Instead of giving in to a gut reaction, you should read the paper. All research must get funding in one way or another. If you exclude research due to funding sources and potential conflicts of interest, you wouldn't be left with much.
If the majority of research is tainted by conflicts, that doesn't somehow make it any less tainted by conflicts.
It's trivially easy to design a study to produce the results you want it to, just by selecting things like accounting methods at the outset.
For example, one method of accounting is to allocate costs based on usage. So if you have a car and you drive it 90% for Uber and 10% for your own usage, allocate 90% of all costs to Uber. That makes Uber look really bad.
But if you would have bought the car either way and the question is the incremental cost of driving for Uber, that accounting method is the sunk cost fallacy. It allocates large fractions of a bunch of fixed costs to the incremental use, like the original purchase price of the car, even though they're sunk and can't be avoided either way.
And the money actually does taint the study, because the funders know the parameters of the study ahead of time and only fund the ones that will produce the results they want, which makes the results tainted by selection bias.
> Of the five sources of cost estimated per mile (Insurance, Maintenance, Repairs, Fuel and Depreciation), approximately 40% of costs are attributable to Insurance, Maintenance and Repairs, 40% to fuel expenses, and 20% to depreciation.
That's because -- and this is the real problem -- finding the flaw in a specific methodology is a lot of work. It's often not just one thing that throws the numbers off by 200%, it's one thing that throws them off by 12%, then another by 7%, and twenty other little things that add and multiply up to an inaccurate overall conclusion. It can take multiple hours to figure out what actually happened and people don't have that kind of time, so the majority of people only have time to read the conclusion and who sponsored it.
In theory the solution to the problem of everyone having to personally evaluate every study is for a trustworthy reporter to do it for you, but that hasn't worked ever since reporters figured out that "Uber eats souls" gets more clicks than "wage study methodology miscalculates wages."
But if you want a real flaw in this study, how about this one -- most other jobs don't let you count transportation expenses against the hourly wage. Obviously an Uber driver's transportation expenses will be higher, but that doesn't mean you can discount them in the other cases when you count them in this case.
If someone has a 30 mile commute (because living closer to work is even more expensive) at $0.53/mile, that's 60 miles a day, so $31.80 in daily incremental expenses to take the job. 8 hours at federal minimum wage ($7.25/hr) is $58, so after transportation expenses their hourly wage is $3.28/hour. For someone with a 50 mile commute it's $0.63/hour. Suddenly Uber doesn't look so bad.
Good job in pointing out this huge flaw in this study. Every minimum wage job I ever worked required me to have a vehicle to get to and from that job. That means that every minimum wage job out there where you need to pay for transportation to get to it is paying below minimum wage as the end the day. Those who live in cities have the benefit of public transportation but also have the time cost of being slower and reduces the number of waking hours they can earn money. What's the hours lost commuting averaged across and subtracted against the number of worked?
Does their methodology ascribe only the incremental aspect of each of these costs?
If I own a car already for my own personal use. I'm going to deal with all of these costs. You're already going to have insurance, maintenance and repairs, they'll just be more frequent. Fuel expenses is the only input that can be accurately estimated and attributed to ridesharing work performed. Even this contains a huge variable. Did they account for drivers using hybrid vehicles, which more and more ridesharing drivers opt for because it's significantly more profitable? Is the median TNC driver using a more fuel efficient vehicle than the one they accounted for in their calculations?
The fact that the study doesn't have an entire section that I found entirely devoted to potential problems with their methodology that we are pointing out here is reason enough to discount it as tainted after also considering its funders.
They're clearly trying to gauge incremental costs, not fixed costs, and each of insurance/maintenance/repairs/fuel/depreciation are incremental. The breakdown between the personal and ride-hail vehicle use is part of the study and appears in the sentence previous to the quoted one as "the vast majority of drivers report that the bulk of the miles they drive are for ride-hailing".
In that case, it's grossly irresponsible to publish what little they have. Society is still struggling to correct the "statistic" that a woman earns 70 cents for every dollar a man earns. This new statistic about how much TNC drivers make is going to be accepted as fact from here on out, when it's grossly misleading.
> "the vast majority of drivers report that the bulk of the miles they drive are for ride-hailing"
That doesn't change the fact that the vast majority of the drivers would already have that car as a sunk cost whether they drove the vast majority of those miles for ridesharing or not.
A car has a resale value. That resale value decreases with each additional mile on the odometer. That's the depreciation incremental cost.
A car requires insurance. Insurance rates fluctuate based on miles driven per period (and also based on whether the car is used in ride-sharing).
A car requires maintenance and repairs are incurred on a per mile basis. For example, when a $30 oil change is needed every 3K miles with 300 miles of trips to the grocery store and 2700 miles of Ubering, it'd be asinine to assign that cost in a way other than $27 of that going to Ubering.
A car requires fuel. The same logic as the oil change; $30 for gas to drive 30 miles to the grocery store and 270 for Uber can be uniformly distributed per-mile.
Is everyone supposed to personally fully read every individual scientific study? That doesn't scale.
The only way it works is to have experts you trust to give you the truth, which is what the scientists are supposed to be, but the funding source undermines the trust.
But if the topic interests you enough for you to comment about it and there is a research paper that you can read, then just read the dang paper or let other people make decisions for you. Nobody has to give you the truth. And, on top of that, even if the results of research are quantified, there may not even be an unquestionable “truth” — the implications of the research can still be up to interpretation.
Then you have to figure it out for yourself. For everything. Which, again, doesn't scale.
At some point you do need to trust some people to make some decisions for you. But what kind of fool trusts someone with a conflict of interest?
Science isn’t a matter of trust, it’s a matter of verifiable results. The results of a scientific research paper must be repeatable by a third party following the same methods. Go see the merits of the research: it’s methodology, its results. If everyone simply looked at a list of funders and went “welp... this research is biased and invalid”, then we would have hardly any “valid” research.
It makes sense to try and figure out what the funders would want the answer to be and see if the research matches that. If it does, shouldn't you be more skeptical?
For example, if Exxon funded research saying that global warming isn't caused by man or Walmart funded research saying that raising the minimum wage causes mass unemployment then it's pretty clear that they would benefit from the result.
In this example it looks like these companies probably just want to be associated with MIT (or they have an interest in other pieces of research) and I can't see any particular reason to suppose that they'd be invested in the outcome of this study in particular.
A) Not always. Bill Gates funding malaria research doesn't benefit him.
B) If you benefit from the research being accurate and truthful that's very different to only benefitting if the research yields a particular answer.
This sort of behavior should not instill confidence in the paper's conclusions. And I think it was completely reckless of the Guardian to run an eye catching headline with absolutely 0 justification given to the numbers other than an implicit appeal to authority by calling it an MIT paper. Though such is the state of the media today.
The paper is fairly open about their methodology, and give access to things like the actual survey being answered, rather than just saying "a survey matching these criteria was used".
Most of their data apparently has come through Harry Campbell [1], and his yearly survey, which has it's own data and methodology out in the open [2].
I'm not remotely qualified to analyse it, but you can see the data from therideshareguy's survey if you're looking to critique the results of this paper. (I've edited this paragraph to try and get less push/offensive. I'm still not happy with it, but the aphasia is kicking my butt. Rest assured, I'm just trying to be helpful, and just wanted to point the direction to discussing the data. Still not happy with this phrasing either.)
[0] http://ceepr.mit.edu/files/papers/2018-005.pdf