How much money can you make on Mechanical Turk? • Hacker News Board

Re: How much money can you make on Mechanical Turk?

#41

Post by Phemist » Mon, Jul 15, 2019, 9:58 AM UTC

Fun anecdote about some of my own dabblings in online psych experiments and MTurk.

7 or 8 years ago, my then supervisor and I experimented with running psychological (behavioural) experiments on MTurk. We had created a method that would run in browsers in native javascript and were looking to validate it. Coming from a testing-1st-year-psych-students-in-the-basement-of-the-faculty-building kind of thing, we naively took what the going wage was for that (something like 8 euros/hour), and put an experiment online that would take around 20 min to complete, and paid 2 euros.

My supervisor used his own credit card and set the spend limit to around 1000 euros; thinking that we'd never hit it. Boy how wrong were we. Apparently this 6 euro/hour wage was _much_ _much_ higher than the going rate, and we hit the spend limit in around 2 hours. Even though we had to throw out around 70% of the completions, we ended up with usable data from around 150 participants.

We went in expecting to run the experiment for a week or 2, and get maybe 50 participants, but came out with 500 in a span of 2 hours. Safe to say, we celebrated a job well done that night over one or two drinks. People were commenting on how nice of a change of pace it was compared to the then-usual MTurk tasks and that they would have done it for free. Some even left their e-mail addresses should we run another online experiment. I've since been out of the field of online behavioural experimentation, but it seems to have taken off quite a bit.

Re: How much money can you make on Mechanical Turk?

#42

Post by blintz » Mon, Jul 15, 2019, 10:06 AM UTC

How do people feel about the research ethics of this? Many academic studies involve the use of MTurk participants, but it seems as though they are not being paid close to a living or even fair wage ($2 an hour is abysmal!).

Should universities require that studies using MTurk pay some minimum pro-rated hourly wage? That is, if a Stanford HCI study (which seem to use MTurk often) wants workers to fill out a 15-minute survey, they would need to pay at least $8 an hour (so $2).

I’ve been to thesis defenses at Stanford where researchers have explicitly stated that they chose to use MTurk because they can pay workers far less than locally recruited participants.

Re: How much money can you make on Mechanical Turk?

#43

Post by dgivney » Mon, Jul 15, 2019, 10:07 AM UTC

A somewhat interesting article which strongly turns into a puff piece of MT. I started scrolling to the bottom to find out who was paying for this content and there it was "In his book, Side Hustle From Home: How To Make Money Online With Amazon Mechanical Turk"..

I wonder just how does he manage a new baby, a side hustle that fills every minute with productive money-making and writing a book!.. The author really should be writing a book about his exceptional time management skills instead.

Re: How much money can you make on Mechanical Turk?

#44

Post by wsc981 » Mon, Jul 15, 2019, 10:18 AM UTC

blintz wrote: post #42

How do people feel about the research ethics of this? Many academic studies involve the use of MTurk participants, but it seems as though they are not being paid close to a living or even fair wage ($2 an hour is abysmal!). Should universities require that studies using MTurk pay some minimum pro-rated hourly wage? That is, if a Stanford HCI study (which seem to use MTurk often) wants workers to fill out a 15-minute…

2 USD per hour can be decent depending the location. Where I live in Thailand people on average make 7 USD per day.

Re: How much money can you make on Mechanical Turk?

#45

Post by turtlecloud » Mon, Jul 15, 2019, 10:37 AM UTC

blintz wrote: post #42

How do people feel about the research ethics of this? Many academic studies involve the use of MTurk participants, but it seems as though they are not being paid close to a living or even fair wage ($2 an hour is abysmal!). Should universities require that studies using MTurk pay some minimum pro-rated hourly wage? That is, if a Stanford HCI study (which seem to use MTurk often) wants workers to fill out a 15-minute…

Globalization means a race to the bottom for the value of labor. $2/hr is great for a dude in India.

Re: How much money can you make on Mechanical Turk?

#46

Post by DoreenMichele » Mon, Jul 15, 2019, 10:40 AM UTC

ptero wrote: post #19

The person described is a "senior business analyst", reportedly making good money, but decides to work on MTurk, doing menial tasks, to help offset extra costs (diapers, etc.) of a baby that was born. This sounds insane to me. Almost anything else sounds better: reduce expenses, figure out if he can get more responsibilities and a raise, even get a low % credit to pay for those diapers for a year, etc. . Is this just…

I guess the article doesn't say it well, but this can be done in small scraps of time from wherever you currently are under circumstances where making money with a "regular job" would not be possible. If you make, say, an extra $5/day on what used to be your smoke breaks and lunch hour at work for an 8-5 job, you have an extra $180 or so a month without any additional overhead (for uniforms, whatever), scheduling conflicts from a second job, additional time taken away from family, etc.

Mechanical Turk generally sucks as a substitute for a regular job, but can make sense as supplemental income, even at nominally very low hourly pay. If you work 8 hours at $10/hour but have a 30 minute commute, you are really getting $80 for 9 hours of your time. That actually puts you below $9/hour to think of it that way.

If you do freelance work, iirc, freelancers chase their pay about 40% of the time. That also drives their real wages down. Plus, there is time involved in getting each assignment.

I have read that you can expect to do one unbillable hour for every billable hour, so you need to charge at least twice as much to make the same wage. Plus, for jobs with benefits, roughly half your compensation can be in the form of benefits.

So freelancers should charge four times as much as the hourly rate they would accept at a salaried job with benefits.

But if you already have benefits and just want to supplement your income, you can accept half as much. If you can eliminate some of the time burden of freelancing, you can halve it again.

Looked at that way, $2/hour is potentially the equivalent of $8/hour as supplemental income. And you can fit it in to otherwise useless scraps of time.

Re: How much money can you make on Mechanical Turk?

#47

Post by ForHackernews » Mon, Jul 15, 2019, 10:43 AM UTC

Thorrez wrote: post #5

senectus1 wrote: post #3

wow, those number suck hard . I wonder what Nationality these "Turker" are from...

> Despite an uptick in foreign users over the past few years, the majority of them (75%) are based in the US.

America is becoming like Brazil: a third world nation with some wealthy enclaves.

Re: How much money can you make on Mechanical Turk?

#48

Post by dsr_ » Mon, Jul 15, 2019, 10:44 AM UTC

wsc981 wrote: post #44

blintz wrote: post #42

How do people feel about the research ethics of this? Many academic studies involve the use of MTurk participants, but it seems as though they are not being paid close to a living or even fair wage ($2 an hour is abysmal!). Should universities require that studies using MTurk pay some minimum pro-rated hourly wage? That is, if a Stanford HCI study (which seem to use MTurk often) wants workers to fill out a 15-minute…

2 USD per hour can be decent depending the location. Where I live in Thailand people on average make 7 USD per day.

Do the people in Thailand who make $7/day generally have reliable access to the Internet for 8 hours a day, with a cost for network access and smartphone under $8/day? And a ready source of loans to cover the initial purchase? And can they read English?

Re: How much money can you make on Mechanical Turk?

#49

Post by sureaboutthis » Mon, Jul 15, 2019, 10:46 AM UTC

blintz wrote: post #42

How do people feel about the research ethics of this? Many academic studies involve the use of MTurk participants, but it seems as though they are not being paid close to a living or even fair wage ($2 an hour is abysmal!). Should universities require that studies using MTurk pay some minimum pro-rated hourly wage? That is, if a Stanford HCI study (which seem to use MTurk often) wants workers to fill out a 15-minute…

No one is forcing anyone to do this.

Re: How much money can you make on Mechanical Turk?

#50

Post by stubish » Mon, Jul 15, 2019, 10:51 AM UTC

blintz wrote: post #42

How do people feel about the research ethics of this? Many academic studies involve the use of MTurk participants, but it seems as though they are not being paid close to a living or even fair wage ($2 an hour is abysmal!). Should universities require that studies using MTurk pay some minimum pro-rated hourly wage? That is, if a Stanford HCI study (which seem to use MTurk often) wants workers to fill out a 15-minute…

I wonder if they actually get meaningful results from MTurk participants; you are getting a very skewed and unnatural demographic, and I don't see how any results based on feedback from MTurk participants can be applied to any normal population. Much more so than the traditional study group of students-who-need-to-pay-rent.

I don't see an ethical problem in how much you pay participants (after all, how much it pays or if it pays at all will be one of the variables the study needs to account for). The ethical problem I see is not caring if the results are meaningful, just publishable.