The promised productivity improvements resulting from using application generators have been described by several authors. Some experiments have also been conducted, but they differ substantially in their research settings and conduct of the experiments. This paper examines these differences by collating the main factors affecting the results of the experiments into an overall framework. Previous research efforts are then analyzed according to this framework.

Using the developed framework, experiments were conducted to evaluate the combined effects of the use of application generators and of development strategy on productivity. The results of these laboratory cases showed that, compared with third generation languages, 30 to 1 productivity improvements are possible in the development of medium-sized administrative systems, when application generators are used together with a prototyping strategy. In addition, programmers experienced more satisfaction in prototyping projects than in linear projects.