Page 111 -
P. 111

4.8  Nonparametric statistical tests  97




                   Table 4.27  Sample Data for Mann-Whitney U test Continued
                   Group            Participants         Rating           Coding
                   Alphanumeric     Participant 4          4                0
                   Alphanumeric     Participant 5          3                0
                   Alphanumeric     Participant 6          2                0
                   Alphanumeric     Participant 7          4                0
                   Alphanumeric     Participant 8          5                0
                   Image-based      Participant 9          4                1
                   Image-based      Participant 10         6                1
                   Image-based      Participant 11         6                1
                   Image-based      Participant 12         7                1
                   Image-based      Participant 13         5                1
                   Image-based      Participant 14         6                1
                   Image-based      Participant 15         4                1
                   Image-based      Participant 16         6                1


                  U value. For this example, U = 10.5 and p < 0.05. Therefore, the null hypothesis is
                  rejected. The data suggests that there is significant difference in the level of perceived
                  frustration between the two authentication techniques. Participants experienced sig-
                  nificantly lower level of frustration when using the image-based password than the
                  alphanumeric password.
                     Let us examine another scenario in which you are interested in the use of the two
                  authentication techniques by people with Down syndrome. Since recruiting partici-
                  pant with Down syndrome from the local area is quite challenging, you only suc-
                  cessfully recruit 10 participants for the study. The small participant size suggests that
                  a within-group design will be more appropriate. So each participant completes the
                  study using both authentication techniques and answers a questionnaire after the in-
                  teraction with each technique. Sample data for the test is demonstrated in Table 4.28.
                  The mean score for the alphanumeric password is 3.9. The mean score for the image-
                  based password is 4.7. Since the data is collected from one group of participants, you
                  can use the Wilcoxon signed-rank test to examine whether there is significant differ-
                  ence in the perceived level of frustration between the two techniques.
                     The result of the Wilcoxon signed-rank test is a normalized z score. For this ex-
                  ample, z = −1.31 and p = 0.19. There is no significant difference in the perceived level
                  of frustration between the two techniques.

                   Table 4.28  Sample Data for Wilcoxon Signed-Rank Test
                   Participants               Alphanumeric           Image-Based
                   Participant 1                   5                     6
                   Participant 2                   3                     4
                   Participant 3                   4                     3
                                                                             Continued
   106   107   108   109   110   111   112   113   114   115   116