next previous contents
Next: Conclusion Up: Measures Previous: Compositional measures

Typing of attributes

We have also already noted that attributes are typed by the sort of values they may accept. We list here some typical examples.

Binary values

Values may be binary. For example, asking whether a spelling checker allows the user to create his own personal dictionary leads to an attribute whose value is simply `yes' or `no'.

Classificatory values

Values may be nominative or classificatory. For example, we might have an attribute which asked how a new word could be added to a spelling checker dictionary. The range of possible values would be inside checker, outside checker, not possible.

Comparative values

Values may be comparative. For example, there is evidence that `guessability' is an important factor in user acceptance of a new piece of software: if he can guess how a particular function works or what a particular icon stands for on the basis of his past experience without having to look it up, he will be happier with the software. It is hard to imagine a measure for guessability other than a simple comparative measure which awards, say, a score on a rating scale.

Numerical values

A value may also be numerical. For example, we might ask not only whether a user can define his own personal dictionary, but also how many personal dictionaries the spelling checker will accept.

Metric values

A value may also be a metric. For example, we might ask in what percentage of cases where an error is detected does the spelling checker give the correct suggestion as the first suggestion offered.



ceditor@tnos.ilc.pi.cnr.it