analyses, a repeated measures ANOVA comparing PANAS-PA preintervention and postintervention scores for Wave 3 yielded a significant interaction effect; F(1, 201) = 4.06, p = .045, with a small effect size (η² = .02). Post hoc testing was completed. Pairwise comparisons using the LSD post hoc criterion for significance showed the mean difference for the treatment group was significant (M = 1.55, p < .01), and the mean difference for the control group was also significant (M = 0.66, p = .036). Therefore, positive affect increased significantly only for the treatment group from preintervention testing to postintervention testing for Wave 3.

Table 3 illustrates mean changes in self-compassion, proactive coping, and depression from pretest to posttest and 1-month follow-up. Mauchly’s test indicated that the assumption of sphericity had been violated for self-compassion, depression, and physical symptoms. Therefore, the Huynh–Feldt correction was taken for these measures. The results of the repeated measures ANOVA for positive and negative affect are shown in Table 4. In addition to a significant within-subjects effect for positive effect across Waves 1–3, there were meaningful Time × Group interactions for positive affect at Wave 1 (F = 26.68, p < .001) and at Wave 3 (F = 4.06, p < .05), demonstrating the treatment group experienced greater positive affect.

Table 3.

Descriptive Statistics for Pretest, Posttest, and 1-Month Follow-Up Scores for SCS-SF, PCS, RDQ (Modified), and CHIPS.

Measure	Group	N	Pretest		Posttest		1-Month
Measure	Group	N	M	SD	M	SD	M	SD
SCS-SF	Control	102	2.31	0.63	2.50	0.69	2.57	0.75
	Treatment	101	2.27	0.65	2.42	0.77	2.55	0.74
PCS	Control	102	35.73	7.20	36.78	7.18	37.88	7.02
	Treatment	101	33.85	7.52	35.36	8.36	36.47	8.15
RDQ	Control	102	77.55	14.32	71.85	15.19	69.41	17.19
	Treatment	101	81.01	13.33	75.40	15.91	72.22	15.95
CHIPS	Control	102	74.58	24.27	67.60	25.44	66.38	26.63
	Treatment	101	74.82	22.64	68.10	23.58	65.67	22.71

Note. SCS-SF = Self-Compassion Scale, Short Form; PCS = Proactive Coping Scale; CHIPS = The Cohen–Hoberman Inventory of Physical Symptoms.

Table 4.

Positive and Negative Affect—Mean Scores and Repeated Measures Analyses of Variance.

Assessment Instrument		Treatment		Control		F		Effect Size (η²)
Assessment Instrument		Prewriting	Postwriting	Prewriting	Postwriting	Within Subjects	Time × Group	Effect Size (η²)
PANAS-PA	Wave 1	9.87 (3.3)	12.48 (3.9)	10.85 (4.4)	11.25 (4.9)	48.95***	26.68***	.12
	Wave 2	11.19 (4.6)	12.38 (4.3)	10.81 (3.9)	11.63 (4.7)	15.46***	0.54
	Wave 3	10.95 (4.3)	12.50 (4.6)	11.02 (4.4)	11.68 (4.9)	24.99***	4.06*	.02
PANAS-NA	Wave 1	10.77 (4.8)	9.51 (4.4)	10.13 (4.9)	8.62 (4.5)	25.40**	0.21
	Wave 2	8.79 (4.8)	8.67 (4.3)	7.81 (3.8)	7.36 (3.9)	1.60	0.55
	Wave 3	8.79 (4.5)	8.55 (4.4)	7.72 (3.6)	7.75 (4.2)	0.21	0.42

Note. PANAS-PA = The Positive and Negative Affect Schedule-Positive Affect; PANAS-NA = The Positive and Negative Affect Schedule-Negative Affect

*p < .05. **p < .01. ***p < .001.

The results for PANAS-NA found a significant within-subjects effect for Wave 1 (F = 25.40, p < .001), showing that both groups decreased in negative affect. There were not any Time × Group interactions for PANAS-NA across any of the waves.

Discussion and Applications to Practice

The analysis of outcome measures for the entire sample of completers indicates that overall, both the treatment and control groups improved significantly over time. Because individuals with mental illness are more vulnerable to stress (Anakwenze & Zuberi, 2013) and trauma (Kilgus et al., 2015; Rudnick & Lundberg, 2012) and write with more negative emotion than nonclinical populations (Fineberg et al., 2016), it may have been harder for participants in the current study to reassess and learn from their stressful experiences, just as it may have been for participants with mental illness to learn from expressive writing. This may have resulted in less improvement or made it more difficult to see significant changes in such a short period of time. In addition, some participants did not complete the 3-day intervention over three consecutive days as directed, so the intervention may not have been concentrated or powerful enough to have a true effect.

Alternatively, both treatment and control participants may have benefited from the writing tasks. Writing about how one spends one’s time may have helped some individuals with mental illness who were overwhelmed with stress build a sense of self-efficacy, improve confidence, and feel better about themselves. This, in turn, may have helped the individuals be more self-compassionate and less self-critical. Anecdotal evidence that writing about everyday activities in a daily journal may help individuals with psychosis create a more coherent sense of self (Stone, 2005) supports this possibility. The symptoms and needs of individuals with mental illness vary over time, and individuals vary in how prepared they are to change and how interested they are in treatment (Prochaska & DiClemente, 1983). From this perspective, it is not surprising that writing about how one spends one’s time and writing about a difficult experience with self-compassion could fill different needs at different times for a group of people who share a clinical diagnosis. For example, research by Himle and Thyer (1989) found that having a client write out a list of intrusive thoughts leads to a reduction of subjective units of distress.

The one outcome that did differentiate between the treatment and control group was positive affect. It may be that this measure of mood intensity was more sensitive to the immediate impact of writing when compared to the other measures of outcomes. This study, like most of the previous self-compassion writing studies and expressive writing studies, did not include a “no treatment” group. Future writing studies should include a “no treatment” group in addition to a neutral writing condition to help tease apart the effects of the treatment.

Notably, in the current study, both groups were encouraged to write continuously and given multiple examples of what to do if they felt stuck, and both the treatment and control groups wrote on average approximately the same number of characters. There are no known expressive writing meta-analyses that explore differences in word count between the treatment and control groups (Pennebaker, 2017); however, in general, control participants tend to write less than treatment participants (Pennebaker, 2017). Niles et al. (2016) found that one of the mechanisms of change for writing interventions is the level of detail with which participants write. This suggests that the more participants write, the more their writing helps build motivation and positive affect, and the more likely it becomes that participants will experience improvement, whether they are writing about how they spend their time or writing with self-compassion about a difficult event.

The current study is limited in that the sample may not be representative of all individuals with mental illness. This study was not preregistered in a clinical trials registry. Moreover, recruitment was limited to individuals signed up to work on the Amazon MTurk platform, and such individuals may be experienced at psychological testing (Necka et al., 2016). In addition, all participants self-identified as having a mental illness and selected which mental illness(es) they had from a list and/or wrote in a diagnosis. While our estimate of the percentage of individuals with mental illness on Amazon MTurk is similar to some national prevalence statistics (Kessler et al., 2005), participants were not screened by a clinical professional, and a considerable number of participants may have misrepresented themselves so they could participate in the study and get paid (Sharpe et al., 2017). Future online studies may prevent this by conducting the study on alternative crowdsourcing platforms that have qualifications for mental illness so researchers can select a setting to automatically prescreen for this qualification, or by putting up a separate, short, prescreen study on the MTurk platform that screens out individuals without mental illness (Sharpe et al., 2017).

In addition, as mentioned above, not all participants completed the interventions within 24 hr after receiving the invitation for the next wave. While the email invitations to the next wave were designed with the intention to make the links active for 24 hrs only, this proved not to be possible. This technical difficulty may have had the effect of diluting the power of the intervention. Additionally, participants were asked to time themselves, and the researchers did not have an accurate way of estimating how much time participants wrote for. Other limitations included the following: The study did not control for differences in diagnostic criteria based on country, racial or ethnic differences, severity of mental illness, or dual diagnosis.

This first study of self-compassion writing for individuals with different types of mental illness showed that both writing how one spends one’s time and writing with self-compassion might improve mental and physical health in this population. More research is needed to determine how different writing interventions can better “meet clients where they are” (Hepworth et al., 2013) to address their specific needs.

Footnotes

Authors’ Note

Debra Urken completed this research as part of her MSW thesis at Arizona State University.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Craig Winston LeCroy

References

Albertson

E. R.

Neff

K. D.

Dill-Shackleford

K. E.

(2014). Self-compassion and body dissatisfaction in women: A randomized controlled trial of a brief meditation intervention. Mindfulness, 6(3), 444–454. https://doi.org/10.1007/s12671-014-0277-3

Anakwenze

Zuberi

(2013). Mental health and poverty in the inner city. Health & Social Work, 38(3), 147–157.

Arch

J. J.

Brown

K. W.

Dean

D. J.

Landy

L. N.

Brown

K. D.

Laudenslager

M. L.

(2014). Self-compassion training modulates alpha-amylase, heart rate variability, and subjective responses to social evaluative threat in women. Psychoneuroendocrinology, 42, 49–58. https://doi.org/10.1016/j.psyneuen.2013.12.018

Baikie

K. A.

Geerligs

Wilhelm

(2012). Expressive writing and positive writing for participants with mood disorders: An online randomized controlled trial. Journal of Affective Disorders, 136(3), 310–319.

Breines

J. G.

Chen

(2012). Self-compassion increases self-improvement motivation. Personality & Social Psychology Bulletin, 38(9), 1133–1143. https://doi.org/10.1177/0146167212445599

Buhrmester

Kwang

Gosling

S. D.

(2011). Amazon’s Mechanical Turk: A new source of inexpensive, yet high-quality, data? Perspectives on Psychological Science, 6(1), 3–5. https://doi.org/10.1177/1745691610393980

Burton

C. M.

Kiing

L. A.

(2007). The health benefits of writing about positive experiences: The role of broadened cognition. Psychology & Health, 24, 867–879.

Casler

Bickel

Hackett

(2013). Separate but equal? A comparison of participants and data gathered via Amazon’s MTurk, social media, and face-to-face behavioral testing. Computers in Human Behavior, 29(6), 2156–2160. https://doi.org/10.1016/j.chb.2013.05.009

Chandler

Shapiro

(2016). Conducting clinical research using crowdsourced convenience samples. Annual Review of Clinical Psychology, 12, 53–81. https://doi.org/10.1146/annurev-clinpsy-021815-093623

10.

Cohen

(1988). Statistical power analysis for the behavioral sciences. Routledge.

11.

Cohen

Hoberman

H. M.

(1983). Positive events and social supports as buffers of life change stress. Journal of Applied Social Psychology, 13(2), 99–125.

12.

Compassion Institute (2017). Compassion cultivation training. https://www.compassioninstitute.com/

13.

Diedrich

Grant

Hofmann

S. G.

Hiller

Berking

(2014). Self-compassion is an emotion regulation strategy in major depressive disorder. Behaviour Research and Therapy, 58, 43–51.

14.

Fineberg

S. K.

Leavitt

Deutsch-Link

Dealy

Landry

C. D.

Pirruccio

Shea

Trent

Cecchi

Corlett

P. R.

(2016). Self-reference in psychosis and depression: A language marker of illness. Psychological Medicine, 46(12), 2605–2615.

15.

Frattaroli

(2006). Experimental disclosure and its moderators: A meta-analysis. Psychological Bulletin, 132(6), 823.

16.

Gilbert

(2009). The nature and basis for compassion focused therapy. Hellenic Journal of Psychology, 6(20), 273–291.

17.

Goetz

J. L.

Simon-Thomas

(2017). The landscape of compassion: Definitions and scientific approaches. In Seppälä

E. M.

Simon-Thomas

Brown

S. L.

Worline

M. C.

Cameron

C. D.

Doty

J. R.

(Eds.), The Oxford handbook of compassion science (pp. 3–15). Oxford University Press.

18.

Greenberg

(2008). Emotion and cognition in psychotherapy: The transforming power of affect. Canadian Psychology/Psychologie Canadienne, 49(1), 49.

19.

Greenglass

Schwarzer

Jakubiec

Fiksenbaum

Taubert

(1999, July 12–14). The Proactive Coping Inventory (PCI): A multidimensional research instrument [Paper presentation] . Stress and Anxiety Research Society (STAR) 20th International Conference, Cracow, Poland.

20.

Helm

(2016). The effect of a brief self-compassion intervention on emotion regulation in individuals with a generalized anxiety disorder [Doctoral dissertation]. Alliant International University.

21.

Hepworth

D. H.

Rooney

R. H.

Rooney

G. D.

Strom-Gottfried

(2013). Direct social work practice: Theory and skills. Brooks.

22.

Himle

Thyer

B. A.

(1989). Clinical social work and obsessive compulsive disorder. Behavior Modification, 13(4), 459–470.

23.

Holschuh

(2007). Humboldt county homeless adult study. Department of Social Work. Arizona State University.

24.

Hope

Koestner

Milyavskaya

(2014). The role of self-compassion in goal pursuit and well-being among first-year university students. Self and Identity, 13(5), 579–593.

25.

Imrie

Troop

N. A.

(2012). A pilot study on the effects and feasibility of compassion-focused expressive writing in Day Hospice patients. Palliative & Supportive Care, 10(2), 115–122.

26.

Jazaieri

McGonigal

Jinpa

Doty

J. R.

Gross

J. J.

Goldin

P. R.

(2014). A randomized controlled trial of compassion cultivation training: Effects on mindfulness, affect, and emotion regulation. Motivation and Emotion, 38(1), 23–35.

27.

Johnson

E. A.

O’Brien

K. A.

(2013). Self-compassion soothes the savage ego-threat system: Effects on negative affect, shame, rumination, and depressive symptoms. Journal of Social and Clinical Psychology, 32(9), 939–963.

28.

Jourard

S. M.

(1971). Self-disclosure: An experimental analysis of the transparent self. Wiley.

29.

Kannan

Levitt

H. M.

(2013). A review of client self-criticism in psychotherapy. Journal of Psychotherapy Integration, 23(2), 166.

30.

Kercher

(1992). Assessing subjective well-being in the old-old: The panas as a measure of orthogonal dimensions of positive and negative affect. Research on Aging, 14(2), 131–168.

31.

Kessler

R. C.

Berglund

Demler

Jin

Merikangas

K. R.

Walters

E. E.

(2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62(6), 593–602.

32.

Kilgus

M. D.

Maxmen

J. S.

Ward

N. G.

(2015). Essential psychopathology & its treatment. WW Norton & Company.

33.

Koschwanez

H. E.

Kerse

Darragh

Jarrett

Booth

R. J.

Broadbent

(2013). Expressive writing and wound healing in older adults: A randomized controlled trial. Psychosomatic Medicine, 75(6), 581–590. https://doi.org/10.1097/PSY.0b013e31829b7b2e

34.

Leary

M. R.

Tate

E. B.

Adams

C. E.

Batts Allen

Hancock

(2007). Self-compassion and reactions to unpleasant self-relevant events: The implications of treating oneself kindly. Journal of Personality and Social Psychology, 92(5), 887.

35.

MacBeth

Gumley

(2012). Exploring compassion: A meta-analysis of the association between self-compassion and psychopathology. Clinical Psychology Review, 32(6), 545–552. https://doi.org/10.1016/j.cpr.2012.06.003

36.

Marsiglia

F. F.

Kulis

(2015). Diversity, oppression & change. (2nd ed.). Lyceum Books.

37.

Matthiesen

Klonoff-Cohen

Zachariae

Jensen-Johansen

M. B.

Nielsen

B. K.

Frederiksen

Schmidt

Ingerslev

H. J.

(2012). The effect of an expressive writing intervention (EWI) on stress in infertile couples undergoing assisted reproductive technology (ART) treatment: A randomized controlled pilot study. British Journal of Health Psychology, 17(2), 362–378.

38.

National Survey on Drug Use and Health. (2017). CBHSQ data review. Substance Abuse and Mental Health Services Administration.

39.

Necka

E. A.

Cacioppo

Norman

G. J.

Cacioppo

J. T.

(2016). Measuring the prevalence of problematic respondent behaviours among MTurk, campus, and community participants. PLOS ONE, 11(6), e0157732.

40.

Neff

K. D.

(2003a). The development and validation of a scale to measure self-compassion. Self and Identity, 2(3), 223–250.

41.

Neff

K. D.

(2003b). Self-compassion: An alternative conceptualization of a healthy attitude toward oneself. Self and Identity, 2(2), 85–101.

42.

Neff

K. D.

Germer

C. K.

(2013). A pilot study and randomized controlled trial of the mindful self-compassion program. Journal of Clinical Psychology, 69(1), 28–44.

43.

Neff

K. D.

Hsieh

Y. P.

Dejitterat

(2005). Self-compassion, achievement goals, and coping with academic failure. Self and Identity, 4(3), 263–287.

44.

Niles

A. N.

Byrne Haltom

K. E.

Lieberman

M. D.

Hur

Stanton

A. L.

(2016). Writing content predicts benefit from written expressive disclosure: Evidence for repeated exposure and self-affirmation. Cognition and Emotion, 30(2), 258–274.

45.

Niles

A. N.

Haltom

K. E.

Mulvenna

C. M.

Lieberman

M. D.

Stanton

A. L.

(2014). Randomized controlled trial of expressive writing for psychological and physical health: The moderating role of emotional expressivity. Anxiety Stress Coping, 27(1), 1–17.

46.

Odou

Brinker

(2013). Exploring the relationship between rumination, self-compassion, and mood. Self and Identity, 13(4), 449–459. https://doi.org/10.1080/15298868.2013.840332

47.

Park-Lee

Lipari

R. N.

Hedden

S. L.

Kroutil

L. A.

Porter

J. D.

(2017). Receipt of services for substance use and mental health issues among adults: Results from the 2016 National Survey on Drug Use and Health. In CBHSQ data review. Substance Abuse and Mental Health Services Administration. https://www.samhsa.gov/data/sites/default/files/NSDUH-DR-FFR2-2016/NSDUH-DR-FFR2-2016.htm

48.

Pascoe

P. E.

(2017). Using patient writings in psychotherapy: Review of evidence for expressive writing and cognitive-behavioural writing therapy. American Journal of Psychiatry Residents’ Journal. https://doi.org/10.1176/appi.ajp-rj.2016.110302

49.

Pennebaker

J. W.

(2004). Theories, therapies, and taxpayers: On the complexities of the expressive writing paradigm. Clinical Psychology: Science and Practice, 11(2), 138–142.

50.

Pennebaker

J. W.

(2017). Expressive writing in psychological science. Perspectives on Psychological Science, 13(2). http://https://doi.org/10.1177/1745691617707315

51.

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.

52.

Petrie

K. J.

Fontanilla

Thomas

M. G.

Booth

R. J.

Pennebaker

J. W.

(2004). Effect of written emotional expression on immune function in patients with human immunodeficiency virus infection: A randomized trial. Psychosomatic Medicine, 66(2), 272–275. https://doi.org/10.1097/01.psy.0000116782.49850.d3

53.

Prochaska

J. O.

DiClemente

C. C.

(1983). Stages and processes of self-change of smoking: Toward an integrative model of change. Journal of Consulting and Clinical Psychology, 51(3), 390.

54.

Raes

Pommier

Neff

K. D.

Van Gucht

(2011). Construction and factorial validation of a short form of the self-compassion scale. Clinical Psychology & Psychotherapy, 18(3), 250–255. https://doi.10.1002/cpp.702

55.

Rudnick

Lundberg

(2012). The stress-vulnerability model of schizophrenia: A conceptual analysis and selective review. Current Psychiatry Reviews, 8(4), 337–341.

56.

Shahar

Henrich

C. C.

(2013). Axis of criticism model (ACRIM): An integrative conceptualization of person–context exchanges in vulnerability to adolescent psychopathology. Journal of Psychotherapy Integration, 23(3), 236.

57.

Shapira

L. B.

Mongrain

(2010). The benefits of self-compassion and optimism exercises for individuals vulnerable to depression. The Journal of Positive Psychology, 5(5), 377–389.

58.

Sharpe Wessling

Huber

Netzer

(2017). MTurk character misrepresentation: Assessment and solutions. Journal of Consumer Research, 44(1), 211–230.

59.

Sloan

D. M.

Marx

B. P.

(2004). Taking pen to hand: Evaluating theories underlying the written disclosure paradigm. Clinical Psychology: Science and Practice, 11(2), 121–137.

60.

Smeets

Neff

Alberts

Peters

(2014). Meeting suffering with kindness: Effects of a brief self-compassion intervention for female college students. Journal of Clinical Psychology, 70(9), 794–807.

61.

Stone

(2005). Speaking of every day: Psychosis and writing. The University of Huddersfield.

62.

Trompetter

H. R.

de Kleine

Bohlmeijer

E. T.

(2017). Why does positive mental health buffer against psychopathology? An exploratory study on self-compassion as a resilience mechanism and adaptive emotion regulation strategy. Cognitive Therapy and Research, 41(3), 459–468. https://doi.org/10.1007/s10608-016-9774-0

63.

Urken

(2018). The effects of an internet-based self-compassion writing intervention for adults with mental illness [Master’s thesis]. Arizona State University.

64.

Waite

Knight

M. T.

Lee

(2015). Self-compassion and self-criticism in recovery in psychosis: An interpretative phenomenological analysis study. Journal of Clinical Psychology, 71(12), 1201–1217.

65.

Wong

C. C.

Mak

W. W.

(2016). Writing can heal: Effects of self-compassion writing among Hong Kong Chinese college students. Asian American Journal of Psychology, 7(1), 74.

66.

World Health Organization. (2017). Scalable psychological interventions for people in communities affected by adversity: A new area of mental health and psychosocial work at WHO (No. WHO/MSD/MER/17.1). http://apps.who.int/iris/handle/10665/254581

67.

Ziemer

(2014). Self-compassion and the expressive writing paradigm: A study of therapeutic effectiveness for chronic pain [Doctoral dissertation]. https://doi.org/10.13016/M2NC70

68.

Zimmerman

Martinez

J. H.

Attiullah

Friedman

Toba

Boerescu

D. A.

Ragheb

(2013). A new type of scale for determining remission from depression: The Reduction from Depression Questionnaire. Journal of Psychiatric Research, 47(1), 78–82.