Coping Scale With Test Anxiety (COSTA) for Adolescent Students: Exploring Coping Profiles and Predictors Through a Latent Class Analysis Approach
Author: Tzu-Yang Chao (Graduate Institute of Learning and Instruction, National Central University)
Vol.&No.：Vol. 70, No. 1
Date：March 2025
Pages：185-221
DOI：https://doi.org/10.6209/JORIES.202503_70(1).0006
Abstract：
Objective
In countries with High-Stakes Testing (HST) systems, student test anxiety has become a widespread phenomenon. For example, Chao and Sung (2019) reported that 22% of ninth-grade students in Taiwan experience high levels of test anxiety. Therefore, researching how students cope with test anxiety is crucial. Earlier researchers classified individual coping strategies into three main categories: problem-focused coping, emotion-focused coping, and avoidance coping. Chao et al. (2014) developed the Coping with Examination Stress Scale (CESS), which also reflected the three-dimensional coping classification. Although CESS aimed to address the context-specific nature of coping, recent studies have increasingly moved beyond the three-dimensional framework to identify specific coping methods for specific situations, aligning with the person-environment interaction model. Additionally, recent research on coping strategies has shifted from a variable-centered approach to a person-centered approach, emphasizing the heterogeneity within subgroups of the same population. Therefore, the primary purpose of this study is to re-develop the Coping Scale with Test Anxiety (COSTA) for secondary school students by expanding beyond the three dimensions of CESS to include a wider range of coping strategies. During the pretest analysis phase, we aim to retain as many meaningful coping strategies as possible. Another objective of this study is to use the LPA approach to identify different coping patterns among students. Furthermore, we employ latent class regression (LCR) to predict different coping pattern categories based on students’ individual attributes.
Methods
During the pretest phase, this study collected potential behaviors or thoughts students might have when facing test anxiety and stress through interviews with several middle and high school students. After organizing and consulting with experts, we developed a 110-item COSTA scale as the preliminary questionnaire. The responses were scored using a five-point Likert scale, where students indicated how well each statement applied to them by selecting one of the following options: “completely untrue,” “mostly untrue,” “somewhat true,” “mostly true,” and “completely true,” scored from 1 to 5 points, respectively. Higher scores indicated a greater likelihood of using that coping strategy. The pretest sample was collected using purposive sampling from 253 secondary school students (grades 7 to 12, 45.8% female) in Taiwan. In the formal testing phase, the study again used purposive sampling, selecting data from 743 secondary school students (grades 7 to 12, 47.1% female) from seven schools in Northern Taiwan. This data included personal attribute information, the formal questionnaire, and external criteria such as the Examination Stress Scale (ExamSS) (Sung & Chao, 2015) and the Adolescents Uncertainty Scale (AUS) (Chao & Sung, 2023). Personal attribute information included gender, grade level, number of siblings, birth order, number of books at home, class rank, days attending tutoring, study time on weekdays and weekends, and other relevant details. For data analysis, exploratory factor analysis (EFA) and item analysis were conducted during the pretest phase. The removing the first factor method was used during EFA to ensure that theoretically meaningful factors were retained. In the formal phase, confirmatory factor analysis (CFA), reliability analysis, correlation analysis with external criteria, LPA, and three-step LCR were conducted to assess the quality of COSTA, identify the number of coping patterns among students, and determine whether individual attributes could predict specific coping pattern categories.
Results
Through the pretest phase, the study finalized a formal COSTA scale comprising 63 items across 12 dimensions: “Seeking Improvement,” “Positive Cognition,” “Abandoning Preparation,” “Emotional Venting,” “Relaxation Behavior,” “Mood Shifting,” “Support for Improvement,” “Accepting Results,” “Gathering with Friends,” “Consoling Thoughts,” “Self-Reward,” and “Writing for Emotional Release.” In the formal phase, CFA indicated that the 12-dimension measurement model had acceptable fit, with composite reliability for each dimension ranging from .56 to .87 and average variance extracted ranging from .18 to .54. In terms of reliability, the Cronbach’s α values for the 12 dimensions ranged from .66 to .88. Overall, the reliability of the 12 dimensions of coping with test anxiety among secondary school students was acceptable, except for slightly lower reliability in the “Consoling Thoughts” and “Writing for Emotional Release” dimensions. In terms of external validity, test anxiety was significantly correlated with “Seeking Improvement” (r = .376), “Relaxation Behavior” (r = .342), and “Positive Cognition” (r = .146), but not significantly correlated with “Abandoning Preparation.” Regarding uncertainty, a significant correlation was found with avoidance coping, such as “Abandoning Preparation” (r = .271), providing external validity evidence for COSTA. Further, LPA on the mean scores of the 12 dimensions identified seven optimal categories, named as “Low Coping” (6.94%), “Abandon and Accept” (3.39%), “Positive Multiple Coping” (19.91%), “Mild Positive Multiple Coping” (26.99%), “Mood Shifting Dominant” (21.68%), “High Positive Multiple Coping” (9.44%), and “Average Multiple Coping” (12.09%). Additionally, LCR was used to predict the seven latent categories based on students’ individual attributes, including gender, grade level, class rank, number of siblings, birth order, number of books at home, days attending tutoring, study time on weekdays and weekends, sleep time, test anxiety, and uncertainty. Results showed that certain individual attributes could indeed predict coping patterns. For example, less desirable coping patterns, such as the “Low Coping” and “Abandon and Accept” groups, were more likely among students with fewer books at home, less study time on weekends, lower test anxiety, higher uncertainty, and lower class ranks.
Discussion
In terms of the theory of test anxiety, this study adopts a theory-driven approach to identify various coping strategies used by students when facing test anxiety. Besides the original three major coping strategies, this study identifies more nuanced categories, such as “Positive Cognition,” which involves cognitive shifts in thinking about exams, “Emotional Venting,” which can be destructive, and “Writing for Emotional Release,” which helps alleviate stress through writing. This study expands on the three-dimensional theory proposed by Chao et al. (2014), presenting 12 dimensions in COSTA that are more refined and highlight the context-specific nature of coping strategies. The results of this study indicate that there is no single coping strategy that stands out as superior when dealing with test anxiety. Therefore, theoretically, future researchers should shift their focus from specific “coping strategies” to specific “coping profiles.” To achieve this, changes in research methods are also necessary. First, researchers should strive to identify more nuanced coping strategies. Second, researchers should use classification statistical methods, as demonstrated in this study. This shift in perspective in coping strategy research is a significant theoretical contribution of this study. Moreover, this study found that more than half of the students employed relatively effective coping profiles, while approximately 10% of the students fell into the less desirable coping profiles of “Low Coping” and “Abandon and Accept.” From a practical application standpoint, this study provides a detailed examination of students’ coping strategies from a profile perspective, allowing counselors to better understand which students might be using less effective coping methods, such as those who study less on weekends, have lower test anxiety, higher uncertainty, and fewer family resources. This study hopes that the information provided will enable practitioners and educators to proactively support these students by offering emotional support and coping interventions.
Keywords：secondary school student, factor analysis, coping scale with test anxiety, latent profile analysis, latent class regression