The Effect of Screen Size on Readability Using Three Different Portable Devices

Discussion

Error Rate

The results showed that there was no statistically significant difference in the error rates among the three different treatments for the one page experiment or the multiple pages experiment. Therefore we can conclude that error rate is not affected by screen size when dealing with either one page of text requiring no scrolling/paging, or with multiple pages of text. Our hypothesis, which stated that error rates would increase as screen size decreased was therefore invalid.

Performance Time

Our results for performance contradicted our hypothesis. Our hypothesis stated that for a one page test there would be no difference in performance times among the devices. However, for the one page experiment, our results showed that screen size influenced performance time. Reading on the Palm Pilot was slower than on the other two devices. This difference was statistically significant. For the multiple pages test our hypothesis stated that as the need for scrolling or paging increased performance time would increase. However, our results did not show any statistically significant difference in performance time for the multiple pages experiment. Therefore, the results show that the hypothesis is incorrect.

We attempted to determine why our hypothesis failed in both cases. We believe that the failure in the one page test may be related to font readability. As can be seen in the analysis of user satisfaction, the t-tests done on font readability produced statistically significant differences. These differences in preference may be reflected in performance time as well.

The failure in the multiple pages is more complicated to explain. We had an outlier in our data. This subject, was both elderly, required glasses to see, and had no previous computer experience of any kind. Her error rates were higher, and her performance times were much longer than other subjects. We suspected that if we recalculated the ANOVA without this outlier we might have different results. In fact, the ANOVA was statistically significant at alpha=0.05 (f(2,48) = 3.34, p = 0.04). A set of paired t-tests were then run to determine whether there were statistically significant differences between devices. The t-test was statistically significant at alpha=0.05 (t(32) = 2.21, p = 0.03) for the Palm Pilot and Laptop. The complete results of the ANOVA and t-tests are available here. Recalculated ANOVA and t-tests. These results suggest that the outlier did, in fact, affect our results. Without the outlier, our hypothesis was supported for the multiple pages experiment. As the need for paging or scrolling increased, performance time increased. A larger number of test subjects might have allowed us to support our hypothesis without having to remove an outlier. We also considered that this outlier might have affected error rates as well, but no statistically significant differences were discovered.

User Satisfaction

We asked six questions on the User Satisfaction form using a scale from 1 to 9, where one was the most difficult or least preferred and 9 was the easiest or most preferred. The six questions are related to "ease of use", "overall difficulty", "scrolling/paging difficulty", "screen size", "font readability" and "portability".

No statistically significant differences appeared in the results of the "ease of use" question. This suggests that users found all three devices similar to use. However, in a paired question, "overall difficulty", users found the Palm Pilot harder to use than the Laptop. The t-tests for the Palm Pilot and the Laptop showed a statistically significant difference in user satisfaction. These two results suggest that while users found all three devices "easy" to use, they found the Palm Pilot more "difficult" to use. We draw two conclusions from this result. First, a smaller device is more difficult to use even though it is straight forward and second, it is important to word user satisfaction questions carefully.

We found a statistically significant difference in the t-tests for "scrolling/paging difficulty" between the Rocket Book and the other two devices. Users preferred the paging mechanism of the Rocket Book. One possible reason for this preference may be the size and position of the paging buttons on the Rocket Book as compared to those on the Palm Pilot and the Laptop.

All three t-test pairs showed statistical significance for screen size and portability. It is very apparent that the size of the screen is the major difference among the three devices. The Palm Pilot has the smallest screen size and the Laptop has the largest screen size. The t-test results for portability are just the opposite. The Palm Pilot is considered the most portable, a direct result of its size, and obviously, the Laptop is considered the least easy to carry, again a direct result of its size.

All three t-tests showed Font readability was statistically significantly different among these three treatments. We made a sincere effort to normalize the font size among the three treatments, and also made an effort to maintain similar light levels on all three devices. However, users felt that the Palm Pilot had the most difficult font to read. This result may be directly related to screen size. Since the screen size of the Palm Pilot is very small, the readability may be influenced by the screen size. The readability for the Rocket Book is close to that for the Laptop. The differences may be due to the contrast factors of the screens. Though the screen size of the Rocket Book is much smaller than the Laptop ( still significantly larger than the Palm Pilot), the contrast may contribute to the readability of the Rocket Book. The Rocket Book is designed for reading and it seems to have good readability according to our results.

At the end of the satisfaction survey we asked subjects for written comments on which devices they liked the most and least, and why. Frequently, subjects discussed all three devices. The user comments can be seen under User Satisfaction in the appendix. the most interesting comments suggest that users liked the laptop best in their comments, but then related that preference to their knowledge of the laptop's power and versatility. They did not base their decision on the experiment recently completed. Users who preferred the Rocket Book discussed ease of use and convenience. One user who preferred the Rocket Book had actually downloaded and read books on his Palm Pilot.

Department of Computer Sciences: Direct questions and comments to the student editorial team