Taylor & Francis Group
泰勒与弗朗西斯集团

To cite this article: Susan P. Besemer (1998) Creative Product Analysis Matrix: Testing the Model Structure and a Comparison Among Products–Three Novel Chairs, Creativity Research Journal, 11:4, 333-346, DOI: 10.1207/s15326934crj1104_7
引用本文：Susan P. Besemer (1998) 创意产品分析矩阵：模型结构的测试及三种新型椅子的比较，创造力研究期刊，11:4，333-346，DOI: 10.1207/s15326934crj1104_7

To link to this article: https://doi.org/10.1207/s15326934crj1104_7
要链接到本文： https://doi.org/10.1207/s15326934crj1104_7

Published online: 08 Jun 2010.
在线发表：2010 年 6 月 8 日。

Submit your article to this journal
将您的文章提交至本期刊

Article views: 1307 文章浏览量：1307

View related articles 查看相关文章

Citing articles: 26 View citing articles
引用文章：26 查看引用文章

Over the past four decades, scholars have used creative products as evidence that can give clues in the investigation of the creative process and insights into the personality. If we take as a premise Guilford’s (1950) assertion that creativity is a normally distributed set of traits, then it follows that the products that are created by artists, inventors, homemakers, and business persons bear the marks of their unique creativity.
在过去的四十年中，学者们将创意产品作为证据，以便在创作过程的研究中提供线索，并洞察个性。如果我们以吉尔福德（Guilford，1950）的主张为前提，即创造力是一组呈正态分布的特征，那么可以推断，艺术家、发明家、家庭主妇和商人所创造的产品都带有他们独特创造力的印记。

Empirical researchers in creativity studies (e.g., MacKinnon, 1978; Simonton, 1991) have long made a case for the importance of studying creativity through the products that arise from the creative process, yet relatively few articles on this subject are published. Among the few empirical studies that have contributed to meeting this need are Basadur and Hausdorf (1996), Besemer and O’Quin (1986, 1987), Howe (1992), Michael and Bachelor (1992), O’Quin and Besemer (1989), and Runco and Charles (1993).
在创造力研究中，实证研究者（例如，MacKinnon，1978；Simonton，1991）长期以来一直主张通过创造过程产生的产品来研究创造力的重要性，但关于这一主题的文章相对较少。在为满足这一需求做出贡献的少数实证研究中，有 Basadur 和 Hausdorf（1996）、Besemer 和 O’Quin（1986，1987）、Howe（1992）、Michael 和 Bachelor（1992）、O’Quin 和 Besemer（1989）以及 Runco 和 Charles（1993）。

Many essays and review papers about the criteria that mark highly creative products have been published in the literature of creativity (Besemer & O’Quin, 1993; Besemer & Treffinger, 1981; Ghiselin, 1958; Jackson & Messick, 1967; Mednick & Mednick, 1964), as well as in the literature of new product development and business practices (Page & Rosenbaum, 1987; Roberts, 1996), education (Pearlman, 1983), and other arts and sciences (Csikszentmihalyi, 1996; Pollock, 1978; Sapp, 1992). Even the literature of the natural sciences has included some very interesting work on the creativity manifested in the products of highly creative scientists (Rothenberg, 1996; Stumpf, 1995; Wolpert, 1996).
关于标志高度创造性产品的标准的许多论文和综述文章已在创造力文献中发表（Besemer & O’Quin, 1993; Besemer & Treffinger, 1981; Ghiselin, 1958; Jackson & Messick, 1967; Mednick & Mednick, 1964），以及在新产品开发和商业实践的文献中（Page & Rosenbaum, 1987; Roberts, 1996）、教育领域（Pearlman, 1983）和其他艺术与科学领域（Csikszentmihalyi, 1996; Pollock, 1978; Sapp, 1992）。甚至自然科学的文献中也包含了一些关于高度创造性科学家所表现出的创造性的非常有趣的研究（Rothenberg, 1996; Stumpf, 1995; Wolpert, 1996）。

Theoretical articles and monographs are more common than empirical studies in the literature of aesthetics and creativity, and since the mid-1950s several such pieces have been published that outline how creative products may be analyzed and evaluated
理论文章和专著在美学与创造力的文献中比实证研究更为常见，自 20 世纪 50 年代中期以来，已经发表了几篇这样的作品，概述了如何分析和评估创造性产品

(Briskman, 1980; Dipert, 1993; Ghiselin, 1958; Pearlman, 1983). These are stimulating and thought-provoking, but without empirical studies that confirm or refute the models they propose, they leave unanswered questions.
（Briskman, 1980；Dipert, 1993；Ghiselin, 1958；Pearlman, 1983）。这些观点引人深思，但由于缺乏实证研究来确认或反驳他们提出的模型，因此留下了未解的问题。

Publications about products are often centered in a discipline like business, education, or technology (Dubuc, 1985; Lobert & Dologite, 1994; Page & Rosenbaum, 1987; Roberts, 1996; Tonemah, 1987) and seek to help practitioners to evaluate or to improve the products they create. Their usefulness is restricted in that they are designed to be limited to certain groups of products or certain groups of persons.
关于产品的出版物通常集中在商业、教育或技术等学科（Dubuc, 1985；Lobert & Dologite, 1994；Page & Rosenbaum, 1987；Roberts, 1996；Tonemah, 1987），旨在帮助从业者评估或改进他们所创造的产品。它们的实用性受到限制，因为它们旨在仅限于某些产品组或某些人群。

Other studies look at “product” as a dependent variable and use it as a measure of a person’s performance on a creative task (Amabile, 1982; Michael & Bachelor, 1992). Although these are each helpful contributions to the ever-growing literature of creativity, the need remains for empirical research that can tie theory and speculation about the qualities of creative products to the real world of everyday life.
其他研究将“产品”视为一个因变量，并将其作为衡量个人在创造性任务上表现的标准（Amabile, 1982; Michael & Bachelor, 1992）。尽管这些都是对不断增长的创造力文献的有益贡献，但仍然需要实证研究将关于创造性产品特质的理论和推测与日常生活的现实世界联系起来。

The study discussed in this article set out to respond to these needs by seeking to answer two research questions. First, does confirmatory factor analysis (CFA) support a hypothesized three-dimensional model for understanding creative products? Second, can a judging instrument elucidate perceived differences among
本文讨论的研究旨在通过回答两个研究问题来满足这些需求。首先，验证性因素分析（CFA）是否支持理解创意产品的假设三维模型？其次，评判工具能否阐明感知差异？
creative products using Norwegian college students as naïve judges of the products? To answer these questions, this study asked participants to evaluate three chairs using a judging instrument based on the three dimensions of the model pictured in Figure 1.
使用挪威大学生作为对产品的天真评审来评估创意产品？为了解答这些问题，本研究要求参与者使用基于图 1 中模型的三个维度的评估工具来评估三把椅子。

The Creative Product Analysis Matrix (CPAM) may be used as a framework for thinking about the creativity manifested in many different kinds of products. For example, the model can be used in respect to works of art, new product ideas in manufacturing, or when considering other types of artifacts of the creative process. The model posits three related factors: Novelty, Resolution, and Elaboration and Synthesis.
创造性产品分析矩阵（CPAM）可以作为思考在多种不同类型产品中表现出的创造力的框架。例如，该模型可以用于艺术作品、制造业中的新产品创意，或在考虑创造过程的其他类型人工制品时。该模型提出了三个相关因素：新颖性、解决方案和细化与综合。

Within the three factors or dimensions, nine component facets are hypothesized. Falling under the rubric of Novelty are the facets for original and surprise. In the category of Resolution, the facets are valuable, logical, useful, and understandable. The third dimension of the model, Elaboration and Synthesis, includes the following facets: organic, elegant, and well crafted.
在这三个因素或维度中，假设有九个组成方面。属于新颖性范畴的方面包括原创性和惊喜。在解决方案类别中，方面包括有价值、合逻辑、有用和易于理解。模型的第三个维度，阐述与综合，包含以下方面：有机性、优雅和精心制作。

The dimension of Novelty addresses various aspects of newness in a product. These include, but are not limited to, conceptual newness, the use of a new
新颖性的维度涉及产品中新颖性的各个方面。这些包括但不限于概念上的新颖性、新技术的使用。

Figure 1. The Creative Product Analysis Matrix.
图 1. 创意产品分析矩阵。
process in making the product, using different materials, or using materials or processes in a new way.
在制造产品的过程中，使用不同的材料，或以新的方式使用材料或工艺。

The dimension called Resolution concerns itself with how well the product does what it is supposed to do. In the case of the unique chairs used as stimulus items for this study, it was of course of interest to a potential user if the chair seemed sturdy, solid, “sit-on-able,” and had qualities that would “work” as a chair. Considering the high level of Novelty in the stimulus chairs, these mundane attributes could not be taken for granted. (See the photographs of chairs in Figure 2.)
所称的“分辨率”维度关注产品执行其预期功能的效果。在本研究中作为刺激物的独特椅子，潜在用户当然会关心椅子是否看起来坚固、结实、可坐，并具备作为椅子的特性。考虑到刺激椅子的高度新颖性，这些平常的属性不能被视为理所当然。（参见图 2 中的椅子照片。）

Elaboration and Synthesis, sometimes called the style dimension, is an especially interesting factor. When we talk about why we especially like certain products, the words that come to mind often refer to our aesthetic appreciation of the product. It may appear to be made with special care; it shows attention to detail; it pleases us because it is beautiful or, on the other hand, fascinates us because it is interestingly “ugly.”
阐述与综合，有时被称为风格维度，是一个特别有趣的因素。当我们谈论为什么特别喜欢某些产品时，脑海中浮现的词语往往与我们对产品的审美欣赏有关。它可能看起来是经过特别用心制作的；它展现了对细节的关注；它让我们感到愉悦，因为它是美丽的，或者，另一方面，它因其有趣的“丑陋”而吸引我们。

The model has been used in previous research since 1986. In early studies, Besemer and O’Quin (1986) established the validity of the Novelty dimension of the CPAM using the Creative Product Semantic Scale (CPSS). In other studies (Besemer & O’Quin, 1987; O’Quin & Besemer, 1989) we found strong, yet inconclusive, support for the Elaboration and Synthesis dimension. Participants in the 1989 study did not perceive significant differences among the products analyzed on the basis of their Resolution. The seemingly obvious dimension of Resolution was not adequately differentiated through multivariate analysis of variance (MANOVA) in earlier studies. Judges were not able to differentiate clearly among products by their level of logicalness, usefulness, and value. Because of the need to address the question of the validity of the Resolution dimension, three highly novel stimulus items were sought, to demonstrate more variance on the Resolution dimension than had been seen in earlier similar studies. Would it be possible, using three finished products, to observe differences in the level of Resolution in the stimulus items?
该模型自 1986 年以来已在先前的研究中使用。在早期研究中，Besemer 和 O’Quin（1986）使用创造性产品语义量表（CPSS）确立了 CPAM 的新颖性维度的有效性。在其他研究中（Besemer & O’Quin, 1987; O’Quin & Besemer, 1989），我们发现对详细性和综合性维度的支持强烈但不确定。1989 年研究中的参与者未能在分析的产品中基于其解决方案感知到显著差异。看似显而易见的解决方案维度在早期研究中未能通过多变量方差分析（MANOVA）得到充分区分。评审者无法根据产品的逻辑性、实用性和价值水平清晰地区分产品。由于需要解决解决方案维度有效性的问题，因此寻求了三个高度新颖的刺激项目，以展示在解决方案维度上比早期类似研究中看到的更多方差。使用三个成品，是否可能观察到刺激项目在解决方案水平上的差异？

The CPSS is an evaluation instrument based on the CPAM, comprised of a series of bipolar adjective item
CPSS 是一种基于 CPAM 的评估工具，由一系列双极形容词项目组成
pairs on 7-point Likert-type scales. Nine subscales measure judged responses to the products being evaluated along each of the nine facets of the model. Each subscale contains four or five item pairs relating to the hypothesized facet. Subscale scores are constructed from the mean of the item pair scores. The CPSS is intended to be useable across domains, and by nonexpert judges. It has been in development and refinement through empirical studies in the United States since 1986 (Besemer & O’Quin, 1986, 1987; O’Quin & Besemer, 1989), but many changes to the items and subscales have been made to the instrument during this time. See the section “Reliability Analyses” for more information about the results of these studies.
在 7 点李克特量表上进行配对。九个子量表测量被评估产品在模型的九个方面上的判断反应。每个子量表包含四到五对与假设方面相关的项目对。子量表得分是通过项目对得分的平均值构建的。CPSS 旨在跨领域使用，并且可以由非专业评审使用。自 1986 年以来，它在美国通过实证研究不断发展和完善（Besemer & O’Quin, 1986, 1987; O’Quin & Besemer, 1989），但在此期间对该工具的项目和子量表进行了许多更改。有关这些研究结果的更多信息，请参见“可靠性分析”部分。

The CPSS had never before been translated into the Norwegian language, Bokmål, although its various versions had been used in research in the United States for more than a decade. Therefore, the reliability of the subscales in Norwegian was computed using Cronbach’s alpha and compared to that of previous studies in the United States. Interrater reliability (Cronbach’s alpha) in the Norwegian sample was good (see Table 2 later) and in some cases exceeded the values found in American samples (O’Quin & Besemer, 1989).
CPSS 从未被翻译成挪威语 Bokmål，尽管其各种版本在美国的研究中使用了超过十年。因此，挪威语子量表的可靠性使用 Cronbach's alpha 进行了计算，并与美国先前研究的结果进行了比较。挪威样本中的评估者间可靠性（Cronbach's alpha）良好（见后面的表 2），在某些情况下超过了美国样本中发现的值（O’Quin & Besemer, 1989）。

Chairs are ubiquitous in western life. Everyone knows what they are, and everyone has had experience with them. Because they are such common objects, chairs were selected as stimulus items. A group of six chairs was selected from illustrations in a book, 397 Chairs (1988) that commemorated an exhibition of chairs as works of art from which three would be chosen for the study. Slides of these illustrations were evaluated by four expert judges. Two were instructors of creativity studies, one a painter, and one an architect. The slides of the three chairs to be used as stimulus items were selected consensually by the expert judges. They selected chairs that were both highly novel and also differentiated on the expected functionality of the chairs for real-life use. Each judge ranked the chairs from most to least creative and from most to
椅子在西方生活中无处不在。每个人都知道它们是什么，每个人都有使用椅子的经验。由于椅子是如此常见的物品，因此被选为刺激物品。选自一本名为《397 把椅子》（1988）的书中的六把椅子的插图，该书纪念了一场将椅子作为艺术作品的展览，其中三把椅子将被选用于研究。这些插图的幻灯片由四位专家评审。两位是创造力研究的讲师，一位是画家，另一位是建筑师。将用于刺激物品的三把椅子的幻灯片由专家评审一致选出。他们选择了既具有高度新颖性又在椅子的预期功能上有所区别的椅子。每位评审将椅子从最具创意到最不具创意进行排名。
© Jennifer Lévy. Chair design: Sylvia Netzner, 1986, “Kitchen Chair.”
© 珍妮弗·莱维。椅子设计：西尔维亚·内茨纳，1986 年，“厨房椅”。

(C Jennifer Lévy. Chair design: Gary Schatmeyer, 1980, “Merylmobile.”
（C Jennifer Lévy. 椅子设计：Gary Schatmeyer，1980 年，“Merylmobile。”）

© Jennifer Lévy. Chair design: Hung-Shu Hu, 1981, “One of the Step-Setter Series.”
© 珍妮弗·莱维。椅子设计：胡宏恕，1981 年，“阶梯设置系列之一。”

Figure 2. Photographs of the stimulus items: Three novel chairs. Note. Photos by Jennifer Lévy. Copyright © 1988 by Jennifer Lévy. Reprinted with permission.
图 2. 刺激物品的照片：三把新颖的椅子。注：照片由詹妮弗·莱维拍摄。版权 © 1988 詹妮弗·莱维。经许可转载。
least functional. Their responses were compared to identify those that were unanimously judged “most creative,” and varying in their level of “functionality.” Because differentiation on Novelty and Elaboration and Synthesis had already been established, this study had Resolution as its focus.
功能性最低。他们的反应被比较，以识别那些被一致评判为“最具创意”的反应，并在“功能性”水平上有所不同。由于新颖性、详细性和综合性的区分已经建立，本研究将解决方案作为其重点。

The chairs (397 Chairs, 1988) were given informal labels for data analysis as follows: #315, Kitchen Chair, by Sylvia Netzner (p. 77), was called Ritz Boxes; #160, Merylmobile, by Gary Schatmeyer, 1980 (p. 43), was called Soft Auto; and #397, One of the Step-Setters Series, by Hung-Shu Hu, 1981 (p. 96), was called Garden Chaise.
这些椅子（397 把椅子，1988 年）在数据分析中被赋予了非正式标签，如下所示：#315，厨房椅，由 Sylvia Netzner（第 77 页）设计，称为 Ritz Boxes；#160，Merylmobile，由 Gary Schatmeyer 于 1980 年设计（第 43 页），称为 Soft Auto；#397，步进系列之一，由 Hung-Shu Hu 于 1981 年设计（第 96 页），称为 Garden Chaise。

Data reported here (see Table 1) were collected in the spring of 1994 from two “folk high schools” (Folkehøgskolene) in western Norway. ${}^1$${}^1$^(1){ }^{1} Participants were 128 students; 89 were women, 38 were men, and 1 individual did not report gender. The average age of the students was 19.54 years ( $SD=1.11$$SD=1.11$SD=1.11S D=1.11 years).
这里报告的数据（见表 1）是在 1994 年春季从挪威西部的两所“民间高等学校”（Folkehøgskolene）收集的。参与者共有 128 名学生；89 名为女性，38 名为男性，1 名未报告性别。学生的平均年龄为 19.54 岁（ $SD=1.11$$SD=1.11$SD=1.11S D=1.11 岁）。

Participants received the CPSS research instrument translated from English into Norwegian and
参与者收到了从英语翻译成挪威语的 CPSS 研究工具

The programs, with variations at many unique institutions in Norway (as well as Denmark, Finland, Iceland, the Faeroe Islands, and Sweden), offer practical education and training in life skills, technical subjects, and personal development in addition to some more traditionally academic subjects. Although some who complete the 1- or 2-year programs then proceed with their education at the university level, the programs are designed to be terminal with the completion of the work done there. Unlike university students, students at the folk high schools may be more talented in crafts or technical subjects than in traditional academic areas. They are from a broader cross-section of the citizenry than are university students.
这些项目在挪威的许多独特机构（以及丹麦、芬兰、冰岛、法罗群岛和瑞典）中有所不同，除了某些更传统的学术科目外，还提供生活技能、技术科目和个人发展的实践教育和培训。尽管一些完成 1 年或 2 年项目的学生随后继续在大学层次上接受教育，但这些项目的设计目的是在完成课程后即为终结。与大学生不同，民众高等学校的学生在工艺或技术科目方面可能比在传统学术领域更有才能。他们来自比大学生更广泛的公民群体。

Table 1. Demographic Description of the Study Sample ${}^{\text{a}}$${}^{\text{a }}$^("a "){ }^{\text {a }}
表 1. 研究样本的人口统计描述 ${}^{\text{a}}$${}^{\text{a }}$^("a "){ }^{\text {a }}

were given oral instructions in Norwegian for completing the form. ${}^2$${}^2$^(2){ }^{2} Time was allowed for clarification of the directions through questions. The students were told that they would be allowed 15 min to complete the form for each of the three stimulus items. Slides of the three chairs were then presented in counterbalanced order. The time was monitored by a stopwatch.
被给予了挪威语的口头指示以完成表格。 ${}^2$${}^2$^(2){ }^{2} 允许通过提问来澄清指示。学生们被告知每个刺激项目将允许 15 分钟来完成表格。然后以平衡顺序展示了三把椅子的幻灯片。时间由秒表监控。

Principle Components Factor Analysis (PCA) was used to examine the factor structure of the data set gathered in Norway to verify that the variables formed coherent subsets. The PCA used in this study was to verify that the items in this use of the CPSS formed logical subsets that were similar to those in earlier studies.
主成分因子分析（PCA）用于检查在挪威收集的数据集的因子结构，以验证变量是否形成一致的子集。本研究中使用的 PCA 旨在验证 CPSS 的项目是否形成与早期研究相似的逻辑子集。

CFA is extremely useful in tests of theoretical models (Nunnally & Bernstein, 1994) and was thus appropriate to this study. CFA tests hypotheses about the factors underlying a set of measured variables. This allows the researcher to specify a particular model or competing models and see how well real-world data fit the expected factor structure. Obviously related to principal components analysis, CFA is generally more useful and complete in the analyses offered (Bryant &
CFA 在理论模型的测试中极为有用（Nunnally & Bernstein, 1994），因此适用于本研究。CFA 检验关于一组测量变量背后因素的假设。这使研究者能够指定特定模型或竞争模型，并观察实际数据与预期因素结构的拟合程度。显然，与主成分分析相关，CFA 在提供的分析中通常更有用且更为完整（Bryant &

Yarnold, 1995) because it provides a way to specify not only factor and variable structure but also allows for specifying factor variances and covariances. CFA also reports errors in the measured variables and simultaneously examines dependence relations. This process helps to develop “a more systematic and holistic view of problems” (Hair, Anderson, Tatham, & Black, 1992, p. 427.) CFA in this study was to test whether a two-factor solution or three-factor solution best accounted for the relations among the factors and the manifest indicators. For the purpose of CFA, the structural modeling program, EQS for Windows Release 5.1 (Bentler & Wu, 1995) was used.
Yarnold（1995）指出，它不仅提供了一种指定因子和变量结构的方法，还允许指定因子的方差和协方差。CFA 还报告测量变量中的误差，并同时检查依赖关系。这个过程有助于形成“对问题更系统和整体的看法”（Hair, Anderson, Tatham, & Black, 1992, 第 427 页）。本研究中的 CFA 旨在测试两因子模型或三因子模型是否更好地解释因子与显性指标之间的关系。为了进行 CFA，使用了结构建模程序 EQS for Windows Release 5.1（Bentler & Wu, 1995）。

The purpose of using MANOVA was to determine if the means of the judged differences among the three chairs were likely to have occurred by chance. Statistical Package for the Social Sciences for Windows Version 6.1 (SPSS; 1994) was used for the MANOVA.
使用多元方差分析（MANOVA）的目的是确定三把椅子之间评判差异的均值是否可能是偶然发生的。使用了社会科学统计软件包（SPSS）Windows 版本 6.1（1994）进行 MANOVA 分析。

Items in each subscale were recoded as necessary to show that a higher score indicated higher levels of the various qualities, such as originality, value, and so on. In order to stabilize the variance and reduce kurtosis and skewness, the data set was transformed by natural logarithmic function. SPSS for Windows Version 6.1 (1994) was used for these procedures and for the reliability analyses.
每个子量表中的项目根据需要进行了重新编码，以表明更高的分数表示更高水平的各种特质，如独创性、价值等。为了稳定方差并减少峰度和偏度，数据集通过自然对数函数进行了转换。使用了 SPSS for Windows 6.1 版（1994）进行这些程序和可靠性分析。

As early as 1986, we were aware of the poor performance of two item pairs: “adequate” and “fresh” (Besemer & O’Quin, 1986; O’Quin & Besemer, 1989). We had retained the items because they had been frequently cited in the literature. Again in the Norwegian study, they continued to be such poor items that they were dropped, and analyses proceeded with this more refined scale. The result of this item pruning was helpful. For example, deletion of “fresh” moved the coefficient alpha for original from .65 to .71 for Ritz Boxes, and the deletion of “adequate” moved the alpha of logical from .63 to .69 on this product. At the item level, only these two changes were made.
早在 1986 年，我们就意识到两个项目对的表现不佳：“足够”和“新鲜”（Besemer & O’Quin, 1986; O’Quin & Besemer, 1989）。我们保留了这些项目，因为它们在文献中被频繁引用。在挪威的研究中，它们仍然表现不佳，因此被删除，分析在这个更精细的量表上继续进行。这次项目修剪的结果是有帮助的。例如，删除“新鲜”使得 Ritz Boxes 的原始系数α从 0.65 提高到 0.71，而删除“足够”使得该产品的逻辑α从 0.63 提高到 0.69。在项目层面，仅进行了这两个更改。

The reliability of each of the subscales was then assessed using Cronbach’s alpha. The mean alphas for each of the three chairs were found to be $.77,.87$$.77,.87$.77,.87.77, .87, and .85 , respectively. The alphas of the Valuable and Useful subscales on the Resolution scale, of special interest in this study, were $>.83$$>.83$> .83>.83.
每个子量表的可靠性随后使用克朗巴赫α系数进行评估。发现三把椅子的平均α值分别为 $.77,.87$$.77,.87$.77,.87.77, .87 和.85。特别关注的解决方案量表中有价值和有用子量表的α值为 $>.83$$>.83$> .83>.83 。

The reliabilities of Novelty, Resolution, and Elaboration and Synthesis and the component subscales were judged to be more than adequate and similar to subscale scores found in other studies using the CPSS in the United States (Besemer & O’Quin, 1986; O’Quin & Besemer, 1989). Table 2 presents the reliabilities of dimensions and scales.
新颖性、解决方案、阐述与综合的可靠性以及各个子量表的可靠性被认为是足够的，并且与其他使用 CPSS 的研究中发现的子量表得分相似（Besemer & O’Quin, 1986; O’Quin & Besemer, 1989）。表 2 展示了各维度和量表的可靠性。

In earlier studies, items had been assembled into 11 subscales. These included the present 9 subscales, plus Complex and Germinal. The subscale Understandable had been hypothesized to be an indicator of Elaboration and Synthesis. In the present study, evidence emerged that suggested that further refinement to the subscales of the CPSS was necessary. There were conceptual, as well as pragmatic, reasons for these modifications. Two subscales, Complex and Germinal, which have posed problems for several years, were yet again problematic. The problem with Germinal was that it had consistently lower alphas than the other Novelty subscales, even in earlier studies in the United States (Besemer & O’Quin, 1986; O’Quin & Besemer, 1989).
在早期的研究中，项目被组装成 11 个子量表。这些包括目前的 9 个子量表，以及复杂性和生发性。子量表“可理解性”被假设为阐述和综合的指标。在本研究中，出现了证据，表明对 CPSS 的子量表进行进一步细化是必要的。这些修改有概念上的原因，也有实际的原因。两个子量表，复杂性和生发性，几年来一直存在问题，再次成为问题。生发性的问题在于，它的α值始终低于其他新颖性子量表，即使在美国的早期研究中也是如此（Besemer & O’Quin, 1986; O’Quin & Besemer, 1989）。

The 55 items of the English language instrument was thus reduced to 43 items in the Norwegian version used in this study. These remaining items produced even greater reliability than the earlier 55-item version. For Ritz Boxes, the mean alpha score for all subscales was .77. Alphas for the three dimensions for Ritz Boxes were $.85,.79$$.85,.79$.85,.79.85, .79, and .87 . The mean alpha for all subscales on Soft Auto was . 87. All subscale alphas were .82 or higher for this chair. There was also clear reliability in judgments about Garden Chaise. The
该研究中使用的挪威版本将 55 个英语量表项目减少至 43 个。这些剩余项目的可靠性甚至超过了早期的 55 个项目版本。对于 Ritz Boxes，所有子量表的平均α值为 0.77。Ritz Boxes 的三个维度的α值分别为 $.85,.79$$.85,.79$.85,.79.85, .79 和 0.87。Soft Auto 所有子量表的平均α值为 0.87。该椅子的所有子量表α值均为 0.82 或更高。对于 Garden Chaise 的判断也显示出明显的可靠性。

Table 2. Reliability (Cronbach’s Alpha) of the Scales by Dimension
表 2. 各维度量表的可靠性（克朗巴赫α系数）

overall coefficient alpha was .85 , and the range of alphas for the dimensions ranged from .84 to .86 .
整体系数α为 0.85，各维度的α值范围为 0.84 至 0.86。

A principal components factor analysis with varimax rotation was used for each product separately to examine the relations among the scales. Following the “a priori criterion” (Hair et al., 1992, p. 237), an effort was made to see how well the subscales sorted into the three proposed dimensions.
对每个产品分别进行了主成分因子分析，并采用方差最大化旋转，以检查量表之间的关系。根据“先验标准”（Hair et al., 1992, p. 237），努力观察子量表如何很好地归类到三个提议的维度中。

Due to the theoretical independence of the three dimensions, earlier tests had been performed with an orthogonal analysis. Therefore, the same analysis was
由于这三个维度的理论独立性，早期的测试采用了正交分析。因此，进行了相同的分析。
used for these new data. A further analysis using oblique rotation was also completed for purposes of comparison. These results may be compared in Table 3.
用于这些新数据。还完成了使用斜旋转的进一步分析以便进行比较。这些结果可以在表 3 中进行比较。

Factor analysis was performed using Principal Components extraction, varimax rotation, and repeated using maximum likelihood extraction and direct oblimin rotation.
因子分析采用主成分提取、方差最大旋转进行，随后使用最大似然提取和直接斜交旋转重复进行。

Using the three-factor format and orthogonal rotation, the three factors’ cumulative percentage of variance accounted for in Ritz Boxes was $74.9\mathrm{\%}$$74.9\mathrm{\%}$74.9%74.9 \% (with eigenvalues of $3.93,1.88$$3.93,1.88$3.93,1.883.93,1.88, and 0.91 , respectively). For Soft Auto 77% of the variance was explained by three factors (with eigenvalues of $4.62,1.60$$4.62,1.60$4.62,1.604.62,1.60, and .70 , respectively). For Garden Chaise the three factors accounted for $79.3\mathrm{\%}$$79.3\mathrm{\%}$79.3%79.3 \% (with eigenvalues of $4.41,1.92$$4.41,1.92$4.41,1.924.41,1.92, and
使用三因子格式和正交旋转，Ritz Boxes 中三个因子的累计方差百分比为 $74.9\mathrm{\%}$$74.9\mathrm{\%}$74.9%74.9 \% （特征值分别为 $3.93,1.88$$3.93,1.88$3.93,1.883.93,1.88 和 0.91）。对于 Soft Auto，三个因子解释了 77%的方差（特征值分别为 $4.62,1.60$$4.62,1.60$4.62,1.604.62,1.60 和 0.70）。对于 Garden Chaise，三个因子占据了 $79.3\mathrm{\%}$$79.3\mathrm{\%}$79.3%79.3 \% （特征值分别为 $4.41,1.92$$4.41,1.92$4.41,1.924.41,1.92 和

Table 3. Results of Factor Analysis of Subscales for Each Product
表 3. 各产品子量表的因子分析结果

Note: Only loadings $>0.45$$>0.45$> 0.45>0.45 were included.
注意：仅包含载荷 $>0.45$$>0.45$> 0.45>0.45 。
${}^2$${}^2$^(2){ }^{2} Loading was split between Factor 1 and Factor 3.
${}^2$${}^2$^(2){ }^{2} 载荷在因子 1 和因子 3 之间分配。

0.78 , respectively). As in studies in the United States (e.g., O’Quin & Besemer, 1989), the best separation of the factors occurred between Novelty and Resolution for Ritz Boxes, Soft Auto, and Garden Chaise.
0.78，分别）。与美国的研究（例如，O’Quin & Besemer，1989）一样，Ritz Boxes、Soft Auto 和 Garden Chaise 之间的因素最佳分离发生在新颖性和解决方案之间。

Turning to the three-factor format with maximum likelihood extraction and oblique rotation, the three factors’ cumulative percentage of variance accounted for was $65.7\mathrm{\%}$$65.7\mathrm{\%}$65.7%65.7 \% (with eigenvalues of $3.93,1.88$$3.93,1.88$3.93,1.883.93,1.88, and 0.91 , respectively). For Soft Auto $68.9\mathrm{\%}$$68.9\mathrm{\%}$68.9%68.9 \% of the variance wasexplained by three factors (with eigenvalues of $4.62,1.60$$4.62,1.60$4.62,1.604.62,1.60, and 0.70 , respectively). For Garden Chaise the three factors accounted for $71.3\mathrm{\%}$$71.3\mathrm{\%}$71.3%71.3 \% (with eigenvalues of 4.41 , 1.92 , and 0.78 , respectively). Again, the best separation of the factors occurred between Novelty and Resolution for Ritz Boxes, Soft Auto, and Garden Chaise.
转向最大似然提取和斜旋转的三因子格式，三个因子的累计方差百分比为 $65.7\mathrm{\%}$$65.7\mathrm{\%}$65.7%65.7 \% （特征值分别为 $3.93,1.88$$3.93,1.88$3.93,1.883.93,1.88 和 0.91）。对于 Soft Auto，三因子解释了 $68.9\mathrm{\%}$$68.9\mathrm{\%}$68.9%68.9 \% 的方差（特征值分别为 $4.62,1.60$$4.62,1.60$4.62,1.604.62,1.60 和 0.70）。对于 Garden Chaise，三个因子占据了 $71.3\mathrm{\%}$$71.3\mathrm{\%}$71.3%71.3 \% （特征值分别为 4.41、1.92 和 0.78）。同样，Ritz Boxes、Soft Auto 和 Garden Chaise 之间的最佳因子分离发生在新颖性和解决方案之间。

Generalized visual representations of the two-factor and three-factor models (path diagrams) are shown in Figures 3 and 4. CFAs were performed to ascertain whether a two-factor or a three-factor solution provided a better fit of the model to the data.
图 3 和图 4 展示了两因子和三因子模型的广义视觉表示（路径图）。进行了确认性因子分析（CFA），以确定两因子或三因子解决方案是否更好地拟合模型与数据。

The two-factor model tested is a representation of the commonly stated idea that a creative solution requires “newness which is useful,” that the product be both new and important, or new and valuable, or that it show “appropriate originality.” Examples of this point of view are found in citations from Mednick and Mednick (1964) to the present day (Csikszentmihalyi, 1996), through the United States patent literature, and the works of scholars like Parnes, Noller, and Biondi (1977), Hayes (1989), Vernon (1989), and Mumford, Baughman, Threlfall, Supinski, and Costanza (1996). The path diagram that represents this model is shown in Figure 3.
所测试的双因素模型是对常见观点的一个表述，即创造性解决方案需要“有用的新颖性”，即产品既要新颖又重要，或新颖且有价值，或显示出“适当的原创性”。这一观点的例子可以在 Mednick 和 Mednick（1964）到现在的引用中找到（Csikszentmihalyi，1996），通过美国专利文献，以及 Parnes、Noller 和 Biondi（1977）、Hayes（1989）、Vernon（1989）和 Mumford、Baughman、Threlfall、Supinski 和 Costanza（1996）等学者的作品中也有体现。表示该模型的路径图如图 3 所示。

The three-factor model, shown in Figure 4 and tested in this study, is a CFA path diagram representation of the theoretical model described in the earliest publication regarding the CPAM (Besemer & Treffinger, 1981). This model also echoes the early important contribution of Jackson and Messick (1967), who introduced the concept of condensation. This relates to the “just-rightness” of the stylistic attributes of the product’s manifestation.
三因素模型，如图 4 所示，并在本研究中进行了测试，是对 CPAM 最早出版物中描述的理论模型的 CFA 路径图表示（Besemer & Treffinger, 1981）。该模型也呼应了 Jackson 和 Messick（1967）的早期重要贡献，他们引入了凝聚的概念。这与产品表现的风格属性的“恰到好处”相关。

In earlier studies (Besemer & O’Quin, 1986; O’Quin and Besemer, 1989) it became clear that there was a correlation among the factors. It is very common
在早期的研究中（Besemer & O’Quin, 1986; O’Quin 和 Besemer, 1989），显然这些因素之间存在相关性。这是非常常见的。

Figure 3. Path diagram of the two-factor generalized model of creativity in products.
图 3. 产品创造力的两因素广义模型路径图。
in scale development to see correlated factors. It should also be noted, however, that the orthogonal principle components analyses shown in Table 3 are superior to those performed with the oblique rotation.
在量表开发中查看相关因素。然而，还应注意，表 3 中显示的正交主成分分析优于采用斜旋转进行的分析。

The three-factor confirmatory model discussed in this study included fixing the first of the regression paths to a value of 1.0 and allowing the others to be freely estimated. The 21 parameters that were estimated corresponded to the three-factor covariances among the three factors or dimensions, the nine regression coefficients between the dependent variables and the factors, and the nine measurement error variances. In order to better understand these parameter estimates, it may be helpful to review the path diagram in Figure 4. The analysis was done on the basis of the cor-
本研究讨论的三因素确认模型包括将第一条回归路径固定为 1.0，并允许其他路径自由估计。估计的 21 个参数对应于三个因素或维度之间的三因素协方差、因变量与因素之间的九个回归系数，以及九个测量误差方差。为了更好地理解这些参数估计，回顾图 4 中的路径图可能会有所帮助。分析是基于相关性进行的。

Figure 4. Path diagram of the three-factor generalized model of creativity in products.
图 4. 产品创造力三因素广义模型的路径图。
relation matrix of the relations among all of the variables in the data set. ${}^3$${}^3$^(3){ }^{3}
数据集中所有变量之间关系的关系矩阵。 ${}^3$${}^3$^(3){ }^{3}

EQS provides several fit indices. These are suitable for different types of samples, but they all range from 0 to 1 , with the minimum acceptable value being 0.90 (Bryant & Yarnold, 1995). For this study, the Robust Comparative Fit Index (CFI), as recommended by Byrne (1994), was used. To provide a statistical correction for any nonnormality in the data set, robust statistics were requested in the EQS analysis. This statistic computes the Satorra-Bentler Scaled ${\chi }^2(\mathrm{S}-\mathrm{B}{\chi }^2$${\chi }^2\left(\mathrm{S}-\mathrm{B}{\chi }^2\right.$chi^(2)((S)-Bchi^(2):}\chi^{2}\left(\mathrm{~S}-\mathrm{B} \chi^{2}\right. ) and robust standard errors. This statistic “has been shown to be the
EQS 提供了几个适合不同类型样本的拟合指数，但它们的范围均为 0 到 1，最低可接受值为 0.90（Bryant & Yarnold, 1995）。在本研究中，使用了 Byrne（1994）推荐的稳健比较拟合指数（CFI）。为了对数据集中的任何非正态性进行统计修正，在 EQS 分析中请求了稳健统计量。该统计量计算 Satorra-Bentler 缩放 ${\chi }^2(\mathrm{S}-\mathrm{B}{\chi }^2$${\chi }^2\left(\mathrm{S}-\mathrm{B}{\chi }^2\right.$chi^(2)((S)-Bchi^(2):}\chi^{2}\left(\mathrm{~S}-\mathrm{B} \chi^{2}\right. ）和稳健标准误差。该统计量“已被证明是
${}^3$${}^3$^(3){ }^{3} Other researchers who wish to replicate this work may contact the author for an electronic copy of the correlation matrix.
其他希望复制此工作的研究人员可以联系作者以获取相关矩阵的电子版。
most reliable test statistic for evaluating covariance structure models under various distributions and sample sizes” (Byrne, 1994, p. 86).
“在各种分布和样本大小下评估协方差结构模型的最可靠检验统计量”（Byrne, 1994, 第 86 页）。

There are three benchmark criteria available in EQS for measuring the adequacy of CFA models. These are the various fit indices: the ratio of the ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} goodness of fit statistic to the degrees of freedom, and the presence or absence of warning condition codes for models that are empirically underidentified.
在 EQS 中，有三个基准标准可用于衡量 CFA 模型的充分性。这些标准包括各种拟合指数： ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} 拟合优度统计量与自由度的比率，以及对于经验上未识别的模型的警告条件代码的存在或缺失。

Two-Factor Model Results of CFA. For the chair that was called Ritz Boxes, using the hypothesized two-factor model, the value of the $\mathrm{S}-\mathrm{B}{\chi }^2$$\mathrm{S}-\mathrm{B}{\chi }^2$S-Bchi^(2)\mathrm{S}-\mathrm{B} \chi^{2} good-ness-of-fit test was 89.52 , based on 27 degrees of freedom. The CFI for this model was 0.86 , which indicates an unacceptable fit. The ratio of the degrees of freedom to the ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} goodness of fit statistic, which should not exceed 2.5 , in this case was 3.31 , which is also unacceptable. Initially, the output of this analysis contained warnings about its usefulness because of certain condition codes. Specifically, two parameters were constrained at the lower bound, which required fixing the error variance at .001 for the indicator original and providing a start value of .001 for surprise. The output for Soft Auto and Garden Chaise were similar to that for Ritz Boxes. The Robust CFI for Soft Auto was 0.84 with 27 degrees of freedom, and the ratio of the ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} to the degrees of freedom was 3.52. The Robust CFI for Garden Chaise was 0.75 , with 26 degrees of freedom, and the ratio of the ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} to the degrees of freedom was 6.42 . All three chairs had unacceptable ratios of the degrees of freedom to the ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} goodness of fit statistic using the two-factor model.
CFA 的双因素模型结果。对于被称为 Ritz Boxes 的椅子，使用假设的双因素模型，适配度检验的值为 89.52，基于 27 个自由度。该模型的 CFI 为 0.86，表明适配度不可接受。自由度与适配度统计量的比率在此情况下为 3.31，应该不超过 2.5，这也是不可接受的。最初，该分析的输出包含关于其有效性的警告，因为某些条件代码。具体而言，两个参数被限制在下限，这要求将指示器原始的误差方差固定为 0.001，并为惊讶提供起始值 0.001。Soft Auto 和 Garden Chaise 的输出与 Ritz Boxes 的输出相似。Soft Auto 的稳健 CFI 为 0.84，具有 27 个自由度，自由度与适配度统计量的比率为 3.52。Garden Chaise 的稳健 CFI 为 0.75，具有 26 个自由度，自由度与适配度统计量的比率为 6.42。 所有三个椅子的自由度与使用双因素模型的拟合优度统计量的比率均不可接受。

Three-Factor Model Results of CFA. Considering the hypothesized three-factor model, the following results were found. Although some kurtosis existed in the sample, these estimates were within the range of acceptability. For each product, the normalized multivariate kurtosis estimates were 6.88 for Ritz Boxes, 7.06 for Soft Auto, and 3.68 for Garden Chaise.
三因子模型的 CFA 结果。考虑到假设的三因子模型，得出了以下结果。尽管样本中存在一些峰度，但这些估计值在可接受范围内。对于每个产品，标准化的多元峰度估计值为：Ritz Boxes 为 6.88，Soft Auto 为 7.06，Garden Chaise 为 3.68。

Using the hypothesized three-factor model, for the chair that was called Ritz Boxes, the value of the $S-B{\chi }^2$$S-B{\chi }^2$S-Bchi^(2)S-B \chi^{2} $=39.87(n=128),24$$=39.87(n=128),24$=39.87(n=128),24=39.87(n=128), 24 degrees of freedom. The Robust CFI for this model was 0.96 , which reflects that the specified model adequately accounted for the observed correlational data. The ratio of the degrees of freedom to the ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} statistic was 1.7 , which is acceptable. Figure 5
使用假设的三因素模型，对于被称为 Ritz Boxes 的椅子，自由度为 $S-B{\chi }^2$$S-B{\chi }^2$S-Bchi^(2)S-B \chi^{2} $=39.87(n=128),24$$=39.87(n=128),24$=39.87(n=128),24=39.87(n=128), 24 。该模型的稳健 CFI 为 0.96，反映出指定模型充分解释了观察到的相关数据。自由度与 ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} 统计量的比率为 1.7，这是可以接受的。图 5

shows the path diagram and standardized solution for Ritz Boxes.
显示了 Ritz Boxes 的路径图和标准化解。

In evaluating the model structure for Soft Auto, it was necessary to remove an outlier and to make some minor adjustments to the generalized path diagram in order for the model to adequately account for the observed data. Because a condition code warning reported that the error variances for both of the Novelty indicators, original and surprise, needed to be restrained at the lower bound, the error variance for original was set at .001, and the error variance for surprising was given a start value of .001 . There were also correlated error variances for the two Novelty indicators and for understandable and elegant, as well as for understandable and logical. There was also one cross-loaded dependent variable, elegant, which
在评估 Soft Auto 的模型结构时，有必要去除一个异常值，并对广义路径图进行一些小的调整，以使模型能够充分解释观察到的数据。由于条件代码警告报告称两个新颖性指标（原始和惊讶）的误差方差需要在下限处受到约束，因此原始的误差方差设定为 0.001，而惊讶的误差方差起始值设定为 0.001。两个新颖性指标之间以及可理解性与优雅性、可理解性与逻辑性之间也存在相关的误差方差。此外，还有一个交叉加载的因变量，即优雅性。

Figure 5. Path diagram and standardized solution for Ritz Boxes.
图 5. Ritz 盒子的路径图和标准化解。
loaded on Novelty, as well as the predicted Elaboration and Synthesis. This cross-loading of subscales between Novelty and Elaboration and Synthesis is not uncommon (e.g., Besemer & O’Quin, 1987) and can cause confusion over the issue whether there are two factors or three factors at work in the CPAM. The effort at fitting the model proved worthwhile, however, because of the greatly improved fit statistics.
加载在新颖性上，以及预测的阐述和综合。这种新颖性与阐述和综合之间的子量表交叉加载并不罕见（例如，Besemer & O’Quin，1987），可能会导致关于 CPAM 中是否存在两个因素或三个因素的问题的混淆。然而，拟合模型的努力是值得的，因为拟合统计显著改善。

For Soft Auto, using a fitted version of the hypothesized three-factor model, the value of the $S-B{\chi }^2$$S-B{\chi }^2$S-Bchi^(2)S-B \chi^{2} good-ness-of-fit test was 22.18 ( $n=127$$n=127$n=127n=127 ), 21 degrees of freedom, $p=.39$$p=.39$p=.39p=.39. The Robust CFI for this model was 0.997, which indicates that the modifications to the specified model greatly improved its adequacy of accounting for the observed data. The ratio of ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} to degrees of freedom was 1.1 , which is another sign of fit. Figure 6 shows the path diagram and standardized solution for Soft Auto.
对于 Soft Auto，使用假设的三因素模型的拟合版本，良度检验的值为 22.18（ $n=127$$n=127$n=127n=127 ），自由度为 21， $p=.39$$p=.39$p=.39p=.39 。该模型的稳健 CFI 为 0.997，这表明对指定模型的修改大大提高了其对观察数据的适应性。 ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} 与自由度的比率为 1.1，这也是适配的另一个标志。图 6 显示了 Soft Auto 的路径图和标准化解。

For Garden Chaise, using the hypothesized three-factor model, the value of the $\mathrm{S}-\mathrm{B}{\chi }^2$$\mathrm{S}-\mathrm{B}{\chi }^2$S-Bchi^(2)\mathrm{S}-\mathrm{B} \chi^{2} good-ness-of-fit test was 60.70, ( $n=128$$n=128$n=128n=128 ), 24 degrees of freedom, $p=.0003$$p=.0003$p=.0003p=.0003. The CFI for this model was 0.94 , which shows that the specified model performed adequately to account for the observed correlational data. The ratio of ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} to degrees of freedom was 2.5 , which is marginally acceptable. Figure 7 shows the path diagram and standardized solution for Garden Chaise.
对于花园躺椅，使用假设的三因素模型，拟合优度检验的值为 60.70，（ $n=128$$n=128$n=128n=128 ），自由度为 24， $p=.0003$$p=.0003$p=.0003p=.0003 。该模型的 CFI 为 0.94，表明指定模型能够充分解释观察到的相关数据。 ${\chi }^2$${\chi }^2$chi^(2)\chi^{2} 与自由度的比率为 2.5，属于边际可接受范围。图 7 显示了花园躺椅的路径图和标准化解。

It is important to point out that it would have been possible to raise the level of the CFI on the models for each of the individual chairs by adding parameters to improve the fit indices for each of the stimulus items, as was done for Soft Auto. Researchers are urged to be cautious about overfitting theoretical models for pragmatic reasons, simply to increase the CFI (Byrne, 1994, p. 86). In this study the analyses were not intended to create the best fit possible between the models and the data. Instead, they were to answer the first research question, “Does confirmatory factor analysis support a three-dimensional approach to understanding creative products?” The answer to that question is yes.
重要的是要指出，通过为每个刺激项目添加参数以改善拟合指数，确实可以提高每个单独模型的 CFI 水平，就像对软汽车模型所做的那样。研究人员被敦促在实际原因上对理论模型的过拟合保持谨慎，仅仅为了提高 CFI（Byrne, 1994, p. 86）。在本研究中，分析并不是为了在模型与数据之间创建最佳拟合。相反，它们旨在回答第一个研究问题：“确认性因素分析是否支持理解创意产品的三维方法？”对此问题的回答是肯定的。

Examination of the path diagrams with estimates in Figures 5, 6, and 7 reveals some interesting insights about the differences in these products. For example, in comparing the relation of surprise to originality for Ritz Boxes and Soft Auto, it is interesting to note that surprise is a stronger manifest indicator of Novelty in Ritz Boxes than is originality, whereas for Soft Auto and for Garden Chaise, the relation is reversed. Origi-
对图 5、6 和 7 中估计的路径图进行检查揭示了关于这些产品差异的一些有趣见解。例如，在比较 Ritz Boxes 和 Soft Auto 中惊讶与原创性的关系时，有趣的是，惊讶在 Ritz Boxes 中是新颖性的一个更强的显性指标，而原创性则相对较弱；而对于 Soft Auto 和 Garden Chaise，这种关系则正好相反。

Figure 6. Path diagram and standardized solution for Soft Auto.
图 6. Soft Auto 的路径图和标准化解。
nality is a stronger manifest indicator of Novelty in those products. Comparing the relation of well crafted to Elaboration and Synthesis for the three chairs, it is possible to see another instance of this phenomenon. Well crafted is a stronger indicator of Elaboration and Synthesis in Garden Chaise than it is in Ritz Boxes or Soft Auto. The path diagrams with their standardized solutions make these relations clear and understandable. Product difference characteristics will also be discussed in the MANOVA results section.
在这些产品中，独特性是新颖性的更强表现指标。比较三把椅子中精心制作与阐述和综合的关系，可以看到这一现象的另一个实例。在花园躺椅中，精心制作是阐述和综合的更强指标，而在丽兹盒子或软汽车中则不是。带有标准化解的路径图使这些关系清晰易懂。产品差异特征也将在 MANOVA 结果部分讨论。

A MANOVA was performed to examine the differences in ratings of the three chairs, with the nine
进行了多元方差分析（MANOVA），以检验对三把椅子评分的差异，涉及九个

Figure 7. Path diagram and standardized solution for Garden Chaise.
图 7. 花园躺椅的路径图和标准化解。
subscale scores as dependent variables for the products. Results showed that the multivariate main effect for chair was significant, $F(18,476)=23.22$$F(18,476)=23.22$F(18,476)=23.22F(18,476)=23.22 (Wilks’ lambda), $p<.001$$p<.001$p < .001p<.001. Detailed results of this analysis are presented in Table 4.
子量表得分作为产品的因变量。结果显示，椅子的多变量主效应显著， $F(18,476)=23.22$$F(18,476)=23.22$F(18,476)=23.22F(18,476)=23.22 （Wilks' lambda）， $p<.001$$p<.001$p < .001p<.001 。该分析的详细结果见表 4。

All of the multivariate $F\mathrm{s}$$F\mathrm{s}$FsF \mathrm{~s} were significant, as is shown in Table 4. Examination of the univariate $F\mathrm{s}$$F\mathrm{s}$FsF \mathrm{~s} revealed that, there, too, all the perceived variance among stimulus items on each of the variables was significant ( $p<.05$$p<.05$p < .05p<.05 ). In fact, all of the variables except those in Novelty (surprising, original) achieved significance at the $p<.001$$p<.001$p < .001p<.001 level. Original, $F(2,246)=4.61$$F(2,246)=4.61$F(2,246)=4.61F(2,246)=4.61, $p<.05$$p<.05$p < .05p<.05, followed by surprise, $F(2,246)=3.34,p<.05$$F(2,246)=3.34,p<.05$F(2,246)=3.34,p < .05F(2,246)=3.34, p<.05, showed the least variance of the nine subscales. Novelty subscales received the highest ratings for each of
所有的多变量 $F\mathrm{s}$$F\mathrm{s}$FsF \mathrm{~s} 都是显著的，如表 4 所示。对单变量 $F\mathrm{s}$$F\mathrm{s}$FsF \mathrm{~s} 的检查显示，在每个变量上，刺激项目之间感知的方差也是显著的（ $p<.05$$p<.05$p < .05p<.05 ）。事实上，除了新颖性（令人惊讶、原创）之外，所有变量在 $p<.001$$p<.001$p < .001p<.001 水平上都达到了显著性。原创、 $F(2,246)=4.61$$F(2,246)=4.61$F(2,246)=4.61F(2,246)=4.61 、 $p<.05$$p<.05$p < .05p<.05 ，其次是惊讶、 $F(2,246)=3.34,p<.05$$F(2,246)=3.34,p<.05$F(2,246)=3.34,p < .05F(2,246)=3.34, p<.05 ，在九个子量表中显示出最小的方差。新颖性子量表在每个方面都获得了最高的评分。

Table 4. Comparisons Among Products: Multivariate Analysis of Variance
表 4. 产品比较：多元方差分析

Note: Originally, scales ranged from 1 to 7 , from $low$$low$lowl o w to high on each variable. Scales were transformed using the natural logrithmic function.
注意：最初，量表的范围从 1 到 7，表示每个变量的 $low$$low$lowl o w 到高。量表使用自然对数函数进行了转换。
${}^{\mathrm{n}}n=125$${}^{\mathrm{n}}n=125$^(n)n=125{ }^{\mathrm{n}} n=125.
${}^{\ast }p<.05.{}^{\ast \ast }p<.001$${}^{\ast }p<.05.{}^{\ast \ast }p<.001$^(**)p < .05.^(****)p < .001{ }^{*} p<.05 .{ }^{* *} p<.001.
the three chairs. Only for the Elaboration and Synthesis subscales of Soft Auto did any ratings reach the level of those for all three of the three chairs’ Novelty scores. This may be a result of the fact that these chairs were intentionally selected to be highly novel. In this sense, Novelty was controlled.
这三把椅子。只有在软自动的阐述和综合子量表中，任何评分达到了这三把椅子的所有新颖性评分的水平。这可能是因为这些椅子是故意选择的，以具有高度的新颖性。从这个意义上说，新颖性是被控制的。

The greatest variance, seen in Table 4, was in usefulness, $F(2,246)=163.77,p<.001$$F(2,246)=163.77,p<.001$F(2,246)=163.77,p < .001F(2,246)=163.77, p<.001