08/03 - Fighting For Freedom Friday!

@ChrisTurk

can you calculate TV hourly as a function of total money earned as opposed to # of hits?

X% of total TV earnings were earned at < $7.25/hour etc.

 

Lets find out!

 

Just got one more Dolan. I only got one the first time around so this helps.

 

someone must of returned one. lucky

 

Disgusting:

$7,545 earned below min-wage
$93,089 earned at or above min-wage

$100,854 total earnings (base + bonus)

93% of earnings on TV have been above minimum wage. Totally fucking crazy dude, thanks for suggesting that method.

 

For people curious: they don't add up because of tasks reviews w/o an hourly rate (ie, completion time nulled out)

 

that's really cool

adjust for slowpokes/speed demons and then recalculate? TV 2.0 pls

% of earnings at > 12 or 15/hour ?

 

% of earnings made while eating a taco?

 

$55,365 > 15/hr
$69,647 > 12/hr

Seems like it makes far more sense to focus on the hourly rate a req pays on avg and attempt to match what TV suggests (or beat, obvi) than it does to putz around with low paying stuff.. I can't believe the % totals on that, they must be horribly skewed by some super high paying HITs or something.

 

Didn't we just skew it to >$20/hr with all those golabs?

 

The hourly doesn't matter, its the % of total earnings inside the higher hourly brackets that are crazy.

It makes some sense that working at a hourly hourly obviously earns you more money, but it doesn't (to me) make sense that it was be THAT crazy high % of TOTAL earned dollars, because its simply too high considering the total # of tasks at that hourly are a much lower percentage.

 

What I'm saying is that if you take the earlier table (25% of reviews are sub min-wage) & compare this, I'd expect to see ~25% [give or take] of total earned be for min-wage.

But its 7/93 instead of 25/75 which is.. idk statistically but just eyeballing it looks super lopsided towards working at the higher hourly.

 

Hey hey! Not all of you, and to be fair - I seem to have messed up too. Man, I'll just start at Turkerhub next time.

 

But, OTOH, it gives reason that a lot of low paying tasks ( total #) are easier to dip down into poor hourly and may be diluting the raw # of reviews at that range. Dunno, was just a half sec glance surprise at it all

 

lol, I was joking. that was a bad worker pool though

 

I never reject HITs in cases where people who were doing what they were supposed to would get rejected unfairly. My experiment is designed in a way that I cannot link the messages to your Worker IDs. Anonymity-preserving steps we take for IRB-approved research. So it was either approve all, or reject all.

 

I never reject HITs in cases where people who were doing what they were supposed to would get rejected unfairly. My experiment is designed in a way that I cannot link the messages to your Worker IDs. Anonymity-preserving steps we take for IRB-approved research. So it was either approve all, or reject all.

 

huh I haven't seen the forums double post in awhile.

@ChrisTurk is running that query to TV hitting the servers hard or something?

 

People aren't really doing low paying HITs that have TH/TV accounts. Unless you think @ThisPoorGuy or @A6_Foul_Out are going to spam TV with Breligs and Ibotta reviews. Also, the nature of TVJS doesn't really allow for batch aggregation. People aren't reviewing low paying batch hits they do after learning of their greatness on r/mturk.

The data skews high because it's from us. Basically our PDO is really high right now and it won't return to the mean value until the average turker starts reviewing HITs.

 

It was quite a mess, sorry about that! Let's all hope it goes better next time.