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

Results

Our project collected information on task error rates, performance time and subjective satisfaction, which we imported into Microsoft Excel for analysis. We did statistical analysis of time to completion and number of errors for the tasks. Each subject performed 6 tasks -- two experiments with 3 treatments in each experiment. For each experiment we used both a one-way ANOVA and t-tests if necessary to analyze the results. We also did a statistical analysis of subjective satisfaction using the results of the post-experiment survey.

Error Rate

A single factor ANOVA on the error rate for the one page experiment showed that there is no relation between the number of errors and the treatments. This result was not statistically significant at alpha=0.05 (f(2,51) <0.001, p > 0.999 ). The t-tests were not necessary since there is no statistically significant difference in the ANOVA's result.

A single factor ANOVA on the error rate for the multiple pages experiment was run to determine whether as the devices differed, the number of errors differed. The result was not statistically significant at alpha=0.05 (f(2,51) = 1.58, p = 0.22). The t-tests were not necessary since there is no statistically significant difference in the ANOVA's result.

The means and standard deviations of the error rates in both experiments are displayed in Figure 1 below.

Tables for error rate of one page experiment
Tables for error rate of multiple pages experiment

Raw Data

Performance Time

A single factor analysis of variance (ANOVA) on performance time for the one page experiment showed that at the smallest screen size, performance time increased. This result was statistically significant at alpha = 0.05 (f(2,51) = 4.07, p = 0.02). A set of paired t-tests were then run to determine whether the time was statistically significant between Palm Pilot and Rocket Book, Palm Pilot and Laptop, and Rocket Book and Laptop in the treatment. The result of the t-test between Rocket Book and Laptop was not statistically significant. The result of the t-test between Palm Pilot and Rocket Book was significant at alpha=0.05 (t(34) = 2.46, p = 0.02) and the result of the t-test between Palm Pilot and Laptop was significant at alpha=0.05 (t(34) = 2.11, p = 0.04).

A single factor ANOVA on performance time for the multiple pages experiment was run to determine whether as the device differed, the performance speed also differed. This result was not statistically significant at alpha=0.05 (f(2,51) = 2.84, p = 0.07). A set of paired t-tests were then run to determine whether the difference in performance rates was statistically significant between Palm Pilot and Rocket Book, Palm Pilot and Laptop, and Rocket Book and Laptop in the variable treatment. All t-test results were not statistically significant at alpha=0.05 (t(34) = 2.00, p = 0.054) for the Palm Pilot and Rocket Book,
(t(34) = 1.96, p = 0.058) for the Palm Pilot and Laptop, and (t(34) = 0.40, p = 0.69) for the Rocket Book and Laptop, respectively.

The means and standard deviations of the performance times in both experiments are displayed in Figure 2 below.

Tables for performance time of one page experiment
Tables for performance time of multiple pages experiment

Raw Data

User Satisfaction

A single factor ANOVA on the six user satisfaction topics showed that as the device differs, user satisfaction ratings for overall difficulty, scrolling/paging difficulty, screen size, font readability ,and portability differ. However, the ease of use does not differ as the device differs. The results for all user satisfaction topics except the ease of use are all statistically significant at alpha=0.05 (ease of use: f(2,51)=2.48, p=0.09; overall difficulty: f(2,51)=3.53, p=0.04; scrolling/paging difficulty: f(2,51)=3.19, p=0.05; screen size: f(2,51)=31.81, p<0.001; font readability: f(2,51)=35.89, p<0.001; portability: f(2,51)=22.26, p<0.001). Five sets of paired t-tests for those topics with statistically significant ANOVA results were then run to determine whether the differences in pairs of devices were statistically significant between Palm Pilot and Rocket Book, Palm Pilot and Laptop, and Rocket Book and Laptop in the fixed treatment. For overall difficulty, the result of the t-test is statistically significant between Palm Pilot and Laptop at alpha=0.05 (t(34)=-2.34, P=0.03). For scrolling/paging difficulty, the results of the t-tests are statistically significant between Palm Pilot and Rocket Book at alpha=0.05 (t(34)=-2.38, p=0.02) and between Rocket Book and Laptop (t(34)=2.35, p=0.02). For screen size, font readability and portability, all t-test results are statistically significant between Palm Pilot and Rocket Book, Palm Pilot and Laptop, and Rocket Book and Laptop. The detailed calculations are in the appendices.

The means and standard deviations of user satisfaction for three devices are displayed in Figure 3 below.

Tables for user satisfaction

Raw Data

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