Nonparametric Statistics: Rank Tests, Spearman & Bootstrap

What you'll build

What you will be able to do

Each lab ends with a concrete result that you build yourself and the checker confirms, cell by cell.

• Measure how skewed Larkspur Delivery's delivery times are, compare mean and median, and draw the histogram that shows why a t-test is a poor fit here.
• Rank the late deliveries of 12 couriers with RANK.EQ and RANK.AVG, rebuild the average rank of ties by hand, and check the rank sum n(n+1)/2.
• Compare delivery times with the old and the new routing app: rank all 22 orders, add up the ranks per group, compute U and test it with the normal approximation.
• Test whether a route training made 12 couriers faster on the same test route: differences, ranks of |d|, signed ranks, W+ and the normal-approximation p-value.
• Compare satisfaction scores in three delivery zones: rank all 20 answers, add up the ranks per zone, compute H and its p-value with CHISQ.DIST.RT.
• Correlate menu photos and weekly orders of 12 restaurants with Spearman and Pearson, and see on a scatter chart why one new restaurant breaks Pearson but not Spearman.
• Test whether 16 customers prefer the new insulated food packaging: signs of the differences, plus/minus counts and an exact binomial p-value with BINOM.DIST.
• Turn 50 bootstrap means of a skewed delivery sample into a bootstrap standard error and a 95% percentile interval with PERCENTILE.INC, and compare it with the classic t-interval.