If your friend is comparing two fuel formulations in a lab, cannot test them on the same engines, and he/she knows very little about the data (e.g., normality characterization, population variance). Which of the following procedures would you recommend to produce a confidence interval on the difference of average the fuel performance? Furthermore, what is the minimal number of tests that should be conducted on each fuel formulation?

(a) Paired sample t-interval.

(b) Paired sample z-interval.

(c) Two-sample t-interval, general procedure.

(d) Two-sample z-interval.

(e) Two-sample t-interval, pooled variance procedure.

Suppose we produce a 90% confidence interval on the true mean of a population. Explain what the 90% level means. In other words 90% corresponds to the probability of what exactly?