10 ms·

Sure, but a person who goes to the store 7 times a week is ~7 times as likely to have been sampled compared to someone who goes only once a week. They are also

Sure, but a person who goes to the store 7 times a week is ~7 times as likely to have been sampled compared to someone who goes only once a week. They are also ~7 times as likely to have contracted covid-19 at the grocery store. It's likely that they are also, in general, isolating less than the other. This sampling would have a (possibly slight, but still unknown) bias towards people who are isolating less (who you would expect to then have a higher than average percent positive for antibodies, assuming isolating is helpful). I don't think anyone knows whether or not this is a more representative sample than the stanford study.

WrtCdEvrydy 6y ago
Unfortunately, unless you go door to door knocking... you're gonna get biased data.
- ineedasername 6y ago
  Or truly random sampling that includes appropriate proportions of sub groups, e.g, socioeconomic status etc. Otherwise you have to know on all of the doors. But it's still a lot: for a 95% confidence level and 2% margin of error, for NYC you would need to test about 2,400 people. Assuming a response rate of 10%, you need to knock on 24,000 doors.
  - austingulati 6y ago
    Wouldn't the ability to refuse taking the test introduce bias? i.e. the 10% that do take it still may not be representative of the entire population. Likely better than the grocery store sample, but still not ideal
    - qqqwerty 6y ago
      Yeah, that is still an issue. You can offer compensation to incentivize participation. And you can pre-select your sample (instead of random door-to-door) and require multiple follow ups with the selected persons to reduce nonresponse bias. You can also do a separate phone survey in addition to surveying the test participants. Questions like "Do you think you have had COVID?" and "How many times per week do you leave the house?". If the responses for the test participants vary significantly from the phone survey participants, you can try and weight your data accordingly.
    - ineedasername 6y ago
      Sure, phone surveys for political purposes (presidential approval ratings etc.) have to deal with that all of the time. There are methods for estimating non-response impact. [0] One method of mitigating it that I've seen it to reach out again to non-responders. You then analyze their results to see how they differ from the baseline responders to estimate the non-responder population. If there's little/no difference, you can be fairly confident the risk of bias is low. It's called non-response follow up, and is a pretty common method. There's also literature that suggests that you don't discard outlier values in the actual responders as they may help approximate the non-responder population, i.e., the non-outliers represent typical responders while outliers are more likely to represent non-responders [1] [0]https://www.warc.com/content/paywall/article/jar/research_note_a_model_for_diagnosing_and_reducing_nonresponse_bias/14027 https://www.warc.com/content/paywall/article/jar/research_no... [1] https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-019-0757-1 https://bmcmedresmethodol.biomedcentral.com/articles/10.1186...
      - ineedasername 6y ago
        Wow, downvoted for providing factual information on how researchers mitigate non-response bias. Didn't think there was anything controversial there. I'm willing to accept the additional downvoted for this comment though.
        austingulati 6y ago
        Not downvoted by me! Appreciate you sharing this information, it is helping me better understand how researchers mitigate non-response bias. Interesting stuff.
- jonas21 6y ago
  Even with door to door knocking, you're going to introduce a bias -- toward people who are around to answer the door, which would tend to undersample essential workers.
  - AznHisoka 6y ago
    Or people who will try to talk to you from their balcony rather than answering the door.
alkibiades 6y ago
going 7 times doesn’t actually increase the likelihood 7x but your point still stands that they would be far more likely to get it
- karagenit 6y ago
  Well, sort of. The increase in likelihood of being sampled due to going to the store N times more than the average person is a function of N and P (the probability of being sampled in any given trip) such that f(N, P): (1 - ((1-P)^N)) / P For a large P you're right, but as P gets very small (let's say that in any given trip you have a 0.01% chance of being sampled) then the increase in likelihood of being sampled approaches N times as much (in this example, it would be 6.999 times as likely). Of course, this assumes they're taking measures to ensure no one gets sampled twice.
  - taeric 6y ago
    Maybe, more rapid trips may reduce your odds of being there when the sample happens on a day. Such that you may have a lower p. :)
  - CyberDildonics 6y ago
    They also said people who went to the grocery store 7x as much were 7x as likely to be infected which is what I think this person was replying to.
ferzul 6y ago
otoh, it gives about the same ifr as the recent study in geneva. at the end of the day, it looks like the first wave will have an ifr of about 0.6% - more than ten times greater than the flu, but not the bubonic plague
- bagacrap 6y ago
  that's not more than 10x...
fyp 6y ago
Even with worst-case scenario of bias, this is still extremely good news. My worry was always with super spreaders who refuse to isolate but the data is suggesting that those spreaders will still soon achieve herd immunity among themselves. As long as the rest of the society behaves we will still hit zero cases relatively quickly. (which wasn't always clear to me before since I initially predicted that this will take years to work out)
- wbl 6y ago
  Superspreaders aren't connected to each other.
- athrowaway3z 6y ago
  What? All of these numbers are inline with what experts have been saying and modeling for more than a month. We will be dealing with this for at least another year. We will cycle through policy to loosen up and close back down a little. In the best case for NYC they are 1/4th of the way to herd immunity in numbers. With the drop in transmissions, this might be 1/8th of the way in terms of time. There will be more deaths in the future than have been recorded so far. Look for the bright side in things, but zero cases is a pipe dream.
  - zaroth 6y ago
    25% prevalence is absolutely not in line with what experts were saying on March 23rd. Nor was an IFR of 0.5%. The CDC reports that flu has an IFR of ~.13% in the US (61,000 deaths out of 45 million cases). That makes 0.5% roughly 3.3 times worse, not 10. Also, herd immunity does not require 100% having positive antibodies, it will show an effect on Ro starting around 65%.
    - hcknwscommenter 6y ago
      It's not 25%. 0.5% versus 0.13% is not the only issue here in terms of how much worse it is. It's the long time in the ICU. The flu kills you fairly quickly or you get better fairly quickly, so you don't take up hospital capacity so long. Herd "immunity" does not require 100%, but that's a decent approximation. Sure, I'll grant that it "starts" to show an effect around 65%, but the effect is not so strong. 70%, much stronger. 80% very strong. Heck, you could probably do containment by then without waiting to get to 100%. Because inadvertent spreading would be so low.
JetSetWilly 6y ago
In the other direction, people who are sick or have recently been sick may well isolate and avoid going to the grocery store, dragging the detection rate back down.

WrtCdEvrydy 6y ago

Unfortunately, unless you go door to door knocking... you're gonna get biased data.

ineedasername 6y ago
Or truly random sampling that includes appropriate proportions of sub groups, e.g, socioeconomic status etc. Otherwise you have to know on all of the doors. But it's still a lot: for a 95% confidence level and 2% margin of error, for NYC you would need to test about 2,400 people. Assuming a response rate of 10%, you need to knock on 24,000 doors.
- austingulati 6y ago
  Wouldn't the ability to refuse taking the test introduce bias? i.e. the 10% that do take it still may not be representative of the entire population. Likely better than the grocery store sample, but still not ideal
  - qqqwerty 6y ago
    Yeah, that is still an issue. You can offer compensation to incentivize participation. And you can pre-select your sample (instead of random door-to-door) and require multiple follow ups with the selected persons to reduce nonresponse bias. You can also do a separate phone survey in addition to surveying the test participants. Questions like "Do you think you have had COVID?" and "How many times per week do you leave the house?". If the responses for the test participants vary significantly from the phone survey participants, you can try and weight your data accordingly.
  - ineedasername 6y ago
    Sure, phone surveys for political purposes (presidential approval ratings etc.) have to deal with that all of the time. There are methods for estimating non-response impact. [0] One method of mitigating it that I've seen it to reach out again to non-responders. You then analyze their results to see how they differ from the baseline responders to estimate the non-responder population. If there's little/no difference, you can be fairly confident the risk of bias is low. It's called non-response follow up, and is a pretty common method. There's also literature that suggests that you don't discard outlier values in the actual responders as they may help approximate the non-responder population, i.e., the non-outliers represent typical responders while outliers are more likely to represent non-responders [1] [0]https://www.warc.com/content/paywall/article/jar/research_note_a_model_for_diagnosing_and_reducing_nonresponse_bias/14027 https://www.warc.com/content/paywall/article/jar/research_no... [1] https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-019-0757-1 https://bmcmedresmethodol.biomedcentral.com/articles/10.1186...
    - ineedasername 6y ago
      Wow, downvoted for providing factual information on how researchers mitigate non-response bias. Didn't think there was anything controversial there. I'm willing to accept the additional downvoted for this comment though.
      - austingulati 6y ago
        Not downvoted by me! Appreciate you sharing this information, it is helping me better understand how researchers mitigate non-response bias. Interesting stuff.
jonas21 6y ago
Even with door to door knocking, you're going to introduce a bias -- toward people who are around to answer the door, which would tend to undersample essential workers.
- AznHisoka 6y ago
  Or people who will try to talk to you from their balcony rather than answering the door.

alkibiades 6y ago

going 7 times doesn’t actually increase the likelihood 7x but your point still stands that they would be far more likely to get it

karagenit 6y ago
Well, sort of. The increase in likelihood of being sampled due to going to the store N times more than the average person is a function of N and P (the probability of being sampled in any given trip) such that f(N, P): (1 - ((1-P)^N)) / P For a large P you're right, but as P gets very small (let's say that in any given trip you have a 0.01% chance of being sampled) then the increase in likelihood of being sampled approaches N times as much (in this example, it would be 6.999 times as likely). Of course, this assumes they're taking measures to ensure no one gets sampled twice.
- taeric 6y ago
  Maybe, more rapid trips may reduce your odds of being there when the sample happens on a day. Such that you may have a lower p. :)
- CyberDildonics 6y ago
  They also said people who went to the grocery store 7x as much were 7x as likely to be infected which is what I think this person was replying to.

ferzul 6y ago

otoh, it gives about the same ifr as the recent study in geneva. at the end of the day, it looks like the first wave will have an ifr of about 0.6% - more than ten times greater than the flu, but not the bubonic plague

bagacrap 6y ago
that's not more than 10x...

fyp 6y ago

Even with worst-case scenario of bias, this is still extremely good news. My worry was always with super spreaders who refuse to isolate but the data is suggesting that those spreaders will still soon achieve herd immunity among themselves. As long as the rest of the society behaves we will still hit zero cases relatively quickly. (which wasn't always clear to me before since I initially predicted that this will take years to work out)

wbl 6y ago
Superspreaders aren't connected to each other.
athrowaway3z 6y ago
What? All of these numbers are inline with what experts have been saying and modeling for more than a month. We will be dealing with this for at least another year. We will cycle through policy to loosen up and close back down a little. In the best case for NYC they are 1/4th of the way to herd immunity in numbers. With the drop in transmissions, this might be 1/8th of the way in terms of time. There will be more deaths in the future than have been recorded so far. Look for the bright side in things, but zero cases is a pipe dream.
- zaroth 6y ago
  25% prevalence is absolutely not in line with what experts were saying on March 23rd. Nor was an IFR of 0.5%. The CDC reports that flu has an IFR of ~.13% in the US (61,000 deaths out of 45 million cases). That makes 0.5% roughly 3.3 times worse, not 10. Also, herd immunity does not require 100% having positive antibodies, it will show an effect on Ro starting around 65%.
  - hcknwscommenter 6y ago
    It's not 25%. 0.5% versus 0.13% is not the only issue here in terms of how much worse it is. It's the long time in the ICU. The flu kills you fairly quickly or you get better fairly quickly, so you don't take up hospital capacity so long. Herd "immunity" does not require 100%, but that's a decent approximation. Sure, I'll grant that it "starts" to show an effect around 65%, but the effect is not so strong. 70%, much stronger. 80% very strong. Heck, you could probably do containment by then without waiting to get to 100%. Because inadvertent spreading would be so low.

JetSetWilly 6y ago

In the other direction, people who are sick or have recently been sick may well isolate and avoid going to the grocery store, dragging the detection rate back down.