>In point of fact, a quarter of the 4,594 people surveyed by Pew Research between May 29 and June 11 report deleting the Facebook app.[1] Of those, half who qualify as young deleted the app.

For a population of 330 million, if you want a 95% confidence level and a 2% margin of error, you need only 2401 people in your sample. This is assuming the survey sampled appropriately (i.e. random selection methodology).

It is standard to take well executed surveys and extrapolate to the population. Calling it out is calling the whole field of statistics out.

Nice try. I'm well versed in the statistics but I'm not blindly criticizing the sample because it's small.

I'm criticizing exactly what you called out as an assumption - the methodology of the sample; namely that they survey's distribution matches the population distribution.

So thanks for the lesson in p values, but it has nothing to do with my point. I take issue with blindly citing sources without doing due diligence, especially when the survey was collected over a very small window of time across a population that most likely doesn't match Americans in toto.

> I'm criticizing exactly what you called out as an assumption - the methodology of the sample; namely that they survey's distribution matches the population distribution.

Yeah, I’m reading your previous comment, and I’m still not seeing it.

You're not seeing the issue with extrapolating a survey that lasted barely a week, from Pew Research, to an authoritative statement about the zeitgeist of the entire country?

Okay. I specifically called out the date range issue in my original comment. But it seems like people can't move past the impulse to (in)correct me about pop statistics. I never said the issue was small sample size.

The reason you're getting downvoted is that your argument is basically "well what if they're wrong?"

Pew has established a reputation as an organization who conducts polls regularly and properly. You are a throwaway account on HN. To be Bayesian about it, the priors are way better for Pew than for you.

If you have specific, substantive criticisms of how Pew conducted that particular poll, by all means, share them and folks can discuss that.

I don't really care why I'm being downvoted, nor did I ask. It's not relevant. I'll continually double down on this point regardless, unless you can actually refute what I'm saying instead of (like the others) trying to point out perceived inconsistencies in what I'm saying, or attacking points I didn't make.

I already made my substantive criticism. You cannot extrapolate the results of a survey evaluated over a 1 - 2 week period to the modal opinion of the American population. To spell it out for you: I vehemently disagree that you can adequately randomly sample a representative subset of the American population in a two week period.

Do you have a substantive rebuttal to that criticism? Evaluate my point on its own merits, not whether or not it's coming from a throwaway account (which this is not).

To put it bluntly, I'm a little stunned no one else is considering that survey as critically as I am. Even if it's ultimately a sound methodology, it's pretty frustrating for that to be taken as an uncontroversial fact without any challenge.

> I never said the issue was small sample size.

> a quarter of the 4,594 people surveyed by Pew Research between May 29 and June 11 report deleting the Facebook app.

Emphasis yours.

Yes, that was an element of what I stated. Now re-read the rest of what you quoted. See the bit about the date range?

Yeah—too bad that wasn't the part you put in emphasis, and the part that you did put in emphasis is something you're claiming wasn't your criticism at all.