This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Mental Health, is properly cited. The complete bibliographic information, a link to the original publication on http://mental.jmir.org/, as well as this copyright and license information must be included.
Self-guided mental health interventions that are disseminated via the Web have the potential to circumvent barriers to treatment and improve public mental health. However, self-guided interventions often fail to attract consumers and suffer from user nonadherence. Uptake of novel interventions could be improved by consulting consumers from the beginning of the development process in order to assess their interest and their preferences. Interventions can then be tailored using this feedback to optimize appeal.
The aim of our study was to determine the level of public interest in a new mental health intervention that incorporates elements of self-help and peer counseling and that is disseminated via a Web-based training course; to identify predictors of interest in the program; and to identify consumer preferences for features of Web-based courses and peer support programs.
We surveyed consumers via Amazon’s Mechanical Turk to estimate interest in the self-help and peer support program. We assessed associations between demographic and clinical characteristics and interest in the program, and we obtained feedback on desired features of the program.
Overall, 63.9% (378/592) of respondents said that they would try the program; interest was lower but still substantial among those who were not willing or able to access traditional mental health services. Female gender, lower income, and openness to using psychotherapy were the most consistent predictors of interest in the program. The majority of respondents, although not all, preferred romantic partners or close friends as peer counselors and would be most likely to access the program if the training course were accessed on a stand-alone website. In general, respondents valued training in active listening skills.
In light of the apparent public interest in this program, Web-disseminated self-help and peer support interventions have enormous potential to fill gaps in mental health care. The results of this survey can be used to inform the design of such interventions.
Self-guided, Web-based health interventions have great promise in reaching consumers and improving public health. Web-based
However, self-guided, Web-based interventions come with their own challenge: they need to be appealing enough to motivate consumer engagement. Internet-based interventions often report low user adherence, with many users discontinuing the intervention prematurely [
In response to these calls, our research team sought consumer input on a new intervention that is called “Crowdsourcing Mental Health” (CMH) because it distributes the task of providing mental health care among the general public. In CMH, 2 individuals who are already acquainted with each other (eg, friends, coworkers) agree to participate in the program as “partners.” Both partners take an asynchronous massive open online course (MOOC) that teaches 2 sets of skills: skills for talking about stressors and skills for listening to others who are stressed. After completing the course, the partners meet weekly for peer counseling, taking turns in the “talker” and “listener” roles. This reciprocal structure allows both partners to gain the benefits of care provision [
Stress can precipitate the onset of, or worsen, a variety of mental illnesses (eg, [
CMH and programs like it have the potential to improve public mental health in principle, but the success of such programs depends upon consumer interest and user adherence. In accordance with recommendations that consumer consultation be the first step in developing public health interventions [
The specific research questions for this investigation were as follows. Although we had tentative hypotheses for some of these questions, this study should be viewed as exploratory, not confirmatory.
Is consumer interest adequate to justify investment in development of the CMH program and interventions like it?
A major goal of the survey was to gauge general public desire to participate in a program like CMH to estimate the potential reach of this program. Deciding upon a precise threshold for “adequate” interest would require knowledge of the cost-effectiveness of this program, which does not yet exist. However, we decided a priori that it would unquestionably be worthwhile to pursue the development of CMH if 30% of participants endorsed willingness to try the program. This was a conservatively high bar: even if the survey overestimated the proportion of users by 90%, this would mean about 10 million persons in the United States would try the program.
Is consumer interest high enough among those who are not accessing or who do not plan to access traditional mental health care, and how does this group’s level of interest compare with that of people who do access traditional care?
Common barriers to seeking mental health help from a professional include structural or practical impediments such as cost, inconvenience, and provider unavailability, as well as attitudinal barriers such as feeling that one’s problem is not severe enough to warrant professional help or wanting to avoid stigma [
What user characteristics are associated with interest in the program?
Characterizing the potential audience is a key step in tailoring health interventions [
Some research has been conducted on predictors of attitudes toward mental health care or eHealth, but findings have not been consistent, and they may not apply to the nontraditional CMH program because it is intended to have broader appeal than traditional mental health services. Thus, further investigation is needed.
On the basis of current literature, one might expect females to be more interested than males [
Whom would prospective users most want to have as a partner?
Most existing peer support programs have paired individuals with strangers, but to our knowledge there is no empirical justification for doing so. In fact, pairing users with individuals with whom they are already acquainted may have advantages, such as eliciting more disclosure [
Which peer counseling skills would prospective users want their partners to learn? Do these skills differ from the skills they would want to learn themselves?
We also inquired about which counseling skills people wanted to learn most, as well as which counseling skills they want their partner to learn. Delivery of appropriate social support is notoriously fraught, and the support provided is often suboptimal, mismatching the situation or the recipient’s characteristics (eg, [
How do 6 possible access channels compare in likelihood of use, appeal, trustworthiness, convenience, and ease of use?
Finally, we investigated the ideal access channel, offering several options for the Web-based platform and the venues through which users would learn about the program. Respondents ranked 6 options (website, app, social media, physician, community center, and school or work), then rated each channel on attributes that have been shown to predict technology and service adoption: appeal [
Although there is evidence that these attributes are important, there is surprisingly little academic research available regarding consumer perspectives on channels of eHealth access (although it is likely that individual organizations and market research firms have collected proprietary data on related questions). With regard to Web-based platforms, we expected that respondents would report being more likely to use a stand-alone website than a mobile app or social networking site, given that a website can be accessed on a wide variety of devices; the percentage of US adults who own a laptop or desktop computer (73%) is slightly higher than those who own a smartphone (68%; [
Survey respondents were drawn from a convenience sample of potential health care consumers: users of Amazon’s Mechanical Turk (MTurk), a work crowdsourcing website where any individual older than 18 years can complete simple tasks, including psychological experiments, for pay [
Access to the survey was limited to MTurk users with Internet protocol (IP) addresses in the United States. Users with the same IP address were prevented from completing the survey to avoid duplicate entries. We required MTurk users to have at minimum a 95% approval rate and at least 50 prior tasks completed on the MTurk website in order to access the survey.
For their data to be included in the analyses, participants need to correctly answer 3 simple comprehension questions after reading a description of the program. Because these questions were extremely straightforward (eg, “What do you learn from the training: How to speak another language, how to be helpful when listening and talking to your partner, or how to improve your memory?”), incorrect responses were indicative of inattention or gross misunderstanding of the description, rendering respondents’ reactions to the program invalid.
Because the majority of survey content was unique to this project and therefore had not been validated in prior research, we pilot-tested and refined the survey through cognitive interviewing [
The survey began with a description of the program accompanied by stick figure illustrations. The description explained that, in the program, 2 people who were already acquainted would take a Web-based course in scientifically supported talking and listening skills, then would meet weekly in person to put the skills to use, and it listed a variety of potential benefits of participating. The complete description appears in
Interest in the program was assessed in 3 ways. Respondents indicated intention to try the program on a dichotomous yes or no item (“Would you try the program?”) and a 4-option forced-choice item. Additionally, participants responded to 12 items measuring attitudes toward the program (eg, “This program could help solve a problem I have”), including 5 reverse-scored items (eg, “This program would be a waste of my time”), rated on a 7-point Likert scale with anchors from “strongly agree” to “strongly disagree.” The mean of the Likert-rated scale items was computed to create a composite continuous attitude score. This attitude scale was internally consistent (coefficient alpha=.94). Finally, we gave respondents the opportunity to provide their email addresses in order to receive more information about the program with the assumption that this, as a behavioral indicator of interest, required more commitment than a verbal claim of hypothetical willingness to try the program.
The survey also included items assessing desired features of the program. Participants were asked to rate their interest in learning 11 social support skills and in having an imagined partner learn the same skills. These skills were selected from those that, according to qualitative studies of social support, are frequently delivered by well-intentioned support providers, although they may or may not be perceived as helpful by support recipients [
Survey respondents also ranked 5 possible types of people they would prefer to have as a partner, 5 possible channels through which to meet with their partner, and 6 possible ways to access the course. They rated their perceptions of the possible ways to access the course on 4 semantic differentials:
Demographic characteristics and information on mental health service use were collected using straightforward, study-specific items. Rather than using total household income in subsequent analyses, we corrected for household size by dividing household income by the square root of the number of people in the household, an adjustment that assumes some economy of scale within the household, such that each additional household member costs the household less than the previous one [
Psychological distress was assessed using the Brief Symptom Inventory (BSI) [
All surveys were approved by the University of Massachusetts Amherst Institutional Review Board and were carried out in accordance with all applicable regulations. Before consenting, participants were informed about the investigator’s identity; the purpose, length, and risks and benefits of taking part in each survey; and the methods of data storage.
We addressed missing data by using the package mice [
To identify demographic and clinical predictors of interest in the program, we conducted a series of regressions for each of the 3 outcome variables: attitude score (continuous), willingness to try the program (dichotomous), and provision of an email address (dichotomous). In the first step of these exploratory analyses, predictors were entered hierarchically in blocks. This allowed for tests of the incremental predictive effect of a group of related variables; for example, all race or ethnicity dummy codes were entered in a block, enabling a test of whether race or ethnicity significantly improved model fit, which would not have been possible if all variables were entered simultaneously. We expected that some of the predictors entered in the same block would be collinear (eg, education and income), but would act as indices of the same construct (eg, socioeconomic status or SES), and we would interpret the test of whether that block improved model fit as an indicator of whether the overall construct predicted interest in the program. The order of the blocks was based on what Cohen and colleagues [
After conducting these hierarchical regressions with all variables, we then repeated the procedure, omitting any blocks of predictors that did not marginally improve model fit (
To compare participants’ interest in their partners versus themselves learning each support skill, we conducted paired
To identify whether specific potential partners, meeting channels, or course access methods were ranked significantly higher than others, we compared each pair of possible choices with a Wilcoxon signed rank test, with a Bonferroni correction for multiple comparisons.
Of the 637 MTurk users who completed the survey, 592 (92.9%) correctly answered all 3 comprehension questions and were therefore eligible for inclusion. Missing data were relatively rare; all 592 individuals responded to the majority of items such that there were no missing values for any attitude items, mental health and distress items, gender, race, marital status, or education; the maximum number of missing values for any variable was 8 (for income).
Demographic, clinical, and service use characteristics of the sample are presented in
Participant characteristics.
Characteristics | Mean (SD) or n (%) (N=592) | |
Age in years, mean (SD) | 37.37 (13.11) | |
Male | 212 (35.8) | |
Female | 375 (63.3) | |
Gender nonconforming (eg, transgender, genderqueer) | 5 (0.8) | |
White, non-Hispanic | 458 (77.4) | |
Black | 44 (7.4) | |
Hispanic | 33 (5.6) | |
Asian | 27 (4.6) | |
Native American | 6 (1.0) | |
Other race or ethnicity | 24 (4.0) | |
Married or cohabiting | 275 (46.5) | |
Never married | 256 (43.2) | |
Separated or divorced | 50 (8.4) | |
Widowed | 11 (1.9) | |
High school or less | 64 (10.8) | |
2-Year degree | 51 (8.6) | |
Some college | 177 (29.9) | |
4-Year degree | 184 (31.1) | |
Some graduate or professional school | 31 (5.2) | |
Graduate or professional degree | 85 (14.4) | |
Median household income | 42K | |
Income per person1/2 | 32.9K (22.7K) | |
Score, mean (SD) | 0.5045 (0.5985) | |
In “clinical” range, n (%) | 142 (24.0) | |
Currently in therapy | 44 (7.4) | |
Ever in therapy | 261 (44.1) | |
Would consider trying therapy | 403 (68.1) | |
Currently prescribed medication | 73 (12.3) | |
Ever prescribed medication | 175 (29.6) | |
Would consider trying medication | 362 (61.1) |
More than half the respondents (378/592, 63.9%; 95% CI 60.0% to 67.7%) indicated on the dichotomous item that they would try the program. When asked to choose from 4 options, 14.9% (88/592) indicated that they “would sign up now,” 46.5% (275/592) indicated that they “might try it in the future,” 29.2% (173/592) indicated that they “would probably not try it,” and 9.5% (56/592) indicated that they “would never try it.” For the behavioral indicator of interest, approximately one-third of the respondents (193/592, 32.6%; 95% CI 28.8% to 37.8%) volunteered their email addresses in order to request more information about the program. The median score on the continuous attitude scale was 5.1 out of a possible range of 1 to 7 (mean 4.9, SD 1.1), corresponding to an anchor of “somewhat agree” on positively worded items.
To assess whether our program would appeal to those who lacked access to or interest in traditional mental health interventions, we computed the percentages of those who had not used psychotherapy or psychiatric medication but who would try our program, as well as the percentages of those who stated that they would not be willing to use those traditional interventions but who would try our program. Of those who never accessed psychotherapy, 62.5% indicated that they would try the program, compared with 65.5% of those who had used therapy, χ21=0.4,
In the first hierarchical linear regression predicting continuous attitude score, the blocks containing race, age, and marital status failed to improve model fit even marginally. Therefore, as planned, these blocks were removed from the subsequent analyses, and the second hierarchical regression included blocks for gender, SES, psychological symptoms, past treatment use, and hypothetical treatment use. At this stage, the following variables significantly predicted more positive attitude toward CMH (at the step in which they were entered): female gender, past use of therapy, and willingness to consider using therapy. Greater psychological symptoms marginally predicted positive attitude, and greater income was marginally associated with worse attitude. Full results of these initial analyses appear in
In a simultaneous regression, the aforementioned predictors with
Simultaneous regression predicting attitude toward the Crowdsourcing Mental Health program.
Predictor | SE |
95% CI |
||
Femalea | 0.200 | 0.095 | 0.013 to 0.386 | .04 |
Incomeb | −0.055 | 0.020 | −0.095 to −0.016 | .006 |
Brief Symptom Inventory | 0.112 | 0.077 | −0.039 to 0.263 | .14 |
Ever used therapy | 0.073 | 0.097 | −0.118 to 0.263 | .45 |
Would consider therapy | 0.586 | 0.101 | 0.387 to 0.784 | <.001 |
aReference category is combined male and gender nonconforming respondents.
bUnit is income per household member1/2in US $10,000 increments.
In the first hierarchical logistic regression predicting whether respondents indicated willingness to participate in the program on the dichotomous item, the following blocks failed to marginally or significantly improve model fit: gender, marital status, and past treatment use. In the next regression, in which those blocks were removed, the block in which race or ethnicity dummy codes were entered marginally improved model fit, apparently driven by Hispanic respondents’ greater likelihood of indicating willingness to participate than non-Hispanic white respondents. All other blocks significantly improved model fit, and all predictors within those blocks were marginal or significant predictors of willingness to try the program except for willingness to consider taking psychiatric medication. See
Therefore, for the simultaneous regression (
Simultaneous regression predicting intention to try the Crowdsourcing Mental Health program.
Predictor | SE |
95% CI |
||
Hispanica | 1.102 | 0.512 | 0.100 to 2.11 | .03 |
Blacka | 0.221 | 0.356 | −0.479 to 0.921 | .53 |
Asiana | 0.682 | 0.461 | −0.224 to 1.587 | .14 |
Native Americana | −0.169 | 0.891 | −1.918 to 1.581 | .85 |
Other racea | −0.484 | 0.443 | −1.354 to 0.386 | .28 |
Age | −0.015 | 0.007 | −0.029 to −0.001 | .04 |
Incomeb | −0.087 | 0.041 | −0.167 to −0.007 | .03 |
Education | −0.123 | 0.066 | −0.255 to 0.007 | .06 |
Brief Symptom Inventory | 0.292 | 0.165 | −0.033 to 0.616 | .08 |
Would consider therapy | 0.936 | 0.194 | 0.556 to 1.317 | <.001 |
aReference category is non-Hispanic white.
bUnit is income per household member1/2in US $10,000 increments.
In the first hierarchical logistic regression predicting whether participants provided an email address to request more information, blocks including gender, race, symptoms, and hypothetical treatment use significantly improved model fit; no other blocks approached significance. In the second hierarchical logistic regression, female gender, symptoms, and openness to using psychotherapy all significantly predicted provision of an email address; Asian participants were significantly less likely to provide an email address than white participants. No other predictors approached significance. The full results of these regression analyses can be found in
In the simultaneous regression predicting email provision, willingness to consider medication and the dummy code for gender nonconforming were dropped (such that the reference category for gender was now non–female-identified rather than male). In this analysis, presented in
Simultaneous regression predicting provision of email address.
Predictor | SE |
95% CI |
||
Femalea | 0.411 | 0.195 | 0.023 to 0.794 | .04 |
Hispanicb | 0.180 | 0.387 | −0.581 to 0.939 | .64 |
Blackb | 0.466 | 0.337 | −0.196 to 1.128 | .17 |
Asianb | −1.230 | 0.642 | −2.491 to 0.031 | .06 |
Native Americanb | 1.207 | 0.933 | −0.626 to 3.040 | .20 |
Other raceb | −0.093 | 0.473 | −1.022 to 0.836 | .84 |
Brief Symptom Inventory | 0.471 | 0.149 | 0.179 to 0.764 | .002 |
Would consider therapy | 0.929 | 0.217 | 0.503 to 1.355 | <.001 |
aReference category is combined male and gender nonconforming respondents.
bReference category is non-Hispanic white.
To summarize the significant results from the final models, female gender predicted favorable attitudes and email provision; Hispanic ethnicity predicted intention to try CMH; older age was associated with lower endorsement of willingness to try CMH; lower income predicted favorable attitudes and intention to try CMH; higher psychological distress was associated with email provision; and willingness to use psychotherapy predicted all 3 dependent variables.
Most preferred peers.
Partner type | Ranked 1, n (%) (N=588) | Ranked 2, n (%) (N=588) | Mean ranka |
Romantic partner or spouse | 339 (57.7) | 88 (15.0) | 1.96 |
Close friend | 102 (17.3) | 232 (39.5) | 2.39 |
Family member | 59 (10.0) | 161 (27.4) | 3.01 |
Acquaintance | 25 (4.3) | 70 (11.9) | 3.55 |
Stranger | 63 (10.7) | 37 (6.3) | 4.08 |
aAll pairwise comparisons between ranks were significantly different,
In order to identify the skills that consumers believe would be most valuable to learn in a peer support program, the survey listed counseling skills and asked respondents to rate how important it was to them that their peer learn that skill, as well as how important it was that they themselves learn that skill, on a 1 to 7 Likert scale from “not at all important” to “extremely important.”
Importance of learning various peer counseling skills.
Peer counseling skills | Rated skill “very” or “extremely” important, n (%) (N=592) | Mean difference in ratings | |||
Want peer to learn | Want self to learn | ||||
How to genuinely listen | 474 (80.6) | 366 (61.9) | 0.58 | 11.06 | <.001 |
How to pay attention | 439 (74.6) | 351 (59.3) | 0.51 | 9.37 | <.001 |
How to show understanding | 416 (70.7) | 377 (63.7) | 0.23 | 4.56 | <.001 |
How to empathize | 403 (68.5) | 366 (61.8) | 0.28 | 5.28 | <.001 |
How to avoid being judgmental | 397 (67.5) | 357 (60.3) | 0.28 | 4.87 | <.001 |
How to be compassionate | 387 (65.9) | 378 (63.9) | 0.14 | 2.94 | .003 |
How to comfort the other person | 357 (60.7) | 363 (64.7) | −0.05 | −0.94 | .35 |
How to help think through decisions | 348 (59.1) | 371 (62.6) | −0.13 | −2.44 | .02 |
How to help solve practical problems | 310 (52.7) | 334 (56.5) | −0.10 | −1.86 | .06 |
How to give advice | 308 (52.4) | 345 (58.3) | −0.11 | −2.03 | .04 |
The most highly prized skills involved simply listening attentively and taking an understanding, nonjudgmental stance. Generally, respondents considered it more important for their peer to learn each skill than themselves. However, this pattern was weaker or even reversed for skills related to intervening to resolve the other person’s distress, for example, by solving problems or comforting the other person. For those “intervention-like” skills, participants wanted to learn to deliver the skills to their peers more than they wanted their peers to learn the skills.
To determine the method of access that would reach the most consumers, we provided respondents with a choice of 6 ways to access the peer counseling training course and intervention: through a stand-alone website, through a social networking website such as Facebook, as a mobile “app,” as a program offered through a doctor’s office or other health care provider, as a program offered through one’s workplace or school, or as a program offered through a community center such as a public library or the YMCA (Young Men’s Christian Association). Respondents were asked to indicate the way they would be most likely to access the program by rank-ordering the options. They also rated each option on several 6-point semantic differential scales: appealing-unappealing, trustworthy-untrustworthy, convenient-inconvenient, and easy to use–hard to use.
The most popular access choice by far was stand-alone website, with 51.2% (303/592) of respondents ranking it as their most likely option, with doctor’s office and mobile app following (see
Most likely method for accessing peer counseling.
Method of access | Ranked 1, n (%) (N=588) | Ranked 2, n (%) (N=588) | Mean ranka |
Website | 301 (51.2) | 111 (18.9) | 2.12 |
Doctor’s office | 86 (14.6) | 95 (16.2) | 3.44b |
Mobile app | 68 (11.6) | 138 (23.5) | 3.66b,c |
Community center | 46 (7.8) | 94 (16.0) | 3.69c |
Work or school | 49 (8.3) | 72 (12.2) | 3.87c |
Social networking site | 38 (6.5) | 78 (13.3) | 4.22 |
aLower mean ranks indicate greater preference.
bThese options did not differ significantly when compared, Wilcoxon signed rank test
cThese options did not differ significantly, Wilcoxon signed rank test
Access methods’ mean ranks on each attribute.
Method of access | Mean ranka | |||
Appealing | Trustworthy | Convenient | Easy to use | |
Website | 2.57 | 3.03b | 2.66 | 1.79 |
Doctor’s office | 3.46b | 2.55 | 4.18c | - |
Mobile app | 3.85c | 4.47c | 3.27b | 2.15 |
Community center | 3.44b | 3.14b | 4.04c | - |
Work or school | 3.74c | 3.32 | 3.54b | - |
Social networking site | 3.93c | 4.49c | 3.30b | 2.06 |
aAttributes not assessed for a particular access method are marked by a dash. Lower mean ranks indicate greater preference.
bThese options did not differ significantly when compared, Wilcoxon signed rank test
cThese options did not differ significantly, Wilcoxon signed rank test
The purpose of this investigation was to assess consumer interest in a Web-based self-help and peer support mental health intervention, to determine demographic and clinical predictors of interest in the intervention, and to evaluate consumer preferences for specific features of Web-disseminated peer support interventions.
Respondents expressed fairly high interest in the CMH program: 63.9% verbally communicated that they would try the program, and 32.6% showed behavioral evidence of interest by offering their email addresses to request more information. More than half of respondents who indicated that they would never consider seeking psychotherapy or psychiatric medication were willing to try the CMH program. This proportion was significantly lower than it was among those who were willing to access these traditional mental health services, which indicates that programs like CMH are not a panacea; they cannot reach every person who is “left out” by traditional services. However, it appears that peer counseling still appeals to a substantive portion of that population.
Several demographic and clinical characteristics were associated with 3 different indicators of interest in the CMH program: a multi-item measure of the program’s appeal (continuous), endorsement of intention to try the program (dichotomous), and provision of an email address in order to receive information about the program (dichotomous). Female gender, lower income, and openness to using psychotherapy were associated with positive attitude toward CMH while controlling for other variables. Younger age, Hispanic ethnicity, lower income, and openness to using psychotherapy predicted willingness to try the program while controlling for other variables. Female gender, higher symptom distress, and openness to using therapy were associated with provision of an email address while controlling for other variables.
We predicted that females would be more interested in the CMH program than males in light of their greater willingness to seek both eHealth and traditional mental health services [
The association between gender and interest in the program suggests that CMH and other peer support programs do not completely avoid the barriers men face to seeking mental health care. This makes sense when one considers the demands of the male gender role. CMH has the potential to reduce stigma because users do not have to identify as mentally ill; however, it still requires that users seek support from one another and disclose their personal experiences, thoughts, and feelings. Such support-seeking and vulnerability violate cultural scripts of traditional masculinity [
The association between interest in CMH and lower income is particularly intriguing in light of previous research showing that people with lower income have more negative attitudes toward seeking professional psychological help (eg, [
Older respondents were less likely to endorse willingness to try the CMH program; however, age did not predict participants’ attitudes toward the program or their likelihood of requesting more information via email. Older age is associated with lower technology adoption [
Respondents who identified as Hispanic were more likely to indicate that they would try the program than non-Hispanic white respondents. (Although the association between Hispanic ethnicity and attitudes did not reach significance, it was in the same direction.) This result requires replication, given that it was not consistent across outcome measures and only 33 respondents selected this ethnic category. Furthermore, there is a great deal of cultural heterogeneity among Hispanic Americans. However, one can speculate that Hispanic respondents’ greater willingness to use the CMH program may be related to the value of
No other racial or ethnic identity group differed from non-Hispanic whites in level of interest in the CMH program. For many racial or ethnic minorities, stigma around mental illness is higher than among white individuals [
Openness to using psychotherapy was the most consistent predictor of interest in CMH across the 3 outcome variables, even when controlling for all demographic and clinical characteristics. This suggests that some of the same, unmeasured factors drive willingness to access therapy and interest in CMH. On the other hand, openness to using therapy accounted for less than 6% of the variance in attitudes toward CMH, and many individuals who indicated they would not use psychotherapy endorsed willingness to try CMH. This suggests that the shared features of professional psychotherapy and Web-based self-help and peer support programs account for
In the hierarchical models, symptom distress was significantly or marginally associated with each index of interest such that people with more symptoms expressed more interest in the CMH program. Distress remained a significant predictor of requesting more information via email when controlling for other variables. The general positive association between distress and interest is heartening—CMH and similar programs will appeal most to those who need them most.
When asked whom they would prefer as a partner in the CMH program, respondents generally favored romantic partners or close friends, although a notable minority of respondents preferred to work with a stranger. These findings on selecting a peer suggest that it is important for anyone creating a peer support intervention to recognize that, although such interventions have typically paired 2 strangers, most people would rather work with someone whom they already know, preferably someone with whom they are close. At the same time, some individuals—perhaps those with narrower social circles or conflictual close relationships—find disclosing to a stranger attractive. Peer support’s reach might be optimized by designing interventions that are flexible enough to allow for a variety of people to act as peers and by building in processes for introducing strangers who do not want to work with friends or family.
Additionally, the popularity of a romantic partner as a peer counselor may also imply that many people are interested in strengthening their romantic relationships, a possibility corroborated by many comments in an open-ended section of the survey about desiring to improve communication with spouses. Web-based interventions for couples or spouses may present an additional opportunity to improve public mental health, given the evidence that interventions for couples can improve individual mental health in addition to relationship functioning [
When presented with a list of counseling skills that they or their peers could learn, respondents expressed the greatest desire that their peers learn skills that involved attentive, nonjudgmental listening. Respondents generally considered it more important for their peer to learn each skill than to learn that skill themselves, perhaps because people are more invested in receiving quality support than providing it or because people tend to overestimate their own abilities [
These findings converge with several lines of work that illuminate helpful and unhelpful ways to react to others’ distress. Well-intentioned peers and loved ones often respond to disclosures of distress by trying to eliminate the stressful stimulus (eg, by solving problems or giving advice) or by trying to change the discloser’s emotional response (eg, by reframing the stressor in a positive light or minimizing its gravity). Yet the recipients of such “support” tend to regard it as unhelpful or even disturbing [
On the basis of this literature and the corroborating results of our survey, those who hope to create new peer support interventions should consider incorporating training in attentive, nonjudgmental listening and reflection skills, as well as education about the potential dangers of common “helping” behaviors such as giving reassurance and advice. However, one must recognize that what support seekers
Selecting from 6 options for learning about and accessing the Web-based training course, the overwhelming majority of respondents indicated that they would be most likely to use the program if it took the form of a stand-alone website. A notable minority indicated that a health care provider’s office or a mobile app would be the ideal point of access. A website and a mobile app were regarded as convenient, and health care providers were regarded as highly trustworthy but inconvenient. These results suggest that a training course is likely to have the greatest reach if offered via a stand-alone website. However, offering a mobile version of the course, or disseminating peer counseling resources through existing health care providers, could expand the population served, given that 11.6% and 14.6% of respondents, respectively, ranked those access options as the most likely way to reach them. Perhaps developers of Web-disseminated self-help or peer support courses could capitalize on the trustworthiness of health care providers by using doctors’ offices as a first point of contact but could increase convenience by enabling users to access the course and meet with peers at any location or on any device of their choosing.
The chief limitation of this study was the use of a nonprobability sample. Although diverse, MTurk users are not representative of the population, and they may be somewhat more interested in this program than the general public because of their comfort with using technology.
An additional limitation of this study is that the findings regarding interest in the program apply to the specific description of the program used in the survey. Interest may have differed if the program were presented in more detail or with other emphases. Furthermore, Web-disseminated self-help or peer support programs that include different features may engender different levels of consumer interest than the CMH program.
Future research on such programs could expand upon the foundation laid here in a variety of ways. This exploratory study established that interest exists; however, it did not explain
The results of this study substantiate the potential for CMH to fill gaps in mental health care. Diverse consumers would be interested in such an intervention, including those who are not already accessing services. Because they minimize stigma and utilize existing social support systems, reciprocal peer support interventions may be especially attractive to some groups who are underserved by professional mental health services.
It appears that CMH’s appeal can be enhanced by allowing a variety of options for peer counselors and by teaching active listening skills. It also seems that the course will reach the greatest number of users if accessed via a website, although access through mobile apps or through health care providers’ offices may also be useful options.
We hope that the results of these surveys will persuade readers of the potential utility of Web-disseminated self-help and peer support programs and will inform the creation and dissemination of other programs.
The description of the program presented in the survey.
The results of the hierarchical linear and logistic regressions.
Brief Symptom Inventory
Crowdsourcing Mental Health
Internet protocol
massive open online course
Mechanical Turk
socioeconomic status
Young Men’s Christian Association
This research was supported in part by a Graduate Research Fellowship awarded to SLB by the University of Massachusetts Amherst Center for Research on Families and by an Honors Research Grant awarded to KB by the University of Massachusetts Amherst Commonwealth Honors College.
SLB conceived the study, designed and administered the survey, conducted cognitive interviews, analyzed the data, and wrote the manuscript. KB contributed to the literature review, participated in development of the survey, conducted cognitive interviews, and participated in preliminary data analysis. GDS contributed to the literature review and revision of the manuscript. MJC collaborated on all aspects of the research design, provided laboratory resources to conduct the surveys, and edited the manuscript.
None declared.