Abstract.Context: Today practitioners have a myriad of methods from which to choose for the development of software applications. However they lack empirical data that characterize these methods in terms of usefulness, ease of use or compatibility, all of them relevant variables to assess the developer's intention to use them. OBJECTIVE: To compare three methods, each following a different paradigm (Model-Driven, Model-Based and the traditional, code-centric, respectively) with respect to its intention to use by junior software developers while developing the business layer of a Web 2.0 application. METHOD: We have conducted an experiment with 26 graduate students of the University of Alicante. The application developed was a Social Network, which was organized in three different modules. Subjects were asked to use a different method for each one of the three modules, and then answer a questionnaire that gathered their perceptions during its use. RESULTS: The results show that the method that followed the Model-Driven development paradigm is regarded as the most useful, although it is also regarded as the more difficult to use. They also show that junior software developers feel comfortable with the use of models, and are likely to use them if accompanied by a model-driven development environment. CONCLUSIONS: Model-driven development methods seem to show a great potential for adoption. However, further experimentation is needed to be able to generalize the results to a different population, different methods, languages and tools, different domains or different application sizes.

BACKGROUND: Model-driven Engineering (MDE) approaches are often acknowledged to improve the maintainability of the resulting applications. However, there is a scarcity of empirical evidence that backs their claimed benefits and limitations with respect to code-centric approaches. OBJECTIVE: To compare the performance and satisfaction of junior software maintainers while executing maintainability tasks on Web applications with two di↵erent development approaches, one being OOH4RIA, a model-driven approach, and the other being a code-centric approach based on Visual Studio .NET and the Agile Unified Process. METHOD: We have conducted a quasi-experiment with 27 graduated students from the University of Alicante. They were aleatory divided into two groups, and each group was assigned to a di↵erent Web application on which they performed a set of maintainability tasks. RESULTS: Maintaining Web applications with OOH4RIA clearly improves the performance of subjects. It also tips the satisfaction balance in favor of OOH4RIA, although not significantly. CONCLUSIONS: Model-driven development methods seem to improve both the developers' objective performance and subjective opinions on ease of use and utility of the method. Further experimentation is needed to be able to generalize the results to di↵erent populations, methods, languages and tools, di↵erent domains and di↵erent application sizes.

No abstract

BACKGROUND: Model-Driven Engineering claims a positive impact on software productivity and satisfaction. However, few efforts have been made to collect evidences that assess its true benefits and limitations. OBJECTIVE: To compare the productivity and satisfaction of junior Web developers during the development of the business layer of a Web 2.0 Application when using either a code-centric, a model-based (UML) or a Model-Driven Engineering approach (OOH4RIA). RESEARCH METHOD: We designed a full factorial, intra-subject experiment in which 26 subjects, divided into five groups, were asked to develop the same three modules of a Web application, each one using a different method. We measured their productivity and satisfaction with each approach. RESULTS: The use of Model-Driven Engineering practices seems to significantly increase both productivity and satisfaction of junior Web developers, regardless of the particular application. However, modeling activities that are not accompanied by a strong generation environment make productivity and satisfaction decrease below code-centric practices. Further experimentation is needed to be able to generalize the results to a different population, different languages and tools, different domains and different application sizes.

No abstract