Amazon Mechanical Research Articles

Convergence of online flow and AI-enabled services: the impact on awe experience in e-tail customer journeys

ABSTRACT This study investigates how control, concentration, and enjoyment impact customers’ awe experience in e-tailing, and examines how AI-enabled services may alter online flow dynamics. The research employed cutting-edge data collection techniques using an AI-driven platform developed by an Indian online survey firm. Participants were recruited via Amazon Mechanical Turk and snowball sampling. Variables were measured using a Likert-type scale, with survey credibility and validity ensured through rigorous measures, including feedback collection and analysis. Through a novel research model and the ‘broaden-and-build theory,’ the study reveals that customers’ awe experience can be significantly influenced by control and concentration. Surprisingly, AI-powered e-retail services negatively affected two flow dimensions and subsequent awe experience. However, customers’ awe experience positively correlated with AI service quality, indicating emotions’ criticality. These findings provide new insights for future research and e-tailers to design effective marketing strategies enhancing customer journeys. The study underscores the importance of fashion retailers embracing AI technology to meet dynamic customer expectations, yielding enhanced efficiency and cost-effectiveness through process automation and waste minimization. By integrating AI, the industry undergoes a significant transformation, effectively redefining the shopping experience.

Trauma, PTSD, and Self-Efficacy: Predictors of Cervical Cancer Screening in Sexual Violence Survivors.

Sexual violence is highly prevalent in the United States and is associated with a host of negative physical and mental health outcomes. Specifically, sexual violence is associated with increased rates of cervical cancer, one of the most common cancers found in women. Furthermore, sexual violence survivors report reduced participation in preventive healthcare behaviors (e.g., Pap tests) which may reduce individuals' risk of developing reproductive health conditions such as cancer. Sexual violence exposure is also associated with increased symptoms of posttraumatic stress disorder (PTSD) and reduced trauma-coping self-efficacy (TCSE), two factors that may impact trauma-exposed individuals' level of avoidance of cervical cancer screens. Current research on the connection between sexual violence and reproductive healthcare often fails to examine potential underlying mechanisms behind this association, nor does it account for confounding factors such as healthcare accessibility and need. Accordingly, the present study tested a proposed moderated mediation model to explore the association between sexual violence and cervical cancer screening participation, including analysis of the indirect effect of TCSE and potential moderation of this effect by PTSD symptoms. Participants were 554 participants who reported experiences of sexual violence on Amazon's Mechanical Turk. Severity of sexual violence was significantly associated with reduced likelihood of participation in recommended cervical cancer screening. TCSE did not mediate nor did PTSD moderate this association. Findings of this study suggest that individuals' reproductive healthcare behaviors are influenced by their experiences of sexual trauma, as well as by structural factors such as insurance and income. Limitations, directions for future research, and clinical implications of study findings are discussed.

Partisan Media, Trust, and Media Literacy: Regression Analysis of Predictors of COVID-19 Knowledge.

The COVID-19 pandemic was a devastating public health event that spurred an influx of misinformation. The increase in questionable health content was aided by the speed and scale of digital and social media and certain news agencies' and politicians' active dissemination of misinformation about the virus. The popularity of certain COVID-19 myths created confusion about effective health protocols and impacted trust in the health care and government sectors deployed to manage the pandemic. This study explored how people's information habits, their level of institutional trust, the news media outlets they consume and the technologies in which they access it, and their media literacy skills influenced their COVID-19 knowledge. We administered a web-based survey using Amazon Mechanical Turk (MTurk) to assess US adults' (n=1498) COVID-19 knowledge, media and news habits, media literacy skills, and trust in government and health-related institutions. The data were analyzed using a hierarchical linear regression to examine the association between trust, media literacy, news use, and COVID-19 knowledge. The regression model of demographic variables, political affiliation, trust in institutions, media literacy, and the preference for watching Fox or CNN was statistically significant (R2=0.464; F24,1434=51.653; P<.001; adjusted R2=0.455) in predicting COVID-19 knowledge scores. People who identified as politically conservative, watched Fox News, and reported lower levels of institutional trust and media literacy, scored lower on COVID-19 knowledge questions than those who identified as politically liberal, did not watch Fox News and reported higher levels of institutional trust and media literacy. This study suggests that the media outlets people turn to, their trust in institutions, and their perceived degree of agency to discern credible information can impact people's knowledge of COVID-19, which has potential implications for managing communication in other public health events.

From Good Citizens to Bad Apples? The Effect of Boredom Proneness: A Moral Licensing Perspective.

Although boredom proneness has been well-studied, there is still an ongoing scholarly debate about whether boredom can move individuals to act counterproductively or productively. Drawing from a moral licensing perspective, this study empirically investigates how boredom proneness influences interpersonal citizenship behavior (ICB), moral meaningfulness, and interpersonal deviance over time. Utilizing a total of 328 three-wave panel data obtained via Amazon Mechanical Turk during a three-month period, we find the following results. First, individuals with a higher initial level of boredom proneness demonstrate a higher initial level of ICB. Second, as boredom proneness increases at a faster rate, ICB increases at a faster rate over time. Third, individuals with a higher initial level of ICB report a higher initial level of moral meaningfulness. Finally, individuals with a higher initial level of moral meaningfulness exhibit a higher initial level of interpersonal deviance. Implications for theory and practice are discussed.

Dynamic payment on microtasking platforms using bee colony optimization

Microtasking involves breaking down a job into smaller tasks and assigning them to a group of workers who voluntarily complete the tasks and receive payment. However, the joint estimation of quality and determination of payment remains an underexplored area in microtask crowdsourcing research. This paper addresses the limitation of unestimated payment on microtasking platforms by introducing a dynamic worker quality and payment estimation algorithm known as Dynamic Quality Payment (DQP). Utilizing the Bee Colony Optimization (BCO) algorithm, the proposed approach integrates worker quality estimation and payment determination. DQP incorporates the Gaussian Process Model (GPM) to initially estimate worker quality and then optimize payments based on the quality of work. Empirical testing of the algorithm is conducted using two datasets from Amazon Mechanical Turk. Additionally, the DQP algorithm is compared against novel microtasking and quality estimation algorithms, demonstrating superior performance. The experimental results illustrate that DQP not only reduces the cost of microtasking but also accurately estimates worker quality.

Generalizing Hate Speech Detection Using Multi-Task Learning: A Case Study of Political Public Figures

Automatic identification of hateful and abusive content is vital in combating the spread of harmful online content and its damaging effects. Most existing works evaluate models by examining the generalization error on train–test splits on hate speech datasets. These datasets often differ in their definitions and labeling criteria, leading to poor generalization performance when predicting across new domains and datasets. This work proposes a new Multi-task Learning (MTL) pipeline that trains simultaneously across multiple hate speech datasets to construct a more encompassing classification model. Using a dataset-level leave-one-out evaluation (designating a dataset for testing and jointly training on all others), we trial the MTL detection on new, previously unseen datasets. Our results consistently outperform a large sample of existing work. We show strong results when examining the generalization error in train–test splits and substantial improvements when predicting on previously unseen datasets. Furthermore, we assemble a novel dataset, dubbed PubFigs, focusing on the problematic speech of American Public Political Figures. We crowdsource-label using Amazon MTurk more than 20,000 tweets and machine-label problematic speech in all the 305,235 tweets in PubFigs. We find that the abusive and hate tweeting mainly originates from right-leaning figures and relates to six topics, including Islam, women, ethnicity, and immigrants. We show that MTL builds embeddings that can simultaneously separate abusive from hate speech, and identify its topics.

Understanding the individual labor supply and wages on digital labor platforms: A microworker perspective

The emergence of crowdsourcing as a new form of work has introduced a paradox among workers who receive small payments for piecemeal microwork yet continue to participate in microwork digital labor platforms (DLPs). To better understand what sustains microworkers’ participation, this study draws upon individual labor supply theory to quantitatively examine the impacts of microworkers’ motivations, perceptions, and preferences on their labor supply and wages. To explore the meaning of monetary rewards for microworkers, a qualitative inquiry explores microworkers’ spending patterns. Based on a survey of 306 microworkers on Amazon Mechanical Turk, our hierarchical regression analysis reveals that while individual motivations for monetary rewards, enjoyment, and microtime structure have some impact on the microwork labor supply and wages, their impact is limited. Our thematic analysis uncovers diverse meanings attached to microwork earnings. The two most noted are meeting subsistence needs and nonessential expenditures, both of which have positive effects on microwork wages. By investigating the elasticity of the microwork labor supply and wages and offering a nuanced understanding of monetary rewards, our study contributes to information management research on DLPs. Moreover, it provides practical insights for various stakeholders, including microworkers, requesters, and DLP operators.

Orthopaedic Surgery and Indusry: What our Potential Patients Believe.

Physician and surgeon involvement in industry has received considerable attention in recent decades. In this study, we outline the perspective of the general US population regarding (1) disclosure, (2) ownership, and (3) compensation between physicians/surgeons and industry. We hypothesize that the general population would be largely supportive of the physician/surgeon-industry relationship. An online, survey-based, descriptive study was conducted through a crowdsourcing platform, Amazon Mechanical Turk. Survey respondents were presented with a seven-item questionnaire inquiring about the physician/surgeon and industry relationship. An "attention check" question was included; those who failed this question were excluded. Descriptive statistics were used to assess the data and a McNemar chi-squared test for paired, dichotomous data. A total of 993 respondents were included. Survey responses are summarized in Table 1. 70.6% of respondents stated that it was "important" or "extremely important" to disclose that the patient be informed whether implants used in surgery had been developed by the operating surgeon. 71.1% of respondents reported that it was "important" or "extremely important" to disclose partial ownership within industry. Seventy-one percent of respondents stated it was "important" or "extremely important" to disclose royalty payments pertaining to surgical implants. 95.6% of respondents suggested that it was acceptable for surgeons to accept free airfare and lodging, and 95.2% of respondents stated that it was acceptable for the surgeon to be compensated for time away from practice to learn about new equipment. In our survey of 993 respondents, we found that relationships with industry are considered acceptable if appropriate disclosure is given to patients. We also found that although respondents suggested that physicians and surgeons may be influenced by a free meal, compensation for trips to try new equipment and time spent away from practice is considered appropriate. 2c, Ecological studies.

Optimizing word embeddings for small datasets: a case study on patient portal messages from breast cancer patients

Patient portal messages often relate to specific clinical phenomena (e.g., patients undergoing treatment for breast cancer) and, as a result, have received increasing attention in biomedical research. These messages require natural language processing and, while word embedding models, such as word2vec, have the potential to extract meaningful signals from text, they are not readily applicable to patient portal messages. This is because embedding models typically require millions of training samples to sufficiently represent semantics, while the volume of patient portal messages associated with a particular clinical phenomenon is often relatively small. We introduce a novel adaptation of the word2vec model, PK-word2vec (where PK stands for prior knowledge), for small-scale messages. PK-word2vec incorporates the most similar terms for medical words (including problems, treatments, and tests) and non-medical words from two pre-trained embedding models as prior knowledge to improve the training process. We applied PK-word2vec in a case study of patient portal messages in the Vanderbilt University Medical Center electric health record system sent by patients diagnosed with breast cancer from December 2004 to November 2017. We evaluated the model through a set of 1000 tasks, each of which compared the relevance of a given word to a group of the five most similar words generated by PK-word2vec and a group of the five most similar words generated by the standard word2vec model. We recruited 200 Amazon Mechanical Turk (AMT) workers and 7 medical students to perform the tasks. The dataset was composed of 1389 patient records and included 137,554 messages with 10,683 unique words. Prior knowledge was available for 7981 non-medical and 1116 medical words. In over 90% of the tasks, both reviewers indicated PK-word2vec generated more similar words than standard word2vec (p = 0.01).The difference in the evaluation by AMT workers versus medical students was negligible for all comparisons of tasks’ choices between the two groups of reviewers (p=0.774\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${\ ext{p}} = 0.774$$\\end{document} under a paired t-test). PK-word2vec can effectively learn word representations from a small message corpus, marking a significant advancement in processing patient portal messages.

Latent profiles of social reactions to sexual assault disclosure among bisexual women.

Bisexual women experience higher rates of sexual assault (SA) and posttrauma distress compared to monosexual women. We examined the patterns of reactions to SA disclosure (positive and two types of negative) and bisexual minority stress (antibisexual stigma from heterosexual and lesbian/gay people, internalized binegativity) experienced by young bisexual women who experienced adult SA. We also examined differences in assault-related experiences, posttraumatic stress disorder and depression symptoms, and hazardous drinking based on the identified patterns. The current sample was drawn from two parent samples of women and included survivors who had disclosed an adult SA: (a) n = 149 bisexual women (98% cisgender, ages 18-35, 87.9% White) recruited via Amazon's Mechanical Turk and (b) n = 81 cisgender bisexual women (ages 18-25, 84% White) recruited via online ads. Latent profile analyses and analyses of variance were conducted. Analyses identified four latent profiles characterized by: (a) low stigmatizing responses, (b) high negative SA disclosure reactions, (c) high antibisexual stigma, and (d) highest stigmatizing responses. The fourth profile had higher posttraumatic stress disorder and depression symptoms and hazardous drinking compared to the other profiles. Results suggest women in the profile characterized by highest negative reactions to SA disclosure and highest bisexual minority stress (antibisexual stigma and internalized binegativity) may experience greatest distress postassault. Results highlight the importance for researchers, policymakers, and clinicians to consider and address both SA-related and identity-related stigma among bisexual survivors. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Social media marketing and brand loyalty: exploring interrelationships through symmetrical and asymmetrical modeling

Purpose This study aims to examine the interrelationships between social media marketing activities, self-brand connections, brand equity, trust and loyalty. Design/methodology/approach A total of 402 valid responses were collected from Amazon MTurk, and the data were subjected to partial least squares structural equation modeling and fuzzy-set qualitative comparative analysis (fsQCA). Findings Findings indicate that social media marketing activities strongly and positively influence self-brand connection, brand equity and brand trust. Moreover, brand loyalty was strongly and positively influenced by self-brand connection, brand equity and brand trust. Moreover, the findings from fsQCA indicate that three causal paths lead to a high level of brand loyalty, and one causal path determines a low level of brand loyalty. Originality/value This research extends current knowledge by bridging the literature between social media marketing activities and branding using self-brand connections. Additionally, this study uses the strength of two complimentary methods – symmetrical and asymmetrical modeling – to uncover how social media marketing activities bridge customer-brand relationships.

Awareness of the potential consequences of alcohol consumption in the context of chronic pain and prescription opioid use

ABSTRACT Background: Alcohol and prescription opioid use are highly prevalent among chronic pain populations. One-fifth of individuals prescribed opioids report same-day use of alcohol and opioids. Alcohol use and alcohol/opioid co-use can have deleterious pain management and health outcomes. The extent to which individuals with chronic pain are aware of these deleterious outcomes is considerably understudied. Objectives: To explore individuals’ understanding of seven health- and pain-related risks of alcohol/alcohol-opioid use. An exploratory aim was to examine whether greater risk awareness was associated with alcohol/opioid use patterns. Methods: Participants included 261 adults age ≥21(36.4% women) endorsing current alcohol use, chronic musculoskeletal pain, and opioid prescription who completed an online survey via Amazon Mechanical Turk. Results: Distribution of the total number of items for which a participant endorsed awareness was as follows: zero (10.7%), one (5.0%), two (13.0%), three (13.8%), four (13.8%), five (11.5%), six (10.0%), and seven items (22.2%). Awareness of the health consequences of alcohol/alcohol-opioid use was positively associated with opioid misuse behaviors (β = .525, ΔR2 = .251, p < .001), and higher-risk alcohol consumption (β = .152, ΔR2 = .021, p = .011). Conclusion: Many adults with chronic pain are unaware of the health consequences of alcohol/alcohol-opioid use. Findings of positive covariation between risk awareness and higher-risk alcohol/opioid use suggest that future interventions among this population should go beyond simple risk education and utilize motivational enhancement to help change decisional balance.

The impact of death priming on climate change denial: a preliminary investigation.

Climate change denial (CCD) has been found to be more pronounced among individuals with a conservative political orientation. Terror Management Theory posits that an individual's worldview serves as a primary defense mechanism in coping with existential threats, such as the reality of climate change. Drawing on these premises, we conducted preliminary research investigating the impact of death priming on CCD from the Terror management theory perspective. We administered a death priming task and measured CCD in its immediate wake or following a delay task. We confirmed among 219 Amazon's MTurk participants that immediately following death priming, CCD was reduced among all participants. In the delayed death priming condition, we acquired preliminary evidence that CCD was heightened among Republican participants. These findings demonstrate the relevance of death priming to CCD tendencies and potentially spawn future research regarding CCD as a particular form of coping with existential threats.

A COVID-19 Paradox of Communication, Ignorance, and Vaccination Intention

This study focused on the role of Americans’ communication about COVID-19 vaccines, and its potential associations with two significant vaccine determinants. Two studies informed by the Situational Theory of Problem Solving (STOPS) were conducted. In the first, the researchers incorporated the intention to receive vaccination against the COVID-19 pandemic and vaccine-specific knowledge into STOPS to examine the associations among these three key variables. The second study introduced a new variable, trust in science/scientists, as a potential conceptual variable that functions together with vaccine knowledge to explain why Americans’ intentions to receive COVID-19 vaccines increased despite their imperfect knowledge about them. Two online surveys were distributed using Amazon Mechanical Turk (Study 1) and Qualtrics (Study 2). The first was administered from December 18 through 22 of 2020, and the second from December 18 through 29 of 2020. An enhanced level of communicative action with respect to COVID-19 vaccines decreased the accuracy of knowledge about the vaccines, which increased the intention to receive these vaccines (Study 1). A significant interaction was also found between COVID-19 vaccine knowledge and participants’ levels of trust in science/scientists on study participants’ future intention to receive COVID-19 vaccines (Study 2). Contrary to previous findings, Americans’ increased activeness in COVID-19 vaccine-specific communication was associated negatively with vaccine knowledge, which in turn, increased COVID-19 vaccination intentions. This negative association reflects the complex and unique nature of information environments with respect to COVID-19 vaccines. Trust in science/scientists served as a significant missing conceptual variable that shed light on the paradoxical finding.

"They Did Not Understand": Exploring Adult Survivors of Childhood Cancer's Memories of Their Experiences With Peers.

Background: Peer relationships are impacted by the diagnosis of childhood cancer. Children with cancer can feel isolation, such as bullying and lack of engagement, from their peers. As the rates of survival increase, one way to further learn about the experiences of childhood cancer is to reflect on the memories of adult survivors. This study examines adult survivors of childhood cancer's memories of their experiences with peers during their cancer. Method: Twenty-seven adult survivors (15 male, 12 female) of childhood cancer were recruited from Amazon Mechanical Turk, a crowdsourcing marketplace for survey distribution, and surveyed about their experiences with peers during childhood cancer. Using inductive thematic analysis, open-ended survey questions were analyzed. Results: Five main themes emerged, including age, perceptions of cancer, acts of kindness, peer responses, and limited peer engagement. Participants' memories included peers' lack of understanding about their diagnosis, prognosis, and treatment, positive and negative responses to their cancer diagnosis, acts of services and gifts from peers, and feelings of isolation during the cancer experience. Discussion: In the current study, adult survivors of childhood cancer had specific memories about how cancer impacted peer relationships. Such findings contribute to the evidence that children with cancer could benefit from psychosocial interventions, including education to peers about the diagnosis and the impact of diagnosis on peer relationships, peer engagement activities, and peer support groups.

Understanding Public Perceptions of Nipple and Scar Characteristics After Chest Wall Masculinization Surgery.

Surgical chest masculinization procedures, especially gender-affirming top surgery (GATS), are becoming increasingly prevalent in the USA. While a variety of surgical techniques have been established as both safe and effective, there is limited research examining ideal aesthetic nipple appearance and incision scar pattern. This study employs patient images to understand the public's perception on top surgery outcomes when adjusting for BMI ranges and Fitzpatrick skin types. Images from RealSelf modified via Adobe Photoshop depicted various scar types and nipple-areolar complex (NAC) sizes/positions. A Qualtrics survey was distributed utilizing Amazon Mechanical Turk. Statistical analysis was performed through JMP Pro 17 for ordinal and categorical values, with a p value less than or equal to 0.05 statistically significant. A moderately sized and laterally placed NAC was preferred. A transverse scar that resembles the pectoral border between the level of the inframammary fold and pectoral insertion was deemed most masculine and aesthetic. Majority of results demonstrated that this is unaffected by Fitzpatrick skin types. Increased BMI images impacted public preferences, as a nipple placed farther from the transverse incision (p=0.04) and a transverse scar position closer to the IMF was preferred in higher BMI patients. An understanding of the most popular NAC and scar choices, as well as how these factors may differ when considering a Fitzpatrick skin type or BMI categorization was attained. This validates the importance of patient-centered approach when employing surgical techniques in GATS. Future studies intend to obtain reports from actual patients considering GATS. This journal requires that authors assign a level of evidence to each submission to which Evidence-Based Medicine rankings are applicable.

Assessing NIAAA’s Definition of Recovery from Alcohol Use Disorder: A Latent Class Analysis of a Heterogeneous Online Sample

ABSTRACT NIAAA’s 2022 definition of recovery from alcohol use disorder (AUD) includes two core components, remission of DSM-5 AUD criteria and cessation of heavy drinking. This study’s purpose was to assess patterns of AUD symptoms and heavy drinking in a heterogeneous national sample, in order to clarify the utility of the definition. Participants who self-reported having resolved an alcohol problem for at least 6 months were recruited through Amazon Mechanical Turk (N = 386) and surveyed about their problem severity, current drinking, and AUD symptomology. We used latent class analysis to discern meaningful clusters of AUD symptoms and heavy drinking, as well as factors associated with class membership. A two-class model was the best fit for the data. The first class, which we termed Less Symptomatic, included 83.4% of the sample. Individuals in this class were unlikely to endorse any of the 10 AUD criteria (<2.5% of the time) and 24.3% reported heavy drinking. In the second class (16.6% of the sample), termed Symptomatic, 45% of respondents endorsed at least one AUD criterion and 88.2% reported heavy drinking. These findings suggest that some individuals in recovery may continue to drink heavily with minimal problems, while others continue to experience AUD symptoms.

Rainbow babies: Trauma and parenting after pregnancy loss.

Parents with a history of pregnancy loss may be at greater risk for less secure attachment (Côté-Arsenault et al., 2020) and likely experience parenting differently than those without pregnancy loss. However, existing literature regarding pregnancy loss and parenting has focused nearly exclusively on attachment and has produced inconsistent findings. The present study aims to expand the scope of previously studied parenting indices by examining the role of pregnancy loss in relation to beliefs about parenthood as well as levels of parental support, satisfaction, involvement, and limit-setting. Cumulative trauma was included in the model to assess the role of trauma exposure. Two hundred twenty-one parents (Mage = 41.69, SD = 12.73; 72.9% female) with at least one living child were recruited from either Amazon's Mechanical Turk or a midwestern university. Pregnancy loss was associated with lower levels of parental support, but did not evince links with parenting perceptions or other aspects of the parent-child relationship. Mothers reported receiving less support than fathers. Racial minorities endorsed less positive child perceptions and lower levels of parental satisfaction and involvement. Cumulative trauma, surprisingly, was unrelated to all of the parenting indices. Pregnancy loss may be related to lower perceptions of parental support, but may not be tied to parenting beliefs or aspects of the parent-child relationship. Future research should further consider context of the loss experience to better unpack how pregnancy loss may be related to specific parenting factors. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Revisiting representativeness heuristic classic paradigms: Replication and extensions of nine experiments in Kahneman and Tversky (1972).

Kahneman and Tversky showed that when people make probability judgements, they tend to ignore relevant statistical information (e.g., sample size) and instead rely on a representativeness heuristic, whereby subjective probabilities are influenced by the degree to which a target is perceived as similar to (representative of) a typical example of the relevant population, class or category. Their article has become a cornerstone in many lines of research and has been used to account for various biases in judgement and decision-making. Despite the impact this article has had on theory and practice, there have been no direct replications. In a pre-registered experiment (N = 623; Amazon MTurk on CloudResearch), we conducted a replication and extensions of nine problems from Kahneman and Tversky's 1972 article. We successfully replicated eight out of the nine problems. We extended the replication by examining the consistency of heuristic responses across problems and by examining decision style as a predictor of participants' use of the representativeness heuristic. Materials, data, and code are available on: https://osf.io/nhqc4/.

Common misperceptions and public knowledge about intrauterine devices among US-based online respondents

ObjectivesMisperceptions about intrauterine devices (IUDs) may dissuade potential users and public beliefs are not well understood. We sought to describe public knowledge of mechanism of action, failure rate, and IUD health risks. Secondary objectives included identifying inaccurate information and information sources about IUDs. Study designWe conducted a convenience-based survey of participants living in the United States 18 years and older using Amazon Mechanical Turk and ResearchMatch. We collected information on demographics, reproductive history, and IUD knowledge. We asked about IUD mechanism of action, failure rate, potential health risks, and information sources. We used multivariable logistic regression to test the association between gender and perceived lower (vs. higher) risks of IUDs. ResultsWe included 1597 individuals with 726 (45.5%) self-reported men and 871 (54.5%) self-reported women. For both hormonal and non-hormonal IUDs, participants most often selected implantation prevention as the primary mechanism of action. In our sample, 55.5% of individuals believed the failure rate was greater than five percent. Weight gain, uterine perforation, and expulsion were cited as health risks likely to occur ≥5% of the time by 42.1%, 34.3%, and 38.4% of participants, respectively. Participants identified healthcare providers as their primary source of contraceptive information. Our multivariable logistic regression analysis revealed women had higher odds of perceiving IUDs as higher risk (aOR=1.35, 95% CI 1.09–1.68) compared to men. ConclusionsIndividuals underestimated IUD effectiveness while overestimating perceived health risks. Our results highlight areas to target education about IUDs to support informed contraceptive decision-making. ImplicationsParticipants misunderstood IUD mechanism of action, underestimated efficacy, and overestimated failure rates and health risks. Providers should begin clinical counseling by assessing baseline knowledge, since mechanism of action, side effect profiles, and concepts like female anatomy should not be assumed to be known.