Hacker News Reader

Comment by ckelly | Hacker News Reader

Hi Max, Survata co-founder here. 164 respondents is actually a fine sample size for a quick read on qualitative feedback. Even if this had been a quantitative study, that would represent a 7.7% margin of error at a 95% confidence level. Counter-intuitively, you don't always need thousands of respondents to measure an audience's behavior!

View on HN

The calculation (1.96 / 2) / sqrt(164) for the margin of error isn't applicable in this case because it's not a Bernoulli pass/fail event, and is instead an open-ended question.

If you framed the question as "Do you think this is a sex object? Yes/No" then the margin of error might be more applicable. I count 10/164 (6%) who did: stating that 94% of people don't think AirBNB's logo is a sex object is a much better lede.

Correct. I had tried to clarify the margin of error calculation wasn't relevant here by saying "Even if this had been a quantitative study"...sorry if that wasn't clear.

I count 17 that classified it as a sex object. Its worth noting that this survey was sampled after it was released to the public and a controversy had already stirred.

I was impressed that one respondent said "ouija board planchette".

The open ended nature of "what do you see" is totally questionable. The 'do you think this is a sex object' is a much better way to ask that question as even though its leading, its also the kind of thing you cant 'unsee'. "Does this look like balls?" It does now that you mention it, yes.

Reply on news.ycombinator.com