1
$\begingroup$

I have daily surveys sent to users of several of my organization´s services, the surveys consists only on NPS question (How likely would recommend X? on a 0 to 10 scale), number of respondents vary by day and service.

I would like to get the NPS score and confidence interval for a service on a monthly scale. Is it ok to joust aggregate the data points and calculate CI as in a cross-sectional survey?

After calculating monthly NPS I would like to compare scores from different services. Wold this be a valid approach?

Supposing the number of respondents is equal or grater than the number of respondents required in a cross-sectional approach. Is this a valid way to obtain a representative sample of the users of such services over a month?

Finally I would like to compare the NPS of a service before and after some significant project is made to improve the service. Projects usually take a few months to implement and have a marked go-live date. Wold comparing the NPS of months after the improvement to the NPS before the improvement be valid?

Thanks for your advice.

Improve this question
$\endgroup$
2
  • $\begingroup$ Have you searched our site? Look for stats.stackexchange.com/search?tab=votes&q=nps. $\endgroup$
    – whuber ♦
    Sep 20, 2018 at 22:40
  • $\begingroup$ Yes I have, It was very useful for calculating CI for NPS, but I think my question is more related to the data acquisition method I am using. $\endgroup$
    – juan alvarez
    Sep 20, 2018 at 23:02

1 Answer 1

Reset to default
-1
$\begingroup$

The problem with NPS is that it’s usually calculated as a ratio (prom - det) / # surveys. This doesn’t lend itself to a CI unless you want to analyze raw opinion rankings. You can use a bootstrap approach to estimate confidence intervals around NPS %, then compare across groups, etc.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Yes as a difference in proportions the sampling distribution is normally distributed, hence bootstrap and CI work as usual. But will this sampling method would yield a valid random sample? $\endgroup$
    – juan alvarez
    Sep 21, 2018 at 17:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.