upa - home page JUS - Journal of usability studies
An international peer-reviewed journal

When Left Might Not Be Right

Xristine Faulkner and Clive Hayton

Journal of Usability Studies, Volume 6, Issue 4, August 2011, pp. 245 - 256

Article Contents


A total of 107 participants took part in the survey. There was a limited attempt to balance the genders by checking how many men and women had completed the experiment, and then attempting to redress the balance by approaching more of the required gender. There was no attempt to balance the direction of initial language acquired or handedness, because this would have required resources not available at the time. The survey ran during one afternoon only for four hours. It took place in one of the university laboratories set aside for this purpose. The experiment produced 107 questionnaires. The data was cleaned by removing any questionnaires that were not accompanied by a complete time log. Incomplete logs were missing a proper start or finish time. This was an issue that was encountered at the start of the experiment when the data for the log wasn’t collected correctly, and this accounted for 21 of the discarded questionnaires. One questionnaire was incomplete. No users abandoned the task because the questionnaire was completed only after the task (the purchase of a tree) was finished. Once the unusable data had been discarded, the collection provided 85 pieces of clean (that is complete) data that were then used in the analysis that follows.

The questionnaire measurement tool was supplemented by a single performance metric in the form of a time span for the completed activity. The timer for this started when the participant launched the homepage and stopped when he or she hit the confirm button after the purchase of their Christmas tree and related products. The confirm button also launched the online questionnaire. Users were asked about their background and experiences and answered two simple usability questions. Answers were provided via mouse click.

In all, a total of 41 participants used the right-hand version of the site, and 44 participants used the left, making a total of 85 participants. Of those, 36 were female and 49 male. Table 1 shows the breakdown for the participants and which version they saw of the Christmas Shop Web site, that is either the left-hand or right-hand version.

Table 1. Participant Gender and Menu Position

Table 1

Table 2 shows the average time taken in minutes by each participant on the two versions of the Web site.

Table 2. Average Times in Minutes Taken to Perform the Task by Gender and Menu Position

Table 2

As can be seen from Table 2, the males took slightly longer on both versions of the Web site (males 4.27 minutes left, 3.93 minutes right; females 3.35 minutes left, 3.74 minutes right), but the times are not widely dissimilar. However, females were faster on the left-hand version and males were faster on the right. Given the very small differences in timings, these findings are in line with other research, though they don’t support the contention that left-hand menus perform faster. In other studies, the findings have shown a bias towards the left-hand placement in terms of speed, though this has not always been a very large one (Kalbach & Bosenick, 2003; Kingsburg & Andre, 2004). Standard t-tests were carried out on this data and there was no statistically significant difference in performance between the right- and left-hand versions. For the male and female times this was t = 0.14, p <.89. For males this was t = 0.69, p < .49 and t = 1.11, p < .27 for the females.

Thus, it can be concluded that the difference between the two genders is very slight indeed and is not statistically significant. Thus the hypothesis was upheld. Gender has no impact on performance times produced by left and right justified menus.

Table 3 shows the breakdown of the participant group by age group, and Table 4 shows their performance time in minutes relative to their age group.

Table 3. Breakdown by Age Group

Table 3

The youngest age group performed the fastest on both the left and right versions, and indeed there is no difference in their performances. The speed gradually decreases through the next two age groups (22 to 30) and (31+). There is a slight difference between left and right menu versions, but for the middle group (22 to 30) the left-hand is faster and for the oldest age group the right-hand is faster. However, the number of participants in this final group is small (only 11 participants overall), and the differences between the performance times are tiny. There is insufficient data to make confident predictions as to what this might mean. To all intents and purpose, the menus perform in very similar ways and certainly do not perform differently enough to suggest that menus should always be on the left-hand side in cases where user performance time is not critical to the Web site and/or the task.

Table 4. Menu Position by Performance Time by Age Group

Table 4

The participants were asked about their Internet experience. This was divided into time spans according to when they first encountered the technology: primary school, secondary school, college, university. Table 5 shows the results.

Table 5. First Encounter with Internet Technologies

Table 5

Most of the participants (44) encountered the Internet at secondary school. The next largest group (24) encountered the Web at primary school, a much smaller number at college (13), and a very few at university (4). These findings correlate very closely to the age groups shown in Table 3. That is the younger the participant the more likely s/he was to encounter Web technologies at primary school.

Table 6 shows the average times taken to perform the purchase by first encounter stage. There is very little difference in performance for those who encountered the Internet at primary and secondary school (average 3.70 at primary school and 3.59 at secondary school), but performance times increases slightly for those who encountered the Internet at college (average 4.88) and university (4.69). The numbers of participants here are very small though (13 overall at college and a mere 4 at university), so not much can be stated with any certainty. It may indicate a trend and something that would be worth looking at in the future.

Table 6. Performance Times According to First Encounter with Internet Technologies

Table 6

There does not seem to be any major differences between the performance of the left and right versions. Incidentally, an identical question was asked about computer usage, that is, when the participant first used a computer. These figures were the same for Internet use so that it can be concluded that these particular participants encountered computers and the Internet at the same time. This would not necessarily be true of older users who are much more likely to have encountered computers before the advent of the Web but it probably reflects the demographics of this particular group of participants.

Average time spent on the Internet per day showed a similar trend to Internet experience with those using the Internet less frequently being slightly slower on both menus placements (3.92 minutes for the left menu and 3.97 minutes for the right menu). The time taken decreased for those who spent more time on the Internet (3.72 left and 3.74 right), but then increased at 11+ hours (4.69 left and 3.90 right). Again, though the number of participants is very small, 8 for 7 to 10 hours and 12 for 11+ hours, as in the earlier case it may indicate a trend. It would have to be tested again using a larger sample. See Tables 7 and 8.

Table 7. Average Internet Usage–Participant Numbers

Table 7

Table 8. Performance Times According to Hours Spent Online

Table 8

Disappointingly, the numbers of left-hand users who took part in the survey turned out to be very small indeed (6). In fact, the numbers are smaller than would be expected by chance so much so that the question over handedness was re-examined in order to consider the possibility that it had been misunderstood by the participants. See Tables 9 and 10.

Estimates for left-handedness seem to range from about 8% to 15% of the world population. (Hardyck & Petrinovich, 1977).  It could therefore be expected about 1 in 10 of the participants would be left-handed so the results below are lower than could be reasonably expected.

Table 9 shows the figures for participation by handedness. Table 8 shows the performance times.

Table 9. Participation by Handedness

Table 9

Left-handed participants took longer than right-handed participants, but the number of participants is too small to say much about that. No notice was taken of which hand they used for moving the mouse so there is no way of explaining that figure. Left-handed computer users quite often use their right-hand for the mouse and the left for typing. They can increase input speeds because using a mouse tends to be easier than typing. However, as this experiment did not require typing there would be no speed advantage over their right-handed counterparts.

Table 10. Performance Time by Handedness

Table 10

Tables 11 and 12 show the attempt to tie language acquisition as a factor in menu placement preference. This question tried to capture all of the right to left languages, but unfortunately had an “other” response to allow for languages that the questionnaire did not list—for example, languages that the questionnaire designers were unaware of. Consequently, a number of participants answered “other,” and there is no way of knowing if that was a right to left language or not. As can be seen, the number of participants who told us they learned to read a right to left language first was very small indeed.

Table 11. Direction of First Learned (Reading) Language

Table 11

Table 12. Performance Time by Direction of First Learned (Reading) Language

Table 12

Although it would be nice to say that menus are better placed on one side or the other, the differences in the performance figures are so slight that it is difficult to conclude that they have any real meaning. It would seem that once users know where the menu is they are quite able to operate either side. This was confirmed by the questions participants were asked about how easy it was to navigate on the site and how easy it was to purchase the products. These usability questions were offered to participants as a scalar question. The participants were asked to rate the ease of navigation and the ease of buying on a scale of 1 to 5 with 1 being easy and 5 being difficult.

Table 13. User Satisfaction with Menu Placement

Table 13

Table 13 shows the results of the user satisfaction portion of the questionnaire for all participants.

Those seeing the left-hand version rated navigation at an average of 1.70 and buying at 1.66. Whilst those seeing the right-hand version rated navigation at an average of 1.76 and buying at an average of 1.54. The statistical test confirms that there is no statistical difference between the two: for the left versus right menu placement for navigation this was t = (0.2003), p <.84; for males this was t = (0.2538), p < .80; and for the females this was t = (0.0000), p < 1.00.

Again the same statistical tests were carried out on male and female participants using the left and right menu placements for buying, and again there was no statistical significance in the findings: for the left versus right menu placement for buying this was t = (0.6279), p <.53; for males this was t = (1.5582), p < .13; and for the females it was t = (0.3847), p < .70.

More research would need to be done, but these findings are in line with other studies of left-hand versus right-hand placement (Kingsburg & Andre, 2004).


Previous | Next