A second screen design issue that Mick Couper and I took up in our GOR paper was Web page size. Some years back the most common monitor size was 800×600 pixels. Today we have larger monitors capable of higher resolution, and the most common size now is 1024×768 pixels, with some Web users on even larger monitors and some still back at 800×600. Understandably, Web survey designers like as large a screen as possible with which to work and there has been some movement to 1024 pixels wide as a standard, but this can create problems for respondents on smaller monitors who may be forced to scroll horizontally in order to view the entire question and answer it. One solution is to use software that will automatically adjust the page size to fit within whatever size window the respondent is using. (The window you are viewing right now has this capability. Go ahead and try it. Whereas many news sites like CNN use a fixed screen width.) The concern then becomes that some respondents will see the questionnaire differently than others–especially on long horizontal scales–and it may change how they answer. In addition, very long lines of text can be more difficult to read than shorter lines, a major reason why newspapers use columns rather than stretch text across the full page.
To add still another complication there is research to show that people with large monitors often do not work in maximized windows, that is full screen. So, for example, some with a very large monitor set to display at a resolution of 1280×1024 may mostly work in smaller windows set to 1000 pixels wide or less.
In our study we randomized respondents into one of three page size conditions: 800×600, 1024×768,and a 1024×768 design that would adjust to the size of the respondent’s browser window. And we used a questionnaire that we knew would be challenging because it had multiple screens with grids containing questions using both long (0-10) and short (1-5) scales. Our key findings:
- Respondents will adjust the size of their browser window–up and down–to fit the survey if they can.
- Respondents on monitors set to 800×600 had difficulty with the wider page size design, terminating more often, taking longer to complete, and indicating lower levels of satisfaction with the survey. In this survey about 16 percent of respondents were using these older, smaller monitors. In other surveys we have seen 30 percent or more in this category. These respondents tend to be older and lower SES.
- While we detected changes in response patterns to questions in wide grids these were mostly due to the age and SES bias described above.
Overall these results support the recommendation of Web usability guru Jakob Nielsen: "Optimize Web pages for 1024×768, but use a liquid layout that stretches well for any resolution, from 800×600 to 1280×1024." We didn’t find any reason to believe that the resulting differences in how respondents viewed the survey questions translated into different response patterns. In fact, we saw some evidence that respondents would change the size of their browser window to what they considered optimal viewing.