中學生性向及興趣之潛在結構分析：整合性研究

在生涯決策時關鍵性的考慮因素包含了性向和興趣，因此學校常用性向測驗和興趣測驗進行生涯輔導輔導工作。本研究採用個人中心取向（person-centered）方法，透過潛在剖面分析（Latent Profile Analysis, LPA）和潛在類別分析（Latent Class Analysis, LCA），將焦點放在辨識國中階段學生性向組型、興趣組型和性向-興趣整合組型的潛在類別。研究一主要目的在分析國中階段學生在八種性向可以形成那些性向組型，並且比較不同性別在組型上是否有所差異。蒐集了18,950名國中三年級學生資料，測量工具為「電腦化適性職涯性向測驗」（宋曜廷，2011）。研究一結果主要有兩項，第一，透過潛在剖面分析可以辨識出24種的性向組型，可歸納為三大類，分別為正向組型、負向組型和未分化組型。約有七成五的學生為正向組型，此代表多數學生都具有一項以上適合發展的潛能。在負向組型和未分化組型中有約二成五的學生。第二，針對24種的性向組型進行性別上的檢定，可以發現男生在空間、邏輯推理、科學能力等組型及負向組型的比例高於女生；女生則是在語文、觀察、美感及未分化組型的比例高於男生。研究二主要在探討國中階段學生在Holland六種興趣類型可以形成那些興趣組型，並同時檢驗Holland（1997）環狀結構、Gati（1979, 1991）的興趣群組和Rounds和Tracey（1996）所提出的三分群模式是否能與本研究所辨識出的組型相呼應。此外，並比較不同性別的興趣組型是否有所差異。最後，比較Holland固定的三碼組型（three-letter code）和本研究所辨識出之組型，何者較為貼近Iachan（1984）所提出的辨識組型的方式。蒐集了32,221名國中三年級學生資料，測量工具為「電腦化情境式職涯興趣測驗」（Sung et al., 2015）。研究二結果主要有三項，透過潛在剖面分析可以辨識出72種的興趣組型，可歸納為三大類，分別為正向組型、負向組型和未分化組型。有高達五成的學生為未分化的組型。部分組型現與有相關理論及研究有一致性的發現。第二，針對各組在性別上的分配進行檢定，研究者發現，男生在實用型（R）和研究型（I）相關的組型上其比例高於女生；女生在藝術型（A）和社會型（S）的組型其比例顯著高於男生。第三，根據LPA所區分出的生涯興趣組型和Holland固定三碼進行比較，透過潛在剖面分析的組型辨識，在個人特徵的辨識度上可達九成的正確率。研究三所欲探討的問題，根據研究一和研究二之研究結果，在取得24組的性向組型和72組興趣組型，可以建構出那些整合組型呢?並進一步探討國中階段學生在15個高職之職群及高中，在性向高低和興趣高低的分布情形。本研究蒐集了研究一和研究二樣本中同時完成了兩個測驗之18,942位國三學生的資料。研究三結果主要有兩項，第一，為能更有效的去解釋及應用組型的特徵，本研究針對研究一和研究二所發現之組型加以分類，透過LCA的分析取得六項組型，其中以「性向為正向組型和興趣為未分化組型」這一個組別的人數最多，占整體樣本超過四成。第二，約有一成五的學生在群科的發展中，有性向和興趣的交集，代表可以選擇到同時具有性向和興趣的群科進行發展，另有八成五的學生在高中及高職群科其興趣與性向是沒有交集的。進一步的分析發現，在無交集的學生中，有近五成八的學生僅具有高性向或是高興趣，代表性向和興趣間產生鴻溝（discrepancy）的現象。此外，有將近一成五的學生在高中及所有群科其所需具備之能力偏低且並不感興趣，未有適合發展的方向，可視為待探索型。綜合三項研究結果，本研究將以國中階段學生之性向組型、興趣組型以及性向-興趣整合組型，提供理論和實務上之建議，期望給予教育學者及諮商人員輔導學生時有所參考。

關鍵字

性向組型；興趣組型；性向-興趣整合組型；潛在剖面分析；潛在類別分析

並列摘要

In countries where streaming takes place in the early or middle stages of education, students have to decide at a young age how to direct their future careers. Since key factors in the making of career choices include aptitude and personal interest, school career counseling programs often use aptitude tests and interest tests. Adopting a person-centered approach, this study uses latent profile analysis（LPA）and latent class analysis（LCA）to identify the latent categories of aptitude profiles, interest profiles, and combined aptitude-interest profiles among students in junior high school. In study 1, this study mainly analyzes the aptitude profiles that can be formed from eight types of aptitude, and seeks to discover whether students of different genders also differ in their prevailing aptitude profiles. The responses of 18,590 jounior high school students from Grade 9 were collected using Sung（2012）Computerized Adaptive Career Aptitude Test（CACAT）. The CACAT consisted of eight sub-tests: Verbal, Numerical, Spatial, Logical Reasoning, Scientific Reasoning, Observation, Aesthetics, and Creativity. Study 1 yields two main results: First, the results show that the data best fit a 24-classes model, 24 aptitude profiles in three categories were identified: positive, negative, and undifferentiated. Positive profiles were divided into those with a single-letter code, double-letter code, three-letter code, and five-letter code or more. Approximately 75% of the students had positive profiles. Second, the proportions of females and males differed signiﬁcantly among the proﬁles. The proportion of males was higher for spatial, logical reasoning, and scientific reasoning profiles and negative profiles than of female proportions. The proportion of female was higher for verbal, observation, and aesthetics profiles than of males proportions. In study 2, this study discusses what interest patterns can be formed from Holland’s six interest types. It also examines whether the profiles identified in this study correspond to Holland’s circular order（1997）, Gati’s hierarchical model（1979,1991）, and Rounds and Tracey’s three-class partition model（1996）, and whether students of different genders have different interest profiles. In addition, Holland’s three-letter code and the profiles identified in this study are compared in order to determine which is closer to Iachan’s index. The sample consisits of 32,221 jounior high school students from Grade 9 were collected using Sung et al.’s (2015) Situation-based Career Interest Assessment (SCIA). SCIA is based on the interest theory of Holland (1997). A latent profile analysis on six interest types revealed several career interest profiles. Results show that: First, the results show that the data best fit a 72-classes model, 72 interest profiles in three categories were identified: positive, negative, and undifferentiated. Positive profiles could be divided into single-letter, double-letter, three-letter, and four-letter codes, while negative profiles could be divided into double-letter, four-letter, and five-letter codes. This study identiﬁed three undifferentiated groups: “like all,” “moderate for all,” and “dislike all”. It is noteworthy that a large percentage (53.05%) of junior-high-school students fell into the undifferentiated group. Some profiles corresponded to Holland’s circular order, Gati’s hierarchical model, and Rounds and Tracey’s three-class partition model. Second, a higher proportion of male had R (Realistic) and I (Investigative) profiles than female, while a significantly higher proportion of female had A (Artistic) and S (Social) profiles than male. A significantly higher percentage of male had undifferentiated profiles than female. Third, a comparison between interest profiles identified by the LPA and Holland’s three-letter code revealed that interest profiles based on codes consisting of a fixed number of holland code cannot truly capture a person’s interest characteristics. By contrast, profiles identified through the LPA had a 90% accuracy rate in capturing an individual’s characteristics. This shows that the profiles identified in the present study can highlight students’ interest profiles while avoiding overproduction of categories, which leads to difficulty in presentation or interpretation. In study 3,the study seeks to answer is based on the results of the first and second parts; that is, after obtaining 24 aptitude profiles and 72 interest profiles, what combined profiles can be formed? This part of the research also looks at the distribution of junior high school students among 15 vocational clusters of vocational high schools and academically oriented high schools as well as the four quadrants of aptitude-interest combinations. Date were collected from 18,942 samples from the Study 1 and Study 2. In order to better interpret and use the characteristics of the profiles, this study combined the profiles identified in Study 1 and Study 2 and using the LCA .The results show that the data best fit a 6-classes model. Of the six profiles, the combination of positive aptitude profiles with undifferentiated interest profiles made up the biggest number, accounting for over 40% of the total samples. About 15% of the students saw intersections of aptitude and interests in specific clusters of subjects, indicating that they could choose to study these subject clusters. 85% of the students did not see any intersections of aptitude and interests for either academically oriented high schools or subject clusters of vocational high schools. Findings highlight the importance of career counseling practitioners’ attention to the individual differences in aptitude profiles, career interest profiles and and combined aptitude-interest profiles .On the basis of these three sets of results, this study provides theory-and practice-based recommendations about career proﬁles. Implications for career practices and future research are proposed.