It seems to reason that people who use SnapChat for sexting are doing so to hide the fact that they're sexting, so they might be less likely to admit to it (even on an anonymous survey). Someone who uses text messaging may feel less compelled to hide it at all. This would be especially true for kids (under 18) since you also ask the age on the survey.
Not sure how big a role this might actually play in the results.
OP here.
Thanks for the feedback.
We only surveyed respondents 18+ for this survey. We agree kids under 18 would have been more likely to lie about this type of behavior.
I've read of a technique where for self-reported statistics you mix in a non-pertinent question with a known probability, using Bayes to extract the original prior probability.
For example, instead of asking kids "do you smoke pot?" you ask that of 80% of people and the other 20% get "do you own a dog?"; the surveyor has no idea which question was asked but knows the overall population of dog owners.
Does this technique have a name that you know of and did you consider it?
I know of a method in which you give each respondent a coin and ask them to flip it in secret. If the coin lands heads, answer "Yes," if not, answer honestly. Then, if you ask a question like "Have you sexted?" or "Do you use drugs?", you may get survey results that look like 60% yes and 40% no. You subtract 50% from the yes count to account for the coin toss and get accurate overall results, but the participants have plausible deniability.
However, I can't remember the name of it, nor the article I read about it, and Google isn't helping.
Interesting. I imagine this method would be particularly helpful for non-anonymized surveys.
Obviously, this would only be helpful if you don't have a good idea of how many people would lie in the first place. It also depends on the assumption that people won't lie given this plausible deniability.
The biggest problem I see is that this could only increase the variance of your survey results. The way I see it you have three binomial distributions base on three random variables:
1. The number of people who would answer yes to the survey question if they were honest.
2. The number of people who lied. (This is clearly not independent to the first random variable)
3. The number of people who flipped heads. (This clearly is independent)
The problem is that coin flipping has the highest possible variance of any binomial distribution for any given sample size. So even if the variance created by people lying is completely eliminated, it would be more than counteracted by the variance introduced by the coin flipping.
I still really like this method since the increased variance is a moot point if giving people plausible deniability is the best way to normalize for lying. And you can always increase your confidence in the resulting proportion by increasing your sample size.
It would be interesting to run anonymized and non-anonymized surveys with the coin flip and without to try to determine how much anonymization reduces lying on various survey questions.
One nitpick: You should subtract 50% from the yes count and then multiply by two. So in your example, you would expect that 20% of those surveyed truly sexted or use drugs.
You are referring to a "randomized response" survey[1]. There are also list experiments and endorsement experiments for eliciting responses to sensitive questions[2].
Not sure how big a role this might actually play in the results.