modelling tools neither between IVA and Interview nor between experts and generalists. The only exception to this tendency is the statement "I could try the task now on my own" which was significantly less agreed by generalists under the IVA condition. This could be an argument for the growing problem awareness through the modelling process.
The task could be modelled on different levels of description. The test persons were free to choose the most suitable level. The experts tended to describe their task from an abstract level, while more generalists outlined the details in their models. When using IVA the differences in abstraction decreased and both groups tried to combine the details with the underlying structure.
Most of the significant results were found in comparing the different interviewer's results. The number of cards used in the models, the mistakes in syntax and semantic, and the shape of the IVA models was compared. No significant difference could be found between the interviewers under the IVA condition. In contrast there was a significant difference in shape, abstraction level, and completeness under the interview condition.
The results show, that there is a better way than plainly asking. Leaving data sampling and modelling to the task owner while supporting him through a suitable modelling technique leads to obtain a decreasing interviewer effect and to more completeness of data. It helps to a more error-free model of reality and extracts a more standardised level of description. No differences were found in the confidence in the modelling. The most important finding is, that a modelling technique in addition to the interview does help to decrease interviewer effects.
To put it all in a nutshell, to make task owners formalise task descriptions takes no effect on the confidence in the correctness of the model, but it saves time and decreases the influence of the interviewer.
Cierjacks, M., Antoni, C., Resch, D. & Mangold, R. ( 1995). Instrument zur Vorgangsanalyse (IVA) : Handanweisung,Unveröffentlichter Forschungsbericht: Mannheim.
Finney, K. ( 1996). "Mathematical Notation.in formal Specification: To difficult for the Masses?" IEEE Transactions on Software Engineering, vol. 9, no. 6, pp. 733-744.