Human Error in SQL Queries

Ray Panko's Human Error Website

Overview

When people engage in simple but nontrivial actions, they usually have error rates of about 1% to 5%. When they engage in more complex cognitive activities, error rates are higher, sometimes much higher. SQL queries are a good example of this. SQL queries have a number of parameters and can incorporate Boolean operations.

Although e+xperiments used college and sometimes high school student, the Chan, Lu, & Wei [1993] survey showed that most professional SQL users do not get a query right the first time they try.

More complex cognitive actions may represent a chain of simpler cognitive actions with lower error rates. The probabability of an error increases as the cognitive action chain length increases because probabilities multiply.

Smelcer (1995) only looked at a particular type of error. However, he measured error rates based on cognitive load. For low memory load, the error rate was only 6%. This tripled under high memory load.

Study Detail Error Rate
Chan, Lu, & Wei [1993] Percentage of 136 surveyed professional SQL users who say they usually take more than one try to do a query. 82%
Gould Cited in Reisner, 1981. Query specification errors in QBW, 39 high school and college students. Per query. 33%
Greenblatt & Waxman [1978] Query specification errors, paper and pencil exercise. Per query. 25% - 27%
Reisner [1975] Query specification errors on final exam, SEQUEL, programmers. Per query. 22%
Smelcer [1995] 20 undergraduates with 80 minutes of training in SQL. Only counted errors in which a required join was not used. Percentage of queries with such errors. 6% if low memory load, 12% if medium memory load, 17% with high memory load. Per query. 14%

References

Chan, H. C., Lu, H. J., & Wei, K. K. (1993). A Survey of SQL Language. Journal of Database Management, 4(4), 4-15.

Reisner, P. (1981). "Human Factors Studies of Database Query Languages: A Survey and Assessment." Computing Survey, 14(1), March, pp. 13-31.

Smelcer, J. B. (1995). User Errors in Database Query Composition. International Journal of Human-Computer Studies, 42(4), 353-381.

Reisner, P., Boyce, R. F., & Chamberlin, D. D. (1975). Human Factors Evaluation of Two Data Base Query Languages--SQUARE and SEQUEL. Proceedings of the National Computer Conference, May 19-22, Arlington, Virginia, pp. 447-452