i-nth logo

Authors

Richard J. Irons

Abstract

The Cell Error Rate in simple spreadsheets averages about 2% to 5%. This CER has been measured in domain free environments.

This paper compares the CERs occurring in domain free and applied domain tasks. The applied domain task requires the application of simple linear algebra to a costing problem.

The results show that domain referent knowledge influences participants' approaches to spreadsheet creation and spreadsheet usage. The conclusion is that spreadsheet error making is influenced by domain knowledge and domain perception.

Qualitative findings also suggest that spreadsheet error making is a part of overall human behaviour, and ought to be analyzed against this wider canvas.

Sample

Spreadsheet error analysis: Summary statistics
Spreadsheet error analysis: Summary statistics

The difference in the number of submitted Wall and Ball spreadsheets indicates that several participants found the Ball task simply 'too hard'.

Where both spreadsheets were submitted there was a difference in the number of errors between both tasks. However, this difference is not statistically significant.

Interestingly, only one participant completed both tasks correctly, whilst 8 participants got the Wall correct and the Ball wrong.

Publication

2003, EuSpRIG

Full article

The wall and the ball: A study of domain referent spreadsheet errors