previous
section  |
 |
next
section  |
As is described in section 3, there were no statistically significant results for any of our hypotheses. For our hypothesis on task time, this simply is consistent with our expectation that there shouldn't be any statistically significant results. However, our other two hypotheses were that there would be statistically significant differences and there were not. The most likely reason for this is the dependent variables we chose to measure errors and retention, respectively. In both cases our measurement was indirect.
We used the total number of hyperlinks followed as the dependent variable measuring error rates. However, observing subjects at work showed that many of the false trails followed were due to ambiguities in where one would look for the answer to a question, rather than errors in carrying out navigation tasks. These ambiguities are discussed in section 4.2.
Similarly, we measured retention by asking what cities appeared on the "Main Cities" page under Population. We were hoping that subjects would remember all four cities on that page, but it turned out that the most common pattern of answers was to circle only Bergen and Oslo. The series of questions in the experimental task included the population of these two cities, so it is natural that they would remember these two.
Section 5.2 suggests refinements that address these and other flaws in the experimental design.