When should participants consist of a mix between genders?

People with a marketing background have been drilled to make sure there is a perfect mix of participants in the test sample to represent the larger user population they are interested in. In user experience, however, it is often possible to get by with a more lenient recruitment strategy. One of the criteria that can often be relaxed is gender.


Men and women can have different perceptions of a brand, different behaviors in their morning routines, different attitudes towards certain sports, etc., but they typically interact with technology in the same way. For example, if men struggle to understand the available balance in their checking account, we can be fairly sure that women would also have issues with this task. Similarly, if females are successful in making an purchase online, men are likely to succeed as well.

Nevertheless, there are some exceptions to this ‘rule’. I have conducted many usability tests and the only time I encountered variations in performance between genders was when we tested a setup of a hardware similar to a printer. Participants were instructed to unpack it from a box and set it up so that they could control it from their smartphone. The box contained an Ethernet cable that was not required for the set up (it was in the box because there may have been some odd cases when it might have been needed). We observed that women read the instructions before they started to plug in all the cables, and realized that the Ethernet cable was not necessary. Around half of the men ignored the instructions and immediately started to plug in all of the available cables, including the unnecessary Ethernet cable.

Some of the men just left the cable hanging from the printer like a tail, while others were determined to plug it in somewhere. One participant found our router (our lab equipment had been moved off to the side in the lab) and plugged it in there. To make sure it did not happen again, we had to hide the router. Another participant saw an Ethernet outlet on the wall and plugged it in there. We had to place a ‘post it’ note on the outlet to hide it. Another male participant plugged the Ethernet cable into the computer we had moved off to the side. We tried to block all of these inputs with cables. Another male came in with his own laptop (he just randomly had it with him) and plugged it into his own laptop. This kept on happening throughout testing and it seemed as nothing we did could prevent the male participants’ from trying to plug in the cable somewhere. We also tried to get the point across by wrapping a tag around the Ethernet cable that said “optional”. Unfortunately, even this was unsuccessful, and the tag ended up being torn off and tossed in the trash before the male participants even read it.

Another example of where gender may have an impact is with fashion or gender-specific items. For example, men probably don’t categorize shoes in as many narrow categories as women might. A man may look at a shoe and say it’s a “flat” or a “heel”, whereas some women can readily categorize a “heel” into “wedge”, “peep toe”, “espadrille”, “pump”, etc.

When recruiting users for testing, always try to get a mix of genders, but don’t put too much emphasis on it unless it is expected that there will be a difference in interaction style between the genders. There are often other criteria that are much more important such as technical expertise and domain expertise. In the shoe example, for instance, women are more likely to have more domain expertise, but there may still be some men who have equal expertise and would have the same type of interaction with the website.

© David Juhlin and www.davidjuhlin.com, 2015