As you might have noted from one of my previous blogs (SUS), I am a big fan of usability analyses and statistics. One thing I have found time and time again, is that it can be very difficult to get participants to take surveys seriously. There often seems to be a sweet spot when the user has less than 5 questions, but over 10 questions and, whether you want to admit it or not, the number of bad data participants increases exponentially. That said, I came across a study recently by researchers at Intel that encountered a similar problem when attempting to incorporate a 10-item, five-point Likert scale called the System Usability Scale (SUS). To solve this, they decided to empirically reduce the number of questions by using only the highest correlating questions from a modified version of the SUS. The SUS was chosen because it had been extensively tested, and has proven to be a reliable representation of user satisfaction and system usability. However, by cutting down the number of questions from the SUS, it would also reduce the amount of data points, to help offset this, the Likert scales increased from 5 to 7 points. Increasing to a 7 point Likert was also heavily backed up by previous research (see Diefenbach et al. (1993), Cox (1980)). The results of the research was titled the Usability Metric for User Experience (UMEX).
It was concluded that the UMEX is a “reliable, valid, and sensitive alternative to the System Usability Scale.” It has a high correlation with SUS at that of above 0.80. Because of this, it is suggested that the UMEX is fully capable as acting as a standalone application for usability testing. More so, because of its compact size and ease of application, the Umex is attributed by a fundamental requirement by the Use Experience community: “in order to measure user experience effectively, its components need to be measured efficiently.”
I have not ever used the UMEX or anything like it. I can not say it should be used, I just think someone doing a quick and dirty study might like using it.