Warning - Statistics may be intentionally misused in this post.
LONG post ahead
This is a somewhat lighthearted look at the 2%/5% statistic - the success side of 92-95% of dieters regain the weight.
First of all, I'm going to use the 5% because it makes the math easier and because it better supports my end claim (hey, I did warn you.

)
This makes the claim "5% of dieters keep the weight off" and since I don't know what the original statement really is "5% of dieters keep the weight off for 5 years or more". Keep in mind 5% is 1 out of 20. (This is a true part, not any of the misused part.)
At any given time when 100 dieters are being interviewed (I'll say interviewed so I don't have to keep saying interviewed, filled out a questionnaire form, answered a survey on line, on the phone, etc.) . . . At any given time when 100 dieters are being interviewed, 5% or 5 of them are reporting success. At any given time when 20 dieters are being interviewed, 5% or 1 of them are reporting success. And keep in mind, they are being interviewed at a particular point in time, not 5 minutes after they stop "dieting" for the very last time.
The others are in various stages of: been on a diet now or in the past and didn't lose anything, didn't lose it all, lost some and gained it back, lost it all and gained it back, lost some and kept it off a short time, lost some and kept it off a long time, lost some and weight crept back up, etc. etc. Some have tried to lose it once, some on their second try, some on their 18th try, 19th, 25th, etc. Remember in the headlines, all those different results are lumped in the same category - didn't lose it all and didn't keep it off for 5 years. People who tried half-heartedly, didn't read the book, didn't really follow the plan, cheated every day, will still be reported as dieting - most likely unsuccessfully.
If the same people are surveyed a year later, some will have tried dieting again or continued dieting with the same or different results.
At some point some people will have given up, some will have tried less than 20 times and some 20 or more. But statistically they all will have been counted at some phase in their dieting life. Only 5% will have been counted at the "completion and maintenance" phase. You have to believe some will have learned something about weight loss and nutrition that they will apply to a future "diet" or will cause them some incremental success in their current "diet".
Now let's say enough time passes that everyone who is going to diet times has done so as many times as they ever will. They've either lost their weight and kept it off, lost some of their weight and kept it off, or are back at original weight or more. At this point the same group of 100 is reinterviewed. Assuming that this is the end stage of dieting, I think you'd have a much higher % of success. You're interviewing the folks at the end of dieting, not all the ones who haven't gotten there yet.
I'm relating this to the research concept that if you interview people in a store on one single day and ask how often they frequent that store, and you use those values straight up, your results are skewed.
How often do you shop?
Daily? 10
Weekly? 10
Monthly? 10
If you think from this that of 30 customers, on average 10 shop daily, you would be wrong. The 10 customers who responded daily will be the same 10 who will be there tomorrow. Assuming your customers are evenly distributed through the week, the next day the same 10 daily customers would be there, but a different 10 who shop weekly and a different 10 who shop monthly. In the end there would be 10 who shop daily, 70 who shop weekly and 300 who shop monthly.
So applying or misapplying that concept back to our 5% of successful dieters: In the end out of 20 people interviewed there are 19 who are on the journey to success and 1 who has reached it AT THIS POINT IN TIME.