The randomized block designs allow the researcher to control for known factors that would contribute to the variability seen in the outcome or potentially confound the experiment. This is done by applying each of the main factor effects on each block. In a non-blocked design, the variability contributed to these factors is absorbed into the error or residual of the statistical test, where when using blocking, the variability due to these blocks is removed from the residual/error terms. Given that most statistical tests are a ratio of a summary statistic and its associated error, decreasing the magnitude of the error increases the likelihood of finding a significant effect.

Provided this information we can easily answer the three parts of the question:

a) Blocking by recreational versus professional runners, allows us to control and eliminate the variability attributed to these two factors.

b) If runners are allowed to choose their shoes, this could result in an unequal distribution of the shoes among those two groups, where professionals choose one type of shoe more prominently than the other, and similarly for recreational runners. Given the uneven distribution, the effect of being a professional or recreational runner would then confound the effect of the shoe type.

c) Replication is achieved through the application of both shoe types within each professional/recreational block. Replication helps us to ensure the outcomes are valid and to eliminate potential confounding variables.