Jamovi 통계 분석 마스터 튜토리얼: GUI 기반 접근에서 고급 모형 구축까지

1. Executive Summary

• 본 보고서는 Jamovi 통계 분석 소프트웨어의 환경 설정부터 구조방정식 모델링(SEM)까지의 전체 과정을 체계적으로 안내합니다. Jamovi는 오픈 소스 기반으로, GUI 인터페이스와 R 코드 통합을 통해 초보자부터 전문가까지 쉽게 통계 분석에 접근할 수 있도록 설계되었습니다. 주요 내용은 Jamovi 설치 및 데이터 준비, 기초 통계량 계산, 가설 검정, 회귀 분석, 고급 통계 기법(신뢰도 분석, 요인 분석), 그리고 실제 여행상품 홍보 사례를 활용한 A/B 테스트 및 SEM 적용을 포함합니다.

• 핵심 내용은 Jamovi의 사용자 친화적인 인터페이스를 활용한 분석 방법과, 통계 모델의 신뢰성 및 타당성 확보를 위한 평가 원칙을 강조합니다. 예를 들어, 크론바흐 알파 계수를 계산하여 측정 도구의 내적 일관성을 평가하고, 탐색적·확인적 요인 분석을 통해 요인 구조를 검증합니다. 또한, 여행상품 홍보 데이터셋을 활용하여 실제 마케팅 전략에 적용 가능한 통계 분석 사례를 제시합니다. 결론적으로, 본 튜토리얼은 통계 분석 역량 강화와 데이터 기반 의사결정 능력 향상을 목표로 하며, 사용자에게 실질적인 가이드라인을 제공합니다.

2. Introduction

• 데이터 기반 의사결정의 중요성이 날로 강조되는 현대 사회에서, 통계 분석은 필수적인 역량으로 자리 잡았습니다. 하지만 복잡한 통계 소프트웨어와 프로그래밍 언어에 대한 부담감으로 인해 많은 이들이 통계 분석에 어려움을 겪고 있는 것이 현실입니다. 이러한 문제점을 해결하기 위해, 본 보고서는 오픈 소스 기반의 통계 분석 소프트웨어인 Jamovi를 활용하여 통계 분석의 문턱을 낮추고, 데이터 기반 의사결정 역량 강화를 목표로 합니다.

• Jamovi는 직관적인 GUI 인터페이스와 강력한 R 기반 엔진을 결합하여, 초보자부터 전문가까지 쉽게 통계 분석에 접근할 수 있도록 설계되었습니다. 본 보고서는 Jamovi의 설치 및 환경 설정부터, 기초 통계 분석, 가설 검정, 회귀 분석, 고급 통계 기법, 그리고 실제 사례 연구까지, 통계 분석의 전 과정을 체계적으로 안내합니다. 특히, 여행상품 홍보 데이터셋을 활용한 A/B 테스트 및 SEM 모델링을 통해, 이론적 지식과 실무 적용 간의 간극을 좁히고, 데이터 기반 의사결정 역량을 강화하는 데 중점을 둡니다.

• 본 보고서를 통해, 독자들은 Jamovi를 활용한 통계 분석 능력을 향상시키고, 데이터 기반 의사결정 역량을 강화하여, 실제 비즈니스 문제 해결에 기여할 수 있을 것입니다. 또한, 통계 분석 결과 해석 및 보고서 작성 능력을 향상시켜, 데이터 기반 의사결정의 효과를 극대화할 수 있을 것입니다.

3. Jamovi의 통계 분석 환경 이해 및 GUI 기반 접근성 확보

• 3-1. Jamovi의 기능과 이점

• This subsection introduces Jamovi, emphasizing its unique advantages as an open-source statistical software tailored for educational and research environments. It highlights Jamovi's cost-effectiveness and user-friendly interface compared to proprietary alternatives like SPSS, establishing its value proposition before diving into practical setup and workflow.

• Jamovi's origin as an open-source project, built upon the R statistical language, positions it uniquely in the statistical software landscape. This contrasts sharply with proprietary software like SPSS, where licensing fees and vendor lock-in can create barriers to access, particularly in educational settings with limited budgets. The open-source nature fosters a collaborative environment, allowing users to contribute to its development and tailor it to specific research or teaching needs.

• The architecture of Jamovi leverages the power of R while providing a graphical user interface (GUI) that simplifies statistical analysis for beginners. This dual structure addresses a key challenge in statistical education: bridging the gap between theoretical concepts and practical application. Users can perform analyses through intuitive point-and-click operations, while simultaneously gaining exposure to the underlying R code. This allows for a gradual progression from GUI-based learning to more advanced, code-based customization and analysis.

• Compared to SPSS, Jamovi offers a compelling value proposition regarding cost and accessibility. While SPSS requires annual license renewals that can be a significant expense for universities and individual researchers [185, 187, 190, 191], Jamovi is free to use and distribute. Furthermore, Jamovi's modular design allows users to install only the specific statistical modules they need, reducing system resource requirements and streamlining the user experience. This combination of cost-effectiveness and ease of use makes Jamovi an attractive alternative for institutions and individuals seeking a powerful yet accessible statistical tool.

• Strategic implications center on Jamovi's potential to democratize statistical education. By removing financial barriers and providing an intuitive interface, Jamovi empowers a broader range of students and researchers to engage with statistical analysis. This can lead to increased statistical literacy and a more data-driven approach to decision-making across various fields.

• For educators, adopting Jamovi can significantly reduce software costs while enhancing the learning experience. Universities can integrate Jamovi into their statistics curricula, providing students with hands-on experience using a modern and versatile tool. Researchers can leverage Jamovi's open-source nature to customize analyses and collaborate with colleagues.

• While Jamovi boasts core functionalities mirroring SPSS, its extensibility through modules developed by the community provides a significant competitive advantage. This is especially relevant considering that users appreciate the freedom to modify and expand functionalities based on their unique research requirements and analytical methods.

• Unlike SPSS, where advanced procedures often require costly add-ons, Jamovi's module system allows users to access cutting-edge statistical techniques for free. The active community development ensures that Jamovi stays abreast of the latest advancements in statistical methodology, offering users a continually expanding toolkit for data analysis. Modules encompassing Bayesian analysis, structural equation modeling (SEM), and meta-analysis extend Jamovi's capabilities far beyond basic statistical procedures [1, 22, 44].

• The rapid growth of the Jamovi module library underscores its adaptability and responsiveness to user needs. As researchers and educators adopt Jamovi, they contribute to its ecosystem by developing modules tailored to specific disciplines and analytical challenges. This collaborative model fosters continuous innovation and ensures that Jamovi remains a relevant and powerful tool for statistical analysis.

• The strategic implication of Jamovi’s modularity is that it offers a future-proof solution for statistical analysis. Users are not limited by the features offered by a single vendor but can leverage the collective intelligence of the open-source community to access and adapt new analytical techniques. This flexibility is particularly valuable in rapidly evolving fields where new statistical methods are constantly emerging.

• To capitalize on Jamovi's modularity, researchers should actively explore the available modules and contribute to their development. Universities can encourage students to create custom modules as part of their coursework, fostering a deeper understanding of statistical methods and promoting innovation. This collaborative approach will ensure that Jamovi remains a dynamic and powerful tool for statistical analysis.

• While Jamovi’s open-source nature and cost-effectiveness are compelling advantages, its performance relative to established software like SPSS is a critical factor for adoption. Anecdotal evidence suggests that Jamovi can offer speed and efficiency gains in certain analytical tasks, particularly when handling large datasets [2, 35].

• Jamovi's underlying R architecture, optimized for statistical computing, can contribute to faster processing times compared to SPSS, which may rely on legacy code. This efficiency is further enhanced by Jamovi's modular design, which avoids loading unnecessary features and reduces system overhead. The specific performance gains will depend on the complexity of the analysis, the size of the dataset, and the hardware configuration, but initial indications suggest that Jamovi can be a competitive alternative for computationally intensive tasks.

• While formal benchmarking studies comparing Jamovi and SPSS performance are limited, user reports and anecdotal comparisons highlight potential advantages for Jamovi in specific scenarios. For example, processing large survey datasets or running complex simulations may be faster in Jamovi due to its optimized R engine and modular design. Further research is needed to quantify these performance differences across a wider range of analytical tasks and hardware configurations.

• The strategic implication of Jamovi's potential performance advantages is that it can free up valuable time and resources for researchers and educators. Faster processing times translate to increased productivity and reduced waiting times, allowing users to focus on interpreting results and drawing insights from their data. This efficiency gain is particularly beneficial in time-sensitive research projects and demanding educational settings.

• To fully realize Jamovi's performance potential, users should optimize their hardware configurations and leverage the software's modular design. Universities can provide students with access to high-performance computing resources to accelerate computationally intensive analyses. Further research is needed to systematically benchmark Jamovi against SPSS and other statistical software packages to provide users with clear guidance on performance expectations.

• 3-2. 두 층위 구조와 확장성

• Building on the discussion of Jamovi's core functionalities and benefits, this subsection delves into its unique dual-layered structure, showcasing how it caters to both beginner and advanced users by seamlessly integrating GUI-based analysis with underlying R code. This structure promotes gradual skill development and customized expansions.

• Jamovi's primary interface offers a user-friendly, click-based workflow that significantly lowers the barrier to entry for statistical analysis. This intuitive GUI allows users to perform common statistical tasks, such as descriptive statistics, t-tests, and ANOVA, without writing a single line of code. This approach contrasts sharply with traditional statistical software like R, which requires users to have a strong understanding of programming concepts [2, 35].

• The strategic advantage of this click-based workflow is that it allows novice users to quickly grasp fundamental statistical concepts and gain hands-on experience with data analysis. By abstracting away the complexities of coding, Jamovi enables users to focus on interpreting results and drawing meaningful conclusions from their data. This is particularly beneficial in educational settings where students may have limited prior experience with statistics or programming [175].

• The '누구나 할 수 있는 jamovi 통계분석' 교재 emphasizes Jamovi's ability to perform analyses with simple clicks after importing data, much like Excel, while still providing results in APA style, a standard for social science publications [1]. This simplifies the reporting process and enables beginners to create professional-looking documents without extensive editing. This feature is especially valuable for students and researchers who need to quickly disseminate their findings.

• For educational institutions, the user-friendly interface of Jamovi translates to reduced training time and increased student engagement. Students can start performing analyses immediately, fostering a more interactive and hands-on learning experience. This can lead to improved understanding of statistical concepts and a greater appreciation for the power of data analysis.

• To maximize the benefits of Jamovi's click-based workflow, educators should design assignments that focus on data interpretation and critical thinking, rather than coding skills. Universities can also provide introductory workshops and tutorials to help students quickly master the software's interface and functionalities.

• While Jamovi excels at providing a user-friendly GUI, it also recognizes the need for advanced users to customize their analyses and access more sophisticated statistical techniques. To address this, Jamovi incorporates a 'More' button that allows users to access the underlying R code for each analysis [1, 2]. This dual-layered structure empowers users to transition from GUI-based analysis to code-based customization as their skills develop.

• The 'More' button serves as a gateway to the vast ecosystem of R packages, allowing users to extend Jamovi's capabilities far beyond its built-in functionalities. By installing and utilizing R packages, users can access cutting-edge statistical methods, perform specialized analyses, and tailor their workflows to specific research questions. This flexibility is particularly valuable in rapidly evolving fields where new statistical techniques are constantly emerging [46].

• Consider the 'jmvtools' package, which enables users with R programming experience to develop their own statistical modules and integrate them into Jamovi [2]. This allows researchers to create custom analyses tailored to their specific needs and share them with the broader community, further expanding Jamovi's capabilities and fostering innovation. The Korean Association for Corpus Linguistics highlights Jamovi's combination of an SPSS-like interface with R programming [111].

• For research institutions, Jamovi's 'More' button provides a pathway for researchers to leverage their existing R skills and contribute to the software's development. By creating and sharing custom modules, researchers can enhance Jamovi's functionality and tailor it to the specific needs of their research community. This collaborative approach fosters continuous innovation and ensures that Jamovi remains a relevant and powerful tool for statistical analysis.

• To fully capitalize on Jamovi's extensibility, universities should encourage researchers to develop custom modules and share them with the community. Educational programs should provide training on R programming and the jmvtools package, empowering students to create their own statistical analyses and contribute to Jamovi's evolving ecosystem.

• Jamovi’s architecture supports a progressive learning curve, allowing users to gradually increase their statistical and programming skills. Starting with the click-based interface, users gain familiarity with statistical concepts and data manipulation techniques [30, 35]. As they become more comfortable, they can explore the underlying R code and customize analyses using the 'More' button, transitioning to more advanced code-based workflows. This step-by-step approach minimizes the learning curve and promotes a deeper understanding of statistical methods.

• This progressive learning structure has major implications for education. Jamovi can be used to teach introductory statistics courses without requiring students to have prior programming experience. As students progress, the same software can be used to introduce them to R programming and more advanced statistical techniques. This avoids the need to learn multiple software packages and provides a seamless transition from basic to advanced analysis.

• The 'KIRD 교육프로그램' highlights Jamovi's use in statistical analysis mentoring programs for graduate students, focusing on understanding and utilizing the program, and implementing statistical analysis methods [30]. This underscores Jamovi's position as a tool for both basic statistical education and advanced research applications.

• For continuous learning, educators can use case studies and real-world datasets to challenge students and encourage them to explore Jamovi's advanced functionalities. By completing progressively more complex assignments, students can solidify their understanding of statistical concepts and develop their programming skills.

• Universities should create integrated curricula that incorporate Jamovi from introductory statistics to advanced research methods courses. This will provide students with a consistent learning experience and equip them with the skills they need to succeed in their chosen fields.

4. 환경 설정 및 데이터 준비 워크플로우

• 4-1. Jamovi 설치 및 데이터 파일 로딩

• This subsection addresses the crucial initial steps for users adopting Jamovi: installation and data loading. It details operating system compatibility, guides users through the installation process, and explores the types of data files that can be imported. Furthermore, it clarifies Jamovi's plugin structure, paving the way for subsequent sections that delve into specific statistical analyses.

• Successfully deploying Jamovi hinges on meeting the necessary system prerequisites. The software is primarily designed for Windows and macOS environments, with compatibility considerations crucial for seamless operation. Windows users should ensure they are running Windows 7 or later, while macOS users need macOS 10.15 or a more recent version [76, 80, 87]. These operating systems should ideally have sufficient RAM (at least 4GB recommended) and a reasonably modern processor (Intel Core i3 or equivalent) to guarantee smooth performance during data analysis [75, 76].

• The Jamovi installation process involves downloading the appropriate installer from the official Jamovi website. The downloaded file is then executed, guiding the user through a standard installation wizard. For Windows, this usually involves accepting the license agreement and choosing an installation directory. macOS installations typically involve dragging the Jamovi application icon to the Applications folder [1, 80]. Ensuring a stable internet connection during the installation is advisable to prevent interruptions during the download of essential components.

• Jamovi's adaptability extends to various devices, such as iPhones, Android devices, and Chrome OS hardware [84]. However, optimal performance is typically achieved on desktop environments due to the computational demands of statistical analyses. Though the reference documents don't explicitly detail mobile optimization, the software's reliance on processing power suggests a more streamlined experience on desktop platforms.

• Strategic implication: Understanding system requirements allows institutions to plan for optimal Jamovi deployment, ensuring that students and researchers have the hardware necessary for effective data analysis education and research. Implementation-focused recommendation: Clearly communicate the minimum and recommended system specifications on course syllabi and research resource pages to mitigate potential technical difficulties.

• Jamovi system requirements should be checked before beginning the installation. This will help to ensure an easy process without future complications.

• A core advantage of Jamovi is its capacity to handle a broad spectrum of data formats, thereby smoothing the transition for users migrating from other statistical software. Jamovi natively supports common formats like CSV (Comma Separated Values) and Excel (.xlsx, .xls), which are widely used for data storage and exchange [1, 174]. Beyond these, Jamovi can also import data directly from SPSS (.sav) files, reducing data conversion hassles and preserving existing datasets [1, 175].

• The data import process in Jamovi is GUI-driven, allowing users to load data via the 'File' menu [1, 35]. For CSV and Excel files, Jamovi automatically detects column headers and data types, though users retain the flexibility to manually adjust these if necessary [1]. When importing SPSS files, Jamovi maintains variable labels and value labels, preserving metadata integrity [1].

• While Jamovi excels in basic data import, challenges may arise with complex or corrupted files. Common issues involve character encoding problems (particularly with non-English datasets), missing values, or inconsistent data types. Addressing these often requires pre-processing the data in external tools like text editors or spreadsheet software before importing into Jamovi [171, 172, 174].

• Strategic implication: Jamovi's broad compatibility lowers the barrier to entry for new users, facilitating wider adoption in educational and research settings. Implementation-focused recommendation: Provide tutorials and troubleshooting guides addressing common data import issues, specifically covering character encoding, missing values, and data type inconsistencies. This proactive approach minimizes user frustration and maximizes learning efficiency.

• For a quicker process, users should verify that the order of the variables is the same as the external source to prevent misplacement when doing statistical work.

• Jamovi's modular architecture allows for extensive functionality expansion through its plugin library, contributed by experts in various fields [2, 175]. These plugins provide advanced statistical methods and specialized tools not included in the base software, catering to diverse research domains.

• Installing plugins in Jamovi is a straightforward process. Users can access the 'Jamovi Library' from the main menu, which displays available plugins with descriptions [238, 239]. Clicking on a plugin and selecting 'Install' downloads and integrates the plugin into the Jamovi environment. Some plugins may require specific R packages as dependencies, which Jamovi automatically handles during installation [2, 175].

• Despite the simplicity of plugin installation, potential challenges include network connectivity issues, plugin compatibility with the current Jamovi version, or conflicts between plugins [2, 30]. Maintaining an updated version of Jamovi and consulting plugin documentation can mitigate these issues. Plugin updates are also managed through the 'Jamovi Library'.

• Strategic implication: The plugin library is a key strategic asset for Jamovi, enabling it to remain competitive and adaptable to evolving statistical needs. Implementation-focused recommendation: Encourage users to actively explore the plugin library and provide clear documentation on plugin compatibility, installation troubleshooting, and conflict resolution. Creating a user forum or knowledge base for plugin-related questions can foster a collaborative and supportive community.

• Updating plug-ins regularly can solve the problem of incompatibility and ensure they run smoothly.

• 4-2. 변수 타입 설정 및 결측값 처리

• This subsection builds upon the previous discussion of data import and installation by addressing the critical steps of variable type definition and missing value management. By ensuring accurate data types and handling missing data effectively, this subsection establishes a strong foundation for subsequent statistical analyses, enhancing the reliability and validity of results obtained using Jamovi.

• Jamovi simplifies the process of converting numeric variables into categorical (factor) variables, a common requirement for many statistical analyses. This is particularly useful when dealing with continuous data that needs to be grouped into meaningful categories for comparison or analysis. For instance, age (a numeric variable) can be recoded into age groups (e.g., young, middle-aged, elderly) [1, 35].

• To perform this conversion, users can access the 'Data' tab, select the variable, and use the 'Transform' option. Jamovi provides a GUI-driven interface to define the recoding rules, allowing users to specify the numeric ranges and the corresponding category labels. This process avoids the need for complex R coding, making it accessible to users with limited programming experience [1, 2, 35].

• A practical example involves survey data where respondents provide numerical ratings on a scale (e.g., 1 to 7). To analyze this data categorically, the ratings can be recoded into 'Low, ' 'Medium, ' and 'High' satisfaction levels. This recoding facilitates the use of frequency tables, chi-square tests, and other analyses appropriate for categorical data. Proper transformation ensures that the dataset accurately represents the intended meaning of the variables, preventing misinterpretation of results [1, 5].

• Strategic implication: By facilitating easy conversion of numeric variables to categorical ones, Jamovi allows for flexible data manipulation to suit various analytical needs. This allows researchers to conduct advanced analyses that require categorical variables, like ANOVA, with minimal coding expertise. Implementation-focused recommendation: Provide detailed, step-by-step tutorials with screenshots to guide users through the variable transformation process. Highlight common use cases and potential pitfalls to ensure accurate and reliable data conversion.

• Recoding can lead to a clearer understanding of the data by focusing on a different perspective for analysis.

• Missing data is a common challenge in statistical analysis, and handling it appropriately is crucial for maintaining data integrity. While Jamovi offers basic missing value handling options like mean or median imputation, the MICE (Multiple Imputation by Chained Equations) extension provides a more sophisticated approach [349, 350, 351]. MICE addresses the limitations of single imputation methods by generating multiple plausible values for each missing data point, creating multiple complete datasets [349, 350].

• The MICE extension in Jamovi leverages the power of R's 'mice' package within a user-friendly GUI. After installing the extension, users can select the variables with missing data and specify the imputation model (e.g., predictive mean matching, logistic regression). MICE then iteratively imputes the missing values based on the relationships between variables, resulting in multiple complete datasets. These datasets can be analyzed separately, and the results can be pooled to obtain more robust and reliable estimates [349, 350].

• For example, in a healthcare dataset, patient records may have missing values for certain lab tests. Using MICE, these missing values can be imputed based on other patient characteristics (e.g., age, gender, medical history). Analyzing the imputed datasets provides a more accurate assessment of treatment outcomes compared to simply removing incomplete cases or using single imputation methods. A study from Anyang University, leveraged MICE in PM-2.5 measurements and was found to have more accurate and stable results [350, 351].

• Strategic implication: The MICE extension enhances Jamovi's analytical capabilities by providing a robust method for handling missing data, which reduces bias and increases the statistical power. Implementation-focused recommendation: Develop comprehensive documentation and training materials on using the MICE extension, including guidelines for choosing appropriate imputation models and interpreting results. Illustrate the benefits of MICE over simpler methods with real-world examples to encourage adoption.

• The multiple imputation technique improves overall model accuracy and the reliability of the dataset.

• Outliers can significantly distort statistical analyses, leading to biased results and inaccurate conclusions. Jamovi provides tools for outlier detection, and the Interquartile Range (IQR) method offers a standardized approach for identifying extreme values. The IQR is calculated as the difference between the 75th percentile (Q3) and the 25th percentile (Q1) of a dataset [419, 420, 421]. Outliers are then defined as values falling below Q1 - 1.5 * IQR or above Q3 + 1.5 * IQR.

• Jamovi's descriptive statistics module allows users to easily calculate the Q1, Q3, and IQR for any variable. Users can then manually filter out the outlier values. To automate this process, users may leverage R code to create more efficient detection rules. While Jamovi does not have a built-in function for automatic IQR outlier removal, this process is designed so users can take an active role in the cleaning process, determining what to remove and how to address it [1, 5].

• Consider a dataset of customer purchase amounts. The IQR method can identify unusually high or low purchase values that may be due to errors, fraudulent transactions, or unique customer behaviors. Removing or adjusting these outliers ensures that subsequent analyses (e.g., calculating average purchase amount) are not skewed by extreme values. For instance, if a digital twin outlier detection algorithm detected the number to be an anomoly, the IQR is useful in further detecting the point [416].

• Strategic implication: By standardizing outlier detection using the IQR method, Jamovi can improve the quality and reliability of statistical analyses, leading to more accurate insights. Implementation-focused recommendation: Create templates and tutorials that provide clear instructions on calculating the IQR and filtering outliers in Jamovi. Emphasize the importance of considering the context of the data and the potential impact of outliers on the results, rather than blindly removing extreme values. In turn, provide ways that outliers can be studied separately, rather than only being deleted.

• The IQR is a useful method for determining data reliability and preventing skewness during data processing.

5. 기초 통계량 및 빈도 분석 실습

• 5-1. 기술 통계량 계산(평균, 분산 등)

• This subsection introduces the fundamental steps for calculating descriptive statistics using Jamovi, building upon the previous section's data preparation workflow. It aims to equip users with the ability to summarize key characteristics of their data through measures like mean, variance, and standard deviation, setting the stage for more advanced statistical analyses.

• Performing descriptive statistics in Jamovi begins with activating the 'Descriptives' module located within the 'Analyses' tab. This module provides a user-friendly interface for calculating various descriptive measures, catering especially to users familiar with GUI-based statistical software. Novices can start by loading their dataset, typically in .csv or .xlsx format, and then navigating to the 'Descriptives' option. The primary challenge lies in understanding which statistics are appropriate for the given data type and research question.

• Within the Descriptives module, users can drag and drop variables of interest into the 'Variables' box. Subsequently, a range of descriptive statistics can be selected from the 'Statistics' dropdown menu. Key statistics include the mean, median, standard deviation, variance, minimum, and maximum values. Jamovi's architecture, built on the R programming language, ensures that these calculations are performed accurately, mirroring the output from traditional R-based analyses, yet simplifying the process through its GUI (Jamovi, 2025).

• For instance, consider a dataset containing customer satisfaction scores for a tourism product. To understand the central tendency and variability of these scores, a user would select the 'Mean' and 'Standard deviation' options. According to 황성동 (2025), this is a standard procedure in introductory statistical analysis, often followed by visual representations to aid interpretation. The selected statistics are then displayed in a table format, providing a concise summary of the data's key attributes.

• The strategic implication here is that Jamovi significantly lowers the barrier to entry for performing initial data exploration. Researchers and educators can quickly gain insights into their data without needing to write code or navigate complex statistical packages. This is particularly beneficial in educational settings where students are learning statistical concepts for the first time.

• To further enhance this initial data exploration, instructors should guide students through the process of selecting the most relevant statistics based on the data's distribution and scale of measurement. This includes understanding when to use the median instead of the mean for skewed data, or the interquartile range in addition to the standard deviation for robust measures of variability.

• While descriptive statistics provide a numerical summary of data, visual representations offer complementary insights into its distribution. Jamovi enables users to generate histograms, box plots, and scatter plots directly from the Descriptives module, enhancing the interpretability of the analysis. However, challenges can arise when exporting these visualizations for inclusion in reports or presentations, as the default settings may not meet professional standards.

• Jamovi facilitates the creation of histograms and scatterplots through a simple interface. For histograms, users can specify the number of bins or let Jamovi automatically determine an appropriate value. For scatterplots, users can drag and drop two variables onto the plot area to visualize their relationship. Moreover, the software automatically generates the APA style to present results (Jamovi, 2025). The underlying R code allows for customization of these plots, such as adding titles, axis labels, and adjusting colors.

• Consider a case study analyzing website traffic data. A user might create a histogram of session durations to identify common engagement patterns or a scatter plot of page views versus bounce rate to explore correlations. The key here is to export these visualizations in a high-resolution format suitable for publication, as well as presenting summary tables of key statistics.

• The strategic implication is that educators should emphasize the importance of effective data visualization in communicating statistical findings. Students should be trained on how to choose appropriate chart types, format them for clarity, and interpret their implications in the context of the research question. The ability to seamlessly integrate these visualizations into reports or presentations significantly enhances the impact of the analysis.

• Recommendations include providing pre-designed templates for exporting visualizations in common publication formats and encouraging students to experiment with different chart types to find the most effective way to communicate their findings. Tutorials should also focus on adjusting aesthetic elements such as color palettes and font sizes to ensure professional-looking outputs.

• 5-2. 빈도표 및 교차표 생성

• This subsection builds upon the foundation laid by descriptive statistics, transitioning into the analysis of qualitative data through frequency tables and cross-tabulations. It extends the user's ability to understand data distributions and relationships, crucial for comprehensive data exploration.

• 빈도분석은 질적 데이터의 분포를 파악하는 데 필수적인 단계이다. Jamovi의 Frequencies 모듈은 클릭 몇 번으로 간단하게 빈도표와 백분율표를 생성할 수 있는 사용자 친화적인 인터페이스를 제공한다. 이 모듈을 통해 사용자는 데이터 내 각 범주형 변수의 출현 빈도와 상대적 비율을 신속하게 확인할 수 있으며, 이는 데이터의 전반적인 특성을 이해하는 데 중요한 역할을 한다.

• Frequencies 모듈을 활성화하려면 'Analyses' 탭에서 'Frequencies' 옵션을 선택한다. 그런 다음 분석하고자 하는 범주형 변수를 'Variables' 상자로 드래그 앤 드롭하면 된다. Jamovi는 자동으로 각 범주의 빈도, 백분율, 유효 백분율, 누적 백분율을 계산하여 표 형태로 제시한다 (황성동, 2025). 사용자는 필요에 따라 백분율 표시 방식이나 누적 백분율 포함 여부 등을 추가적으로 설정할 수 있다.

• 예를 들어, 한 여행 상품에 대한 고객 만족도 조사에서 '만족', '보통', '불만족'과 같은 범주형 변수의 빈도분석을 수행한다고 가정해 보자. Frequencies 모듈을 사용하면 각 응답 범주에 해당하는 응답자 수와 백분율을 쉽게 파악할 수 있으며, 이를 통해 어떤 범주가 가장 일반적인 응답인지, 그리고 전반적인 만족도 분포는 어떠한지 등을 신속하게 파악할 수 있다.

• 이러한 빈도분석 결과는 마케팅 전략 수립에 중요한 시사점을 제공한다. 예를 들어, 불만족 응답이 높게 나타난다면 해당 상품의 개선이 필요하다는 신호로 해석할 수 있으며, 만족 응답이 높다면 해당 상품의 강점을 더욱 부각하는 마케팅 전략을 고려할 수 있다.

• 실습 시, 사용자는 다양한 범주형 변수를 사용하여 빈도분석을 수행하고, 결과를 해석하는 연습을 해야 한다. 특히, 백분율과 유효 백분율의 차이점을 이해하고, 결측값 처리 방식에 따라 결과가 어떻게 달라지는지 등을 고려하는 것이 중요하다.

• 빈도분석이 단일 변수의 분포를 파악하는 데 유용하다면, 교차표 분석은 두 개 이상의 범주형 변수 간의 관계를 탐색하는 데 효과적인 도구이다. Jamovi의 Frequencies 모듈은 교차표 생성 기능도 제공하며, 이를 통해 사용자는 두 변수 간의 연관성을 시각적으로 확인할 수 있다.

• 교차표를 생성하려면 Frequencies 모듈에서 두 개의 범주형 변수를 각각 'Rows'와 'Columns' 상자로 드래그 앤 드롭한다. Jamovi는 자동으로 각 셀에 해당하는 빈도와 백분율을 계산하여 표 형태로 제시하며, 사용자는 필요에 따라 행 백분율, 열 백분율, 전체 백분율 등을 선택적으로 표시할 수 있다. 또한, 카이제곱 검정과 같은 통계적 검정을 통해 두 변수 간의 독립성 여부를 판단할 수도 있다 (Jamovi, 2025).

• 예를 들어, 한 여행 상품에 대한 고객 만족도와 연령대 간의 관계를 분석한다고 가정해 보자. 교차표 분석을 통해 각 연령대별 만족도 분포를 비교하고, 특정 연령대에서 불만족 응답이 특히 높게 나타나는지 등을 확인할 수 있다.

• 이러한 교차표 분석 결과는 타겟 마케팅 전략 수립에 유용한 정보를 제공한다. 예를 들어, 특정 연령대에서 불만족 응답이 높게 나타난다면 해당 연령대에 특화된 상품 개선 또는 마케팅 캠페인을 고려할 수 있다.

• 실습 시, 사용자는 다양한 범주형 변수 쌍을 사용하여 교차표 분석을 수행하고, 결과를 해석하는 연습을 해야 한다. 특히, 카이제곱 검정 결과를 통해 두 변수 간의 관계가 통계적으로 유의미한지 판단하고, 연관성의 방향과 강도를 파악하는 것이 중요하다.

• 빈도표와 교차표는 데이터의 분포와 관계를 수치적으로 보여주지만, 파이 차트와 막대 그래프는 이를 시각적으로 표현하여 더욱 직관적인 이해를 돕는다. Jamovi는 Frequencies 모듈에서 생성된 빈도표를 기반으로 파이 차트와 막대 그래프를 쉽게 생성할 수 있는 기능을 제공한다.

• 파이 차트는 각 범주의 비율을 원형으로 표현하며, 막대 그래프는 각 범주의 빈도를 막대의 높이로 표현한다. Jamovi에서 차트를 생성하려면 Frequencies 모듈에서 원하는 변수를 선택한 후 'Plots' 옵션을 선택하고, 'Pie chart' 또는 'Bar plot'을 선택하면 된다 (김승재, 2025). 사용자는 필요에 따라 차트의 색상, 제목, 축 레이블 등을 변경하여 시각적인 효과를 높일 수 있다.

• 예를 들어, 한 설문조사에서 응답자의 학력 수준을 파악하기 위해 빈도분석을 수행한 후, 결과를 파이 차트로 표현하면 각 학력 수준별 응답자의 비율을 한눈에 파악할 수 있다. 또는, 특정 제품에 대한 선호도를 막대 그래프로 표현하면 각 제품별 선호도 수준을 쉽게 비교할 수 있다.

• 시각화된 빈도 분포는 의사결정 과정에 중요한 정보를 제공한다. 예를 들어, 특정 제품에 대한 선호도가 낮게 나타난다면 해당 제품의 디자인, 기능, 가격 등을 개선해야 할 필요성을 시사할 수 있다.

• 실습 시, 사용자는 다양한 데이터셋을 사용하여 파이 차트와 막대 그래프를 생성하고, 각 차트의 장단점을 비교하는 연습을 해야 한다. 특히, 어떤 유형의 데이터에 어떤 차트가 더 적합한지, 그리고 차트를 통해 어떤 정보를 효과적으로 전달할 수 있는지 등을 고려하는 것이 중요하다.

6. 가설 검정: t-검정 및 카이제곱 검정 실습

• 6-1. 독립표본 t-검정 수행

• This subsection expands on the practical application of the independent samples t-test within Jamovi, focusing on essential pre-analysis checks and result interpretations to ensure the robustness and validity of statistical findings. It builds upon the introduction to t-tests by incorporating Levene's test for equality of variances, Cohen's d for effect size, and Shapiro-Wilk for normality.

• Before conducting an independent samples t-test, verifying the assumption of equal variances between groups is crucial for ensuring the reliability of results. Levene's test serves this purpose by assessing whether the variances of the two groups are significantly different. In Jamovi, Levene's test can be easily implemented within the T-Tests module.

• To execute Levene's test in Jamovi, first, access the 'T-Tests' menu and select 'Independent Samples T-Test'. Assign the grouping variable (the independent variable distinguishing the two groups) and the dependent variable (the continuous variable being compared). Under the 'Assumption Checks' section, tick the box labeled 'Equality of variances test'. Jamovi will then automatically compute Levene's test statistic and its corresponding p-value.

• The output from Jamovi will provide the Levene's test statistic (F) and the p-value. If the p-value is less than the significance level (typically 0.05), the null hypothesis of equal variances is rejected, indicating that the variances are significantly different. Conversely, if the p-value is greater than 0.05, the assumption of equal variances is tenable. For instance, if we're comparing test scores between two teaching methods and Levene's test yields a p-value of 0.03, this suggests unequal variances, requiring adjustments to the t-test interpretation or alternative non-parametric tests.

• If Levene's test indicates unequal variances, it is crucial to report this violation when presenting the t-test results. In Jamovi, selecting the 'Welch's' option automatically adjusts the t-test to account for unequal variances, providing a more accurate assessment of the group difference. Furthermore, understanding this assumption and how to address it in Jamovi equips users to make more informed and reliable statistical inferences.

• For educators and researchers using Jamovi, familiarity with Levene's test is essential for conducting valid t-tests. Including this step in the analysis workflow ensures that assumptions are checked and appropriate adjustments are made, ultimately bolstering the credibility of research findings. For example, '누구나 할 수 있는 jamovi 통계분석 (빈도분석에서 구조방정식까지 | 2 판)' (ref_idx 1) likely covers these assumption checks within its t-test discussions.

• While the p-value from a t-test indicates the statistical significance of a group difference, it doesn't convey the magnitude of that difference. Cohen's d is a standardized effect size measure that quantifies the practical significance of the difference between two group means, independent of sample size.

• To calculate Cohen's d in Jamovi, conduct the independent samples t-test as described earlier. Under the 'Effect Size' section, select 'Cohen's d'. Jamovi will automatically calculate Cohen's d, providing a measure of the standardized mean difference between the two groups.

• Cohen's d is interpreted using established guidelines. Cohen (1988) suggests that d values around 0.2 represent a small effect, 0.5 a medium effect, and 0.8 a large effect. For instance, if comparing customer satisfaction scores after two marketing campaigns, a Cohen's d of 0.6 indicates a moderate practically significant difference, even if the p-value is marginal.

• Presenting Cohen's d alongside the t-test results provides a more complete picture of the group difference. Reporting both the statistical significance (p-value) and practical significance (Cohen's d) allows readers to assess the real-world implications of the findings. It would also help decide what A/B testing design should be used in real world setting.

• By incorporating Cohen's d into the Jamovi analysis workflow, educators and researchers can move beyond simply identifying statistically significant differences and begin to understand the substantial impact of those differences. Relevant sections in 'jamovi로 통계 배우기 — Learning statistics with jamovi' (ref_idx 5) likely discuss effect sizes, providing additional context for interpretation.

• The independent samples t-test assumes that the data within each group are approximately normally distributed. While the t-test is relatively robust to violations of this assumption, particularly with larger sample sizes, it's still important to assess normality to ensure the validity of the results. The Shapiro-Wilk test is a commonly used method to formally test for normality.

• To perform the Shapiro-Wilk test in Jamovi, while not directly available within the T-Tests module, it can be implemented in the 'Exploration' -> 'Descriptives' menu. Drag the variable of interest into the variables box, then under 'Plots', select 'Q-Q plot'. Also, under 'Statistics' select 'Shapiro-Wilk'.

• The Shapiro-Wilk test outputs a W statistic and a p-value. The null hypothesis is that the data are normally distributed; therefore, a p-value greater than the chosen alpha level (e.g., 0.05) suggests that the data do not significantly deviate from normality. Conversely, a p-value less than 0.05 indicates a violation of the normality assumption. For example, when testing the normality of student test scores, a Shapiro-Wilk p-value of 0.10 suggests that the normality assumption is reasonable.

• If the Shapiro-Wilk test indicates a violation of normality, several options are available. Consider using non-parametric alternatives to the t-test, such as the Mann-Whitney U test, which does not assume normality. Alternatively, explore data transformations (e.g., logarithmic or square root) to potentially improve normality before applying the t-test.

• Including the Shapiro-Wilk test in the analysis process provides a more thorough assessment of the t-test assumptions, promoting more responsible and informed data analysis. By checking and addressing normality, researchers enhance the reliability and interpretability of their results. Online resources and textbooks on statistical analysis will provide detailed explanations and examples of Shapiro-Wilk test implementation and interpretation (ref_idx 260, 261).

• 6-2. 카이제곱 검정을 통한 범주형 변수 관계 분석

• This subsection builds upon the previous discussion of the independent samples t-test by transitioning to the Chi-square test, a crucial method for analyzing relationships between categorical variables. It expands on the basic application of the Chi-square test by incorporating Cramer's V for effect size, guidelines for expected frequencies, and Yates' correction for small sample sizes, ensuring a more nuanced and accurate analysis.

• While the Chi-square test determines if a statistically significant association exists between categorical variables, it doesn't quantify the strength of that association. Cramer's V is a post-hoc effect size measure that addresses this by indicating the magnitude of the relationship, regardless of sample size, thereby offering a clearer interpretation of practical significance.

• To calculate Cramer's V in Jamovi after performing the Chi-square test, navigate to the 'Contingency Tables' module. Select the 'Independent Samples' option, assign your row and column variables, and then, under the 'Statistics' section, tick the box labeled 'Cramer's V'. Jamovi will automatically compute Cramer's V, which ranges from 0 to 1, where values closer to 1 indicate a stronger association.

• Interpreting Cramer's V depends on the size of the table. For a 2x2 table, guidelines suggest values around 0.1 indicate a small effect, 0.3 a medium effect, and 0.5 a large effect. However, in larger tables, these thresholds may need adjustment. For instance, analyzing the relationship between customer segment (e.g., age groups) and product preference, a Cramer's V of 0.4 suggests a moderately strong association, influencing targeted marketing strategies.

• Reporting Cramer's V alongside the Chi-square results provides a comprehensive understanding of the relationship between categorical variables. Presenting both the statistical significance (p-value) and practical significance (Cramer's V) allows readers to evaluate the real-world implications of the findings, informing decisions on targeted interventions or marketing strategies.

• For educators and researchers, including Cramer's V in the Jamovi analysis workflow enhances the interpretability of Chi-square test results, moving beyond mere significance to understanding the strength of the relationship. References like 'jamovi로 통계 배우기 — Learning statistics with jamovi' (ref_idx 5) likely cover effect sizes for categorical data, providing additional context for interpretation.

• The Chi-square test relies on comparing observed frequencies with expected frequencies, and the validity of the test is contingent upon meeting certain criteria for these expected frequencies. A common guideline stipulates that no more than 20% of the cells should have expected frequencies less than 5, and all cells should have expected frequencies of at least 1.

• To assess whether these criteria are met in Jamovi, after conducting the Chi-square test, examine the output table. Jamovi displays the expected frequencies for each cell. If a substantial number of cells (more than 20%) have expected frequencies below 5, the Chi-square test may not be appropriate, and alternative tests or data pooling should be considered.

• Violations of the expected frequency criteria can lead to inaccurate p-values and unreliable conclusions. For example, if analyzing the relationship between product type and purchase channel (online vs. in-store) and many cells have expected frequencies below 5, the Chi-square result should be interpreted with caution. A potential remedy is to combine categories with low frequencies to increase the expected cell counts.

• When presenting Chi-square test results, it's crucial to report whether the expected frequency criteria were met. If violations occurred, describe the steps taken to address them or acknowledge the limitations of the analysis. Ensuring adherence to these guidelines strengthens the validity and interpretability of the research findings.

• By considering expected frequency guidelines in the Jamovi analysis process, educators and researchers promote more responsible and informed use of the Chi-square test, bolstering the credibility of their statistical inferences. Introductory statistics textbooks and online resources provide detailed explanations of these guidelines (ref_idx 1, 5).

• The standard Chi-square test can be inaccurate when dealing with small sample sizes, particularly in 2x2 contingency tables. Yates' correction for continuity is an adjustment applied to the Chi-square statistic to mitigate this issue, providing a more conservative and accurate assessment of the association between categorical variables.

• To perform the Chi-square test with Yates' correction in Jamovi, after accessing the 'Contingency Tables' module and selecting 'Independent Samples', assign your row and column variables. Under the 'Statistics' section, tick the box labeled 'Yates' continuity correction'. Jamovi will then compute the corrected Chi-square statistic and its corresponding p-value.

• Yates' correction involves subtracting 0.5 from the absolute difference between observed and expected frequencies in the Chi-square formula. This adjustment reduces the magnitude of the Chi-square statistic, leading to a higher p-value. For example, analyzing the relationship between treatment and outcome with a small sample, applying Yates' correction might change the p-value from 0.04 (significant) to 0.06 (non-significant), altering the conclusion.

• When reporting Chi-square test results, clearly indicate whether Yates' correction was applied and justify its use based on the sample size. In cases where the correction significantly alters the conclusion, discuss the implications of this adjustment and consider potential limitations. Also, in cases where the sample size is big enough, it would be better not to use Yates' correction because it's too conservative.

• By incorporating Yates' correction into the Jamovi analysis workflow, educators and researchers enhance the accuracy and reliability of Chi-square tests when working with small samples, promoting more responsible and informed data analysis. Online resources and statistical textbooks often discuss Yates' correction in the context of Chi-square tests (ref_idx 1).

7. 관련성 분석 및 회귀 모델 구축

• 7-1. 피어슨/스피어맨 상관계수 계산

• This subsection delves into the practical aspects of calculating Pearson and Spearman correlation coefficients within Jamovi, focusing on their interpretation, the necessary sample sizes for reliable analysis, and methods for assessing the linearity assumption crucial for Pearson's r. It serves as a bridge between basic statistical knowledge and advanced regression modeling, providing actionable insights for researchers and analysts using Jamovi.

• Pearson's correlation coefficient (r) in Jamovi is a measure of the strength and direction of a linear relationship between two continuous variables. The interpretation of 'r' values varies, but generally, benchmarks help contextualize the effect size. A common guideline suggests that r values around ±0.1 indicate a small effect, ±0.3 a medium effect, and ±0.5 or greater a large effect [26]. However, these thresholds are not absolute and should be considered alongside the specific context of the research.

• The statistical significance of Pearson's r, indicated by the p-value, determines whether the observed correlation is likely due to a true relationship or random chance. Jamovi directly displays this p-value alongside the correlation coefficient. A p-value less than the significance level (typically 0.05) suggests the correlation is statistically significant. This means that there is a less than 5% chance that the observed correlation occurred by chance, assuming there is no true correlation in the population [1, 5].

• To effectively interpret Pearson’s r in Jamovi, consider both the magnitude of 'r' (effect size) and the p-value (statistical significance). For example, a correlation of r = 0.35 with p < 0.01 indicates a statistically significant medium positive correlation. Conversely, a correlation of r = 0.15 with p > 0.05 suggests a small, non-significant correlation. Researchers should avoid overemphasizing the statistical significance of small correlations in large samples, as these may not have practical importance [25]. Jamovi’s clear output of both 'r' and 'p' facilitates a nuanced understanding of variable relationships.

• Strategic implications for interpreting Pearson's r lie in understanding the predictive power and practical relevance of the findings. A strong, statistically significant correlation can inform predictive models and interventions, but a weak or non-significant correlation may suggest the need for alternative analyses or the inclusion of additional variables. For instance, in tourism research, a correlation between satisfaction scores and revisit intention might inform marketing strategies, but only if the effect is both strong and statistically significant [6].

• To enhance the interpretation of Pearson’s r in Jamovi, include confidence intervals for the correlation coefficient. These provide a range of plausible values for the true correlation in the population. Also, always report the sample size (n) to provide context for the statistical power of the analysis. Visual aids, such as scatterplots with regression lines, can help illustrate the nature of the relationship between variables, ensuring a comprehensive understanding for both researchers and stakeholders [1, 5].

• Determining the minimum sample size for correlation analysis is critical to ensure the reliability and validity of the findings. Insufficient sample sizes can lead to unstable correlation estimates and increase the risk of Type II errors (failing to detect a true correlation). Conversely, excessively large samples can detect trivially small correlations that lack practical significance [127].

• Several rules of thumb exist for determining minimum sample sizes. A commonly cited guideline suggests having at least 30 cases for correlation analysis [22]. However, more sophisticated approaches consider the desired statistical power, the expected effect size, and the level of significance. Power analysis, a statistical method for estimating the required sample size, can be conducted using dedicated software or online calculators. These tools require specifying the desired power (typically 0.80), the significance level (typically 0.05), and an estimate of the expected correlation coefficient [127].

• For instance, if researchers anticipate a medium effect size (r = 0.3) and desire a power of 0.80 with a significance level of 0.05, power analysis might indicate a minimum sample size of approximately 85 cases. This ensures a high probability of detecting a true correlation if it exists. Studies employing structural equation modeling (SEM) or factor analysis often require even larger sample sizes due to the complexity of these techniques [128, 129]. Tanaka (1984) and Hadow (1985) suggest a sample size of at least 400 or 500 is needed for SEM.

• Strategic implications for sample size determination involve balancing the costs of data collection with the need for reliable results. Pilot studies can help estimate the expected effect size, informing a more accurate power analysis. Researchers should also consider the heterogeneity of the population being studied, as more diverse populations typically require larger samples. Failing to meet minimum sample size requirements can undermine the credibility of the research and limit its practical applications [134].

• To address sample size considerations in Jamovi, conduct a thorough power analysis before data collection to determine the appropriate sample size based on expected effect sizes and desired statistical power. When reporting correlation results, always include the sample size (n) and justify the adequacy of the sample based on established guidelines or power analysis. Sensitivity analyses can also assess the robustness of the findings to variations in sample size or effect size assumptions. Transparent reporting enhances the rigor and credibility of the research [1, 5].

• Pearson's correlation coefficient (r) assumes a linear relationship between two variables. Violating this assumption can lead to an underestimation of the true strength of the relationship, especially if the relationship is curvilinear or monotonic but non-linear. Therefore, assessing linearity is crucial for valid interpretations of Pearson's r [1].

• Jamovi does not have a built-in function for directly testing linearity, but the assumption can be assessed through visual inspection of scatterplots. Create a scatterplot of the two variables and examine the pattern of points. If the points appear to cluster around a straight line, the linearity assumption is likely met. Deviations from linearity, such as curves or clusters, suggest that Pearson's r may not be appropriate. Additionally, examine residual plots from linear regression models, where non-random patterns suggest violations of linearity [1, 5].

• In cases of non-linearity, consider transforming one or both variables to achieve linearity. Common transformations include logarithmic, exponential, or square root transformations. Alternatively, consider using non-parametric correlation coefficients, such as Spearman's rho, which do not assume linearity and can capture monotonic relationships [1, 5]. Spearman's rho assesses the degree to which variables tend to change together, but does not require a linear form.

• Strategic implications for linearity assessment involve selecting appropriate statistical methods and interpretations based on the nature of the relationship between variables. Ignoring non-linearity can lead to misleading conclusions and ineffective decision-making. Always examine scatterplots and residual plots to visually assess linearity, and consider transformations or non-parametric methods if necessary. Document the assessment process and justify the choice of correlation coefficient in the research report [24].

• To assess linearity in Jamovi, visually inspect scatterplots of the variables to assess whether the data is approximately linear. If non-linearity is suspected, transform the data using Jamovi's built-in transformation functions or consider Spearman's rho. Compare the results of Pearson's r and Spearman's rho; if they differ substantially, it suggests that non-linearity is affecting Pearson's r. Ensure transparent reporting of the linearity assessment process and the rationale for selecting the appropriate correlation coefficient [1, 5].

• 7-2. 단순 및 다중 회귀 모델 구축

• Building on the foundation laid by correlation analysis, this subsection provides a step-by-step guide to constructing and interpreting regression models in Jamovi. It focuses on practical implementation, diagnostic procedures, and understanding the statistical outputs, thus equipping readers with the skills to build predictive models and assess relationships between variables.

• In Jamovi, the p-values for regression coefficients are readily accessible in the results table generated after running the linear regression analysis. Specifically, when performing a linear regression, Jamovi displays a table that includes the estimated coefficients, their standard errors, t-values, and corresponding p-values for each predictor variable in the model [1, 5]. This p-value indicates the statistical significance of each predictor's effect on the dependent variable.

• The p-value associated with each regression coefficient tests the null hypothesis that the coefficient is equal to zero, meaning the predictor has no effect on the dependent variable. A p-value less than the chosen significance level (commonly 0.05) suggests that the predictor has a statistically significant effect. Conversely, a p-value greater than 0.05 indicates that the predictor’s effect is not statistically significant [1].

• For example, consider a regression model predicting customer satisfaction based on advertising spending and customer service ratings. If the p-value for the advertising spending coefficient is 0.02, it suggests that advertising spending has a statistically significant positive or negative impact on customer satisfaction. Conversely, if the p-value for the customer service rating is 0.10, its effect is not statistically significant. This informs decision-making by prioritizing resources towards advertising if the p-value is significant or improving customer service if it is not [5].

• Strategic implications for regression coefficient p-value interpretation involve differentiating between statistical significance and practical importance. A statistically significant predictor with a small coefficient may not have substantial practical effects, particularly in large samples. Reporting both the coefficient magnitude and its p-value offers a nuanced view. Also, consider the direction of the coefficient (positive or negative), indicating the nature of the relationship between the predictor and the dependent variable [1, 5].

• To effectively use Jamovi, identify the coefficients table in the output to find the p-value for each predictor. Report both the coefficient estimate and its p-value, along with confidence intervals, to provide a comprehensive picture of the predictor’s effect. Interpret the p-value considering both statistical significance and practical implications, and consider potential confounding factors or interactions that may influence the predictor’s effect [1].

• Residual plots are diagnostic tools to assess the assumptions of linear regression, including linearity, homoscedasticity (constant variance of errors), and independence of errors. Creating residual plots in Jamovi involves running a linear regression model and then selecting the appropriate options in the results panel to generate the plots [1].

• To create a residual plot in Jamovi, first, perform a linear regression by selecting the ‘Regression’ → ‘Linear Regression’ menu. Specify the dependent and independent variables. In the ‘Plots’ section, check the boxes for ‘Residuals plot’ and ‘QQ plot of residuals.’ Jamovi will generate these plots as part of the regression output [5].

• The residual plot displays residuals (the differences between observed and predicted values) on the y-axis and predicted values on the x-axis. A random scatter of points around zero suggests that the assumptions of linearity and homoscedasticity are met. Patterns, such as a funnel shape (heteroscedasticity) or a curve (non-linearity), indicate violations of these assumptions. The QQ plot of residuals compares the distribution of residuals to a normal distribution. Deviations from the straight line indicate non-normality [1].

• Strategic implications for residual plot analysis include identifying potential model misspecifications. If the residual plots reveal patterns indicating violations of regression assumptions, consider transforming variables, adding interaction terms, or using different modeling techniques. For instance, logarithmic transformation can address non-linearity, and weighted least squares regression can handle heteroscedasticity [1].

• To use Jamovi effectively for residual diagnostics, generate the residual plots as part of the regression output. Visually inspect the plots for patterns indicating violations of regression assumptions. If violations are detected, take corrective actions, such as variable transformations or alternative modeling techniques. Document the diagnostic process and any corrective actions taken in the research report to ensure transparency and rigor [1, 5].

• Multicollinearity, a condition where predictor variables in a regression model are highly correlated, can inflate the standard errors of regression coefficients, making it difficult to determine the individual effects of predictors. Variance Inflation Factor (VIF) is a common measure for detecting multicollinearity, quantifying how much the variance of an estimated regression coefficient increases due to multicollinearity [399].

• While Jamovi does not directly compute VIF, it can be calculated by installing the 'jstats' module. After installation, run your regression model, then, using the jstats module, calculate the VIF for each predictor. A VIF value greater than 5 or 10 (depending on the source) indicates substantial multicollinearity [24, 392].

• As an example, in a model predicting sales revenue based on advertising spending, pricing, and promotional discounts, high correlations between advertising spending and promotional discounts might lead to high VIF values for both predictors. If the VIF values exceed the threshold, it suggests that these predictors are redundant, and their individual effects on sales revenue are difficult to disentangle [399].

• Strategic implications for multicollinearity assessment involve addressing the issue to improve model interpretability and stability. Options include removing one of the correlated predictors, combining them into a single variable, or using regularization techniques that penalize large coefficients. The choice depends on the context and research objectives [24, 392].

• To address multicollinearity in Jamovi, install the 'jstats' module. Run the regression, calculate VIFs using the module functions, and identify predictors with high VIF values. Take appropriate corrective actions, such as removing or combining correlated predictors. Re-run the regression and reassess VIF values to ensure that multicollinearity has been adequately addressed. Document the assessment and corrective actions in the research report [399].

8. 측정 도구의 신뢰성과 타당성 확보

• 8-1. 크론백 알파 계수 산출

• This subsection delves into the practical aspects of ensuring measurement tool reliability within Jamovi, specifically focusing on calculating Cronbach's alpha. It addresses the practical UI navigation for conducting the analysis and strategies for improving reliability by examining individual items, thereby bridging the gap between theoretical understanding and actionable steps in data analysis.

• Calculating Cronbach's alpha in Jamovi is crucial for assessing the internal consistency of measurement scales. However, many users, especially those new to Jamovi, find it difficult to locate the correct menu path. The challenge lies in efficiently accessing the Reliability Analysis module within Jamovi's GUI.

• To compute Cronbach’s α, users should navigate to the 'Analyses' tab, then select 'Reliability' followed by 'Reliability Analysis'. Within this module, drag and drop the items intended to measure a single construct into the 'Variables' box. Ensure that the 'Cronbach’s α' option is checked under the 'Statistics' dropdown. Jamovi then automatically calculates and displays the Cronbach’s α coefficient (refer to ref_idx 15).

• For instance, a study examining the reliability of a 10-item smartphone overdependence scale (ref_idx 22) successfully used this method in Jamovi 2.2.5, yielding a Cronbach’s α of .896. Similarly, a study on service quality dimensions used Jamovi to assess the reliability of scales like 'assurance, ' obtaining a Cronbach’s α of .935 (ref_idx 73). These cases demonstrate the module's utility in varied research contexts.

• Strategically, clear documentation and training materials must emphasize this navigation path to reduce user friction. Providing tooltips or brief tutorials directly within the Jamovi interface can further enhance usability. This reduces the learning curve and allows users to focus on interpreting results rather than struggling with the software's interface.

• We recommend integrating a step-by-step guide with screenshots directly into the Jamovi help documentation. Additionally, creating short video tutorials demonstrating the process from data import to Cronbach’s α calculation would significantly improve user adoption and confidence.

• A critical challenge in reliability analysis is addressing situations where Cronbach's alpha is below the acceptable threshold (typically .70). This often requires identifying and removing problematic items that reduce the scale's internal consistency. However, deciding which items to remove and executing this process efficiently in Jamovi requires a clear methodology.

• Jamovi provides item-level statistics within the Reliability Analysis module to guide item removal. After performing the initial reliability analysis, examine the 'item statistics' table. This table displays the 'alpha if item is removed', indicating how Cronbach's alpha would change if a specific item were deleted. Focus on items that, when removed, substantially increase the overall alpha.

• For example, in a study assessing humanistic competence, researchers analyzed item-total correlations and Cronbach's alpha to identify problematic items. Items with low item-total correlations or those that, when removed, significantly increased alpha were considered for deletion (ref_idx 62). Similarly, studies on service quality assessment have used Cronbach’s alpha to iteratively refine measurement scales (ref_idx 63, 65).

• From a strategic perspective, item removal should be guided by both statistical and theoretical considerations. While an item's removal may increase alpha, it is crucial to ensure that the remaining items still adequately represent the construct being measured. Removing too many items can compromise the scale's content validity.

• We recommend a phased approach to item removal: (1) Identify items that substantially increase alpha upon removal. (2) Review the content of these items to assess their relevance to the construct. (3) If theoretically justifiable, remove the item and rerun the reliability analysis. (4) Repeat this process iteratively until an acceptable alpha level is achieved, ensuring content validity is maintained. Adding warnings or best-practice tips to the Jamovi UI could further assist researchers.

• 8-2. 탐색적·확인적 요인분석 수행

• This subsection transitions from ensuring the reliability of measurement tools to establishing their validity through Exploratory Factor Analysis (EFA). It focuses on guiding users through Jamovi's EFA functionalities, emphasizing how to effectively use the KMO test, Bartlett's test, and various rotation methods to ensure the data are suitable for factor analysis and that the resulting factors are interpretable and meaningful.

• Before conducting EFA, it's critical to determine if the dataset is suitable for factor analysis. Many users, especially those new to Jamovi, struggle with interpreting the Kaiser-Meyer-Olkin (KMO) measure and Bartlett's test of sphericity and locating these tests within Jamovi's interface. This can lead to inappropriate factor analysis and misleading results, undermining the validity of the research.

• To assess data suitability, users should first navigate to the 'Factor' analysis tab, then select 'Exploratory Factor Analysis'. Within this module, drag and drop the variables into the 'Variables' box. Under the 'Assumption Checks' dropdown, ensure that both 'KMO' and 'Bartlett test of sphericity' options are checked (see ref_idx 331). The KMO value should be greater than 0.6 (ideally above 0.8), indicating adequate sampling adequacy. A significant Bartlett's test (p < 0.05) suggests that the correlation matrix is not an identity matrix, meaning there are significant relationships between the variables to justify factor analysis.

• For example, in a study validating a data literacy measurement tool for university students, a KMO of .93 and a significant Bartlett’s test (p < .001) confirmed the suitability of the data for EFA (ref_idx 273). Similarly, research on collaborative measurement tools achieved a KMO of .955 and a significant Bartlett’s test (p < .001), validating the appropriateness of their data (ref_idx 276). However, if the KMO is too low or the Bartlett's test is not significant, researchers should consider collecting more data, revising the variables, or using a different analysis technique.

• Strategically, user-friendly documentation and training materials must provide clear guidance on how to locate and interpret these tests. Interactive tutorials directly within the Jamovi interface can further enhance usability, allowing users to quickly assess data suitability and avoid common pitfalls. This reduces the risk of inappropriate factor analysis and ensures more reliable research outcomes.

• We recommend integrating a checklist within Jamovi that prompts users to assess KMO and Bartlett's test results before proceeding with EFA. Additionally, providing tooltips or brief explanations of these tests directly within the interface would significantly improve user adoption and confidence.

• After confirming data suitability, the next crucial step in EFA is choosing an appropriate rotation method to simplify the factor structure and enhance interpretability. Many Jamovi users find it challenging to navigate the rotation options and understand the implications of each method. This can lead to poorly defined factors and difficulty in interpreting the results.

• To perform Varimax rotation, after dragging the variables into the 'Variables' box in the EFA module, navigate to the 'Rotation' dropdown menu. Select 'Varimax' for orthogonal rotation, which assumes uncorrelated factors, or select an oblique rotation method like 'Direct Oblimin' if factors are expected to be correlated (ref_idx 276). Varimax rotation maximizes the variance of the factor loadings, simplifying the factor structure by making high loadings higher and low loadings lower.

• For instance, a study developing a scale for K-12 teachers utilized Varimax rotation in EFA to ensure clear factor separation, resulting in interpretable factors with high loadings (ref_idx 328). Similarly, researchers exploring factors influencing Gen Z's perceptions of AI in recruiting used Varimax rotation to identify underlying constructs (ref_idx 327). These cases demonstrate the module's utility in varied research contexts.

• Strategically, training materials should emphasize the importance of selecting the appropriate rotation method based on theoretical assumptions and research goals. Providing a decision tree or flowchart within Jamovi can help users choose the best method. This reduces the risk of misinterpreting factor structures and ensures more valid research outcomes.

• We recommend a phased approach to rotation selection: (1) Begin with Varimax rotation for simplicity. (2) Examine the factor correlations. (3) If correlations are high (e.g., > .30), consider oblique rotation. (4) Compare the interpretability of the factor structures and select the most meaningful solution. Adding best-practice tips to the Jamovi UI could further assist researchers.

• After conducting EFA, researchers often perform Confirmatory Factor Analysis (CFA) to validate the factor structure identified. A significant challenge is efficiently executing CFA and interpreting model fit indices, particularly for users new to lavaan integration in Jamovi. This section provides actionable guidance to bridge this gap.

• To execute CFA, the lavaan module must be installed in Jamovi. Once installed, navigate to the 'Analyses' tab, select 'Factor', and choose 'Confirmatory Factor Analysis'. Specify the factor structure based on the EFA results by defining the items that load onto each factor (ref_idx 1). Evaluate model fit using indices such as CFI, TLI, RMSEA, and SRMR. Acceptable model fit typically includes CFI and TLI values close to .95 or higher, RMSEA values below .08, and SRMR values below .08 (Hu & Bentler, 1999).

• For instance, in validating a scale for cybersecurity behavior, researchers used CFA and achieved a good fit (χ2 (df = 258) = 520.788, CFI = 0.946, TLI = 0.937, RMSEA = 0.050) (ref_idx 368). Similarly, a study analyzing university teachers’ well-being used CFA (χ2 = 248.46, df = 49, p < 0.01, CFI = 0.97, TLI = 0.96, RMSEA = 0.073) (ref_idx 369).

• Strategically, lavaan documentation needs to be integrated within Jamovi to guide CFA execution and model fit interpretation. This reduces the learning curve and allows users to confidently validate their factor structures. Clear guidelines are added to the interface to assist users in running and validating models.

• We recommend improving the lavaan integration in Jamovi by adding model syntax generators and automated fit index summaries. Short video tutorials demonstrating the CFA process, from model specification to fit index interpretation, can be created to significantly improve user confidence.

9. 여행상품 홍보 사례: A/B 테스트 및 SEM 적용

• 9-1. 방문객 만족도 영향 요인 분석

• This subsection delves into a practical application of Jamovi in the tourism industry, specifically focusing on analyzing factors influencing visitor satisfaction and modeling service value structures. It builds upon the previous sections by demonstrating how basic and advanced statistical techniques can be applied to real-world data, bridging the gap between theoretical knowledge and actionable insights. This section serves as a case study to exemplify the utility of Jamovi in strategic decision-making.

• Understanding the factors that drive tourist satisfaction is crucial for optimizing marketing strategies and improving service delivery. In the context of tourism promotion, factors such as price, content quality, and ease of booking significantly impact visitor experiences. Identifying and quantifying these relationships allows tourism operators to allocate resources effectively and tailor offerings to meet customer needs. However, simply assuming these relationships exist without empirical validation can lead to misdirected efforts and wasted resources.

• Jamovi offers a streamlined approach to generating correlation matrices, enabling analysts to quickly assess the strength and direction of relationships between key variables. By inputting data related to price perception, content engagement metrics, booking process efficiency, and overall satisfaction scores, a correlation matrix can reveal which factors are most strongly associated with positive visitor outcomes. This goes beyond simple observation and provides a quantitative basis for prioritizing improvement initiatives. According to '누구나 할 수 있는 jamovi 통계분석 (빈도분석에서 구조방정식까지 | 2 판)' [1], Jamovi simplifies this process, providing an intuitive interface for variable selection and matrix generation.

• Consider a scenario where a tourism company has collected data on visitor satisfaction, alongside metrics for price competitiveness, content engagement (e.g., time spent on promotional videos, click-through rates), and booking process completion rates. A Jamovi correlation matrix might reveal a strong positive correlation (e.g., r > 0.7) between content engagement and overall satisfaction, but a weaker correlation with price perception. Conversely, '여행상품 홍보 전략 : 전략/기획 | 한국관광공사 관광e배움터' [6] suggests that failing to address price concerns can erode the positive impact of high-quality content.

• The strategic implication is clear: investing in high-quality, engaging promotional content may yield a higher return in terms of visitor satisfaction compared to simply offering lower prices. However, a weak negative correlation between booking ease and satisfaction should not be ignored either, as a smooth booking process is crucial for converting interest into confirmed bookings. These findings should guide strategic decisions related to content creation, website optimization, and pricing strategies.

• To effectively implement these findings, tourism operators should prioritize improvements to their online content, ensuring it is visually appealing, informative, and easy to navigate. This might involve creating high-quality videos showcasing destinations, writing compelling blog posts highlighting unique experiences, and optimizing website landing pages for engagement. Simultaneously, efforts should be directed towards streamlining the booking process, reducing friction points, and providing clear, concise information to prospective visitors. Jamovi facilitates the quick and effective translation of data into these actionable strategies.

• While correlation matrices provide valuable insights into the relationships between variables, linear regression allows for a more nuanced understanding of the predictive power of individual factors on overall visitor satisfaction. Specifically, it enables quantification of the extent to which changes in price, content, or booking ease influence satisfaction levels, holding other factors constant. This is critical for making informed decisions about resource allocation and strategic prioritization, as it moves beyond mere association to establish a causal relationship.

• Jamovi simplifies the process of setting up and running linear regression models, making it accessible to users with varying levels of statistical expertise. The user interface allows for easy assignment of dependent and independent variables, and provides options for examining model fit statistics (R-squared) and regression coefficients. This functionality enables analysts to quickly assess the overall explanatory power of the model, as well as the individual contribution of each predictor variable. '누구나 할 수 있는 jamovi 통계분석 (빈도분석에서 구조방정식까지 | 2 판)' [1] details Jamovi's regression capabilities, emphasizing ease of use and comprehensive output.

• Imagine, for instance, a tourism agency wants to understand how specific aspects of their service – website usability, content relevance, and customer support responsiveness – predict overall customer satisfaction. Using Jamovi, they can run a multiple regression model with overall satisfaction as the dependent variable and the other three factors as independent variables. The regression output might reveal that website usability has the strongest positive influence (beta = 0.6, p < 0.01), while content relevance has a moderate effect (beta = 0.3, p < 0.05), and customer support responsiveness has a non-significant impact (p > 0.05).

• Strategically, this implies that improving website usability should be the agency's top priority, as it has the greatest potential to enhance customer satisfaction. Investment in enhancing content relevance may also be worthwhile, but resources should be directed towards areas most likely to yield tangible results. The non-significant effect of customer support responsiveness may warrant further investigation; it could be that the current support channels are not effectively addressing customer needs, or that responsiveness is not a primary driver of satisfaction in this particular context.

• To translate these insights into action, the tourism agency should conduct a comprehensive review of its website usability, identifying areas for improvement based on user feedback and best practices. This could involve simplifying navigation, optimizing page load speeds, and ensuring mobile responsiveness. Resources could then be allocated to content enhancement, ensuring that promotional materials are relevant, engaging, and tailored to the target audience. These actionable strategies, derived directly from Jamovi-driven regression analysis, allow for targeted improvements that are more likely to boost customer satisfaction and drive business growth.

• A/B testing is a crucial methodology for optimizing marketing campaigns and website elements to maximize conversion rates and visitor engagement. In the tourism context, A/B testing can be used to compare different versions of website landing pages, promotional emails, or ad creatives to determine which performs best in terms of click-through rates, booking completion rates, or overall visitor satisfaction. However, designing and analyzing A/B tests can be complex, requiring careful consideration of statistical significance and sample size.

• While Jamovi doesn't offer dedicated A/B testing modules, its hypothesis testing capabilities can be leveraged to analyze the results of A/B experiments effectively. By conducting t-tests or chi-square tests, analysts can determine whether observed differences between the control and treatment groups are statistically significant, and not simply due to random chance. This ensures that decisions about which version to implement are based on solid evidence, rather than gut feeling. Though '누구나 할 수 있는 jamovi 통계분석 (빈도분석에서 구조방정식까지 | 2 판)' [1] does not directly cover A/B testing, the fundamental statistical principles for comparison are illustrated well.

• For instance, a hotel chain might want to test two different versions of their booking confirmation email – one with a promotional offer for spa services, and one without. They randomly assign new bookings to receive either version A (control) or version B (treatment). After a set period, they can use Jamovi to compare the average spend on spa services between the two groups. If the t-test reveals a significant difference (p < 0.05), with version B leading to higher average spend, the hotel chain can confidently implement version B for all future booking confirmations.

• The strategic implication is clear: by rigorously testing different elements of their marketing campaigns, tourism operators can continuously optimize their messaging and offers to maximize visitor engagement and revenue. This approach moves beyond intuition and guesswork, allowing for data-driven decisions that lead to more effective marketing strategies.

• To implement A/B testing effectively, tourism companies should establish clear goals and metrics, design experiments with appropriate sample sizes, and carefully track and analyze the results using Jamovi. This requires a systematic approach to marketing optimization, with continuous monitoring, analysis, and refinement based on data-driven insights.

• 9-2. SEM을 통한 서비스 가치 구조 모델링

• This subsection builds on the previous analysis of visitor satisfaction drivers, now delving into the application of Structural Equation Modeling (SEM) to model service value structures and understand the complex relationships between service quality, emotional satisfaction, and revisit intention. This showcases Jamovi's advanced analytical capabilities, enabling a deeper understanding of the underlying mechanisms driving tourist behavior. This section serves as a bridge to illustrate how EFA and CFA results can be practically implemented in SEM using Jamovi.

• Exploratory Factor Analysis (EFA) is crucial for identifying the underlying structure of a dataset and reducing its dimensionality, especially when dealing with complex constructs like service value. It helps determine the number of factors that explain the variance in a set of observed variables. However, navigating Jamovi's interface to perform EFA can be challenging for new users, potentially leading to errors in analysis and misinterpretation of results. Clearly defined menu paths are essential for ensuring accurate and efficient analysis.

• To perform EFA in Jamovi, users should follow these steps. First, open the data file in Jamovi. Second, navigate to the ‘Factor’ menu. Third, select ‘Exploratory Factor Analysis’. Fourth, drag and drop the variables of interest into the ‘Variables’ box. Fifth, under the ‘Extraction’ section, choose the extraction method (e.g., Principal components). Sixth, under the ‘Rotation’ section, select the rotation method (e.g., Varimax). Seventh, review the output, including the factor loadings, eigenvalues, and scree plot, to determine the number of factors to retain. '누구나 할 수 있는 jamovi 통계분석 (빈도분석에서 구조방정식까지 | 2 판)' [1] provides a comprehensive guide on how to interpret these outputs.

• Consider a scenario where a tourism company has collected data on various aspects of service quality, such as staff friendliness, cleanliness, and efficiency. By performing EFA, they can identify whether these variables load onto a smaller number of underlying factors, such as ‘Overall Service Quality’ or ‘Tangibles’. Identifying these factors helps the company understand what aspects of service quality are most important to visitors. The '여행상품 홍보 전략 : 전략/기획 | 한국관광공사 관광e배움터' [6] emphasizes that understanding these underlying factors is key to crafting effective marketing messages and improving service delivery.

• Strategically, a clear understanding of Jamovi's EFA menu paths allows analysts to quickly and accurately identify the underlying structure of their data, enabling them to focus on the most critical factors driving visitor satisfaction. Without this clarity, companies risk misinterpreting their data and making suboptimal decisions about resource allocation.

• To improve accessibility, creating a visual guide or checklist with screenshots of each menu selection would be beneficial. This could be integrated into training materials for new Jamovi users. Additionally, providing sample datasets and step-by-step instructions would allow users to practice the EFA procedure and gain confidence in their ability to perform the analysis accurately.

• Confirmatory Factor Analysis (CFA) is used to test hypotheses about the factor structure of observed variables, confirming or disconfirming the relationships identified in the EFA stage. However, executing CFA within Jamovi, particularly using the 'lavaan' module, requires precise steps and can be a hurdle for users unfamiliar with structural equation modeling. Incorrect specification of the model can lead to inaccurate results and misleading conclusions.

• To execute CFA using the lavaan module in Jamovi, users should first install the ‘lavaan’ module from the Jamovi library. Second, specify the model syntax, defining the relationships between latent variables and observed indicators. Third, navigate to the ‘Analyses’ tab and select ‘SEM’. Fourth, input the model syntax into the ‘Model Syntax’ box. Fifth, specify the estimator (e.g., maximum likelihood). Sixth, review the output, including model fit indices (e.g., CFI, RMSEA, SRMR) and parameter estimates, to assess the model fit and significance of the relationships. According to '빅 데이터와 기계 학습의 시대 심리학 연구 모형의 평가 원칙과 ...' [12], understanding the fit indices is essential for evaluating model validity.

• Consider a scenario where a tourism company has used EFA to identify three factors related to visitor experience: 'Service Quality', 'Emotional Satisfaction', and 'Revisit Intention'. Using CFA, they can test a model where 'Service Quality' predicts 'Emotional Satisfaction', which in turn predicts 'Revisit Intention'. The '여행상품 홍보 전략 : 전략/기획 | 한국관광공사 관광e배움터' [6] suggests that if the CFA model shows a good fit, it provides strong evidence for the proposed relationships between these constructs.

• The strategic advantage of clearly outlining the steps for executing CFA in Jamovi is that it allows researchers and analysts to rigorously test their hypotheses about the relationships between key constructs, leading to more informed decisions about service improvement and marketing strategies. Without this rigorous testing, companies risk relying on assumptions that may not be supported by empirical evidence.

• To enhance clarity, provide sample model syntax for common CFA models. Include screenshots of the lavaan module interface within Jamovi, highlighting key settings. Offer training sessions or webinars demonstrating the CFA execution process from start to finish. This allows users to follow step-by-step instructions, from installing the module to running and interpreting results.

• Visualizing SEM results through path diagrams is essential for communicating complex relationships to stakeholders, facilitating understanding, and informing strategic decisions. However, Jamovi's default SEM output may not always provide clear and visually appealing path diagrams. Therefore, specific guidelines are needed to enhance the visualization of SEM path diagrams within Jamovi, enabling effective communication of research findings.

• After running SEM in Jamovi, users can visualize the path diagram by selecting the ‘Path Diagram’ option under the ‘Plots’ section. To enhance the diagram, users can adjust the layout, node positions, and arrow styles. Furthermore, users can customize the appearance of the diagram by changing the colors, fonts, and labels. The ‘여행상품 홍보 전략 : 전략/기획 | 한국관광공사 관광e배움터' [6] emphasizes the importance of clear visualizations for conveying complex information to non-technical audiences.

• For example, a tourism agency might want to present the results of their SEM analysis to the board of directors, highlighting the relationships between service quality, emotional satisfaction, and revisit intention. A well-designed path diagram can effectively communicate these relationships, showing the strength and direction of each path. Without a clear visualization, the board members may struggle to understand the complex statistical output, hindering their ability to make informed decisions.

• The strategic implication is that providing clear guidelines for visualizing SEM path diagrams in Jamovi allows analysts to effectively communicate their findings to a wider audience, leading to better-informed strategic decisions and increased buy-in for data-driven initiatives.

• Offer templates for creating visually appealing path diagrams, demonstrating effective use of colors, fonts, and labels. Provide step-by-step instructions on how to customize the layout, node positions, and arrow styles within Jamovi. Share examples of well-designed path diagrams from published research to illustrate best practices. These actions ensure effective visual communication, turning intricate data models into accessible and impactful insights.

10. 종합 결론: 완결형 프로젝트 연습 제언

• 10-1. 환경 설정부터 SEM까지의 완결형 프로젝트

• 이 소절에서는 Jamovi를 활용한 통계 분석 튜토리얼의 최종 단계로, 환경 설정부터 구조방정식 모델링(SEM)까지 전 과정을 아우르는 완결형 프로젝트 학습을 제안합니다. 이는 단순 기능 습득을 넘어 실제 데이터 분석 상황에 대한 적응력을 높이고, 실질적인 문제 해결 능력을 함양하는 데 목표를 둡니다. 특히, 여행상품 홍보 데이터셋을 활용한 사례 연구를 통해 이론적 지식과 실무 적용 간의 간극을 좁히고, 데이터 기반 의사결정 역량을 강화하는 방안을 모색합니다.

• Jamovi를 이용한 완결형 프로젝트 실습을 위해, 실제 비즈니스 상황을 반영한 여행상품 홍보 데이터셋 활용을 제안합니다. 이 데이터셋은 잠재 고객의 특성, 상품 속성, 마케팅 채널 효과 등 다양한 변수를 포함하여, 데이터 분석 및 의사결정 과정을 현실적으로 경험할 수 있도록 설계되었습니다.

• 데이터셋 컬럼의 예시로는 가격, 콘텐츠 매력도, 예약 용이성, 고객 만족도 점수 등이 있으며, 이를 통해 상관관계 분석, 회귀분석, A/B 테스트 등 다양한 통계 분석 기법을 적용해 볼 수 있습니다(Ref. 6, 1). 특히, 구조방정식 모델링(SEM)을 통해 서비스 품질, 감성 만족, 재방문 의도 간의 복잡한 인과 관계를 분석함으로써, 데이터 기반 의사결정 역량을 강화할 수 있습니다.

• 여행상품 홍보 데이터셋을 활용한 완결형 프로젝트는 Jamovi 설치부터 데이터 준비, 기초 통계 분석, 가설 검정, 회귀 분석, SEM 모델링, 결과 해석 및 보고서 작성의 전체 과정을 포함합니다. 이를 통해, 데이터 분석 도구 활용 능력뿐만 아니라, 문제 정의, 데이터 수집 및 전처리, 분석 결과 해석, 의사결정 제안 등 데이터 기반 의사결정 역량을 종합적으로 향상시킬 수 있습니다.

• 제언: 데이터셋 컬럼 정의 및 데이터 분석 목표를 명확히 설정하고, Jamovi를 이용한 단계별 분석 과정을 체계적으로 수행하여 분석 결과 해석 능력과 실질적인 의사결정 역량을 강화합니다. 특히, A/B 테스트와 SEM 모델링을 통해, 데이터 기반 의사결정의 효과를 직접 경험하고, 분석 결과를 바탕으로 실질적인 마케팅 전략을 도출하는 연습을 수행합니다.

• Jamovi 완결형 프로젝트의 성공적인 수행을 위해, 단계별 체크리스트와 프로젝트 보고서 샘플 목차를 제공합니다. 이는 학습 과정을 체계화하고, 결과물의 완성도를 높이는 데 기여합니다. 체크리스트는 Jamovi 설치 및 환경 설정, 데이터 준비 및 전처리, 기초 통계 분석, 가설 검정, 회귀 분석, SEM 모델링, 결과 해석 및 보고서 작성 등 각 단계별 주요 작업 항목과 확인 사항을 포함합니다.

• 프로젝트 보고서 샘플 목차는 서론, 이론적 배경, 데이터 및 분석 방법, 분석 결과, 결론 및 제언, 참고 문헌 등으로 구성됩니다. 서론에서는 연구 배경 및 목적, 연구 문제, 연구 가설 등을 제시하고, 이론적 배경에서는 관련 이론 및 선행 연구를 검토합니다. 데이터 및 분석 방법에서는 데이터 출처, 데이터 설명, 변수 설정, 분석 방법 등을 상세히 기술하고, 분석 결과에서는 통계 분석 결과 및 해석을 제시합니다. 마지막으로, 결론 및 제언에서는 연구 결과 요약, 정책적 시사점, 연구의 한계점 및 향후 연구 방향 등을 제시합니다.

• 단계별 체크리스트를 통해, 학습자는 각 단계별 필수 작업 항목을 빠짐없이 수행하고, 문제 발생 시 적절한 해결 방안을 모색할 수 있습니다. 프로젝트 보고서 샘플 목차를 통해, 학습자는 보고서 작성의 틀을 이해하고, 분석 결과를 체계적으로 정리하여 제시할 수 있습니다.

• 제언: 단계별 체크리스트를 활용하여 프로젝트 진행 상황을 주기적으로 점검하고, 프로젝트 보고서 샘플 목차를 참고하여 보고서 작성의 완성도를 높입니다. 특히, 분석 결과 해석 및 제언 도출 과정에서, 비판적 사고 능력과 창의적 문제 해결 능력을 발휘하여 실질적인 가치를 창출하는 데 집중합니다.

• 자기 점검을 위한 통계 모형 평가 원칙 적용은 Jamovi 완결형 프로젝트의 중요한 부분입니다. 모형 평가 시, 편향-분산 균형(Bias-Variance Trade-off) 원칙을 적용하여 모형의 적합성을 판단하고, 과적합 또는 과소적합 문제를 해결해야 합니다(Ref. 7).

• 편향-분산 균형은 모형의 복잡도와 예측 정확도 간의 관계를 나타내는 개념입니다. 편향이 높은 모형은 데이터의 패턴을 제대로 학습하지 못해 예측 정확도가 낮고, 분산이 높은 모형은 훈련 데이터에 지나치게 적합하여 새로운 데이터에 대한 예측 정확도가 낮습니다. 따라서, 모형의 복잡도를 적절히 조절하여 편향과 분산 간의 균형을 맞추는 것이 중요합니다.

• Jamovi 결과 해석 가이드라인은 통계 분석 결과의 의미를 명확하게 이해하고, 실질적인 의사결정에 활용할 수 있도록 돕습니다. 가이드라인은 t-검정, ANOVA, 회귀 분석, 구조방정식 모델링 등 다양한 통계 분석 기법에 대한 해석 방법과 주의 사항을 포함합니다.

• 제언: 편향-분산 균형 원칙을 적용하여 모형의 적합성을 평가하고, 필요에 따라 모형의 복잡도를 조절합니다. Jamovi 결과 해석 가이드라인을 참고하여 통계 분석 결과의 의미를 명확하게 파악하고, 실질적인 의사결정에 활용합니다. 특히, 통계적 유의성뿐만 아니라, 효과 크기, 신뢰 구간 등 다양한 지표를 함께 고려하여 분석 결과를 해석합니다.

11. Conclusion

• 본 보고서는 Jamovi 통계 분석 소프트웨어의 환경 설정부터 구조방정식 모델링(SEM)까지의 전 과정을 체계적으로 안내하며, 데이터 기반 의사결정 역량 강화를 목표로 하였습니다. Jamovi는 오픈 소스 기반으로, GUI 인터페이스와 R 코드 통합을 통해 초보자부터 전문가까지 쉽게 통계 분석에 접근할 수 있도록 설계되었으며, 본 튜토리얼을 통해 다양한 통계 분석 기법을 실무에 적용할 수 있음을 확인하였습니다.

• 여행상품 홍보 데이터셋을 활용한 사례 연구에서는 A/B 테스트와 SEM 모델링을 통해, 데이터 기반 의사결정의 효과를 직접 경험하고, 분석 결과를 바탕으로 실질적인 마케팅 전략을 도출하는 연습을 수행하였습니다. 이를 통해, 독자들은 통계 분석 도구 활용 능력뿐만 아니라, 문제 정의, 데이터 수집 및 전처리, 분석 결과 해석, 의사결정 제안 등 데이터 기반 의사결정 역량을 종합적으로 향상시킬 수 있을 것입니다.

• 본 보고서에서 제시된 내용들을 바탕으로, 독자들은 Jamovi를 활용한 완결형 프로젝트를 수행하고, 자기 점검을 위한 통계 모형 평가 원칙(편향·분산 균형)을 적용하여 모형의 적합성을 판단하고, 과적합 또는 과소적합 문제를 해결할 수 있을 것입니다. 또한, Jamovi 결과 해석 가이드라인을 참고하여 통계 분석 결과의 의미를 명확하게 파악하고, 실질적인 의사결정에 활용할 수 있을 것입니다. 향후에는 다양한 산업 분야의 데이터셋을 활용하여, Jamovi 튜토리얼의 적용 범위를 확장하고, 데이터 기반 의사결정 역량 강화를 위한 노력을 지속할 필요가 있습니다.

Source Documents

• 누구나 할 수 있는 jamovi 통계분석 (빈도분석에서 구조방정식까지 | 2 판)https://gimssine.com/products/9788999729324?srsltid=AfmBOoqMowlQJf1GCf0faST2wU2PMKkhmmuiYn8qyGjgrY7aWyhwDufw
• Jamovi(자모비)란?http://kocw-n.xcache.kinxcdn.com/data/document/2021/honam/kimseungjae0113/1.pdf
• jamovi로 통계 배우기 — Learning statistics with jamovihttps://lsj.readthedocs.io/ko/latest/
• 여행상품 홍보 전략 : 전략/기획 | 한국관광공사 관광e배움터https://touredu.visitkorea.or.kr/common/C202300341/0000/0
• 빅 데이터와 기계 학습의 시대 심리학 연구 모형의 평가 원칙과 ...https://accesson.kr/kpageneral/assets/pdf/16123/journal-40-4-389.pdf
• 빅 데이터와 기계 학습의 시대 심리학 연구 모형의 평가 원칙과 ...https://accesson.kr/kpageneral/assets/pdf/16123/journal-40-4-389.pdf
• 기초통계분석 원데이 클래스 #4.신뢰도 분석 using JAMOVIhttps://passme.co.kr/%EA%B8%B0%EC%B4%88%ED%86%B5%EA%B3%84%EB%B6%84%EC%84%9D-%EC%9B%90%EB%8D%B0%EC%9D%B4-%ED%81%B4%EB%9E%98%EC%8A%A4-4%EC%8B%A0%EB%A2%B0%EB%8F%84-%EB%B6%84%EC%84%9D-using-j/
• Effect of Adolescent Stress on Smartphone Overdependencehttp://journal.dcs.or.kr/xml/36517/36517.pdf
• Risk Simulator 2012 User Manual - Koreanhttps://www.crystalballservices.com/Portals/0/eng/ROV/ROVSim/rsmanual-korean.pdf
• 초등학령기 자녀를 둔 기혼 여성의 양육스트레스가 주관적 만족 ...https://journal.kslrp.net/xml/35238/35238.pdf
• 체육교수학습론http://att.eduspa.com/etcdata/pto/exam_prof12.pdf
• KIRD 교육프로그램https://alpha-campus.kr/pdf/2025_KIRD_%EA%B5%90%EC%9C%A1%EC%95%88%EB%82%B4%EC%B1%85%EC%9E%90.pdf
• Jamovi(자모비) 활용(1)http://kocw-n.xcache.kinxcdn.com/data/document/2021/honam/kimseungjae0113/3.pdf
• 여행 트렌드와 여행사 MD : 전략/기획 | 한국관광공사 관광e배움터https://touredu.visitkorea.or.kr/common/C202300350/0000/0
• Jamovihttps://en.wikipedia.org/wiki/Jamovi
• 인문역량 검사도구의 개발: 신뢰도 및 타당도 분석*https://scholarworks.bwise.kr/hanyang/bitstream/2021.sw.hanyang/146312/1/kyobo_10511689.pdf
• 서비스스케이프와 인적 서비스가 고객 반응에 미치는 영향의 비교https://scholarworks.bwise.kr/hanyang/bitstream/2021.sw.hanyang/202881/1/KCI_FI001568688.pdf
• 온라인 쇼핑몰의 AI 챗봇 서비스 품질이 지속이용의도에 미치는 영향: 사회적 실재감, 몰입 및 지각된 유용성의 매개효과https://koreascience.kr/article/JAKO202408643205340.pdf
• The Influence of the Service Quality of Eyelash Beauty Treatment and Perceived Value on Behavior Intentionhttps://www.e-jkc.org/journal/view.php?number=2538
• INDUSTRIAL METROLOGY-GRADE - Artec 3Dhttps://cdn.artec3d.com/pdf/Artec3D-SpaceSpider.pdf
• 참조 설명서https://download.nikonimglib.com/archive1/AlcGv007gfWP01MpO4691DBEl901/D5200RM_NT(Kr)03.pdf
• 사용설명서https://download.nikonimglib.com/archive1/eEgkF00h5L9C01QjsUK90oDi9D91/D3200UM_NT(Kr)04.pdf
• My Cloud EX2 Ultra - 개인 클라우드 스토리지https://documents.westerndigital.com/content/dam/doc-library/ko_kr/assets/public/wd/product/nas/my_cloud/ex2_ultra/user-manual-my-cloud-expert-series-ex2-ultra.pdf
• 의학 통계 코드 저장소https://medistat.tistory.com/?page=6
• KACL|한국코퍼스언어학회http://corpus.or.kr/html.php
• eISSN: 2093-758X - Journal of Korean Academy of Nursinghttps://jkan.or.kr/upload/pdf/jkan-55-1_full.pdf
• Structural Equation Modeling in Practice: A Review and ...https://www3.nd.edu/~kyuan/courses/sem/readpapers/anderson.pdf
• 제28권 제4호 (통권 97호) / 2023. 12 - 기독교세계관학술동역회https://www.worldview.or.kr/library/series/faith-and-scholarship/download/313/15567/FaS%2028-4%20%ED%95%A9%EB%B3%B8.pdf
• Studying Scientific Collaboration Networks Using Social ...https://www.iosrjournals.org/iosr-jhss/papers/Vol.29-Issue3/Ser-4/J2903045870.pdf
• A Meta-Analysis of the Effects of Instructional Interventions on ...https://www.nro.nl/wp-content/uploads/2014/05/PROO-A-Meta-Analysis-of-the-Effects-of-Instructional-Interventions-on-Students%E2%80%99-Mathematics-Achievement-Egbert-Harskamp-ea.pdf
• A Comprehensive Guide to REDCaphttps://www.unmc.edu/vcr/_documents/unmc_redcap_usage.pdf
• 관리자 설명서 - Lexmarkhttps://publications.lexmark.com/publications/lexmark_solutions/Cloud/AG/pdf/Lexmark_CloudServices_AdminGuide_ko.pdf
• Create Lists | IFLS Library Systemhttps://iflsweb.org/knowledge-base/create-lists
• G-TELP 측정통계연구소https://www.g-telp.co.kr:335/PA/sub/sub2_2.html
• 사 업 보 고 서 - 상장공시시스템(KIND) - 한국거래소https://kind.krx.co.kr/external/2025/03/12/000923/20250312002424/11011.htm
• 사 업 보 고 서 - 상장공시시스템https://kind.krx.co.kr/external/2023/03/09/001018/20230309002308/11011.htm
• 사 업 보 고 서 - 상장공시시스템(KIND)https://kind.krx.co.kr/external/2024/03/08/000071/20240308000175/11011.htm
• 분 기 보 고 서 - 상장공시시스템(KIND)https://kind.krx.co.kr/external/2024/05/16/000898/20240516002110/11013.htm
• FmX Integrated Display User Guide - Elite Ag Solutionshttps://eliteagsolutions.com/uploads/firmware/FmX_UserGuide_72A.pdf
• 클라이언트 라이브러리 | Cloud Speech-to-Text Documentation | Google Cloudhttps://cloud.google.com/speech-to-text/docs/client-libraries?hl=ko
• An Introduction to the Shapiro-Wilk Test for Normalityhttps://builtin.com/data-science/shapiro-wilk-test
• Shapiro-Wilk Normality Test | shapiro.test in R | R-bloggershttps://www.r-bloggers.com/2024/10/shapiro-wilk-normality-test-shapiro-test-in-r/
• 대학생의 데이터 리터러시 측정 도구 타당화 및 신입생 데이터 ...https://pdfs.semanticscholar.org/284b/b7ed7e0e6cf1558981faf8f08fa1aa8c1721.pdf
• 초연결사회 집단지성의 기본 단위로서 협업 을, ' ' 측정하는 도구 ...https://english.ksfme.or.kr/xml/12408/12408.pdf
• Bridging Technology and Talent: Gen Z's Take on AI in Recruiting and Hiringhttps://ceeconference.vse.cz/wp-content/uploads/ceecbe-2025-proceedings-poenaru-diaconescu.pdf
• Development and validation of a generic self-assessment scale for K-12 teachers as feedback givers: Insights from item response theory and factor analysis - Humanities and Social Sciences Communicationshttps://www.nature.com/articles/s41599-025-04927-4
• Exploratory Factor Analysis — Learning statistics with jamovihttps://lsj.readthedocs.io/ru/latest/Ch15/Ch15_FactorAnalysis_1.html
• 에이아이트릭스, 임상 데이터 결측치 중요 예측 정보 활용 입증http://www.bosa.co.kr/news/articleView.html?idxno=2251219
• 안양대학교 미세먼지관리 특성화대학원, 세계 최대 지구과학 학술대회 'EGU 2025'참가 우수 연구성과 발표 : 네이트 뉴스https://news.nate.com/view/20250520n15449
• 안양대 미세먼지관리 특성화대학원, ‘EGU 2025’서 연구성과 발표https://dhnews.co.kr/news/view/1065577922536763
• The impact of corporate social responsibility on cybersecurity behavior: The crucial role of organizationally-prescribed perfectionism - Humanities and Social Sciences Communicationshttps://www.nature.com/articles/s41599-025-04511-w
• Frontiers | Examining the Relationships Between Job Characteristics, Emotional Regulation and University Teachers’ Well-Being: The Mediation of Emotional Regulationhttps://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2020.01727/full
• tax structure and economic growth: an oecd analysishttps://ninercommons.charlotte.edu/record/3209/files/Raynor_uncc_0694N_11248.pdf
• Multicollinearity in Machine Learning | Restackiohttps://www.restack.io/p/machine-learning-multiscale-modeling-answer-multicollinearity-cat-ai
• 디지털 트윈 환경에서의 이상치 탐지 알고리즘 비교분석http://journal.dcs.or.kr/xml/35378/35378.pdf
• Data anomalies: Definition, examples, and best practices for finding and resolving them | Metaplanehttps://www.metaplane.dev/blog/data-anomalies-definition-examples
• Essential Data Cleaning Techniques for Accurate Machine Learning Modelshttps://www.kdnuggets.com/essential-data-cleaning-techniques-accurate-machine-learning-models
• DataEther: Data Exploration Framework For Ethereumhttps://web.eecs.umich.edu/~chenang/papers/icdcs-2019.pdf