4 ms·

Mining The Data Behind Toy Safety

30 points by thingsilearned 13y ago

thingsilearned 13y ago
Link directly to data dashboard: https://chartio.com/project/14404/dash/18166/public/ https://chartio.com/project/14404/dash/18166/public/
kevinpet 13y ago
Strikes me as pretty meaningless because while they compare, for example, the number of recalls of toy animals vs. musical instruments, there's no scale showing how many different types of each product are sold. It's entirely possible that musical instruments were recalled at 10x the rate of animals, which seems to be the useful question. It seems that as the tools have gotten strong, the focus has been on creating prettier graphs, not making sure the units are chosen to be most meaningful.
- Loughla 13y ago
  After working with complicated statistics in that GOD-DAMNED SPSS PROGRAM during graduate school, I have a four-part theory about why this happens (and it applies to many, many areas as well): The better the tools are, the more people use them. The more people use them, the higher the ratio of junk. The higher the ratio of junk, the more people feel the need to set themselves apart. The more people feel the need to set themselves apart, the prettier they will make something, regardless of how useful, functional or interesting the actual content happens to be. I call it the 3D-Movie Theory of Attrition.
- shawn-furyan 13y ago
  This was my reaction as well. The article calls out high selling manufacturers directly, implying that they are the worst offenders, but without sales data you can't determine recall rate (something like recalled toys per 1000 sold), which is what consumers need to act upon this data. It strikes me that the largest manufacturers with the most mature design and fabrication processes at least have all the tools to outperform smaller manufacturers in this regard, and so the article may well be nudging readers to manufacturers with higher recall rates. I mean, one of the principal challenges of empirical data analysis is putting together data from disparate sources in order to come to actionable analysis. One doesn't just publish the data that's easy to come by and then call it a day. And the poor analysis presented in this article is a good illustration of why.
  - Zancarius 13y ago
    > I mean, one of the principal challenges of empirical data analysis is putting together data from disparate sources in order to come to actionable analysis. One doesn't just publish the data that's easy to come by and then call it a day. I thought it was curious how the most dangerous "toy" is the bicycle. I suppose that's not surprising, but other charts indicate that the greatest hazard to children is choking. I'd wager it's difficult to choke on a bicycle save for disassembling it and ravenously ingesting the smaller parts. Given the CPSC's recent history (think Buckyballs), I also wonder if the relative number of product recalls has increased due to changes in regulation, changes specific to the CPSC itself (organizational, leadership, mission, etc), or increased pressure from consumer/physician groups. Products containing lead appear to be responsible for a sizable portion of the recalls, but I'm not completely sure I agree with the implicit suggestion that modern toys are somehow more dangerous based on recalls alone. The US is a litigocracy (and increasingly so) after all.
lectrick 13y ago
This data is completely meaningless without it being as a percent of total sales. Otherwise it's essentially the same as a popularity chart. I cannot fathom how this simple understanding was missed.
ye 13y ago
You should probably work on the embeddable charts, instead of taking a screenshot, saving-optimizing, uploading, CDN'ing, linking to the image.
tobyjsullivan 13y ago
"Fisher Price is the most high-risk manufacturer in terms of recalls" High-risk implies that for every toy released, there is a high probability that it will be recalled. The post fails to provide any data indicating this is true for Fisher Price. It is most likely that Fisher Price and Mattel simply sell a significantly wider array of toys. For all we know, based on this data, Fisher Price could just as possibly be the lowest-risk manufacturer.
bonemachine 13y ago
Perhaps a better title would be "Everything Wrong With Our Insatiable Big Data Fetish In 4 Easy Charts".

Link directly to data dashboard: https://chartio.com/project/14404/dash/18166/public/ https://chartio.com/project/14404/dash/18166/public/

kevinpet 13y ago

Strikes me as pretty meaningless because while they compare, for example, the number of recalls of toy animals vs. musical instruments, there's no scale showing how many different types of each product are sold. It's entirely possible that musical instruments were recalled at 10x the rate of animals, which seems to be the useful question. It seems that as the tools have gotten strong, the focus has been on creating prettier graphs, not making sure the units are chosen to be most meaningful.

Loughla 13y ago
After working with complicated statistics in that GOD-DAMNED SPSS PROGRAM during graduate school, I have a four-part theory about why this happens (and it applies to many, many areas as well): The better the tools are, the more people use them. The more people use them, the higher the ratio of junk. The higher the ratio of junk, the more people feel the need to set themselves apart. The more people feel the need to set themselves apart, the prettier they will make something, regardless of how useful, functional or interesting the actual content happens to be. I call it the 3D-Movie Theory of Attrition.
shawn-furyan 13y ago
This was my reaction as well. The article calls out high selling manufacturers directly, implying that they are the worst offenders, but without sales data you can't determine recall rate (something like recalled toys per 1000 sold), which is what consumers need to act upon this data. It strikes me that the largest manufacturers with the most mature design and fabrication processes at least have all the tools to outperform smaller manufacturers in this regard, and so the article may well be nudging readers to manufacturers with higher recall rates. I mean, one of the principal challenges of empirical data analysis is putting together data from disparate sources in order to come to actionable analysis. One doesn't just publish the data that's easy to come by and then call it a day. And the poor analysis presented in this article is a good illustration of why.
- Zancarius 13y ago
  > I mean, one of the principal challenges of empirical data analysis is putting together data from disparate sources in order to come to actionable analysis. One doesn't just publish the data that's easy to come by and then call it a day. I thought it was curious how the most dangerous "toy" is the bicycle. I suppose that's not surprising, but other charts indicate that the greatest hazard to children is choking. I'd wager it's difficult to choke on a bicycle save for disassembling it and ravenously ingesting the smaller parts. Given the CPSC's recent history (think Buckyballs), I also wonder if the relative number of product recalls has increased due to changes in regulation, changes specific to the CPSC itself (organizational, leadership, mission, etc), or increased pressure from consumer/physician groups. Products containing lead appear to be responsible for a sizable portion of the recalls, but I'm not completely sure I agree with the implicit suggestion that modern toys are somehow more dangerous based on recalls alone. The US is a litigocracy (and increasingly so) after all.

lectrick 13y ago

This data is completely meaningless without it being as a percent of total sales. Otherwise it's essentially the same as a popularity chart. I cannot fathom how this simple understanding was missed.

ye 13y ago

You should probably work on the embeddable charts, instead of taking a screenshot, saving-optimizing, uploading, CDN'ing, linking to the image.

tobyjsullivan 13y ago

"Fisher Price is the most high-risk manufacturer in terms of recalls" High-risk implies that for every toy released, there is a high probability that it will be recalled. The post fails to provide any data indicating this is true for Fisher Price. It is most likely that Fisher Price and Mattel simply sell a significantly wider array of toys. For all we know, based on this data, Fisher Price could just as possibly be the lowest-risk manufacturer.

bonemachine 13y ago

Perhaps a better title would be "Everything Wrong With Our Insatiable Big Data Fetish In 4 Easy Charts".