Hacker News Reader

Provided the people were chosen truly randomly/make up a representative sample of the population, 863 is more than enough to get an accurate estimate of the infection rate. ~1000 is a big enough sample size for the entire US, let alone just Los Angeles

View on HN

I used to work on survey designs in my old job. I am having a hard time coming up with a way that they could possibly control for selection bias. We are in the middle of a pandemic, the people willing to risk leaving their homes to get tested is going to be highly skewed towards individuals that believe they have already had the infection. It's possible that they asked the participants about this, and adjusted their weights accordingly, but considering that this study shares authors with the Santa Clara study, and that study did not control for selection bias, I have my doubts.

Where's the marketing company's criteria for sample selection? How are the communicating and choosing the people? Are they all within the same locality?

And no 863 is not representative of LA. LA is made up of individual communities and some never leave those communities. Some communities are pass through for commuters. Really important where those 863 are from. All are 863 from a small area? Are some from Pasadena and other from Santa Monica. What about Covina or Redondo? LA or Beverly Hills.

Reply on news.ycombinator.com