March 22, 2007

User Research Doesn’t Prove Anything :: UXmatters

Samples and Stats

There's some good stuff in this UXmatters article: User Research Doesn’t Prove Anything. The sections on sampling and statistics are very nice summaries for UX practitioners. I've met many UX practitioners who either:
- never learned these subjects in school
- forgot the relevant lessons
- think that you can gloss over these points in business

Being accurate is important. There is a difference between saying "78% of users think…" versus "78% of the research participants think…”. Yet, I've seen people who think that in order to write things understandably in "plain language", you need to sterilize the message of anything sounding statistical. They try to convey the "jist" of the statistics without sounding like a statistician.

Sloppy research, sloppy statistics and inaccurate wording of findings will lead to haphazard business decisions. It's not good enough that you get your client to make the "right" decision. "Professionals" shouldn't have the attitude that the "ends justify the means".

We have an ethical responsibility to our clients, and to our users, to report accurate statistics from our research. You don't have to kill your clients with stats...just be accurate with the stats you report.

We should help our clients understand our research, the data, and the statistics that are used in the analysis. Not only will they better appreciate the skills we bring to the table, but they will be smarter clients when confronted by other opinions or seemingly contradictory "research". In short, be accurate, and educate your clients rather than dumbing things down for them.

Additional resources
- UPA Code of Professional Conduct
- Measuring Usability

No comments: