Paired data or not...


I'm conducting an investigation into daily water use at an establishment before and after water saving devices were installed. We have considerably more readings taken prior to installation than after. Would this be treated as unpaired data for statistical tests?

Thanks for any advice!
In this case, you would like to make an inference about daily water use before and after a treatment where measurements are made from one facility. With the information you have provided, there appears to be no reason why measurements specific measurements after the treatment would be dependent on measurements after the treatment. Hence, you may use an unpaired analysis. A paired scenario would occur when (for example) there were many facilities for which you would like to measure daily water use before and after treatment. The two measurements from a single facility would be a dependent pair. Hence, a paired analysis is appropriate.