Unless I'm missing something, this article isn't even beginning to live up to it's headline, doesn't disclose it's data sources for Tesla fatalities, and seems like someone with an ax to grind or...
Neither IIHS nor NHTSA have yet published summary quantitative real-world fatality data on Tesla cars.
Unless I'm missing something, this article isn't even beginning to live up to it's headline, doesn't disclose it's data sources for Tesla fatalities, and seems like someone with an ax to grind or who wants to make a name for themselves. According to the author's bio on medium they don't have any formal experience with the auto industry. The first third of the article deals with SolarCity's problems...
If I'm being generous this article makes a case for why IIHS and NHTSA should be looking at Tesla in their stats.
EDIT: Not an attack on OP at all. I have posted some things that looking back might not have been so great myself. Just a heads up to anyone who might be thinking about devoting time to this "article"
EDIT2: @shiruken pointed out that they do disclose there data for the Tesla crashes. My bad!
Probably worth nothing the author is a hedge fund manager that is short on Tesla That being said, they do list their sources for the Tesla fatalities There were a total of 11 fatalities through...
Probably worth nothing the author is a hedge fund manager that is short on Tesla
Lest anyone believe that because we are short Tesla shares, our analysis somehow doesn’t matter, it’s important to note that we are short Tesla precisely because of our open-minded analysis of the stock’s prospects.
That being said, they do list their sources for the Tesla fatalities
To build our list of Tesla driver fatalities, we searched the NHTSA-FARS database using various methods, including for vehicles of make 29 & model 5 (Tesla), and separately for VINs beginning with 5YJ. This search found five Tesla driver fatalities. We conducted a Google search for Tesla driver fatalities reported by English-language sources, using a variety of search terms, and reviewed and fact-checked the crowdsourced Tesla mortality file maintained by ElonBachman (EB) on Twitter, first published in May 2018. The EB list included three driver fatalities prior to December 31, 2016 not found in our NHTSA-FARS search. Our NHTSA-FARS search identified one fatality not found on the EB file (and incidentally also identified one additional Tesla-associated pedestrian fatality in a separate accident).
There were a total of 11 fatalities through December 2016, all of which were summarized in these two tables
Well, shares to short at least There's some pretty clear bias though, or a lack of understanding of statistics. The data set for Tesla is much, much smaller. That's why his confidence interval on...
seems like someone with an ax to grind
Well, shares to short at least
There's some pretty clear bias though, or a lack of understanding of statistics. The data set for Tesla is much, much smaller. That's why his confidence interval on deaths per million miles is roughly 5x the size of his luxury pool.
Perhaps more importantly, he spends a ton of time figuring out when deaths could be attributed to a Tesla, but does not do the same for other luxury makes. If people are forgetting to input the make but inputting just the model year for Tesla, why wouldn't they make the same mistake for other luxury cars? Why do they only search google for Tesla deaths, do they not think the same absence of data could apply to other luxury models?
The lack of statistical thinking makes me wonder how this guy ever became a hedge fund manager.
Well that wouldn't help short shares...always follow the money. I must say that I really appreciate your comment. I don't trust my knowledge of statistics enough to have felt confident enough to...
The lack of statistical thinking makes me wonder how this guy ever became a hedge fund manager.
Well that wouldn't help short shares...always follow the money.
I must say that I really appreciate your comment. I don't trust my knowledge of statistics enough to have felt confident enough to state the first part, but your second part about the deep dive on Tesla and not others, I should have identified that and am embarrassed I didn't.
Now I need to go find my critical thinking cap before I make any more comments!
EDIT: I already used my "exemplary " tag or I would have given it to you
Haha no reason to be embarrassed! This kind of insight comes from reading far too many healthcare articles and isn't so much an issue of statistics (technically speaking it is, but it's obscured...
am embarrassed I didn't.
Haha no reason to be embarrassed! This kind of insight comes from reading far too many healthcare articles and isn't so much an issue of statistics (technically speaking it is, but it's obscured by the math) as it is study design.
I think the opinionated parts of this article were definitely problematic, and it’s hard to refute Musk’s claims without good data. That said, I was thrilled to see some journalistic investigation...
I think the opinionated parts of this article were definitely problematic, and it’s hard to refute Musk’s claims without good data. That said, I was thrilled to see some journalistic investigation of some public/government data sources. I strongly believe that data like this is essential to a modern society. We cannot make informed decisions about infrastructure, laws, and regulations unless we have good data about the world and our society. I’m enthused that there are journalists out there pointing out flaws in data, data gathering methods, and data analysis. We need more of that, but with less of the editorializing, and also a lot less of the uninformed data analysis methods this author used, such as mixing datasets and taking naive averages when they are not necessarily informative.
Edit: What someone really ought to do is do a reliability analysis of the underlying raw data. What is the observed error rate in the raw data reporting on vehicle fatalities? Considering the sparsity of vehicle fatalities in general, even small errors in reporting/classification could have extreme effects on downstream analysis.
Why can’t journalistic investigation be published on Medium? I guess it’s a matter of semantics, but I would consider this bad journalism, but journalism, nonetheless. How does one draw the line...
Why can’t journalistic investigation be published on Medium?
I guess it’s a matter of semantics, but I would consider this bad journalism, but journalism, nonetheless. How does one draw the line where something that is low quality journalism becomes unfit to be designated as journalism at all? It seems like the author’s effort in publishing this was not totally in bad faith (though that is also just my subjective judgment). I can entertain the idea that the author thought they were doing journalistic investigation, even if their effort was flawed.
Well apparently they're shorting Telsa according to @shiruken so that pretty much the definition of bad faith I never said it couldn't. Citizen journalism is real and medium is certainly a...
It seems like the author’s effort in publishing this was not totally in bad faith
Well apparently they're shorting Telsa according to @shiruken so that pretty much the definition of bad faith
Why can’t journalistic investigation be published on Medium?
I never said it couldn't. Citizen journalism is real and medium is certainly a publishing platform so I can't see why it couldn't be used.
How does one draw the line where something that is low quality journalism becomes unfit to be designated as journalism at all?
A source of debate since before the internet was a thing. There is no clear answer. It's like porn vs art....you know it when you see it.
Yeah, that is pretty damning. I stand by my statement that more people investigating public data and data sources and identifying issues with them is important. Not much one can do about bad...
Well apparently they're shorting Telsa according to @shiruken so that pretty much the definition of bad faith
Yeah, that is pretty damning. I stand by my statement that more people investigating public data and data sources and identifying issues with them is important. Not much one can do about bad actors pushing narratives about/interpretations of the data, though.
I don’t see it very frequently, and usually they just report aggregates and don’t dive into the raw data enough to even notice issues with it, much less surface and discuss them.
it's pretty standard fair for journalists
I don’t see it very frequently, and usually they just report aggregates and don’t dive into the raw data enough to even notice issues with it, much less surface and discuss them.
Hmmm, you may have to search for it a bit depending on your preferred news sources. I think Vox does a good job of this. Their podcast the Weeds gets into "the weeds" on policy issues which...
Hmmm, you may have to search for it a bit depending on your preferred news sources.
I think Vox does a good job of this. Their podcast the Weeds gets into "the weeds" on policy issues which includes deep dives into thinktank white papers, policy proposals, etc. https://www.vox.com/the-weeds
Most published datasets in academia will be scrutinized thoroughly by the community before being analyzed. I find outside of the sciences, there is much less scrutiny of the raw data itself, and...
Most published datasets in academia will be scrutinized thoroughly by the community before being analyzed. I find outside of the sciences, there is much less scrutiny of the raw data itself, and much more scrutiny of higher order analyses of the data. Higher order analyses are all well and good. But investigation of data quality/reliability (which is something I do for a living, granted, within a specific domain) is also important. You can apply state-of-the-art analysis methods to a dataset. But, if that underlying dataset contains flaws, such as errors, or gaps, the usefulness of the analysis on it is of questionable value. Journalists oftentimes assume that the data they are analyzing is the 'ground truth', but such assumptions are problematic because they are rarely sound.
Yes, that's really what I was thinking of. I don't think we really want journalists to play the role of academics, scientific experts, statisticians, data scientists etc.
much more scrutiny of higher order analyses of the data.
Yes, that's really what I was thinking of. I don't think we really want journalists to play the role of academics, scientific experts, statisticians, data scientists etc.
Interesting, because I think it’s problematic when journalists are given the license to not be informed about a domain that they report on. A lot of prominent newspapers have their own data...
I don't think we really want journalists to play the role of academics, scientific experts, statisticians, data scientists etc.
Interesting, because I think it’s problematic when journalists are given the license to not be informed about a domain that they report on. A lot of prominent newspapers have their own data science teams now precisely because it would be irresponsible of them to not have in-house expertise in the domain of data analysis.
Well science reporting is always a place where the vast knowledge required to understand the domain runs right up against a lot of practical obstacles that a journalist faces.
Well science reporting is always a place where the vast knowledge required to understand the domain runs right up against a lot of practical obstacles that a journalist faces.
I mean honestly just some basic knowledge of how to interpret a study's design and conclusion is all I really want... not another title about how we cured cancer for the 400th time.
I mean honestly just some basic knowledge of how to interpret a study's design and conclusion is all I really want... not another title about how we cured cancer for the 400th time.
Oh sure, that's totally fair. Journalism/reporting is a spectrum like anything else and there are ones that do a good job and are motivated by the right reasons and there are those who do a poor...
Oh sure, that's totally fair. Journalism/reporting is a spectrum like anything else and there are ones that do a good job and are motivated by the right reasons and there are those who do a poor job and are motivated by all the wrong reasons. My guess is that the science writers for the NYT have that understanding whereas those for MotherJones or BrietBart or some other rag do not.
The field of journalism is an interesting one and I hope to see a return to higher standards in the post-Trump era (but that may be overly optimistic as what most people think of as "objective journalism was more of a blip in journalism's long history).
A journalist's job at it's most basic level is to be a jack of all trades and a master of none. While reporters certainly specialize and learn their "beat" they generally aren't supposed to be an...
A journalist's job at it's most basic level is to be a jack of all trades and a master of none. While reporters certainly specialize and learn their "beat" they generally aren't supposed to be an expert in the fields they cover. If that were not the case then only someone would a Ph.D in physics would cover the field of physics, only a football coach would cover football, etc. You can see how silly that would be.
A lot of prominent newspapers have their own data science teams now precisely because it would be irresponsible of them to not have in-house expertise in the domain of data analysis.
I don't doubt that, but that is not the norm nor should it be.
I really don’t think that would be silly. I think that would be wonderful. I think reality is complex enough that we should want people who we trust to inform us about it to have some depth of...
If that were not the case then only someone would a Ph.D in physics would cover the field of physics, only a football coach would cover football, etc. You can see how silly that would be.
I really don’t think that would be silly. I think that would be wonderful. I think reality is complex enough that we should want people who we trust to inform us about it to have some depth of knowledge. I realize that it’s impractical for all journalists to have the same depth of knowledge in a field as professionals in that field, but I don’t think it’s unrealistic for journalists to aspire to that, nor do I think it’s impossible to find professionals who shift careers from some specific domain into journalism, and that those individuals would be more ideal journalists because of their previous experience. I also think journalists who lack in-depth knowledge should consult experts much more often than relying on their own interpretations of their own research. Basically, I just think that journalism is really pretty bad mostly, and if journalists could take it upon themselves to increase their depth, and if audiences appreciated that depth more, journalism would be improved.
I doubt it. Though, there is a transitional period where there are some ’smart’ features such as lane detection, collision avoidance systems etc. that are permeating the market from the top down....
Are luxury cars that much safer than normal cars or does it have something to do with who are driving luxury cars?
I doubt it. Though, there is a transitional period where there are some ’smart’ features such as lane detection, collision avoidance systems etc. that are permeating the market from the top down. That is one potential factor.
I assume luxury car drivers tend to be older, more educated, commute less and take less risks.
I think that’s a fairly safe assumption.
Maybe when/where/how much people are driving luxury cars is important? I.e., if you can afford to live out in the suburbs and you commute to work over a small distance (i.e., you don’t even need to get on a highway), you may be more likely to drive a luxury brand, and thus your risk of being in a fatal collision is lower by virtue of your vehicle never achieving high speeds for much of your driving time or not being in close proximity to other vehicles at high speeds. (In that case, looking at total collision rates, not just fatal rates would be informative.) Whereas, if your commute stretches further, or includes stretches of roads/highways that are more dangerous, your likelihood of a fatality increases. Other factors I can think of might include how careful you are as a driver which is influenced by a wide range of variables (including age as you mentioned). If you are financially secure, you may be less stressed and more able to focus on your driving. Or, if you are less financially secure, you may work multiple jobs which may mean you are more likely stressed while driving, or getting less sleep.
This is all speculation, obviously, but it’s not hard to imagine any number of variables that might be conflating here.
Unless I'm missing something, this article isn't even beginning to live up to it's headline, doesn't disclose it's data sources for Tesla fatalities, and seems like someone with an ax to grind or who wants to make a name for themselves. According to the author's bio on medium they don't have any formal experience with the auto industry. The first third of the article deals with SolarCity's problems...
If I'm being generous this article makes a case for why IIHS and NHTSA should be looking at Tesla in their stats.
EDIT: Not an attack on OP at all. I have posted some things that looking back might not have been so great myself. Just a heads up to anyone who might be thinking about devoting time to this "article"
EDIT2: @shiruken pointed out that they do disclose there data for the Tesla crashes. My bad!
Probably worth nothing the author is a hedge fund manager that is short on Tesla
That being said, they do list their sources for the Tesla fatalities
There were a total of 11 fatalities through December 2016, all of which were summarized in these two tables
Thank you. I definitely missed that. I'll amend my comment.
Yeah, I totally believe someone who has a direct financial incentive to get people to believe him. No conflict of interest there whatsoever.
Well, shares to short at least
There's some pretty clear bias though, or a lack of understanding of statistics. The data set for Tesla is much, much smaller. That's why his confidence interval on deaths per million miles is roughly 5x the size of his luxury pool.
Perhaps more importantly, he spends a ton of time figuring out when deaths could be attributed to a Tesla, but does not do the same for other luxury makes. If people are forgetting to input the make but inputting just the model year for Tesla, why wouldn't they make the same mistake for other luxury cars? Why do they only search google for Tesla deaths, do they not think the same absence of data could apply to other luxury models?
The lack of statistical thinking makes me wonder how this guy ever became a hedge fund manager.
Well that wouldn't help short shares...always follow the money.
I must say that I really appreciate your comment. I don't trust my knowledge of statistics enough to have felt confident enough to state the first part, but your second part about the deep dive on Tesla and not others, I should have identified that and am embarrassed I didn't.
Now I need to go find my critical thinking cap before I make any more comments!
EDIT: I already used my "exemplary " tag or I would have given it to you
Haha no reason to be embarrassed! This kind of insight comes from reading far too many healthcare articles and isn't so much an issue of statistics (technically speaking it is, but it's obscured by the math) as it is study design.
I think the opinionated parts of this article were definitely problematic, and it’s hard to refute Musk’s claims without good data. That said, I was thrilled to see some journalistic investigation of some public/government data sources. I strongly believe that data like this is essential to a modern society. We cannot make informed decisions about infrastructure, laws, and regulations unless we have good data about the world and our society. I’m enthused that there are journalists out there pointing out flaws in data, data gathering methods, and data analysis. We need more of that, but with less of the editorializing, and also a lot less of the uninformed data analysis methods this author used, such as mixing datasets and taking naive averages when they are not necessarily informative.
Edit: What someone really ought to do is do a reliability analysis of the underlying raw data. What is the observed error rate in the raw data reporting on vehicle fatalities? Considering the sparsity of vehicle fatalities in general, even small errors in reporting/classification could have extreme effects on downstream analysis.
Please don't confuse blogs on medium with actual journalism. It's insulting to journalists.
Why can’t journalistic investigation be published on Medium?
I guess it’s a matter of semantics, but I would consider this bad journalism, but journalism, nonetheless. How does one draw the line where something that is low quality journalism becomes unfit to be designated as journalism at all? It seems like the author’s effort in publishing this was not totally in bad faith (though that is also just my subjective judgment). I can entertain the idea that the author thought they were doing journalistic investigation, even if their effort was flawed.
Well apparently they're shorting Telsa according to @shiruken so that pretty much the definition of bad faith
I never said it couldn't. Citizen journalism is real and medium is certainly a publishing platform so I can't see why it couldn't be used.
A source of debate since before the internet was a thing. There is no clear answer. It's like porn vs art....you know it when you see it.
Yeah, that is pretty damning. I stand by my statement that more people investigating public data and data sources and identifying issues with them is important. Not much one can do about bad actors pushing narratives about/interpretations of the data, though.
Yeah, I'm definitely on the same page with you on that.
It's curious to me that this is exciting to you as it's pretty standard fair for journalists...picking apart government data and questioning it.
I don’t see it very frequently, and usually they just report aggregates and don’t dive into the raw data enough to even notice issues with it, much less surface and discuss them.
Hmmm, you may have to search for it a bit depending on your preferred news sources.
I think Vox does a good job of this. Their podcast the Weeds gets into "the weeds" on policy issues which includes deep dives into thinktank white papers, policy proposals, etc.
https://www.vox.com/the-weeds
I also find that 538 looks pretty closely at all sorts of data with polling being the big one. Here's a good article about tariffs:
https://fivethirtyeight.com/features/trumps-tariffs-could-actually-work-if-he-has-a-plan/
Lastly many financial sources look very deeply into things, as they're used to make investment decisions. For example, here's a deeper look into a labor report from "themarket" (I don't know much about them):
https://realmoney.thestreet.com/articles/01/11/2016/deep-dive-fridays-jobs-report
I may have misunderstood to some degree. I'm really talking about analysis of data.
Most published datasets in academia will be scrutinized thoroughly by the community before being analyzed. I find outside of the sciences, there is much less scrutiny of the raw data itself, and much more scrutiny of higher order analyses of the data. Higher order analyses are all well and good. But investigation of data quality/reliability (which is something I do for a living, granted, within a specific domain) is also important. You can apply state-of-the-art analysis methods to a dataset. But, if that underlying dataset contains flaws, such as errors, or gaps, the usefulness of the analysis on it is of questionable value. Journalists oftentimes assume that the data they are analyzing is the 'ground truth', but such assumptions are problematic because they are rarely sound.
Yes, that's really what I was thinking of. I don't think we really want journalists to play the role of academics, scientific experts, statisticians, data scientists etc.
Interesting, because I think it’s problematic when journalists are given the license to not be informed about a domain that they report on. A lot of prominent newspapers have their own data science teams now precisely because it would be irresponsible of them to not have in-house expertise in the domain of data analysis.
now if only they'd apply the same scrutiny to healthcare science reporting.... 😞
Well science reporting is always a place where the vast knowledge required to understand the domain runs right up against a lot of practical obstacles that a journalist faces.
I mean honestly just some basic knowledge of how to interpret a study's design and conclusion is all I really want... not another title about how we cured cancer for the 400th time.
Oh sure, that's totally fair. Journalism/reporting is a spectrum like anything else and there are ones that do a good job and are motivated by the right reasons and there are those who do a poor job and are motivated by all the wrong reasons. My guess is that the science writers for the NYT have that understanding whereas those for MotherJones or BrietBart or some other rag do not.
The field of journalism is an interesting one and I hope to see a return to higher standards in the post-Trump era (but that may be overly optimistic as what most people think of as "objective journalism was more of a blip in journalism's long history).
A journalist's job at it's most basic level is to be a jack of all trades and a master of none. While reporters certainly specialize and learn their "beat" they generally aren't supposed to be an expert in the fields they cover. If that were not the case then only someone would a Ph.D in physics would cover the field of physics, only a football coach would cover football, etc. You can see how silly that would be.
I don't doubt that, but that is not the norm nor should it be.
I really don’t think that would be silly. I think that would be wonderful. I think reality is complex enough that we should want people who we trust to inform us about it to have some depth of knowledge. I realize that it’s impractical for all journalists to have the same depth of knowledge in a field as professionals in that field, but I don’t think it’s unrealistic for journalists to aspire to that, nor do I think it’s impossible to find professionals who shift careers from some specific domain into journalism, and that those individuals would be more ideal journalists because of their previous experience. I also think journalists who lack in-depth knowledge should consult experts much more often than relying on their own interpretations of their own research. Basically, I just think that journalism is really pretty bad mostly, and if journalists could take it upon themselves to increase their depth, and if audiences appreciated that depth more, journalism would be improved.
I doubt it. Though, there is a transitional period where there are some ’smart’ features such as lane detection, collision avoidance systems etc. that are permeating the market from the top down. That is one potential factor.
I think that’s a fairly safe assumption.
Maybe when/where/how much people are driving luxury cars is important? I.e., if you can afford to live out in the suburbs and you commute to work over a small distance (i.e., you don’t even need to get on a highway), you may be more likely to drive a luxury brand, and thus your risk of being in a fatal collision is lower by virtue of your vehicle never achieving high speeds for much of your driving time or not being in close proximity to other vehicles at high speeds. (In that case, looking at total collision rates, not just fatal rates would be informative.) Whereas, if your commute stretches further, or includes stretches of roads/highways that are more dangerous, your likelihood of a fatality increases. Other factors I can think of might include how careful you are as a driver which is influenced by a wide range of variables (including age as you mentioned). If you are financially secure, you may be less stressed and more able to focus on your driving. Or, if you are less financially secure, you may work multiple jobs which may mean you are more likely stressed while driving, or getting less sleep.
This is all speculation, obviously, but it’s not hard to imagine any number of variables that might be conflating here.