0
$\begingroup$

I’m trying to determine a valid method for quantifying the maximum spread of values I can statistically expect to see.

I’ve collected 12 sensor values from 12 different sensors. I’ve calculated the mean of the data and the standard deviation and then calculated 1.96 standard deviations from the mean, this gives me a wide spread of the data. But this assumes the mean remains stationary. Is it not therefore fairer to calculate the 95% confidence interval for the mean and the calculate the spread of data (1.96 standard deviations) from the max and min of the means within the confidence bound? Is this valid given the sensor measurements are not mean sensor values, but rather just a one off reading from the sensor?

$\endgroup$
0
$\begingroup$

The question is a little ambiguous. You collected data from 16 different sensors (1 measurement from each sensor) and now you are trying to estimate a mean and confidence interval for the global mean of all the sensors? That's what the confidence interval does. For ex. a 95 % CI will return an interval which will contain the true (global, in this case) population mean 95 times out of 100 (if you were to repeat this sampling procedure 100 times). Note: a CI returns an interval for the mean, not the distribution of your data. It does assume that the mean is "stationary", in that the mean does not change over time. If you are worried about the fact that you only have 1 measure for each sensor and this might affect your estimate, for ex. if each sensor is very variable, then there is a way to correct for this using within and between variance, if you had more than 1 measurement. Since you don't, there is not much you can do about it.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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