6 ms·

Pet peeve: choosing the axis to make yourself look better. Model 3 is around 8-10% lower than the Model S, but on that graph appears to be 50% lower. These ar

by mabbo 8y ago

Pet peeve: choosing the axis to make yourself look better. Model 3 is around 8-10% lower than the Model S, but on that graph appears to be 50% lower. These are great scores, and fantastic news, but this is purposeful dishonesty to exaggerate and mislead.

rcMgD2BwE72F 8y ago
Hiding the labels of the axis is dishonest. Choosing a good frame to compare different values is not. Would you consider this chart misleading: https://tradingeconomics.com/united-states/gdp https://tradingeconomics.com/united-states/gdp ? edit: they gain nothing by making the 3 appear safer than the S and X. They make much bigger margin on the most expensive models, and they want to increase demand for them (while they're ramping up the 3 production and go through the huge backlog). Although they've stopped anti-selling the 3, they certainly don't want people to think of the 3 has a newer/better version of the S.
- ianai 8y ago
  I think they chose their axis range to make the differences starker than they are. Their graph doesn’t include a value for the highest probabilities, for instance. I’m all for zooming a graph to maximize information content, but a range of 5-8% doesn’t present a clear heuristic tie to any widely known phenomenon. A range of 5-10% would have the benefit of being easier to tie to 1/20 to 1/10 odds. Having said that, this is Tesla’s website. Of course it’s all sales copy to move product.
- crististm 8y ago
  The one you mention is not misleading: they put the Y values everywhere in the chart. The other one is starting at 5% and ending at 8%. While a full range 0-100% would bury the relevant data, a picture with 0-10% would tell a different story.
- superlopuh 8y ago
  I do, and I don't think I'm the only one. There is no hint of the bars continuing below the axis, and the effective result is the impression that the relative difference is larger than it really is. Even the gradients seem poorly chosen, the bar gets to full white by the bottom. My intuitive first impression is that if the gradient cannot continue below the chosen cut off point, then the bar cannot continue either.
  - rcMgD2BwE72F 8y ago
    >There is no hint of the bars continuing below the axis Aren't the numbers big and bold enough? Should we only design chart while presuming that viewers can't read numbers?
  - coryfklein 8y ago
    Scaling graphs is not so out of the norm that you need to explicitly call out when you are doing it. Just present your axes and label them correctly.
chapium 8y ago
I agree. The axis chosen exagerrates differences between the values. Also, why are all the non-tesla products the same color as the background? Is this another distortion technique?
- cptskippy 8y ago
  > Also, why are all the non-tesla products the same color as the background? Is this another distortion technique? That's convention when you're calling attention to specific data points. When you have a graph comparing a dozen items but you're really only concerned with how 1 or 2 items fall in the pack, you generally color them to stand out from the rest. Take a look at any computer hardware review site and you'll see similar coloring. (https://images.anandtech.com/graphs/graph13254/light-bw.png https://images.anandtech.com/graphs/graph13254/light-bw.png)
Someone1234 8y ago
They're also comparing them to "top 50 vehicles since 2011." Since they don't list which specific vehicles and years that includes, it is hard to draw any conclusions but it is odd contrasting vehicles which are up to seven years old. I cannot even locate "Lowest Probability of injury" on NHTSA's site. So I'm not sure what the percentage are referring to. Can someone point me towards it? PS - I don't doubt the Model 3 is a very safe vehicle. My post is about the way Tesla is conveying that information and what information they're conveying.
- hef19898 8y ago
  Same hear, I tried to get at least some official ranking of results by model and year, if not the raw data, and didn't find anything. Maybe that's non-public, which wouldn't make sense. But as it stands it is hard to recreatw the ranking Tesla is using and to verify the numbers. Without sources, and a scetchy presentation, it is just a markwting statement from one manufacturer. So no different to everybody else out there. Cherry picking numbers and setting context alsways works.
- SmellyGeekBoy 8y ago
  > it is odd contrasting vehicles which are up to seven years old. How so? Some of the top 50 safest cars are seven years old, hence they appear on the chart.
  - cptskippy 8y ago
    Exactly. That's like saying the Saturn 5 shouldn't be on the list of most powerful rockets because it's 50+ years old. If a 7 year old car is still one of the top 50 safest cars on the road then you should either be applauding whomever made that car or condemning whomever ranked beneath it. Not questioning why it's on the list.
  - Someone1234 8y ago
    It didn't say the top 50 SAFEST cars, it say the "top 50 cars." Totally different things.
trumped 8y ago
There is also large weight differences between a Model S P100D and Model S 40 and they don't specify which one they are talking about. I would not be surprised if weight was an important factor in the results (there is an even a larger weight difference between basic Model 3 and Model X 100D)

by mabbo 8y ago

rcMgD2BwE72F 8y ago

Hiding the labels of the axis is dishonest. Choosing a good frame to compare different values is not. Would you consider this chart misleading: https://tradingeconomics.com/united-states/gdp https://tradingeconomics.com/united-states/gdp ? edit: they gain nothing by making the 3 appear safer than the S and X. They make much bigger margin on the most expensive models, and they want to increase demand for them (while they're ramping up the 3 production and go through the huge backlog). Although they've stopped anti-selling the 3, they certainly don't want people to think of the 3 has a newer/better version of the S.

ianai 8y ago
I think they chose their axis range to make the differences starker than they are. Their graph doesn’t include a value for the highest probabilities, for instance. I’m all for zooming a graph to maximize information content, but a range of 5-8% doesn’t present a clear heuristic tie to any widely known phenomenon. A range of 5-10% would have the benefit of being easier to tie to 1/20 to 1/10 odds. Having said that, this is Tesla’s website. Of course it’s all sales copy to move product.
crististm 8y ago
The one you mention is not misleading: they put the Y values everywhere in the chart. The other one is starting at 5% and ending at 8%. While a full range 0-100% would bury the relevant data, a picture with 0-10% would tell a different story.
superlopuh 8y ago
I do, and I don't think I'm the only one. There is no hint of the bars continuing below the axis, and the effective result is the impression that the relative difference is larger than it really is. Even the gradients seem poorly chosen, the bar gets to full white by the bottom. My intuitive first impression is that if the gradient cannot continue below the chosen cut off point, then the bar cannot continue either.
- rcMgD2BwE72F 8y ago
  >There is no hint of the bars continuing below the axis Aren't the numbers big and bold enough? Should we only design chart while presuming that viewers can't read numbers?
- coryfklein 8y ago
  Scaling graphs is not so out of the norm that you need to explicitly call out when you are doing it. Just present your axes and label them correctly.

chapium 8y ago

I agree. The axis chosen exagerrates differences between the values. Also, why are all the non-tesla products the same color as the background? Is this another distortion technique?

cptskippy 8y ago
> Also, why are all the non-tesla products the same color as the background? Is this another distortion technique? That's convention when you're calling attention to specific data points. When you have a graph comparing a dozen items but you're really only concerned with how 1 or 2 items fall in the pack, you generally color them to stand out from the rest. Take a look at any computer hardware review site and you'll see similar coloring. (https://images.anandtech.com/graphs/graph13254/light-bw.png https://images.anandtech.com/graphs/graph13254/light-bw.png)

Someone1234 8y ago

They're also comparing them to "top 50 vehicles since 2011." Since they don't list which specific vehicles and years that includes, it is hard to draw any conclusions but it is odd contrasting vehicles which are up to seven years old. I cannot even locate "Lowest Probability of injury" on NHTSA's site. So I'm not sure what the percentage are referring to. Can someone point me towards it? PS - I don't doubt the Model 3 is a very safe vehicle. My post is about the way Tesla is conveying that information and what information they're conveying.

hef19898 8y ago
Same hear, I tried to get at least some official ranking of results by model and year, if not the raw data, and didn't find anything. Maybe that's non-public, which wouldn't make sense. But as it stands it is hard to recreatw the ranking Tesla is using and to verify the numbers. Without sources, and a scetchy presentation, it is just a markwting statement from one manufacturer. So no different to everybody else out there. Cherry picking numbers and setting context alsways works.
SmellyGeekBoy 8y ago
> it is odd contrasting vehicles which are up to seven years old. How so? Some of the top 50 safest cars are seven years old, hence they appear on the chart.
- cptskippy 8y ago
  Exactly. That's like saying the Saturn 5 shouldn't be on the list of most powerful rockets because it's 50+ years old. If a 7 year old car is still one of the top 50 safest cars on the road then you should either be applauding whomever made that car or condemning whomever ranked beneath it. Not questioning why it's on the list.
- Someone1234 8y ago
  It didn't say the top 50 SAFEST cars, it say the "top 50 cars." Totally different things.

trumped 8y ago

There is also large weight differences between a Model S P100D and Model S 40 and they don't specify which one they are talking about. I would not be surprised if weight was an important factor in the results (there is an even a larger weight difference between basic Model 3 and Model X 100D)