1,000 is more than enough people to make a measurement if they are selected sufficiently randomly. You only need to poll ~400 people to have a 5% margin of error for a binomial variable across all of the US (with 95% confidence). 1100 makes that margin 3%.

Sure, but I would argue that 1,000 registered voters is going to make that a lot harder, not to mention it's actually "1000 registered voters who self selected to go through a 20 minute poll about facebook," which is why they said "no estimates of sampling error can be calculated."

https://accountabletech.org/wp-content/uploads/2020/07/Accou...

your rebuttal depends on the premise that people don't view facebook as harmful, but you yourself have not yet presented any data to that effect -- criticizing the study saying that people view it as harmful isn't the same as providing your own study saying they don't

tl;dr his evidence, even if we assume the issues you bring up are significant, outweighs your lack of any, so I am leaning in his favor

Speaking from a survey design point of view, he's more correct here. Especially with topics like this, the self selection effect and subsequent sampling bias can be very strong.

The evidence here is sufficiently poor to provide minimal value. Poorly collected data can often be worse than no data, as it can mislead you.

> your rebuttal depends on the premise that people don't view facebook as harmful.

No it doesn't. It depends on the fact that I don't know the opinions of other people at all. But I can infer that the selection biases of "registered voters" (which I think is a far bigger confounding factor than anyone is giving it credit for), and self-selection make it too difficult to extrapolate to the population of US citizens.

I also don't know. So I have to look to the evidence.

On the no-harm side, I see nothing.

On the harm side, I see a study about which someone has raised some concerns.

I understand you have concerns about the study design, but I'm not confident that they affect the outcome in such a way that the study provided zero evidence for Facebook being viewed as harmful.

Is there a study which doesn't run into the same concerns which says the opposite of this one?

I think it's important to note that the study is not about whether or not Facebook was harmful, but whether the average person thinks Facebook is harmful (or rather if the average registered voter who would volunteer to spend 20 minutes on a survey on Facebook thinks Facebook is harmful). It was just a survey of public opinion.

I personally also think Facebook is harmful.

It is possible to think a product is terrible and harms society and still use it. A lot of commenters on HN make that argument: Yea, I know their products are bad, but I simply can't stop using them because of [not so great] Reasons.

It tracks pretty well with my personal experience. I have plenty of non-tech friends who generally aren't following Facebook's every move, but there is a strong across-the-board sentiment that FB is a malevolent entity.

But not all for the same reason. Part of them think FB is evil because they only care about engagement and profits regardless of the outcome to society. And the other part thinks FB is biased and is censoring conservatives and pushing too liberal of an agenda.

Also there are liberals convinced that FB got Trump elected.