Hacker News Reader

Comment by dogorman | Hacker News Reader

ring ring "Hello sir, do you own any firea-" click

I would hang up on that call and I don't even own a gun. Surely some people will answer, but wide swaths of the American public, gun owning or otherwise, would refuse to answer these sort of questions on basic principle and/or common sense. If you could get an estimate for what portion of gun owners relative to non-owners would refuse to answer those questions then you might have something for the statisticians to work with, but I don't see how you could estimate even that.

Surveys give you data, but sometimes that data is worth less than dirt.

View on HN

Yes. It's amazing to me that people give any credence on telephone surveys about sensitive issues.

Do you own any firearms?

Do you use illegal drugs?

Have you ever had an abortion?

How do you not expect a substantial portion of your respondents to either refuse to answer or simply lie when you ask questions like that?

> How do you not expect a substantial portion of your respondents to either refuse to answer or simply lie when you ask questions like that?

Surveyers do expect it. And any errors are measurable and have been studied:

> Polling potential or existing consumers is often the first step when it comes time for new businesses to enter the market or existing ones to add new products. And in today’s tele-connected world, no longer do researchers have to rely on focus groups, in-depth interviews or direct mail surveys to get the job done. Instead, a host of remotely accessible options exist that enables savvy market researchers the choice of connecting to the market without physically having to mobilize it. One of the oldest and most staid of all the research methods is the ubiquitous phone survey. It’s a no frills, “tried and true” alternative for companies looking for robust information quickly. But it does have its limitations. Take a look at the phone survey’s advantages and disadvantages:

* https://www.cfrinc.net/cfrblog/phone-surveys-data-collection

Also: "A comparison of reliability between telephone and web-based surveys"

* http://dx.doi.org/10.1016/j.jbusres.2007.02.015

There's a whole plethora of papers on the reliability (or lack thereof) of various surveying methods, including over the phone.

> https://www.cfrinc.net/cfrblog/phone-surveys-data-collection

This doesn't say anything about the likelihood of people lying when a stranger calls them up and asks questions about a controversial/sensitive topic.

> http://dx.doi.org/10.1016/j.jbusres.2007.02.015

This paper compares the (claimed) reliability of web surveys with the (claimed) reliability of phone surveys. The paper is behind a paywall, but at least in the abstract, it does not appear to directly address the likelihood of people lying when a stranger calls them up and asks questions about a controversial/sensitive topic.

> There's a whole plethora of papers on the reliability (or lack thereof) of various surveying methods, including over the phone.

That may be true, but you don't appear to have cited any of them. Care to try again?

How about "The Impact of Repeated Lying on Survey Results"?

* https://journals.sagepub.com/doi/pdf/10.1177/215824401247234...

The general point is that surveyors know people aren't always completely honest when asked questions.

Most human beings in general are aware that other humans will lie to them at times when asked questions, so why should surveyors being any different? Whereas people use 'heuristics' to judge another's honesty, surveyors will use quantitate methods.

> https://journals.sagepub.com/doi/pdf/10.1177/215824401247234...

This paper talks about figuring out how to tell when dishonest people are taking the same survey more than once. Also, it's discussing people who voluntarily take internet surveys repeatedly to attempt to cook the numbers, not people who are involuntarily cold-called at an identifiable telephone number, and wish to avoid answering a sensitive question.

Again, not even close to what you're claiming. That's the third non-relevant citation you've put forward. I think we're done here.

> The general point is that surveyors know people aren't always completely honest when asked questions.

Of course they know that. They just don't talk about it.

And no, there's no "quantitate method" to tell how many people are lying, barring an external verification source independent of the survey.

Tell me the approximate percentage of people who will lie when a stranger calls them up and asks them how many firearms they own, then tell me how much money you'd be willing to bet on that being the correct number.

Reply on news.ycombinator.com