Hacker News new | ask | show | jobs
by SkittlesNTwix 4715 days ago
I've taken a Pew Research Poll before. What struck me immediately was the inane and terrible format of the questions. I actually said to the individual giving the poll, "None of these questions really allow me to confer how I feel about the issues that I know you're asking me about. They're basically nonsense."
2 comments

Is it because it sets up a situation where there are only two sides? That thought crossed my mind when it was the state or religion, and excluding a possible third option that could describe the way people feel.
I wonder if there is room in the market to disrupt polling then. I have never been part of a poll but have heard similar bad things about their structure, so I wonder if anyone is out there trying to make it more accurate.
On the technical side you need two things for disruption:

1. A sane test design. Before trying to pull an MVP here, please take some psychology classes on that topic since there's already a huge body of work on how to design a test in a way that lets you infer reliable data, even if participants think the test is nuts (and try to sabotage it), or if they try to second-guess "desired" answers ("Of course I'm not racist").

2. Comparable data sets over time/regions. People are interested in how things change. That's the incumbent advantage in this market, they have all that stuff.

#2 also helps to get _some_ useful information out of insane questions (since they're consistently insane over the years).