[godelski]

More specifically

> Datta told BuzzFeed News that his company, on average, has access to location data for 30% to 60% of people in any given location in the United States.

And getting down to it more: which 30-60% of people? I find it odd that when HN is talking about Covid they are so keen on finding the slightest bias in data/report and arguing over that, yet in posts like this most of the comments presume accuracy. I would expect the same scrutiny (which I think is good!) everywhere.

[ThrowawayR2]

Not that I have an opinion about Mobilewalla's analysis but polling gets accurate results with far less than 30-60% sampling, so that, in and of itself, is not necessarily a problem.

[godelski]

> polling gets accurate results with far less than 30-60% sampling, so that, in and of itself, is not necessarily a problem.

I'd like to refer you to the sentence after I quoted the article.

>> And getting down to it more: which 30-60% of people?

Polling works hard to ensure that their data set is a representative demographic. We don't know that here. Considering that there are socioeconomic correlations between race, it isn't out of the question that this data is not representative. You could have 60% but if your data set isn't representative, you aren't going to draw accurate conclusions.

[ashtonkem]

Polling has random sampling. We don’t know if Mobilewalla’s sampling is truly random or not.

[ashtonkem]

And is there any bias inherent in that data along economic or racial lines? Given that black people are poorer on average in the US, any collection of data that’s biased based on cell phone cost might incidentally pick up a racial bias too.