新视野:不确定性容忍度发展的演进方法。

IF 5.2 1区 教育学 Q1 EDUCATION, SCIENTIFIC DISCIPLINES
Michelle D. Lazarus, Melanie Farlie, Md Nazmul Karim
{"title":"新视野:不确定性容忍度发展的演进方法。","authors":"Michelle D. Lazarus,&nbsp;Melanie Farlie,&nbsp;Md Nazmul Karim","doi":"10.1111/medu.15732","DOIUrl":null,"url":null,"abstract":"<p>The paper titled “Tolerance for uncertainty and medical students' specialty choices: A myth revisited” by Wegwarth and colleagues<span><sup>1</sup></span> provides additional evidence that the existing approaches for uncertainty tolerance scale development and evaluation in health professional learners<span><sup>2</sup></span> and professionals<span><sup>3</sup></span> may not be the optimal approach for measuring this complex construct. In this commentary, we discuss the predominant approaches currently taken in developing and evaluating uncertainty tolerance using scales and suggest ways to evolve these strategies to optimise measurement validity of this construct.</p><p>Uncertainty tolerance is defined as individuals' perceptions and responses to sources of uncertainty (such as ambiguity, probability or complexity).<span><sup>4</sup></span> An individuals' response to uncertainty is expressed by how they think, feel and act. Uncertainty tolerance construct modelling by Hillen and colleagues (2017)<span><sup>4</sup></span> describes a multi-component construct comprised of (1) the source of uncertainty, (2) how this source is perceived by an individual and (3) each subsequent response (emotions, thoughts and behaviours).<span><sup>4</sup></span> However, the degree to which these elements contribute to the overall measurement of this construct remains (ironically) uncertain.<span><sup>2, 3</sup></span> Further complicating this is the question as to the changeability of one's uncertainty tolerance, which influences measurement approaches. In the mid-20th Century, uncertainty tolerance was conceptualised as a static personality trait.<span><sup>4</sup></span> Mounting contemporary evidence, however, suggests that uncertainty tolerance is a dynamic and changeable state-based construct influenced by contextual and personal factors such as prior experiences, geographic location and reflective capacity.<span><sup>5</sup></span></p><p>There is evidence that uncertainty tolerance is, at least in part, state-based in learners. Learners can feel negatively about an experience of uncertainty but can act adaptively (e.g. with uncertainty tolerance) in response through their cognition and behaviour.<span><sup>6</sup></span> For example, a medical student evaluating a patient with fatigue, fever and joint pain may initially feel anxious, due to the broad range of potential differential diagnoses (stimulus). However, by relying on their training and prior experiences (e.g. moderators) they can adaptively respond—despite their feelings of anxiety (emotional response)—by gathering a history, recommending appropriate tests and consulting with peers and colleagues about the next step to take (behavioural responses) and feeling confident about these next steps in supporting the patient (cognitive response). If this students' uncertainty tolerance was being measured, would they be evaluated as intolerant of uncertainty because of their anxiety or tolerant of uncertainty because of their adaptive behaviours and cognition? This is where the field remains conflicted and where existing scales may be falling short.<span><sup>7</sup></span></p><p>Most uncertainty tolerance scales were developed and tested using classical test theory (CTT) approaches and prior to the modern multi-component conceptual model of uncertainty tolerance.<span><sup>4</sup></span> Due to this, existing scales have a tendency towards numerous items focusing on one element of the construct—that of emotional responses and often fail to capture the complex and nuanced nature of the construct.<span><sup>3</sup></span></p><p>As managing uncertainty becomes increasingly recognised as a health professions' competency,<span><sup>8-11</sup></span> the desire to measure learners' uncertainty tolerance has equally gained attention—yet multiple studies, including the paper by Wegwarth et al (2025),<span><sup>1</sup></span> call into question the utility of these scales.</p><p>Psychological and educational measurement instruments are primarily situated within two psychometric worldviews: CTT and latent trait modelling (LTM), such as Item Response Theory (IRT). Both CTT and IRT approaches aim to establish the reliability and validity of scales.<span><sup>12</sup></span> While CTT is the most widely used approach for developing and testing existing uncertainty tolerance scales, it presents significant limitations when applied to assessments of health professions learners' uncertainty tolerance,<span><sup>13</sup></span> and these limitations may contribute to the challenges we have in measuring uncertainty tolerance in this population.</p><p>CTT assumes that an observed score reflects a fixed true score plus random error, with the error considered normally distributed and unrelated to the true score. Under this model, score fluctuations across repeated testing are attributed to external factors—such as test conditions or distractions—rather than actual changes in the trait being measured.<span><sup>14</sup></span> For instance, a student may perform differently on the same test when delivered in a formative assessment context versus under invigilated, high-stakes conditions. As a result, a scale validated in one context may not maintain its validity in another, requiring revalidation of the same scale in new contexts despite using identical items.</p><p>CTT-based approaches are also sample-dependent,<span><sup>15</sup></span> meaning reliability and validity estimates are specific to the population in which the scale was developed and may not generalise to other groups. This poses a challenge for applying uncertainty tolerance scales across a diversity of health professions learners. Meta-analyses of scales developed using CTT methods (e.g. syntheses of Cronbach's alpha) show wide variability in scale performance across populations, necessitating revalidation in each new setting.<span><sup>2</sup></span> Additionally, a scale developed using a CTT approach is not designed to account for context-driven response differences—such as those between low- (e.g. uncertainty with completing a puzzle) and high-stakes environments (e.g. patient care)—further limiting its ability to provide valid, transferable insights into learners' uncertainty tolerance across contexts.</p><p>A further conceptual limitation is the reliance on total scores, which assumes that all items equally reflect the construct, without accounting for item-level characteristics such as difficulty or discriminatory power—features we consider crucial for measuring the complex construct of uncertainty tolerance. This ‘total score reliance’ limits the sensitivity and precision of existing uncertainty tolerance scales in distinguishing between varying levels of the construct. Uncertainty tolerance often also varies between individuals, even when they encounter the same source of uncertainty. One person may respond confidently to a minor clinical unknown, while another may react with extreme anxiety. This highlights the subjective, dynamic nature of uncertainty tolerance. Scales developed using CTT approaches do not account for differences in test-takers based on item difficulty or relevance. As a result, instruments developed using CTT approaches are limited in their ability to reflect the nuanced, individualised and varied ways individuals experience and respond to different sources of uncertainty.<span><sup>15</sup></span></p><p>LTMs, including IRT and Rasch models, represent psychometric frameworks that model the relationship between an individual's latent traits and their responses to specific test items. These models estimate the probability of a particular response based on the respondent's underlying trait level and the item characteristics, such as item difficulty and discrimination. Unlike CTT approaches, LTMs offer advantages that make them well-suited for measuring complex, multidimensional constructs like uncertainty tolerance.</p><p>A key advantage of LTMs in measuring constructs like uncertainty tolerance is their emphasis on item-level analysis rather than reliance on total scores. In the context of uncertainty tolerance, different items may tap into diverse cognitive, emotional or behavioural reactions to uncertainty. LTMs, particularly IRT, allow researchers to evaluate how effectively each item distinguishes between individuals with varying levels of the trait. For example, an item reflecting emotional unease about incomplete information may be highly discriminative among mid-level uncertainty tolerance respondents but less so at the extremes (those intolerant of uncertainty, for instance). This type of item-level insight supports more precise scale development opportunities by allowing for retaining items that contribute meaningfully to construct measurement, rather than assuming all items are equally informative.</p><p>Another strength of LTMs is their ability to produce sample-independent, context-transferable measures. Once items are calibrated using IRT or Rasch analysis with good model fit, their psychometric properties remain stable across populations and settings. This is valuable for developing uncertainty tolerance assessments applicable across diverse health professions learners. For instance, a scale developed with Australian medical students may be extended to physiotherapy or nursing students in other countries. However, cross-group validation—including tests for measurement invariance and differential item functioning—remains essential to ensure fair and accurate interpretation.<span><sup>16</sup></span></p><p>LTMs allow for nuanced insights into individual responses to different sources of, for instance, uncertainty by estimating each respondent's position on the latent trait continuum in a manner that is independent of the test form. For instance, two learners may give the same response to an item on diagnostic ambiguity, but IRT can reveal meaningful differences in overall tolerance based on scale response patterns. This precision supports identifying subgroups and tailoring feedback or interventions. Given the nuanced nature of uncertainty tolerance responses—when emotional, cognitive and behavioural aspects of uncertainty tolerance vary across individuals—this could prove valuable in the field.</p><p>We encourage uncertainty tolerance researchers to consider LTM holistically from conception and design of the uncertainty tolerance scale through to evaluation. In this phase of transition, however, we suggest that researchers explore existing uncertainty tolerance scales, often developed through CTT approaches, with LTM evaluation strategies.</p><p>“You cannot swim for new horizons until you have courage to lose sight of the shore”—William Faulkner.</p><p>As William Faulkner suggests, the recent paper by Wegworth and colleagues<span><sup>1</sup></span> can serve as a ‘north star’ that inspires the field to lose sight of the shore (e.g. uncertainty tolerance scales developed and evaluated using CTT approaches) and move in the direction of a new horizon (e.g. development and evaluation of these scales using LTM approaches). When we take a closer look at the historical uncertainty tolerance scale “shore”, we see some significant challenges. These include a primary focus on measuring one feature of the construct (i.e. emotional responses), as well as the inability to tease out potential moderating factors (e.g. related to individual items) influencing an individual's perception and responses to uncertainty.<span><sup>17-19</sup></span></p><p>Given advances in measurement theory applicable to health professions education, there is an opportunity now to adjust our course towards a new horizon of LTM. LTM approaches could consider known relevant features of uncertainty tolerance including the emotional, cognitive and behavioural elements as well as the moderating factors which may be influencing an individual learners' response to uncertainty—and even consider how different sources of uncertainty may play a role in these responses. In this way, LTM approaches offer a way to advance the measurement of this complex construct.</p><p>Given the widespread inclusion of uncertainty tolerance as a health profession graduate attribute and the desire to measure uncertainty tolerance across health professions, there is a clear imperative to embrace more nuanced psychometric approaches. The assumptions of LTM demonstrate congruence with the modern conceptual uncertainty tolerance modelling and could allow for scale development and measurement which are theoretically grounded, empirically robust and responsive to the realities of modern health professions education. The field is ready to lose sight of the shore and swim for new horizons.</p><p><b>Michelle D. Lazarus:</b> Conceptualisation; writing—original draft; writing—review and editing. <b>Melanie Farlie:</b> Conceptualisation; writing—original draft; writing—review and editing. <b>Md Nazmul Karim:</b> Conceptualisation; writing—original draft; writing—review and editing.</p>","PeriodicalId":18370,"journal":{"name":"Medical Education","volume":"59 8","pages":"787-791"},"PeriodicalIF":5.2000,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1111/medu.15732","citationCount":"0","resultStr":"{\"title\":\"New horizons: Evolving approaches to uncertainty tolerance scale development\",\"authors\":\"Michelle D. Lazarus,&nbsp;Melanie Farlie,&nbsp;Md Nazmul Karim\",\"doi\":\"10.1111/medu.15732\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>The paper titled “Tolerance for uncertainty and medical students' specialty choices: A myth revisited” by Wegwarth and colleagues<span><sup>1</sup></span> provides additional evidence that the existing approaches for uncertainty tolerance scale development and evaluation in health professional learners<span><sup>2</sup></span> and professionals<span><sup>3</sup></span> may not be the optimal approach for measuring this complex construct. In this commentary, we discuss the predominant approaches currently taken in developing and evaluating uncertainty tolerance using scales and suggest ways to evolve these strategies to optimise measurement validity of this construct.</p><p>Uncertainty tolerance is defined as individuals' perceptions and responses to sources of uncertainty (such as ambiguity, probability or complexity).<span><sup>4</sup></span> An individuals' response to uncertainty is expressed by how they think, feel and act. Uncertainty tolerance construct modelling by Hillen and colleagues (2017)<span><sup>4</sup></span> describes a multi-component construct comprised of (1) the source of uncertainty, (2) how this source is perceived by an individual and (3) each subsequent response (emotions, thoughts and behaviours).<span><sup>4</sup></span> However, the degree to which these elements contribute to the overall measurement of this construct remains (ironically) uncertain.<span><sup>2, 3</sup></span> Further complicating this is the question as to the changeability of one's uncertainty tolerance, which influences measurement approaches. In the mid-20th Century, uncertainty tolerance was conceptualised as a static personality trait.<span><sup>4</sup></span> Mounting contemporary evidence, however, suggests that uncertainty tolerance is a dynamic and changeable state-based construct influenced by contextual and personal factors such as prior experiences, geographic location and reflective capacity.<span><sup>5</sup></span></p><p>There is evidence that uncertainty tolerance is, at least in part, state-based in learners. Learners can feel negatively about an experience of uncertainty but can act adaptively (e.g. with uncertainty tolerance) in response through their cognition and behaviour.<span><sup>6</sup></span> For example, a medical student evaluating a patient with fatigue, fever and joint pain may initially feel anxious, due to the broad range of potential differential diagnoses (stimulus). However, by relying on their training and prior experiences (e.g. moderators) they can adaptively respond—despite their feelings of anxiety (emotional response)—by gathering a history, recommending appropriate tests and consulting with peers and colleagues about the next step to take (behavioural responses) and feeling confident about these next steps in supporting the patient (cognitive response). If this students' uncertainty tolerance was being measured, would they be evaluated as intolerant of uncertainty because of their anxiety or tolerant of uncertainty because of their adaptive behaviours and cognition? This is where the field remains conflicted and where existing scales may be falling short.<span><sup>7</sup></span></p><p>Most uncertainty tolerance scales were developed and tested using classical test theory (CTT) approaches and prior to the modern multi-component conceptual model of uncertainty tolerance.<span><sup>4</sup></span> Due to this, existing scales have a tendency towards numerous items focusing on one element of the construct—that of emotional responses and often fail to capture the complex and nuanced nature of the construct.<span><sup>3</sup></span></p><p>As managing uncertainty becomes increasingly recognised as a health professions' competency,<span><sup>8-11</sup></span> the desire to measure learners' uncertainty tolerance has equally gained attention—yet multiple studies, including the paper by Wegwarth et al (2025),<span><sup>1</sup></span> call into question the utility of these scales.</p><p>Psychological and educational measurement instruments are primarily situated within two psychometric worldviews: CTT and latent trait modelling (LTM), such as Item Response Theory (IRT). Both CTT and IRT approaches aim to establish the reliability and validity of scales.<span><sup>12</sup></span> While CTT is the most widely used approach for developing and testing existing uncertainty tolerance scales, it presents significant limitations when applied to assessments of health professions learners' uncertainty tolerance,<span><sup>13</sup></span> and these limitations may contribute to the challenges we have in measuring uncertainty tolerance in this population.</p><p>CTT assumes that an observed score reflects a fixed true score plus random error, with the error considered normally distributed and unrelated to the true score. Under this model, score fluctuations across repeated testing are attributed to external factors—such as test conditions or distractions—rather than actual changes in the trait being measured.<span><sup>14</sup></span> For instance, a student may perform differently on the same test when delivered in a formative assessment context versus under invigilated, high-stakes conditions. As a result, a scale validated in one context may not maintain its validity in another, requiring revalidation of the same scale in new contexts despite using identical items.</p><p>CTT-based approaches are also sample-dependent,<span><sup>15</sup></span> meaning reliability and validity estimates are specific to the population in which the scale was developed and may not generalise to other groups. This poses a challenge for applying uncertainty tolerance scales across a diversity of health professions learners. Meta-analyses of scales developed using CTT methods (e.g. syntheses of Cronbach's alpha) show wide variability in scale performance across populations, necessitating revalidation in each new setting.<span><sup>2</sup></span> Additionally, a scale developed using a CTT approach is not designed to account for context-driven response differences—such as those between low- (e.g. uncertainty with completing a puzzle) and high-stakes environments (e.g. patient care)—further limiting its ability to provide valid, transferable insights into learners' uncertainty tolerance across contexts.</p><p>A further conceptual limitation is the reliance on total scores, which assumes that all items equally reflect the construct, without accounting for item-level characteristics such as difficulty or discriminatory power—features we consider crucial for measuring the complex construct of uncertainty tolerance. This ‘total score reliance’ limits the sensitivity and precision of existing uncertainty tolerance scales in distinguishing between varying levels of the construct. Uncertainty tolerance often also varies between individuals, even when they encounter the same source of uncertainty. One person may respond confidently to a minor clinical unknown, while another may react with extreme anxiety. This highlights the subjective, dynamic nature of uncertainty tolerance. Scales developed using CTT approaches do not account for differences in test-takers based on item difficulty or relevance. As a result, instruments developed using CTT approaches are limited in their ability to reflect the nuanced, individualised and varied ways individuals experience and respond to different sources of uncertainty.<span><sup>15</sup></span></p><p>LTMs, including IRT and Rasch models, represent psychometric frameworks that model the relationship between an individual's latent traits and their responses to specific test items. These models estimate the probability of a particular response based on the respondent's underlying trait level and the item characteristics, such as item difficulty and discrimination. Unlike CTT approaches, LTMs offer advantages that make them well-suited for measuring complex, multidimensional constructs like uncertainty tolerance.</p><p>A key advantage of LTMs in measuring constructs like uncertainty tolerance is their emphasis on item-level analysis rather than reliance on total scores. In the context of uncertainty tolerance, different items may tap into diverse cognitive, emotional or behavioural reactions to uncertainty. LTMs, particularly IRT, allow researchers to evaluate how effectively each item distinguishes between individuals with varying levels of the trait. For example, an item reflecting emotional unease about incomplete information may be highly discriminative among mid-level uncertainty tolerance respondents but less so at the extremes (those intolerant of uncertainty, for instance). This type of item-level insight supports more precise scale development opportunities by allowing for retaining items that contribute meaningfully to construct measurement, rather than assuming all items are equally informative.</p><p>Another strength of LTMs is their ability to produce sample-independent, context-transferable measures. Once items are calibrated using IRT or Rasch analysis with good model fit, their psychometric properties remain stable across populations and settings. This is valuable for developing uncertainty tolerance assessments applicable across diverse health professions learners. For instance, a scale developed with Australian medical students may be extended to physiotherapy or nursing students in other countries. However, cross-group validation—including tests for measurement invariance and differential item functioning—remains essential to ensure fair and accurate interpretation.<span><sup>16</sup></span></p><p>LTMs allow for nuanced insights into individual responses to different sources of, for instance, uncertainty by estimating each respondent's position on the latent trait continuum in a manner that is independent of the test form. For instance, two learners may give the same response to an item on diagnostic ambiguity, but IRT can reveal meaningful differences in overall tolerance based on scale response patterns. This precision supports identifying subgroups and tailoring feedback or interventions. Given the nuanced nature of uncertainty tolerance responses—when emotional, cognitive and behavioural aspects of uncertainty tolerance vary across individuals—this could prove valuable in the field.</p><p>We encourage uncertainty tolerance researchers to consider LTM holistically from conception and design of the uncertainty tolerance scale through to evaluation. In this phase of transition, however, we suggest that researchers explore existing uncertainty tolerance scales, often developed through CTT approaches, with LTM evaluation strategies.</p><p>“You cannot swim for new horizons until you have courage to lose sight of the shore”—William Faulkner.</p><p>As William Faulkner suggests, the recent paper by Wegworth and colleagues<span><sup>1</sup></span> can serve as a ‘north star’ that inspires the field to lose sight of the shore (e.g. uncertainty tolerance scales developed and evaluated using CTT approaches) and move in the direction of a new horizon (e.g. development and evaluation of these scales using LTM approaches). When we take a closer look at the historical uncertainty tolerance scale “shore”, we see some significant challenges. These include a primary focus on measuring one feature of the construct (i.e. emotional responses), as well as the inability to tease out potential moderating factors (e.g. related to individual items) influencing an individual's perception and responses to uncertainty.<span><sup>17-19</sup></span></p><p>Given advances in measurement theory applicable to health professions education, there is an opportunity now to adjust our course towards a new horizon of LTM. LTM approaches could consider known relevant features of uncertainty tolerance including the emotional, cognitive and behavioural elements as well as the moderating factors which may be influencing an individual learners' response to uncertainty—and even consider how different sources of uncertainty may play a role in these responses. In this way, LTM approaches offer a way to advance the measurement of this complex construct.</p><p>Given the widespread inclusion of uncertainty tolerance as a health profession graduate attribute and the desire to measure uncertainty tolerance across health professions, there is a clear imperative to embrace more nuanced psychometric approaches. The assumptions of LTM demonstrate congruence with the modern conceptual uncertainty tolerance modelling and could allow for scale development and measurement which are theoretically grounded, empirically robust and responsive to the realities of modern health professions education. The field is ready to lose sight of the shore and swim for new horizons.</p><p><b>Michelle D. Lazarus:</b> Conceptualisation; writing—original draft; writing—review and editing. <b>Melanie Farlie:</b> Conceptualisation; writing—original draft; writing—review and editing. <b>Md Nazmul Karim:</b> Conceptualisation; writing—original draft; writing—review and editing.</p>\",\"PeriodicalId\":18370,\"journal\":{\"name\":\"Medical Education\",\"volume\":\"59 8\",\"pages\":\"787-791\"},\"PeriodicalIF\":5.2000,\"publicationDate\":\"2025-07-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://onlinelibrary.wiley.com/doi/epdf/10.1111/medu.15732\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Medical Education\",\"FirstCategoryId\":\"95\",\"ListUrlMain\":\"https://asmepublications.onlinelibrary.wiley.com/doi/10.1111/medu.15732\",\"RegionNum\":1,\"RegionCategory\":\"教育学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"EDUCATION, SCIENTIFIC DISCIPLINES\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Medical Education","FirstCategoryId":"95","ListUrlMain":"https://asmepublications.onlinelibrary.wiley.com/doi/10.1111/medu.15732","RegionNum":1,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"EDUCATION, SCIENTIFIC DISCIPLINES","Score":null,"Total":0}
引用次数: 0

摘要

基于ct的方法也是样本依赖的,15意味着信度和效度估计是特定于开发量表的人群的,可能不会推广到其他群体。这对在各种卫生专业学习者中应用不确定性容忍度量表提出了挑战。使用CTT方法开发的量表的荟萃分析(例如,Cronbach’s alpha的合成)显示,不同人群的量表表现存在很大差异,需要在每个新设置中重新验证此外,使用CTT方法开发的量表并不是为了考虑情境驱动的反应差异而设计的,例如低风险环境(例如完成谜题的不确定性)和高风险环境(例如患者护理)之间的反应差异,这进一步限制了其提供有效的、可转移的洞察力来了解学习者跨情境的不确定性容忍度的能力。进一步的概念限制是对总分的依赖,它假设所有项目都平等地反映了结构,而没有考虑项目层面的特征,如难度或歧视性权力特征,我们认为这对测量不确定性容忍的复杂结构至关重要。这种“总分依赖”限制了现有不确定性容忍度量表在区分不同层次结构时的灵敏度和精度。即使遇到相同的不确定性来源,个体对不确定性的容忍度也常常不同。一个人可能对一个小小的临床未知有自信的反应,而另一个人可能会极度焦虑。这突出了不确定性容忍的主观性和动态性。使用CTT方法开发的量表不考虑基于项目难度或相关性的考生差异。因此,使用CTT方法开发的工具在反映个体经历和应对不同不确定性来源的细微、个性化和多样化方式方面的能力有限。ltm,包括IRT和Rasch模型,代表了模拟个体潜在特征和他们对特定测试项目的反应之间关系的心理测量框架。这些模型根据被调查者的潜在特质水平和项目特征(如项目难度和歧视)来估计特定反应的概率。与CTT方法不同,ltm提供的优势使它们非常适合测量复杂的多维结构,如不确定性容忍。ltm在测量不确定性容忍度等结构方面的一个关键优势是它们强调项目层面的分析,而不是依赖于总分。在不确定性容忍的背景下,不同的项目可能会对不确定性产生不同的认知、情感或行为反应。ltm,尤其是IRT,使研究人员能够评估每个项目如何有效地区分具有不同特征水平的个体。例如,一个反映对不完整信息的情绪不安的项目可能在中等水平的不确定性容忍受访者中具有高度的歧视性,但在极端情况下(例如,那些不能容忍不确定性的人)则不那么具有歧视性。这种类型的项目级洞察力通过允许保留对构建度量有意义的贡献的项目来支持更精确的规模开发机会,而不是假设所有项目都具有相同的信息。ltm的另一个优点是它们能够产生与样本无关的、上下文可转移的度量。一旦项目使用具有良好模型拟合的IRT或Rasch分析进行校准,它们的心理测量特性在人群和环境中保持稳定。这对于开发适用于不同卫生专业学习者的不确定性容忍度评估是有价值的。例如,澳大利亚医科学生制定的量表可以推广到其他国家的理疗或护理专业学生。然而,跨组验证——包括测量不变性和差异项目功能的测试——仍然是确保公平和准确解释的必要条件。ltm允许对个体对不同来源的反应进行细致入微的洞察,例如,通过以独立于测试形式的方式估计每个被调查者在潜在特征连续体上的位置来确定不确定性。例如,两个学习者可能对诊断歧义的一个项目给出相同的反应,但IRT可以揭示基于量表反应模式的总体耐受性的有意义的差异。这种精确性支持识别子群体和定制反馈或干预。考虑到不确定性容忍反应的微妙性质——当不确定性容忍的情感、认知和行为方面因人而异时——这可能在该领域证明是有价值的。我们鼓励不确定性容忍度研究者从不确定性容忍度量表的构思、设计到评估,从整体上考虑长期管理。 然而,在这个过渡阶段,我们建议研究人员利用LTM评估策略探索现有的不确定性容忍量表,这些量表通常是通过CTT方法开发的。“除非你有勇气离开海岸,否则你无法游向新的地平线。”——威廉·福克纳。正如威廉·福克纳(William Faulkner)所建议的那样,Wegworth及其同事最近的论文1可以作为一颗“北极星”,激励该领域失去对海岸的关注(例如,使用CTT方法开发和评估的不确定性容忍量表),并朝着新的地平线方向前进(例如,使用LTM方法开发和评估这些量表)。当我们仔细审视历史上的不确定性容忍尺度“海岸”时,我们看到了一些重大挑战。其中包括主要关注于测量结构的一个特征(即情绪反应),以及无法梳理出影响个人感知和对不确定性反应的潜在调节因素(例如与单个项目相关)。17-19 .鉴于适用于卫生专业教育的测量理论的进步,现在有机会调整我们的课程,朝着LTM的新视野发展。LTM方法可以考虑不确定性容忍的已知相关特征,包括情绪、认知和行为因素,以及可能影响个体学习者对不确定性反应的调节因素,甚至考虑不确定性的不同来源如何在这些反应中发挥作用。通过这种方式,LTM方法提供了一种方法来推进这种复杂结构的测量。鉴于不确定性容忍度被广泛纳入卫生专业毕业生的属性,以及衡量卫生专业不确定性容忍度的愿望,显然有必要采用更细致入微的心理测量方法。LTM的假设证明了与现代概念不确定性容忍模型的一致性,并且可以允许规模开发和测量,这些开发和测量在理论上是有基础的,经验上是强大的,并且对现代卫生专业教育的现实作出反应。田野即将失去海岸的视线,向新的地平线游去。Michelle D. Lazarus:概念化;原创作品草案;写作-审查和编辑。Melanie Farlie:概念化;原创作品草案;写作-审查和编辑。Md Nazmul Karim:概念化;原创作品草案;写作-审查和编辑。
本文章由计算机程序翻译,如有差异,请以英文原文为准。

New horizons: Evolving approaches to uncertainty tolerance scale development

New horizons: Evolving approaches to uncertainty tolerance scale development

New horizons: Evolving approaches to uncertainty tolerance scale development

New horizons: Evolving approaches to uncertainty tolerance scale development

The paper titled “Tolerance for uncertainty and medical students' specialty choices: A myth revisited” by Wegwarth and colleagues1 provides additional evidence that the existing approaches for uncertainty tolerance scale development and evaluation in health professional learners2 and professionals3 may not be the optimal approach for measuring this complex construct. In this commentary, we discuss the predominant approaches currently taken in developing and evaluating uncertainty tolerance using scales and suggest ways to evolve these strategies to optimise measurement validity of this construct.

Uncertainty tolerance is defined as individuals' perceptions and responses to sources of uncertainty (such as ambiguity, probability or complexity).4 An individuals' response to uncertainty is expressed by how they think, feel and act. Uncertainty tolerance construct modelling by Hillen and colleagues (2017)4 describes a multi-component construct comprised of (1) the source of uncertainty, (2) how this source is perceived by an individual and (3) each subsequent response (emotions, thoughts and behaviours).4 However, the degree to which these elements contribute to the overall measurement of this construct remains (ironically) uncertain.2, 3 Further complicating this is the question as to the changeability of one's uncertainty tolerance, which influences measurement approaches. In the mid-20th Century, uncertainty tolerance was conceptualised as a static personality trait.4 Mounting contemporary evidence, however, suggests that uncertainty tolerance is a dynamic and changeable state-based construct influenced by contextual and personal factors such as prior experiences, geographic location and reflective capacity.5

There is evidence that uncertainty tolerance is, at least in part, state-based in learners. Learners can feel negatively about an experience of uncertainty but can act adaptively (e.g. with uncertainty tolerance) in response through their cognition and behaviour.6 For example, a medical student evaluating a patient with fatigue, fever and joint pain may initially feel anxious, due to the broad range of potential differential diagnoses (stimulus). However, by relying on their training and prior experiences (e.g. moderators) they can adaptively respond—despite their feelings of anxiety (emotional response)—by gathering a history, recommending appropriate tests and consulting with peers and colleagues about the next step to take (behavioural responses) and feeling confident about these next steps in supporting the patient (cognitive response). If this students' uncertainty tolerance was being measured, would they be evaluated as intolerant of uncertainty because of their anxiety or tolerant of uncertainty because of their adaptive behaviours and cognition? This is where the field remains conflicted and where existing scales may be falling short.7

Most uncertainty tolerance scales were developed and tested using classical test theory (CTT) approaches and prior to the modern multi-component conceptual model of uncertainty tolerance.4 Due to this, existing scales have a tendency towards numerous items focusing on one element of the construct—that of emotional responses and often fail to capture the complex and nuanced nature of the construct.3

As managing uncertainty becomes increasingly recognised as a health professions' competency,8-11 the desire to measure learners' uncertainty tolerance has equally gained attention—yet multiple studies, including the paper by Wegwarth et al (2025),1 call into question the utility of these scales.

Psychological and educational measurement instruments are primarily situated within two psychometric worldviews: CTT and latent trait modelling (LTM), such as Item Response Theory (IRT). Both CTT and IRT approaches aim to establish the reliability and validity of scales.12 While CTT is the most widely used approach for developing and testing existing uncertainty tolerance scales, it presents significant limitations when applied to assessments of health professions learners' uncertainty tolerance,13 and these limitations may contribute to the challenges we have in measuring uncertainty tolerance in this population.

CTT assumes that an observed score reflects a fixed true score plus random error, with the error considered normally distributed and unrelated to the true score. Under this model, score fluctuations across repeated testing are attributed to external factors—such as test conditions or distractions—rather than actual changes in the trait being measured.14 For instance, a student may perform differently on the same test when delivered in a formative assessment context versus under invigilated, high-stakes conditions. As a result, a scale validated in one context may not maintain its validity in another, requiring revalidation of the same scale in new contexts despite using identical items.

CTT-based approaches are also sample-dependent,15 meaning reliability and validity estimates are specific to the population in which the scale was developed and may not generalise to other groups. This poses a challenge for applying uncertainty tolerance scales across a diversity of health professions learners. Meta-analyses of scales developed using CTT methods (e.g. syntheses of Cronbach's alpha) show wide variability in scale performance across populations, necessitating revalidation in each new setting.2 Additionally, a scale developed using a CTT approach is not designed to account for context-driven response differences—such as those between low- (e.g. uncertainty with completing a puzzle) and high-stakes environments (e.g. patient care)—further limiting its ability to provide valid, transferable insights into learners' uncertainty tolerance across contexts.

A further conceptual limitation is the reliance on total scores, which assumes that all items equally reflect the construct, without accounting for item-level characteristics such as difficulty or discriminatory power—features we consider crucial for measuring the complex construct of uncertainty tolerance. This ‘total score reliance’ limits the sensitivity and precision of existing uncertainty tolerance scales in distinguishing between varying levels of the construct. Uncertainty tolerance often also varies between individuals, even when they encounter the same source of uncertainty. One person may respond confidently to a minor clinical unknown, while another may react with extreme anxiety. This highlights the subjective, dynamic nature of uncertainty tolerance. Scales developed using CTT approaches do not account for differences in test-takers based on item difficulty or relevance. As a result, instruments developed using CTT approaches are limited in their ability to reflect the nuanced, individualised and varied ways individuals experience and respond to different sources of uncertainty.15

LTMs, including IRT and Rasch models, represent psychometric frameworks that model the relationship between an individual's latent traits and their responses to specific test items. These models estimate the probability of a particular response based on the respondent's underlying trait level and the item characteristics, such as item difficulty and discrimination. Unlike CTT approaches, LTMs offer advantages that make them well-suited for measuring complex, multidimensional constructs like uncertainty tolerance.

A key advantage of LTMs in measuring constructs like uncertainty tolerance is their emphasis on item-level analysis rather than reliance on total scores. In the context of uncertainty tolerance, different items may tap into diverse cognitive, emotional or behavioural reactions to uncertainty. LTMs, particularly IRT, allow researchers to evaluate how effectively each item distinguishes between individuals with varying levels of the trait. For example, an item reflecting emotional unease about incomplete information may be highly discriminative among mid-level uncertainty tolerance respondents but less so at the extremes (those intolerant of uncertainty, for instance). This type of item-level insight supports more precise scale development opportunities by allowing for retaining items that contribute meaningfully to construct measurement, rather than assuming all items are equally informative.

Another strength of LTMs is their ability to produce sample-independent, context-transferable measures. Once items are calibrated using IRT or Rasch analysis with good model fit, their psychometric properties remain stable across populations and settings. This is valuable for developing uncertainty tolerance assessments applicable across diverse health professions learners. For instance, a scale developed with Australian medical students may be extended to physiotherapy or nursing students in other countries. However, cross-group validation—including tests for measurement invariance and differential item functioning—remains essential to ensure fair and accurate interpretation.16

LTMs allow for nuanced insights into individual responses to different sources of, for instance, uncertainty by estimating each respondent's position on the latent trait continuum in a manner that is independent of the test form. For instance, two learners may give the same response to an item on diagnostic ambiguity, but IRT can reveal meaningful differences in overall tolerance based on scale response patterns. This precision supports identifying subgroups and tailoring feedback or interventions. Given the nuanced nature of uncertainty tolerance responses—when emotional, cognitive and behavioural aspects of uncertainty tolerance vary across individuals—this could prove valuable in the field.

We encourage uncertainty tolerance researchers to consider LTM holistically from conception and design of the uncertainty tolerance scale through to evaluation. In this phase of transition, however, we suggest that researchers explore existing uncertainty tolerance scales, often developed through CTT approaches, with LTM evaluation strategies.

“You cannot swim for new horizons until you have courage to lose sight of the shore”—William Faulkner.

As William Faulkner suggests, the recent paper by Wegworth and colleagues1 can serve as a ‘north star’ that inspires the field to lose sight of the shore (e.g. uncertainty tolerance scales developed and evaluated using CTT approaches) and move in the direction of a new horizon (e.g. development and evaluation of these scales using LTM approaches). When we take a closer look at the historical uncertainty tolerance scale “shore”, we see some significant challenges. These include a primary focus on measuring one feature of the construct (i.e. emotional responses), as well as the inability to tease out potential moderating factors (e.g. related to individual items) influencing an individual's perception and responses to uncertainty.17-19

Given advances in measurement theory applicable to health professions education, there is an opportunity now to adjust our course towards a new horizon of LTM. LTM approaches could consider known relevant features of uncertainty tolerance including the emotional, cognitive and behavioural elements as well as the moderating factors which may be influencing an individual learners' response to uncertainty—and even consider how different sources of uncertainty may play a role in these responses. In this way, LTM approaches offer a way to advance the measurement of this complex construct.

Given the widespread inclusion of uncertainty tolerance as a health profession graduate attribute and the desire to measure uncertainty tolerance across health professions, there is a clear imperative to embrace more nuanced psychometric approaches. The assumptions of LTM demonstrate congruence with the modern conceptual uncertainty tolerance modelling and could allow for scale development and measurement which are theoretically grounded, empirically robust and responsive to the realities of modern health professions education. The field is ready to lose sight of the shore and swim for new horizons.

Michelle D. Lazarus: Conceptualisation; writing—original draft; writing—review and editing. Melanie Farlie: Conceptualisation; writing—original draft; writing—review and editing. Md Nazmul Karim: Conceptualisation; writing—original draft; writing—review and editing.

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
Medical Education
Medical Education 医学-卫生保健
CiteScore
8.40
自引率
10.00%
发文量
279
审稿时长
4-8 weeks
期刊介绍: Medical Education seeks to be the pre-eminent journal in the field of education for health care professionals, and publishes material of the highest quality, reflecting world wide or provocative issues and perspectives. The journal welcomes high quality papers on all aspects of health professional education including; -undergraduate education -postgraduate training -continuing professional development -interprofessional education
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信