Methodology

Recruiting and Sampling Participants

The table below displays key details for each of our four stakeholder group samples. We contracted with Qualtrics to recruit our non-clinician participants. This survey company specializes in drawing samples of survey participants from large internet panels designed to capture the demographic and regional diversity of the US adult population. Our team recruited our clinician participants, contacting selected psychiatrists listed in the American Board of Psychiatry and Neurology directory multiple times via postal mail and email.

Key Stakeholder Group Samples

Stakeholder Group	Target Population	Sampling Design	Participation Dates in 2020	Median Completion Time	# of Completed Surveys
general public	adults in US general public	quota sample from diverse national online panel (quotas for age, sex, race, & income)	April 9 to April 15	16 minutes, 7 seconds	1,022
caregivers	adults in US general public who have served as caregiver to family member or friend with depression in recent years	quota sample from diverse national online panel (screened for caregiving experience in prior two years)	April 24 to April 28	16 minutes, 51 seconds	1,026
patients	adults in US general public with prior clinical diagnosis of depression	quota sample from national online panel of adults with prior depression diagnosis (quotas for age, sex, & race)	April 1 to April 3	16 minutes, 13 seconds	1,050
psychiatrists	49,431 US psychiatrists who were board-certified in mid-October 2019	stratified random sample of psychiatrists from American Board of Psychiatry & Neurology directory (stratified by state)	May 16 to July 14	20 minutes, 41 seconds	505

Experimental Design

To answer our key research questions, we designed an experiment that we embedded within our national survey. Our experiment has a 4*2 full factorial design. The two factors—PEI modality (ECT vs. rTMS vs. DBS vs. ABIs) and depression severity (moderate vs. severe)—produced eight experimental conditions into which participants were randomly assigned. The table below displays the total number of participants in each condition.

Total Participants Randomly Assigned to Eight Experimental Conditions

We carefully crafted and filmed a video vignette featuring professional actors playing a patient with treatment-resistant depression (TRD) receiving information about a PEI from her psychiatrist. Across these eight video vignettes, the psychiatrist presented information about one of the PEIs, and the patient displayed either moderate or severe depressive symptoms. As an example, the following is the short video vignette for our “DBS & severe TRD” condition.

Standardized Survey

We developed our national survey over a year and a half, after we secured a human subjects exemption from a university Institutional Review Board.  During this time, we drew insights from the published scientific literature and from our earlier in-depth interviews with representatives from each stakeholder group. We created and pilot tested multiple versions of our survey, refining our questions and directions in the process. Incorporating feedback we received from our Scientific Advisory Board, we eventually completed the final version of our survey for our general public, caregiver, patient, and psychiatrist samples.

All participants received the same set of core questions (about 95% of the survey), with each stakeholder group receiving a few additional questions unique to them. After providing informed consent, non-clinicians answered a few preliminary questions before viewing their randomly assigned video vignette; psychiatrists proceeded directly to their randomly assigned video vignette after providing informed consent. After answering a few questions assessing their understanding of the content in their assigned video vignette, all participants answered a series of questions measuring their views about the PEI featured in their video, before reporting their demographic, social, and political characteristics.

Because of the novelty of our study, we created new measures of several PEI views, which are displayed in the table below. For each new scale, we present the verbatim text for the beginning of the question (i.e., the question stem), one example of the items comprising the scale, the lower range and upper range of possible responses, and the Cronbach’s Alpha value of the scale items. The latter is an indicator of consistency across the scale items; a value of 0.70 or higher suggests a scale has sufficient measurement reliability.

Novel Composite Measures of Key PEI Views

Scale Name	Question Stem Wording	Example Item	Scale Range	Cronbach’s Alpha
8-item General Affect Scale	“In your opinion, [PEI] seems”	“disgusting” ↔ “pleasant”	negative trait (1) to positive trait (7)	0.93
7-item Perceived Influence on Self Scale	“In your view, how much of a negative or positive influence would getting [PEI] have on the following?”	“Mary’s agency or free will”	“strong negative influence” (1) to “strong positive influence” (7)	0.90
6-item Perceived Benefits Scale	“In your view, to what extent would each of the following be a benefit for Mary getting [PEI]?”	“it quickly relieves depression symptoms”	“no benefit at all” (1) to “great benefit” (6)	0.94
5-item Perceived Riskiness Scale	“In your view, to what extent would each of the following be a risk for Mary getting [PEI]?”	“it produces cognitive impairment (e.g., memory loss)”	“no risk at all” (1) to “great risk” (5)	0.87
6-item Perceived Invasiveness Scale	“In your view, how much would getting [PEI] interfere with the following aspects of Mary’s life?”	“interferes with her expression of emotions”	“not interfere at all” (1) to “greatly interfere” (6)	0.87
3-item Support for Expanded Use Scale	“Consider each of the following policies or actions about using [PEI] to treat clinical depression.  Indicate whether you oppose or support each one in your state.”	“use state tax revenue to fund more research on using [PEI] for treating depression”	“strongly oppose” (1) to “strongly support” (7)	0.74
3-item Support for Reduced Use Scale	“Consider each of the following policies or actions about using [PEI] to treat clinical depression.  Indicate whether you oppose or support each one in your state.”	“ban all use of [PEI] for treating depression”	“strongly oppose” (1) to “strongly support” (7)	0.65

Data Management and Analysis

We managed and conducted the analyses of our survey data with IBM SPSS 26.0. We performed a series of diagnostic tests to check the quality of our data. We then conducted a conventional suite of bivariate and multivariate statistical analyses to answer our project’s research questions. After the conclusion of our NIH grant, we will make our dataset (with accompanying documentation) available for analysis by other scholars.