This paper discusses the roles that devices which we term templates and tar
iffs can play in controlling empirical studies of programmers. Having been
involved in a number of empirical investigations of programming, some based
on observational studies of large groups and others based on protocols of
individual subjects, we briefly consider problems which are inherent in the
se various methodologies and consider how in our recent work, we have attem
pted to ameliorate some of these problems using templates and a number of t
ariffs. We outline the nature and mode of using these methods and suggest t
hat they can play important roles in giving us a better picture of program
construction. (C) 1998 Elsevier Science B.V. All rights reserved.