Hacker News Reader

Comment by nickjj | Hacker News Reader

What was their sample size to report that number? They mentioned it's based on a survey but don't list how many people they surveyed in the article. 97% is a high number to claim. They definitely didn't survey every single adult in the US (I know this for sure because I didn't take their survey).

View on HN

The sample is "1,502 U.S. adults from Jan. 25 to Feb. 8, 2021, by cellphone and landline phone" if this is the same survey as linked in this article

https://www.pewresearch.org/fact-tank/2022/01/13/share-of-th...

it links to this methodology document in the "How we did this" dropdown:

https://www.pewresearch.org/wp-content/uploads/2022/01/Tech-...

I agree with your point. Do they account for the number of Americans that don't have a phone at all somehow or should the real headline be "97% of American adults that can be called by cellphone or landline own a cellphone or smartphone"?

The vast majority of error in modern polling comes from systematic bias when doing the sampling. The sample size is essentially irrelevant; they could ask a million people and their number would not be significantly more reliable.

They probably called people to find out :)

Mind-bogglingly, that’s exactly what they did. Calling people to research how many of them own a particular subset type of telephone apparently passes for research now.

In my own family (considering only direct ancestors), I can think of 4 adults who don’t have any mobile phone (all are over 70). That’s over 50% of adults in our case.

That's what they did:

"The analysis in this report is based on telephone interviews conducted Jan. 25-Feb. 8, 2021, among a national sample of 1,502 adults ages 18 and older, living in all 50 U.S. states and the District of Columbia (300 respondents were interviewed on a landline telephone, and 1,202 were interviewed on a cellphone, including 845 who had no landline telephone)."

TLDR completely meaningless study, it's like doing car ownership study at petrol station.

Yes and no. It will definitely inform about cell phone usage among landline users. You won’t cover population groups with neither a landline nor a cell phone though.

In the us there is thus famous example about the election exit polls that went horribly wrong because cell phones were new and so the phone survey ignored the laborers

One problem is that landline-only users are often not reachable by phone during large spans of the day.

With appropriate sample sizes and data collection protocols you can rebalance against such a bias.

Maybe it was an online survey, but 3% of them having internet access, being enough literate to answer the survey and not having a phone would be a big number then.

Notably, 3% is close to the Lizardman Constant[1].

[1] https://slatestarcodex.com/2013/04/12/noisy-poll-results-and...

TLDR completely meaningless study, it's like doing car ownership study at petrol station.

Reply on news.ycombinator.com