No matter how I look at it, the whole rating system - for both paxs and drivers - is horribly flawed

The values that Uber uses for its keep/kill decisions is completely irrelevant because the paxs rate the drivers on an interval scale, but then Uber attempts to apply that interval data to an ordinal scale. Any first-year Stats student can explain to you the folly in that. You have two completely different scales that have not a thing to do with one another.

Uber is fully aware that the rating system is flawed and statistically irrelevant. I know that because I have explained it to a couple Uber execs in the past six months. But because no one in the Uber exec ranks has any grounding in stats, they are laboring under the belief that they have developed and implemented a valid driver rating system.

So, here’s all you need to know about ratings. For Uber, the keep/kill threshold is 4.6, or so we’re told. Because the rating system is statistically insignificant, a rating of 4.61 is exactly the same as a rating of 5.0. Likewise, a rating of 4.59 is exactly the same as a rating of 1.0, 2.0, 3.0, etc. You get the picture.

Someone gave me good advice never ever pick up a pax with 4.6 and under and they where right…they have a low score for a reason…and they will bring your scores down…just started myself driving uber…you have to start out strong and maintain your high rating until it does matter anymore…I have a 4.86 with 125 trips…last week hit 4.91 for the week.

I keep seeing people saying that nce you get a lot of rides under your belt, a bad rating doesn’t matter. This unfortunately is untrue!! In the wonderful Fuber rating system, they monitor your last 100 rides.

I think you guys are right that I made too big a deal of this, especially to the pax, should have just downplayed it instead of hyping it up. Glad to see some people saying it was no big deal, I was feeling pretty frazzled afterwards.

You can’t. You can just guess/assume based on your rating.

I think I was actually wrong about that girl giving me a *1

Also I’ve seen my overall ranking increase directly after a good first ride of the day while my 1 day has remained showing 0.0 - so I think Uber does some extra stuff to obfuscate any possibility of gaining direct info / thwart any attempts at correlating ratings with individual riders.

You feel bad? For the rate she was paying in a storm she got more than she deserved. Don’t ever apologize. It just makes you look weak. Drivers look weak enough just driving for the Uber gig already. Don’t depreciate yourself more than you already have.

Regardless of what your overall average is, if you drop below 4.6 on your last 100 rides (in a mature market!) they send you the warning email.

So, all you have to be concerned about is staying at or above 4.6. It makes no difference at all how far above 4.6 you are. A rating of 4.61 in this statistically flawed system is the same as a 4.9.

All we really have here is a crudely developed PASS/FAIL system that Uber believes produces meaningful data, but that does nothing of the sort.