Technical Translation: Usability Strategies for Translating Technical Documentation

Task Times 
Total time taken to complete tasks
: A two-sided Wilcoxon-Mann-Whitney 
exact test of the null hypothesis of no difference between total times for the 
p-value of 0.0079
indicating that there is a statistically significant difference 
between the times for the Control group and the Experimental group. 
240
ing, the more detailed analyses we perform the more likely we are to
Control group and Experimental group was carried out. This yielded a 
Due to the limited sample involved in this study (n = 5), the usual asymp-
results which can look credible but which are in fact wrong or misleading. 
encounter a Type I error (Norman 2003). Type I errors refer to spurious

---

Results of Usability Evaluation
Figure 6: Average Task Times for Each Group in Main Study 
From the above graph it is clear that the Experimental group performed the 
tasks significantly faster than the Control group with the Experimental 
group completing tasks on average 44.4% faster than the Control group. As 
already explained, these figures do not include Task 3 which was the log-
ging task. The reason for this is that the task was of a fixed length, i.e. 302 
seconds. As such, including this task would achieve nothing other than in-
flating the group times of each group by the same amount.
An interesting issue is raised by the results for Task 4. When each task 
time was analysed individually, two-sided Wilcoxon-Mann-Whitney exact 
tests yielded p-values of 0.0079 for both Task 1 and Task 2. However, the 
p-value for Task 4 was 0.254, indicating that there was no statistically sig-
nificant difference between the average times for each. Indeed, Figure 6 
clearly shows that while there is a difference between the two groups, the 
difference is not as pronounced as in the other tasks. There are several pos-
sible explanations for this. Firstly, it is possible that analysing each task time 
individually resulted in a Type I error or spurious result.
Another, more probable, explanation is that, bearing in mind this task 
involved selecting a method for saving a completed log file to a specific lo-
cation, the task is not as complex as the other tasks and as such there is a 
smaller likelihood of serious problems arising. We could argue that this task 
had more to do with users’ knowledge of the Windows operating system 
than their knowledge of DigiLog and as such, this - and not the user 
guide - played a greater role in the outcome. Nevertheless, analysing the 
241

---

Assessing Usability 
overall time taken by each group to complete all tasks, we can see that the 
Experimental group performed the tasks faster than the Control group.
An interesting issue raised by the results for the main study is that the 
times are significantly lower than those observed in the pilot study. This can 
be attributed to the fact that, in comparison to the pilot group, the partici-
soft
were a close match to the profile of real users, the participants in the main 
study were an even better match.