Assign 50 chickens to groups randomly, record eggs per unit time.
Use all hens on the new feed for a given unit of time, then switch to old feed and compare results.
Second option
Show the same group of poeple some imiages of sexy products and others of normal products, and record results, have randomizaiton and indepndence within images.
Have two groups and show one group sexy images and anohter normal images, record results.
Pair groups, since the groups were the same through both time periods (same locations)
There is statsitically significant difference between the two time period, given that there is only a 2% chance we saw the results we saw due to random sampling variation alone.
Paired since the two groups of consequtive fridays are not independent
We can conclude that less people drive on the road on friday the 13th, since there is only a 2% chance that we saw the results we saw due to random sampling variation alone.
NPP is suitably straight and we can assume that these fridays were chose at random.