Software Modeling in Industry: A Practitioner Survey Dataset

Overview

This dataset provides empirical evidence on how software modeling, particularly the Unified Modeling Language (UML), is perceived and used by software practitioners in industrial settings. It includes 416 survey responses collected between 2013 and 2023, capturing practitioners' professional backgrounds, modeling experience, and perspectives on UML adoption, benefits, limitations, and practical challenges. The survey investigates several dimensions of industrial software modeling, including communication and collaboration, model comprehension, modeling complexity, consistency maintenance, tool support, productivity, and the perceived return on investment in modeling activities. It also examines factors that discourage UML adoption and potential improvements that could encourage wider use in software development. The dataset combines structured questionnaire responses with open-ended questions exploring how practitioners use UML in practice, the main difficulties they encounter, and the circumstances under which modeling provides value. This dataset supports reproducible empirical research, secondary analyses, and comparative investigations into industrial modeling practices. It provides researchers and practitioners with an evidence base to understand the factors influencing software modeling adoption and to identify opportunities to improve modeling languages, tools, and development practices.

Survey Instrument and Research Opportunities

The dataset comprises practitioner survey responses collected between 2013 and 2023 to investigate software modeling adoption, use, perceived benefits, and challenges in industrial software development, with particular emphasis on the Unified Modeling Language (UML).

The survey examines several complementary dimensions:

• Practitioner Background: Demographic characteristics, educational background, professional roles, and experience in software development and modeling.
• UML Adoption and Usage: Perceptions of industrial UML adoption, selective and informal modeling, model-driven development, and code generation.
• Communication and Comprehension: The perceived role of UML in supporting communication, shared understanding, and software design discussions.
• Modeling Complexity and Consistency: Challenges involving abstraction, model completeness, diagram consistency, and model synchronization.
• Barriers to Adoption: Technical, organizational, economic, and human factors potentially discouraging UML usage.
• Modeling Tool Improvements: Perceptions of feature-oriented diagrams, automatically generated system overviews, bidirectional model-code synchronization, and collaborative modeling environments.
• Industrial Experiences: Open-ended responses concerning how UML is used in practice, difficulties encountered, and circumstances under which modeling is considered worthwhile.

Potential Research Applications

The dataset supports secondary empirical investigations into:

1. Conditions under which software modeling is perceived as valuable.
2. Technical, organizational, and economic barriers to UML adoption.
3. Differences in modeling perceptions across practitioner profiles and collection periods.
4. Selective and informal modeling practices in industrial settings.
5. The perceived impact of model inconsistency and synchronization challenges.
6. Requirements for improving industrial software modeling tools.
7. The role of UML in communication and shared understanding.
8. Relationships between professional experience, UML adoption, and modeling perceptions.

The combination of structured responses and open-ended questions supports quantitative, qualitative, and mixed-methods investigations. Researchers can use the dataset to examine associations, identify recurring perceptions, develop empirically grounded taxonomies, and formulate hypotheses for future studies.

Methodological Considerations: The dataset primarily captures self-reported experiences and perceptions. Consequently, findings should not be interpreted as direct measurements of modeling effectiveness or causal effects. Comparisons across collection periods should account for differences in sample composition and data collection coverage.