《Did I Buy the Wrong Gadget?How the Evaluability of Technology Features Influences...》中英文对比文献翻译

引用：Valacich J S , Wang X , Jessup L M . Did I Buy the Wrong Gadget? How the Evaluability of Technology Features Influences Technology Feature Preferences and Subsequent Product Choice[J]. MIS Quarterly, 2018, 42(1):633-644.

Abstract

Prior usability assessment research has paid little attention to how product and feature ratings are influenced by the evaluation context. However, the evaluability hypothesis, which guides this research, suggests that the evaluation context is a vital factor in shaping user’s assessments and perceptions about technology features. pacifically, the evaluability hypothesis proposes that technology feature perceptions, and ultimately technology choices, will change when evaluating a single technology in isolation versus when simultaneously comparing more than one. To demonstrate the evaluability hypothesis effect in the context of consumer technology product evaluations, two experiments were conducted. Both studies support the evaluability hypothesis effect, showing that when two IT products are compared, hard-to-evaluate but easy-to-compare features are perceived to be more important and therefore have a larger influence on product preferences. Alternatively, when evaluating a single product in isolation, easy-to-evaluate features are perceived to be more important and therefore have a larger influence on product preferences. Consequently, different product preferences emerge (i.e., preference reversals) in different evaluation contexts. The results demonstrate that this theoretical lens is robust to the technology evaluation context, providing important theoretical and practical insights for technology design, usability assessments, and, ultimately, product acceptance.

摘要：

之前的可用性评估研究很少关注评估环境如何影响产品和特性评估。然而，指导这项研究的可评估性假设表明，评估环境是形成用户对技术特性的评估和感知的重要因素。具体来说，可评估性假说提出，当单独评估一项技术时，与同时比较多项技术时相比，技术特征感知以及最终的技术选择将发生变化。为了验证消费者技术产品评估中的可评估性假设的影响，我们进行了两个实验。这两项研究都同意可评估性假设的影响，表明当两个IT产品进行比较时，难以评估但易于比较的特征被认为更重要，因此对产品偏好的影响更大。另外，当单独评估一个产品时，容易评估的特性被认为更重要，因此对产品偏好有更大的影响。因此，在不同的评估环境中会出现不同的产品偏好（即偏好逆转）。结果表明，该理论视角对技术评估环境具有很强的作用，为技术设计、可用性评估以及最终的产品接受度提供了重要的理论和实践见解。

1.Introduction

Guided by research and practice in human–computer interaction (HCI), technology designers pursue high usability, which depends on technology attribute categories such as content, ease of use, and promotion (Agarwal and Venkatesh 2002). To achieve high usability, designers often conduct evaluations to examine the relative importance of various technology features (Benlian and Hess 2011; Keil and Tiwana 2006). However, because people often have different preferences for technology features in different evaluation contexts (Valacich et al. 2007), understanding the context of a product’s evaluation is an important part of the usability assessment process. For example, if a user perceives the Fitbit Alta to have better social media support than the Nike FuelBand, a person most interested in using social media to share exercise updates with friends may prefer the Alta, which provides strong features to support social activities including leaderboards, status updates, automatic posts to Facebook, and so on. Alternatively, if another person perceives the Nike FuelBand to have greater movement tracking accuracy than the Alta, a person most interested in activity tracking accuracy would likely prefer the FuelBand (all other things being equal).

1.引言

在人机交互（HCI）的研究和实践的指导下，技术设计师追求高可用性，这取决于诸如内容，易用性和推广等的技术属性类别（Agarwal和Venkatesh 2002）。为了实现高可用性，设计人员经常进行评估，以检验各种技术特性的相对重要性（Benlian和Hess，2011； Keil和Tiwana，2006）。然而，由于人们在不同的评估环境中对技术特性有不同的偏好(Valacich et al. 2007)，了解产品评估的环境是可用性评估过程中的重要部分。例如，如果用户认为Fitbit Alta比Nike FuelBand具有更好的社交媒体支持，那么最有兴趣使用社交媒体与朋友分享锻炼最新信息的人可能会喜欢Alta，它提供强大的功能来支持包括排行榜的社交活动，状态更新，自动发布到Facebook等。或者，如果另一个人认为Nike FuelBand具有比Alta更高的运动跟踪精度，那么对活动跟踪精度最感兴趣的人可能会更喜欢FuelBand（在所有其他条件相同的情况下）。

Another important contextual factor influencing usability evaluations relates to whether a person is considering a product in isolation (e.g., “Should I buy a Fitbit?”) or comparing multiple products, which is often the case in more mature product categories (e.g., “Do I like Fitbit Alta or Nike FuelBand?”). As such, it is possible that people may perceive one technology feature highly important when considering a single technology in isolation, but may not value that feature as much when simultaneously comparing multiple IT products. This issue is further exacerbated given that multiple generations of products frequently coexist in the market (Xu et al. 2010). Consequently, designers must consider various contextual factors (e.g., isolated evaluation, product comparison, primary use) when assessing product features or else misleading evaluations may occur, leading to costly errors in the design, production, and marketing of products.
影响可用性评估的另一个重要的环境因素涉及到一个人是否在单独考虑一个产品(例如，“我应该买一个Fitbit吗?”)或比较多个产品，这种情况通常出现在更成熟的产品类别(例如，“我喜欢Fitbit Alta还是Nike FuelBand?”) 因此，在单独考虑单个技术时，人们可能会认为一个技术特性非常重要，但是在同时比较多个it产品时，可能不会那么重视该特性。这个问题进一步恶化了，因为在市场上，多代产品经常共存(Xu et al. 2010)。因此，设计师在评估产品特性时必须考虑各种环境因素(例如，孤立的评价、产品比较、主要用途)，否则可能会产生误导性的评价，导致产品在设计、生产和营销中出现代价高昂的错误。

Prior research on technology adoption has focused primarily on identifying various factors, such as technology features, that influence why individuals adopt a single technology (Venkatesh et al. 2003; Venkatesh, Thong and Xia 2012; Venkatesh et al. 2016; Wixom and Todd 2005). Relatively little is known, however, regarding how people evaluate technology when more than one product option is being considered. Examining how technologies are evaluated in different contexts can extend the prior research on HCI design, usability, and technology adoption. To address these gaps, our guiding research question is: Does the relative importance of the technology features influencing an IT product evaluation change when evaluating a single product in isolation versus when evaluating two products in comparison?
先前对技术采用的研究主要集中在确定影响个人采用单一技术的原因的各种因素（例如技术特性）（Venkatesh等人，2003； Venkatesh，Thong和Xia，2012； Venkatesh等，2016； Wixom和Todd，2005）。 ）。然而，关于人们在考虑多种产品选择时如何评价技术，人们所知相对较少。研究如何在不同的环境中评估技术，可以扩展之前关于HCI设计、可用性和技术采用的研究。为了解决这些差距，我们的指导研究问题是:在单独评估单个产品时，与在比较中评估两个产品时，影响IT产品评估的技术特性的相对重要性是否会发生变化?

To understand how people’s preference toward technology features and subsequent product evaluation differ when making an isolated evaluation versus when comparing multiple products, we draw upon the evaluability hypothesis (EH) (Hsee 1996; Hsee et al. 1999; Hsee et al. 2013). The EH suggests that when evaluating a product in isolation, people tend to perceive easy-to-evaluate features (e.g., aesthetics2) as being more important and therefore more heavily consider those features. However, when simultaneously comparing multiple products, people tend to perceive features that are hard to evaluate in isolation, but easy to compare (e.g., screen resolution), as being more important and therefore more heavily consider those features. Consequently, when a product is being assessed in isolation versus when being compared to another, different technology feature, preferences can emerge. To explore the EH within the context of IT-based consumer products, we report two experimental studies.
为了了解人们对技术特性和后续产品评价的偏好在进行单独评价和比较多个产品时的差异，我们采用了可评估性假设（EH）（Hsee 1996; Hsee et al。1999; Hsee et al。2013）。EH表明，当单独评估一个产品时，人们倾向于将易于评估的功能(如审美)视为更重要的特性，因此更重视这些特征。然而，当同时比较多个产品时，人们倾向于认为很难单独评估但容易比较的特性(如屏幕分辨率)更重要，因此更重视这些特性。因此，当单独评估一个产品和与其它产品对比评估的情况出现时，对于不同的技术特性，可能会出现偏好。为了在基于IT的消费品环境中探索EH，我们报告了两个实验研究。

Our work makes two important contributions. First, we contribute to the existing HCI and usability literature by highlighting how differences in the evaluation context can shape assessments and outcomes. Specifically, we show that people’s preferences toward technology features differ when considering a single product in isolation versus when considering two products in comparison. This in turn helps us provide important practical guidelines for IT design. Second, our work contributes to the existing technology adoption literature by demonstrating how IT product evaluation and preferences change in different contexts. The implications of our research for the broader technology adoption literature could be profound, providing both an important boundary condition for existing theoretical perspectives and methodological insights for the design of future studies.
我们的工作做出了两个重要贡献。首先，我们通过强调评估环境中的差异如何影响评估和结果，为现有的HCI和可用性文献做出贡献。具体地说，我们表明，人们对技术特性的偏好在单独考虑单个产品时与在比较中考虑两个产品时是不同的。这反过来又帮助我们为IT设计提供重要的实用指南。其次，我们的工作通过展示IT产品评估和偏好如何在不同的环境中变化，为现有的技术采用文献做出了贡献。我们的研究对更广泛的技术采用文献的影响可能是深远的，既为现有的理论观点提供了重要的边界条件，也为未来研究的设计提供了方法上的见解。

2.Literature Review and Hypotheses Development

During usability testing, evaluators assess various product features（ According to ISO/IEC 9126 (ISO 1991), a feature represents "a set of attributes … which bear on the effort needed for use and on the individual assessment of such use " (Bevan and Macleod 1994, p. 136).） to determine their perceived importance in a specific context. As discussed, the evaluation context can significantly influence the assessment of a technology and its various features. Carefully matching the evaluation criteria, and the relative weightings of such criteria, to the context is crucial for gaining an accurate assessment. Unfortunately, much of the prior relevant literature has failed to adjust the criteria and weightings in differing contexts, such as when evaluating a single technology in isolation versus when comparing multiple technologies. For example, Venkatesh and Ramesh (2006) examined various criteria in a desktop versus device evaluation context. In their study, multiple websites were presented to study participants, and it is not clear whether assessments would hold if only one website was evaluated. In other words, the results from one evaluation context (e.g., evaluating and comparing multiple technologies) may not transfer to other evaluation contexts (e.g., evaluating a single technology).

2.文献回顾与假设发展

在可用性测试期间，评估人员评估各种产品特性(根据ISO/IEC 9126 (ISO 1991)，一个特性代表“一组属性……取决于使用所需的工作量以及对此类使用的单独评估”（Bevan和Macleod 1994，第136页）以确定它们在特定环境下的重要性。如前所述，评价环境对技术及其各种特性的评价有显著影响。仔细地将评价标准和这些标准的相对权重与环境相匹配，对于获得准确的评价是至关重要的。然而，许多现有的相关文献未能在不同的环境中调整标准和权重（例如在单独评估单个技术时与在比较多个技术时）。例如，Venkatesh和Ramesh（2006）在台式机与设备评估环境中检查了各种标准。在他们的研究中，研究人员向参与者展示了多个网站，如果只评估一个网站，评估是否有效还不清楚。换句话说，来自一种评估环境（例如，评估和比较多种技术）的结果可能不会转移到其他评估环境（例如，评估一种技术）。

Additionally, depending upon the context, some features are easier to evaluate than others. For instance, according to Hsee et al. (1999), a technology feature can be hard or easy to evaluate independently, which is referred to as its evaluability. Specifically, the evaluability of a feature is determined by a person’s ability to independently map a given value assessment of a feature onto an evaluation scale (Hsee et al. 1999). Thus, “hard-to evaluate” means that “the evaluator does not know how good a given value on the attribute is without comparison,” and “easy-to-evaluate” refers to conditions where “the evaluator knows how good the value is” without comparison (Hsee 1996, p. 249). Consequently, hard to-evaluate does not mean that individuals do not know the value of a feature but that individuals have “difficulty determining the desirability of its value in the given decision context” (Hsee et al. 1999, p. 580). Thus, people can ascertain the value of a hard-to-evaluate feature but are unable to determine its worth without comparison (Hsee and Zhang 2004).
此外，根据上下文，有些特性比其他特性更容易评估。例如，Hsee et al.(1999)认为，技术特征可以是难以评价的，也可以是容易评价的，这被称为技术特征的可评价性。具体来说，一个特征的可评估性取决于个人将要素的给定价值评估独立映射到评估量表上的能力（Hsee等，1999）。因此，“难以评估”意味着“在没有比较的情况下评估者不知道特征上的给定值有多好”，而“易于评估”指的是“在没有比较的情况下评估者知道值有多好”(参见1996，第249页)。因此，难以评价并不意味着个体不知道特征的价值，而是个体“在给定的决策环境中难以确定其价值的可取性”(Hsee et al. 1999, p. 580)。因此，人们可以确定一个难以评估的特征的价值，但是没有比较就无法确定其价值(Hsee和Zhang 2004)。

Hsee and colleagues (1996; 1999; 2013) further argue that how individuals assess a hard- versus an easy-to-evaluate feature is different, depending on the evaluation context. When assessing a single product (i.e., a separate evaluation (SE))— because a hard-to-evaluate feature without comparison is hard to assess – people are more likely to base their evaluation on an easy-to-evaluate feature (e.g., aesthetics). Alternatively, when comparing two products (i.e., a joint evaluation (JE)), people can compare both easy- and hard-to-evaluate features. An example of a JE evaluation from Venkatesh and Ramesh is where participants evaluated multiple websites. As such, because hard-to-evaluate features (e.g., screen resolution, processing clock speed, and storage capacity) become more salient when being compared, they have a greater influence on any product assessment (Hsee et al. 1999). Because of the difference in evaluation focus between JE and SE contexts, a person’s perceptions toward the same product can change due to the shift of the focal product feature used in making the assessment. Ultimately, such changes in focus can lead to preference reversals (PRs). The EH theoretical lens has been found to be robust to a wide variety of evaluation contexts, such as payoffs of dispute settlements, job offers, and servings of ice cream (e.g., Bazerman et al. 1999; GonzálezVallejoa and Moran 2001; Hsee and Rottenstreich 2004; Todorov et al. 2007; Wilson and Arvai 2006). As such, the EH can provide insights for designers when conducting a heuristic evaluation or other types of usability evaluations in different contexts. Specifically, assessment of a feature can change across evaluation contexts (SE versus JE), even if everything else remains the same.4
Hsee及其同事(1996年;1999;2013年)进一步认为，个体如何评估一个困难的和一个容易评估的特征是不同的，这取决于评估的环境。评估单一产品（即单独的评估（SE））时-因为一个难以评估的特征如果不进行比较是很难评估的-人们更倾向于基于一个易于评估的特征(例如，审美)来进行评估。另外，在比较两种产品（即联合评估（JE））时，人们可以比较易于评估和难以评估的功能。Venkatesh和Ramesh对JE进行评估的一个例子是，参与者评估了多个网站。这样一来，由于在比较时难以评估的功能（例如屏幕分辨率，处理时钟速度和存储容量）变得更加突出，因此它们对任何产品评估都具有更大的影响力（Hsee et al.1999）。由于JE和SE环境下评估重点的不同，一个人对同一产品的看法可能会因进行评估时使用的重点产品特性的改变而改变。最终，这种关注点的改变会导致偏好逆转(PRs)。人们发现，EH理论的视角在各种各样的评估环境中都是稳健的，比如解决争端的报酬、工作机会和冰淇淋的分量(例如，Bazerman等人1999年;GonzalezVallejoa和Moran 2001;Hsee和Rottenstreich 2004;Todorov et al. 2007;Wilson和Arvai 2006)。这样，当在不同环境中进行启发式评估或其他类型的可用性评估时，EH可以为设计师提供见解。具体而言，即使其他条件保持不变，对特征的评估也会在评估环境（SE与JE）之间发生变化。

In sum, the EH proposes that individuals focus on different technology features in different contexts (i.e., when evaluating a single product in isolation versus when simultaneously evaluating two products). When evaluating a single product, individuals rely more on easy-to-evaluate features and perceive those features more important. In such a scenario, technology products with more favorable easy-to-evaluate features will more likely receive a higher evaluation. In contrast, when evaluating two products, individuals rely more on hard-to-evaluate, but easy-to-compare, features and perceive those features more important. In such a scenario, technology products with more favorable hard-to-evaluate features will likely receive a higher evaluation. Such changes in focus can lead to a PR. Thus, we propose
综上所述，EH建议人们在不同的环境下关注不同的技术特性（即，单独评估一个产品时与同时评估两个产品时）。在评估单个产品时，个人更多地依赖于易于评估的特性，并且认为这些特性更重要。在这种情况下，具有更易于评估的特性的技术产品更有可能获得更高的评价。相比之下，在评估两种产品时，人们更多地依赖于难以评估但易于比较的特征，以及那些更重要的特征。在这种情况下，具有更佳难以评估特性的技术产品可能会得到更高的评价。焦点的这种变化可能导致PR。 因此，我们建议

H1: When simultaneously evaluating two IT products, the IT product with the more favorable hard-to-evaluate, but easy-to-compare feature, will be evaluated higher.
H2: When independently evaluating two IT products, the IT product with the more favorable easy-to-evaluate feature will be evaluated higher.
H1:在同时评价两个IT产品时，有较难评价但易于比较的特征的IT产品评价较高。
H2:在独立评价两个IT产品时，有更容易评价特征的IT产品评价更高。

3.Methodology and Results

We first describe two exploratory pilot studies. We then present the two primary experiments that build on the pilot studies and directly test our hypotheses.

3.方法与结果

我们首先描述两个探索性试点研究。 然后，我们介绍基于试点研究提出两个主要实验，并直接检验我们的假设。

3.1Exploratory Pilot Studies

Prior to conducting the main experiments, we ran two exploratory pilot studies to examine how people would perceive different technology features when evaluating Microsoft PowerPoint (PPT) versus Visio and the Apple iPad versus the Motorola Xoom (see Appendices A and B for the details). These pilot studies allowed us to refine our tasks, experimental procedures, and measures for the main studies, and are also reported for both completeness and for aiding future research. In both pilot studies, we found the EH to be partially supported. When both options were simultaneously compared, as expected, both PPT and the iPad were evaluated higher than Visio and Xoom, respectively; however, we did not see a product preference reversal.

3.1探索试点研究

在进行主要实验之前，我们进行了两个探索性的试点研究，以检验人们在评价Microsoft PowerPoint (PPT)与Visio、苹果iPad与摩托罗拉Xoom时如何看待不同的技术特性(详见附录A和B)。这些试点性研究使我们能够完善主要研究的任务，实验程序和措施，并报告其完整性和对未来研究的帮助。在两项试点研究中，我们发现EH均得到部分支持。 如预期的那样，当同时比较这两个选项时，PPT和iPad的评估分别高于Visio和Xoom。 但是，我们没有看到产品偏好反转。

In hindsight, a problem with utilizing “known” products (e.g., that have different brand awareness, market shares, and features) is that we could not control, or adequately account for, a subject’s prior relative knowledge or opinions of these commercial products. Additionally, because we highlighted real product features in these evaluations, whether a feature was easy- or hard-to-evaluate was assumed based on the guidance of the prior literature rather than being carefully manipulated. Thus, because participants were less familiar with Visio and Xoom, they were also less likely to fully understand the difficulty to evaluate features and/or simultaneously suffer from a bias toward the more popular products (Erdem and Swait 1998). These results suggest that it may be difficult to overcome differences in brand awareness without a more sophisticated research approach (e.g., anonymize the brand). To overcome the limitations of the pilot studies and following the guidance of the existing EH literature, we chose a “brandless” product category (i.e., wireless Internet service) where brand awareness could be controlled and the easy- and hard to-evaluate product features could be carefully manipulated.
事后看来，使用“已知”产品（例如具有不同的品牌知名度，市场份额和功能）的问题是，我们无法控制或充分考虑受试者对这些商业产品的先前相对知识或看法。另外，因为我们在这些评估中强调了真实的产品特性，所以一个特性是容易还是难以评估是基于先前文献的指导而不是仔细操作的。因此，由于参与者不太熟悉Visio和Xoom，他们也不太可能完全理解评估功能的困难和/或同时对更受欢迎的产品有偏好(Erdem和Swait 1998)。这些结果表明，如果没有更复杂的研究方法(例如，将品牌匿名化)，可能很难克服品牌知名度的差异。为了克服试点研究的局限性，在现有EH文献的指导下，我们选择了一种“无品牌”产品类别（即无线互联网服务），在该类别中可以控制品牌知名度，并且产品特性的评估难易可以得到精心的处理。

3.2 Experiment 1

Task Context Development
While there are many product features that could influence a person’s choice making, the EH literature recommends that a hard-to-evaluate feature be “sufficiently important” so that the product with the superior performance on this attribute will be favored when making a product comparison assessment (i.e., a JE context) (González-Vallejo and Moran 2001). The hard to-evaluate feature should also be sufficiently difficult for most consumers to assess without comparison. For example, from the EH literature, hard-to-evaluate features have routinely been operationalized as numeric values related to a particular product feature (e.g., a screen resolution of 1920 × 1080 versus 1366 × 768; a product weight of 2.5 versus 1.2 kg). Likewise, when selecting an attribute when evaluating a product in isolation (i.e., a SE context), the feature should again be important and relatively easy to evaluate without comparison. Again, using guidance from the literature (Hsee 1996), easy-to-evaluate features are typically operationalized as dichotomous factors (e.g., exists versus doesn’t exist; low versus high).

3.2实验1

任务环境开发
虽然有许多产品特性可能会影响一个人的选择，但是EH文献建议一个难以评估的特性是“足够重要的”，这样在进行产品比较评估时，在这个属性上性能更好的产品将会受到青睐(例如，(Gonzalez-Vallejo和Moran 2001)。难以评估的特性对于大多数消费者来说，如果不进行比较，也很难进行评估。例如，在EH文献中，难以评估的特性通常被操作化为与特定产品特性相关的数值(例如，屏幕分辨率为1920×1080，而不是1366×768;产品重量为2.5千克对1.2千克)。同样，在单独评估产品（即SE环境）时选择属性时，该特性应该同样重要，并且无需比较即可轻松评估。再次，根据文献(Hsee 1996)的指导，易于评估的特征通常被操作化为二分因素(例如，存在与不存在;低和高)。

Based on lessons learned from our pilot studies, Experiment 1 examined how participants evaluated different wireless Internet services. To understand how technology features influence product choice decisions within the context of wireless Internet services for our target population (i.e., what are the important features), we surveyed 33 undergraduate students from a college-wide, junior-level business class at a large public university in the northeast region of China. Students were asked to list at least two features which they thought were important for wireless Internet services (see Table 1 for a list of features and their frequency of occurrence). Based on the results of this survey and following the guidance of the extant literature, we chose connection speed as the hard-to-evaluate feature because this feature was an extremely important feature, it is almost always reported numerically, and would be hard to evaluate without comparison (e.g., 100 versus 50 mbps). Additionally, given the timeliness of cybercrime and its importance as well as the ability to represent this feature as a dichotomous variable, we chose secured connection as the easy-to evaluate feature (e.g., secured versus not secured connection).
根据我们从试点研究中获得的经验教训，实验1研究了参与者如何评估不同的无线互联网服务。在无线互联网服务范围内，为了了解技术特征如何影响目标人群的产品选择决策（即重要特征是什么），我们调查了中国东北地区一所大型公立大学的33名本科生，他们来自一个全学院的初级商务班。要求学生列出至少两个他们认为对无线Internet服务很重要的特性(特性及其出现频率的列表见表1)。根据调查结果，并根据现有文献的指导，我们选择连接速度作为难以评估的特性，因为该特性是非常重要的特性，几乎总是以数字形式报告，并且很难评估无需比较（例如100与50 mbps）。此外，考虑到网络犯罪的及时性及其重要性，以及将该特征表示为二分类变量的能力，我们选择了安全连接作为易于评估的特征(例如，安全连接与不安全连接)。

Consider a context where two wireless Internet service options, Plan A and Plan B, are assessed. Here, Plan A is faster, but has no embedded security; Plan B is slower, but provides integrated security. When evaluating either Plan A or Plan B in isolation, individuals are more likely to perceive the security feature (easy-to-evaluate) more important. However, when evaluating both plans simultaneously, individuals are more likely to perceive the connection speed differences (hard-to-evaluate) more important. Thus, Plan A will be evaluated higher when jointly compared (H1); Plan B will be evaluated higher when evaluated separately (H2).
考虑一个环境，其中评估了两个无线Internet服务选项，方案a和方案B。在这里，方案A更快，但没有嵌入式安全;方案B比较慢，但是提供了集成的安全性。当单独评估方案A或方案B时，个人更可能认为安全特性(易于评估)更重要。然而，当同时评估两个计划时，个体更可能认为连接速度差异(难以评估)更重要。因此，联合比较时，方案A的评价会更高(H1);方案B单独评估时的评估值更高(H2)。

Sample
The participants were from a college-wide, junior-level business class at a large public university in the northeast region of China; 69 students voluntarily participated in the study. Participants’ age ranged between 20 and 22, and their average age was 21.14 (SD .72); 59.15% were women. Each participant received about 1% of their final course grade for participating in the experiment.
样本
参与者来自中国东北地区一所大型公立大学的一个全学院范围的初级商务班;69名学生自愿参加了研究。参加者的年龄介乎20至22岁，平均年龄为21.14岁(SD .72);59.15%是女性。每个参与实验的学生都得到了他们最终课程成绩的1%。

Measures
The technology evaluation measure was developed by following the guidance from the extant literature (e.g., Dishaw and Strong 1999; Venkatesh et al. 2003); that is, “the degree to which a person has a favorable or unfavorable evaluation or appraisal” (Ajzen 1991, p. 188). Technology experience was collected to access participants’ self-reported background and experience with the focal technology (see Appendix C).
指标
技术评价指标是根据现有文献（例如Dishaw和Strong，1999； Venkatesh等，2003）的指导制定的。就是说，“一个人对评价或评价的好坏程度”（Ajzen 1991，第188页）。收集技术经验以获取参与者的自我报告背景和对焦点技术的经验（请参阅附录C）。

Task and Experimental Procedure
The task asked participants to imagine that their school planned to implement a new wireless Internet service and to evaluate either Plan A or Plan B or both (see Appendix D, Main Study 1). Participants were also asked to primarily focus on the plan’s connection speed and security features, with everything else (e.g., price) being equal.
After arriving at a computer classroom where the study was administered, participants were briefly introduced to the study and then directed to a secure website to finish the task and fill out a short background questionnaire. Following prior EA research (Hsee 1996; Hsee and Leclerc 1998), participants were randomly assigned to one of three groups (see Figure 1). Group 1 received and read Plan A and then completed the evaluation survey. Group 2 received and read Plan B and then completed the evaluation survey. Group 3 received and read both Plan A and Plan B and then completed the evaluation survey for both technologies.

任务与实验程序
该任务要求参与者想象他们的学校计划实施一项新的无线互联网服务，并评估方案A或方案B或两者进行评估（请参阅附录D，主要研究1）。还要求他们主要集中在连接速度和安全特性,两个方案的其他特性(如价格)是相同的。
到达负责研究的计算机教室后，向参与者简要介绍了研究，然后将他们引导到安全的网站以完成任务并填写简短的背景调查表。根据先前的EA研究（Hsee 1996； Hsee和Leclerc 1998），参与者被随机分配为三组（见图1）。第一组接受并阅读方案A，然后完成评估调查。第二组接受并阅读方案B，完成评估调查。第三组分别接收并阅读方案A和方案B，完成两种技术的评估调查。

Manipulation Check
A manipulation check was used to assess whether Internet connection speed was a harder to evaluate feature than was security. In this check, participants in Groups 1 and 2 were asked (1) “Do you have any idea how fast Plan A/B is?” and (2) “Do you have any idea how secure Plan A/B is?” When answering, participants chose among five options, ranging from (1) = “I don’t have any idea.” To (5) = “I have a clear idea.” The mean evaluability score for speed (2.82) was significantly lower than that for security (3.50) (t (43) = -3.03; p < 0.01), confirming that security was perceived easier to evaluate than speed.
操作检查
使用一个操作检查来评估互联网连接速度是否比安全性更难评估。在这项检查中，第一组和第二组的参与者被问到(1）“您是否知道方案A / B有多快？” （2）“您是否知道方案A / B有多安全？” 回答时，参与者从5个选项中进行选择，这些选项从(1)=“我不知道”到 (5) =“我有一个清晰的想法。” 速度的平均可评估性分数（2.82）显着低于安全性的平均可评估性分数（3.50）（t（43）= -3.03； p <0.01），这证明安全性比速度更容易评估。

Analysis and Results
The Cronbach’s alpha of the technology evaluation measure was 0.94. A one-way ANOVA found no differences for gender, age, and wireless Internet experience across the three treatment groups. Two t-tests were conducted to compare the evaluation differences between Plan A and Plan B. First, a paired sample t-test was run using the data from Group 3; this analysis found the Plan A (with a faster connection speed) (Mean = 4.83; S.D. = 1.78) to be rated significantly higher than Plan B (Mean = 3.26; S.D. = 1.54) (t (23) = 2.47; p < 0.05; Cohen’s d = 0.94; a large effect). Thus, H1 was supported. Next, an independent sample t-test was run for Groups 1 and 2; this analysis found Plan A (Mean = 4.96; S.D. = 1.22) and Plan B (Mean = 5.82; S.D. = 0.97) to be significantly different (t (43) = -2.622; p # 0.05; Cohen’s d = 0.78; a medium effect), with improved security rated higher. Thus, H2 was also supported. Additionally, the results demonstrated a product PR across evaluation modes (see Table 2).
分析和结果
技术评价指标的克朗巴赫（Cronbach）alpha值为0.94。单因素方差分析发现，在三个实验组中，性别、年龄和无线上网体验没有差异。进行了两次t检验以比较方案A和方案B之间的评估差异。首先，使用来自第3组的数据进行配对样本t检验。该分析发现方案A（连接速度更快）（平均值= 4.83; SD = 1.78）的评分明显高于方案B（平均值= 3.26; SD = 1.54）（t（23）= 2.47; p <0.05 ；科恩（Cohen）d = 0.94；影响很大）。因此，支持H1。然后，对第1组和第2组进行独立样本t检验；该分析发现方案A（平均值= 4.96; SD = 1.22）和方案B（平均值= 5.82; SD = 0.97）有显着差异（t（43）= -2.622; p＃0.05; Cohen d = 0.78;中等）效果），改进的安全性评分更高。因此，支持H2。此外，结果证明了跨评估模式的产品PR（请参见表2）。

Discussion
Experiment 1 supports the EH within the context of IT adoption, demonstrating that people’s perceptions of a feature’s importance are influenced by the evaluation contexts. However, Experiment 1 has some limitations. First, network security was framed as a dichotomous variable: Plan A having no security and Plan B having security. Therefore, it is possible that the evaluability of the security attribute was limited by its dichotomous nature. Second, the security feature’s evaluability was a characteristic of the feature per se and was not empirically manipulated. As discussed above, when making a joint (JE) versus separate (SE) evaluation, PRs occur because one feature is hard-to-evaluate and the other is easy-to-evaluate. Thus, to eliminate such concerns, if both connection speed and security features are hard-to evaluate, the relative importance of those two features should not change between the two evaluation contexts (i.e., JE versus SE) and a PR will not occur. To explore this rival hypothesis, we conducted a second experiment where security was operationalized as a continuous variable (i.e., security level 2 versus security level 5), thus empirically manipulating the security feature’s evaluability.
讨论
实验1在IT采用的背景下支持EH，表明人们对特性重要性的感知受到评价背景的影响。然而，实验1有一些局限性。首先，将网络安全性定义为一个二分变量:方案A没有安全性，方案B有安全性。因此，安全属性的可评估性可能受到其二分性的限制。其次，安全性的可评估性本身就是该特性的一个特征，没有经验地加以操纵。如上所述，在进行联合(JE)和单独(SE)评估时，发生PRs是因为一个特性难以评估，而另一个易于评估。因此，为了消除这些问题，如果连接速度和安全性都很难评估，则这两个特性的相对重要性在两个评估环境（即JE与SE）中不应该改变，并且不会出现PR。为了探究这个对立的假设，我们进行了第二个实验，其中安全性被作为一个连续变量(即，安全级别2对安全级别5)，从而经验地操作安全特性的可评估性。

3.3Experiment 2

Task Context
Like Experiment 1, participants assessed two different wireless Internet services where Plan A is faster but is less secure; Plan B is slower but is more secure (see Appendix D, Experiment 2). However, in this experiment, security was operationalized in two ways, with low and high evaluability. Using this modification, we had two evaluability conditions: hard/hard and hard/easy. In the hard/hard condition, both the connection speed and security features were hard-to-evaluate; specifically, participants were provided with a meaningless number for establishing the quality of the security feature. In the hard/easy condition, speed remained hard-to-evaluate, whereas security was relatively easy-to-evaluate (i.e., participants were told the meaning of the security rating number). Therefore, we predict that there will not be a joint-separate evaluation for the hard/hard condition and thus no PR will occur. However, for the hard/easy condition, a classic EH condition, a joint-separate evaluation will occur, and a PR will occur. As such, we propose that a PR will not occur in the hard/hard condition, but will occur in the hard/easy condition.

3.3实验2

任务环境
与实验1一样，参与者评估了两种不同的无线互联网服务，其中方案A速度更快，但安全性更差;方案B较慢，但更安全(见附录D，实验2)。但在本实验中，安全性以两种方式实现，低可评估性和高可评估性。使用这个修改，我们有两个评估条件:hard/hard（难/难）和hard/easy（难/易）。在hard/hard条件下，连接速度和安全特性都难以评估;具体来说，为确定安全特性的质量，向参与者提供了一个没有意义的数字。在困难/容易的情况下，速度仍然难以评估，而安全性相对容易评估(即，参与者被告知安全评级数字的含义)。因此，我们预计不会对hard/hard条件进行联合-单独评估，因此不会发生PR。但是，对于hard/easy的情况，会发生经典的EH情况，联合-单独评估会发生，PR也会发生。因此，我们假设PR不会在hard/hard件下发生，而会在hard/easy状态下发生。

Sample
The participants were from a college-wide, junior-level business class at the same university as Experiment 1; 82 students voluntarily participated in the study. Participants’ age ranged between 19 and 23, and their average age was 21.41 (SD .75); 59.76% were women. Each participant received about 1% of their final course grade for participating in the experiment.
样本
参与者来自与实验1所在大学相同的大学范围内的初级商务班； 82名学生自愿参加了研究。 参与者的年龄介于19到23岁之间，平均年龄为21.41（SD .75）；59.76%是女性。每个参与实验的学生都得到了他们最终课程成绩的1%。

Task and Experimental Procedure
The measures were the same as Experiment 1. The tasks and procedures, however, were slightly different from Experiment 1 (see Figure 2). The participants were again divided into three groups, but the survey for each group contained two parts, which constituted three evaluation contexts by two evaluability conditions (i.e., six total conditions). Three evaluation contexts were the same as those for Experiment 1. The two evaluability conditions were hard/hard and hard/easy. In the hard/hard condition, both connection speed and security level were rated numerically and therefore both were hard to evaluate without comparison. In the hard/easy condition, connection speed remained hard to evaluate without comparison but security was made relatively easy to evaluate by telling participants the meaning of the numerical rating. Following Hsee (1996), the two evaluability conditions were designed within-subjects.
任务与实验流程
指标与实验1相同。但是，任务和过程与实验1略有不同（请参见图2）。 参与者再次分为三组，但每组的调查包含两个部分，由两个可评估条件（即六个总条件）构成三个评估环境。 三个评估环境与实验1相同。这两个评估条件分别为hard/hard和hard/easy。在hard/hard条件下，连接速度和安全级别都是用数字来评估的，因此不进行比较很难评估。在hard/easy条件下，如果不进行比较，连接速度仍然很难评估，但是通过告诉参与者数值评级的含义，安全性相对容易评估。根据Hsee（1996），这两个可评估性条件是在受试者内部设计的。

Manipulation Check
A manipulation check was used to assess whether Internet connection speed was a harder to evaluate feature than security in the two evaluability conditions. In this check, participants in Groups 1 and 2 were asked (1) “Do you have any idea how fast Plan A/B is?” and (2) “Do you have any idea how secure Plan A/B is?” When answering, participants chose among five options, ranging from (1) = “I don’t have any idea” to (5) = “I have a clear idea.” In the hard/hard condition, the mean evaluability score for connection speed (2.40) was not significantly different from that for security (2.52) (t (54) = -0.73; p > 0.05), confirming that both features were deemed relatively hard to evaluate. In hard/easy condition, the mean evaluability score for connection speed (2.84) was significantly lower than that for security (3.33) (t (54) = -3.46; p < 0.05), confirming that security was perceived easier to evaluate than connection speed.
操作检查
使用一个操作检查来评估互联网连接速度是否比安全性更难评估。在这项检查中，第一组和第二组的参与者被问到(1）“您是否知道方案A / B有多快？” （2）“您是否知道方案A / B有多安全？” 回答时，参与者从5个选项中进行选择，这些选项从(1)=“我不知道”到(5) =“我有一个清晰的想法。” 在hard/hard条件下，连接速度(2.40)与安全(2.52)的平均可评估得分没有显著差异(t (54) = -0.73;(p > 0.05)，证实这两个特征都被认为是相对难以评估的。在hard/easy条件下，连接速度(2.84)的平均可评估得分显著低于安全(3.33)(t (54) = -3.46;(p < 0.05)，确认安全性比连接速度更容易评估。

Analysis and Results
The Cronbach’s alpha of the technology evaluation measure was 0.97. No differences for gender, age, and wireless Internet experience were found across the three treatment groups. Two t-tests were used to compare the evaluation differences between Plan A and Plan B. In the hard/hard condition, a paired sample t-test was run using the data from Group 3; this analysis found Plan A (Mean = 5.91; S.D. = 1.07) to be rated significantly higher than Plan B (Mean = 2.85; S.D. = 0.69) (t (26) = 10.67; p < 0.05; Cohen’s d = 4.19; a large effect). Next, an independent sample t-test was run for Groups 1 and 2; this analysis found Plan A (Mean = 5.50; S.D. = 1.63) and Plan B (Mean = 5.72; S.D. = 1.44) to be equivalent (t (53) = -0.52; p > 0.05 Cohen’s d = 0.14). As proposed, a PR did not occur in the hard/hard condition (see Table 3).
Next, in the hard/easy condition, a paired sample t-test was run using the data from Group 3; this analysis found the Plan A (Mean = 4.94; S.D. = 1.55) to be rated significantly higher than Plan B (Mean = 3.56; S.D. = 1.48) (t (26) = 2.61; p < 0.05; Cohen’s d = 1.09; a large effect). Next, an independent sample t-test was run for Groups 1 and 2; this analysis found Plan A (Mean = 4.86; S.D. = 1.50) and Plan B (Mean = 5.75; S.D. = 1.01) to be significantly different (t (53) = -2.59; p < 0.05; Cohen’s d = 0.70; a medium effect), with improved security rated higher. Also, as proposed, a PR occurred in the hard/easy condition (see Table 3).
分析和结果
技术评价指标的克朗巴赫（Cronbach）alpha值为0.97。单因素方差分析发现，在三个实验组中，性别、年龄和无线上网体验没有差异。进行了两次t检验以比较方案A和方案B之间的评估差异。在hard/hard条件下，使用来自第3组的数据进行配对样本t检验；该分析发现方案A（平均值= 5.91; SD = 1.07）的评分显着高于方案B（平均值= 2.85; SD = 0.69）（t（26）= 10.67; p <0.05; Cohen d = 4.19;大影响）。接下来，对第1组和第2组运行独立的样本t检验；该分析发现，方案A（平均值= 5.50；标准差= 1.63）和方案B（平均值= 5.72；标准差= 1.44）是等效的（t（53）= -0.52； p> 0.05科恩d = 0.14）。按照提议，在hard/hard条件下不会发生PR（请参见表3）。
接下来，在hard/easy条件下，使用来自第3组的数据进行配对样本t检验；该分析发现方案A（平均值= 4.94; SD = 1.55）的评分明显高于方案B（平均值= 3.56; SD = 1.48）（t（26）= 2.61; p <0.05; Cohen d = 1.09; a效果大）。接下来，对第1组和第2组运行独立的样本t检验；该分析发现方案A（平均值= 4.86; SD = 1.50）和方案B（平均值= 5.75; SD = 1.01）有显着差异（t（53）= -2.59; p <0.05; Cohen d = 0.70;中等）效果），改进的安全性评分更高。而且，如所提出的，PR在hard/easy状态下发生（参见表3）。

4.Discussion

Based on the EH, when evaluating a single IT product, the easy-to-evaluate product feature was perceived more important. When two IT products were compared, the hard-to evaluate feature was perceived more important, resulting in a PR between the two evaluation contexts. Our two experiments showed that joint-separate evaluation PRs happen not only when one feature is dichotomous and the other is continuous, but also when both features are continuous. Also, Experiment 2 demonstrated that joint-separate evaluation PRs can be turned on and off by changing the evaluability of the features. Thus, as suggested by this theoretical lens, when comparing two IT products, individuals will more favorably evaluate the product with more desirable easy-to-compare features; when evaluating one IT product, individuals will more favorably evaluate the product with more desirable easy-to-evaluate features.

4.讨论

基于EH，当评估单个IT产品时，容易评估的产品特性被认为更重要。当两个IT产品进行比较时，难以评估的特征被认为更重要，从而导致两个评估上下文之间的PR。我们的两个实验表明，联合-单独评价PRs不仅发生在一个特征是二分的，另一个特征是连续的情况下，而且发生在两个特征都是连续的情况下。实验2还表明，通过改变特征的可评价性，可以实现联合-单独评价PRs的开启和关闭。因此，正如该理论观点所建议的，当比较两种IT产品时，个人会更倾向于评估具有更理想的易于比较的特性的产品。当评估一个IT产品时，个人会更倾向于评估具有更理想的易于评估的特性的产品。

5.Theoretical Contributions

This research makes two important theoretical contributions. First, these results contribute to HCI literature, suggesting that people’s perceived importance toward technology features can change in different evaluation contexts. Specifically, when only one IT product is evaluated, individuals are likely to rely more on easy-to-evaluate features. When there are at least two IT products being evaluated, individuals are more likely to rely more on hard-to-evaluate features. In other words, different product features are salient in different evaluation contexts. Thus, when examining the relative importance of features for a specific technology, researchers should be aware of the evaluation context and its potential influence on assessments. For example, Keil and Tiwana (2006) used conjoint analysis (a multi-attribute judgment analysis to derive the underlying structure of individuals’ decision rules) to explore the relative importance of various features within eight enterprise resource planning (ERP) systems. They found that participants rated system functionality higher than they rated ease of customization. From the EH perspective, ease of customization was relatively easy to assess because evaluators could rapidly identify whether the system could, or could not, be easily customized (i.e., a system was described as easy to customize or not). Alternatively, understanding which systems had all desired functionality was relatively harder to evaluate without comparison (i.e., which system [i.e., A through H], has the desired and/or better set of features?). Thus, the EH suggests that participants perceived those hard-to-evaluate but easy-to-compare criteria (e.g., system functionality) as more important than those easy-toevaluate features (e.g., ease of customization). Similarly, Venkatesh, Chan et al. (2012) found that usability is more important for service adoption than security provision. Here, the usability of e-government services could be judged based on the number of steps required by a consumer. Therefore, usability was hard-to-evaluate without comparison, playing a larger role when individuals have multiple e-government services in mind. To summarize, our study can help explain results from previous literature in terms of why certain technology features are perceived to be more important in certain contexts.

5.理论贡献

本研究有两个重要的理论贡献。首先，这些结果对HCI文献有贡献，表明人们对技术特性的感知重要性可以在不同的评估环境中改变。具体来说，当只评估一个IT产品时，个人可能更依赖于易于评估的特性。当至少有两个IT产品被评估时，个体更可能更多地依赖于难以评估的特性。换句话说，不同的产品特性在不同的评估环境中很重要。因此，在检查特性对于特定技术的相对重要性时，研究人员应了解评估环境及其对评估的潜在影响。例如，Keil和Tiwana(2006)使用联合分析(一种多属性判断分析，得出个人决策规则的底层结构)来探索八个企业资源规划(ERP)系统中各种特征的相对重要性。他们发现，参与者对系统功能的评价要高于对自定义易用性的评价。从EH的角度来看，自定义易用性相对是容易评估的，因为评估人员可以快速确定系统是否可以轻易地进行定制（即，系统被描述为容易进行定制）。另外，如果不进行比较的话，理解哪些系统具有所有需要的功能相对比较困难（即哪个系统[即A至H]具有期望的和/或更好的特征集合？）。因此，EH表明参与者认为那些难以评估但易于比较的标准(例如，系统功能)比那些易于评估的特征(例如，易于定制)更重要。类似地，Venkatesh, Chan等人(2012)发现，对于服务采用来说，可用性比安全性更重要。在这里，电子政府服务的可用性可以根据用户所需的步骤数量来判断。因此，如果没有比较，可用性很难评估，当个人考虑多个电子政府服务时，可用性扮演着更大的角色。总而言之，我们的研究可以帮助解释先前文献的结果，即为什么某些技术特性在某些上下文中被认为更重要。

Second, our study contributes to the technology adoption literature, suggesting that researchers need to be aware of the way in which technology evaluation is operationalized in future studies. Specifically, technology evaluation studies yield misleading results if participants are using a SE context, but the study assumes a JE context (i.e., if researchers are interested in the technology evaluation of a single technology, care must be taken to assure that participants are not considering other technologies). Our results show that the mechanisms that people follow when evaluating one IT product in isolation are fundamentally different from those when comparing two IT products. As such, the product features that influence technology evaluation when only one IT product is considered can be completely different from those driving evaluation and adoption when multiple IT products are being compared. For example, Wixom and Todd (2005) found accessibility to be important while timeliness was not in the context of evaluating a single data warehousing system. From the perspective of the EH, a possible explanation may be that accessibility is relatively easy-to-evaluate (i.e., accessibility relates closely with how easy the system is to use), while timeliness is relatively hard-to-evaluate (i.e., timeliness is hard to assess without comparison to another system). Therefore, accessibility is perceived more important than timeliness in such a context. To account for different evaluation processes, future research must take care to clearly understand and control the various adoption contexts5 in order to gain accurate usability assessments.
其次，我们的研究为技术采用文献做出了贡献，表明研究人员需要了解在未来的研究中技术评估的实施方式。具体地说，如果参与者使用的是SE环境，但该研究假设的是JE环境，技术评估研究就会产生误导结果(即，如果研究人员对某一项技术的技术评估感兴趣，必须注意确保参与者没有考虑其他技术)。我们的结果表明，人们在单独评估一个IT产品时所遵循的机制与在比较两个IT产品时所遵循的机制是完全不同的。因此，当只考虑一个IT产品时，影响技术评估的产品特性可能与在比较多个IT产品时驱动评估和采用的产品特性完全不同。例如，Wixom和Todd(2005)发现可访问性很重要，而及时性在评估单个数据仓库系统时并不适用。从EH的角度来看，一种可能的解释可能是可访问性是相对容易评估的，（即，可访问性与系统使用的容易程度密切相关），而及时性则相对难以评估（即及时性如果不与其他系统进行比较则很难评估）。因此，在这种情况下，可访问性比及时性更为重要。为了说明不同的评估过程，未来的研究必须注意清楚地理解和控制各种采用环境（其他的环境因素，比如一个人对产品的主要使用，也可能会影响产品特性的相对价值以及对它们的评估方式。），以便获得准确的可用性评估。

6.Practical Contributions

These results have three important practical implications. First, our work provides useful guidelines for designers when assessing usability. Depending upon whether a single or multiple products exist in a category, the relative importance of specific features may vary. As such, designers should be aware of how the salience of features may change depending upon the number of products in a particular category. Likewise, designers may need to adjust evaluation criteria and their relative weightings in different contexts to gain the most meaningful evaluations of features and products. For example, when designing a new technology product category (e.g., when Apple first released its iWatch), designers should focus more on easy-to-evaluate features to receive higher user evaluations. In contrast, in more mature technology product markets, designers should focus more on hard-to-evaluate features to receive higher evaluations from users.

6.实际贡献

这些结果具有三个重要的实际意义。 首先，我们的工作为设计师评估可用性时提供了有用的指导。根据一个类别中存在单个产品还是多个产品，特定功能的相对重要性可能会有所不同。因此，设计师应该意识到特性的显著性是如何随着特定类别产品的数量而变化的。同样，设计师可能需要调整评估标准及其在不同环境中的相对权重，以获得对特性和产品最有意义的评估。例如，在设计一个新的技术产品类别时(例如，当苹果首次发布iWatch时)，设计师应该更多地关注易于评估的功能，以获得更高的用户评价。相比之下，在更成熟的技术产品市场中，设计师应该更多地关注难以评估的功能，以获得用户更高的评价。

Second, consumers of IT products will benefit by becoming aware of the tendency to rely more heavily on hard-to-evaluate features when comparing multiple products. According to Hsee and Zhang (2004), people have better experiences with options with the more favorable easy-to-evaluate features. To limit evaluation bias, people may want to first evaluate a technology in isolation, then evaluate the product against others to see whether and how initial evaluations change. In other words, while consumers may have better usage experiences with technology products with the more favorable easy-to-evaluate features, they may ultimately choose technology products with the more favorable hard-to-evaluate features by comparing multiple products. Consumers need to be made aware of how the evaluation context shapes product evaluation perceptions.
其次，在比较多个产品时更加倾向于难以评估的功能，这将使IT产品的消费者受益。Hsee和Zhang(2004)认为，人们对更易于评估的选项有更好的体验。为了限制评估偏差，人们可能想要首先单独评估一项技术，然后将产品与其他产品进行比较，以了解初始评估是否以及如何发生变化。换句话说，当消费者对易于评价的特征较好的技术产品有较好的使用体验时，他们可能会通过比较多个产品最终选择较难于评价的特征较好的技术产品。消费者需要了解评价环境是如何塑造产品评价感知的。

Third, for companies launching new IT products, our findings have implications for their marketing campaigns. For example, when no similar products are available in the marketplace, companies should emphasize the easy-to-evaluate features because these features are most salient to assessment perceptions. However, when multiple similar products are available, companies should emphasize their strongest hard-to-evaluate features. For example, in 2015, Apple released two new products, the iWatch 1 and iPhone 6s. For the iWatch 1, wearables were an emerging market with few similar products for comparison. Thus, Apple likely benefitted by emphasizing its easy-to-evaluate features, such as aesthetics and style. In contrast, for the iPhone 6s, a relatively mature market, Apple benefitted most by emphasizing its hard-to-evaluate features, such as battery quality, processing performance, resolution, or app store. In fact, Apple followed this guidance in the official Apple advertising videos for the release of the iWatch 1 and iPhone 6s. For the iWatch 1, you will find scant information about its detailed capabilities. In contrast, for the iPhone 6s, advertising is packed with new technical features that are hard to evaluate without comparison.6 Apple seems to have embraced the EH in its marketing strategy.
第三，对于推出新IT产品的公司，我们的发现对他们的营销活动有影响。例如，当市场上没有类似的产品时，公司应该强调易于评估的特性，因为这些特性对于评估感知来说是最突出的。但是，当有多个类似的产品可用时，公司应该强调它们最强大的难以评估的特性。例如，2015年，苹果发布了两款新产品，iWatch 1和iPhone 6s。对于iWatch 1来说，可穿戴设备是一个新兴市场，几乎没有类似产品可供比较。因此，苹果可能通过强调美学和风格等易于评估的功能而获益。相比之下，iPhone 6s是一个相对成熟的市场，苹果受益最大的是强调其难以评估的功能，比如电池质量、处理性能、分辨率或应用商店。事实上，苹果在发布iWatch 1和iPhone 6s的官方广告视频中就遵循了这一指导。对于iWatch 1，你会发现关于其详细功能的信息很少。相比之下，对于iPhone 6s来说，广告充满了新技术特性，如果不进行比较，很难对其进行评估。苹果似乎已经接受了EH的营销策略。

7.Limitations and Opportunities for Future Studies

While making a significant contribution to our understanding of technology choice, our work has some limitations. Although broadly applied (e.g., Bazerman et al. 1999; González-Vallejoa and Moran 2001; Hsee and Rottenstreich 2004; Todorov et al. 2007; Wilson and Arvai 2006), the EH has several characteristics which can limit its application. First, the theory deals with the comparison of two products; clearly, consumers often face more than two product options (McCracken 2010). Nevertheless, our results help to build a foundation for future studies investigating more than two options. Second, the theory focuses on two products that have a trade-off between easy- and hard-to-evaluate features. Future studies can focus on various mixes of easy- and hard-to-evaluate features. Third, the theory assumes that individuals have reasonable information on the products being evaluated so that they can objectively evaluate each option. When having limited information, individuals may make their decisions based on other factors, such as the visual appeal, brand, and even extrinsic factors such as shopping atmospherics (Kotler 1973) and website quality (Wells et al. 2011).7 Future studies can focus on these interesting scenarios. Another limitation relates to the experimental methodology with primarily student participants. Given our objective for clear-cut theory testing, we carefully controlled the manipulation of the technology features in order to maximize measurement precision, limiting generalizability and realism. However, by using three different product categories and a relatively diverse set of participants across the pilot studies and main experiments, the results were consistent across these varying conditions. Nevertheless, the design of our experiments was artificial; thus, the results may be limited to the specific IT products, tasks utilized, and subject populations in these studies. Clearly, these limitations must be considered when interpreting our results. Future studies can explore other IT products and associated features with more diverse samples (e.g., samples drawn with Amazon Mechanical Turk) to examine the robustness of various contingencies and contexts.

7.局限性和未来研究方向

在为我们对技术选择的理解做出重大贡献的同时，我们的工作也有一些局限性。尽管EH已得到广泛应用（例如Bazerman等，1999;González-Vallejoa和Moran 2001; Hsee和Rottenstreich 2004; Todorov等，2007; Wilson和Arvai 2006），但EH具有一些可能会限制其应用的特征。首先，该理论涉及两个产品的比较;显然，消费者经常面临两种以上的产品选择(McCracken 2010)。尽管如此，我们的研究结果有助于为未来研究更多的选择奠定基础。其次，该理论侧重于在易于评估和难以评估的功能之间进行权衡的两种产品。未来的研究可能集中在易于评估和难以评估的功能的各种组合上。第三，该理论假设个人对被评估的产品有合理的信息，因此他们可以客观地评估每个选项。当信息有限时，个人可能会根据其他因素做出决定，如视觉吸引力、品牌，甚至外部因素，如购物氛围(Kotler 1973)和网站质量(Wells et al. 2011)。未来的研究可以专注于这些有趣的场景。另一个限制涉及主要是学生参加的实验方法。鉴于我们的目标是进行清晰的理论测试，我们仔细控制了技术功能的操纵，以最大程度地提高测量精度，从而限制了通用性和真实性。但是，通过在试点研究和主要实验中使用三种不同的产品类别和一组相对不同的参与者，结果在这些变化的条件下是一致的。然而，我们的实验设计是人为的。因此，结果可能仅限于这些研究中的特定IT产品，所使用的任务和主题人群。显然，在解释我们的结果时必须考虑这些限制。未来的研究可以探索其他IT产品和与更多样化的样本(例如，使用Amazon Mechanical Turk绘制的样本)相关的特性，以检查各种意外情况和环境的稳健性。

In Experiment 2, we explored nuances of how to operationalize easy- and hard-to-evaluate product features within the bounds of the EH. Specifically, we demonstrated that joint-separate evaluation PRs can be turned on and off and that hard-to-evaluate features can be switched into easy-to-evaluate features by providing more information about the feature. Of course, it is likely that not all hard-to-evaluate features can be manipulated in this manner. For example, in the marketing literature, Nelson (1974) proposes that product features can be divided into search and experience features. Search features include a product’s color, size, the number of calories or ingredients, and can be obtained through second hand sources such as advertisement and word of mouth. Alternatively, experience features, typically obtained via exposure to the product, include a product’s design quality, aesthetics, fit, taste, or ease of use. Therefore, search features tend to be more objective and diagnostic, whereas experience features tend to be more subjective, characterized by uncertainty and equivocality (Hoch and Deighton 1989). Thus, it may be possible to convert hard-to-evaluate search features into easy-to-evaluate search features by providing more descriptive or comparative information. However, it may be challenging to transform hard-to-evaluate experience features into easy-to-evaluate features by simply providing more feature-related details. Clearly, additional research is needed to explore how technology features can be transformed into easy-to-evaluate features. It may be possible that different types of HCI environments (i.e., virtual product experiences) can be used to improve the evaluability of many experiential product features.
在实验2中，我们探讨了如何在EH范围内操作容易和难以评估的产品特性的细微差别。具体而言，我们证明了可以打开和关闭联合-单独的评估PR，并且通过提供有关特征的更多信息来可以将难以评估的特征转换为易于评估的特征。例如，在市场营销文献中，尼尔森（Nelson）（1974）提出可以将产品功能分为搜索特性和体验特性。搜索特性包括产品的颜色，大小，卡路里或成分的数量，并且可以通过二手来源（例如广告和口碑）获得。另外，体验特性（通常通过接触产品获得）包括产品的设计质量，美学，合身性，品味或易用性。因此，搜索特性趋向于更客观和更具诊断性，而体验特征趋向于更加主观，以不确定性和模棱两可为特性（Hoch and Deighton 1989）。因此，通过提供更多描述性或比较性信息，可以将难以评估的搜索特性转换为易于评估的搜索特性。但是，仅通过提供更多与特性相关的详细信息，将难于评估的体验特性转换为易于评估的特性可能是一项挑战。显然，还需要进行其他研究来探索如何将技术特性转换为易于评估的功能。可能可以使用不同类型的HCI环境（即虚拟产品体验）来提高许多体验产品特性的可评估性。

8.Conclusion

Consumers are facing an ever increasing number of technology product choices. Organizations want to design better technology products to best meet the needs of current and future customers. The EH provides a useful theoretical lens for improving our understanding of people’s perceived importance toward different features and overall technology product evaluations. The results from two experiments showed that individuals indeed perceived different technology features to have different levels of importance and perceived the same technology differently in different evaluation contexts (i.e., JE versus SE). These results help to explain how designers can develop technology products with higher usability, why individuals sometimes choose the wrong technology, and how organizations can best position their products in differing market contexts.

8.结论

消费者面临着越来越多的科技产品选择。组织想要设计更好的技术产品来最好地满足当前和未来客户的需求。EH提供了一个有用的理论视角来提高我们对人们感知到的对不同特性和整体技术产品评估的重要性的理解。两个实验的结果表明，在不同的评价情境中，个体确实感知到不同的技术特征具有不同的重要性，对同一技术的感知也存在差异(即JE对SE)。这些结果有助于解释设计师如何开发可用性更高的技术产品，为什么个人有时会选择错误的技术，以及组织如何在不同的市场环境中最好地定位他们的产品。