What differences are between one-factor-at-a-time method and split-plot method? I think I miss something here. It seems to me that one-factor-at-a-time method and split-plot method both change the values of multiple factors, one factor at a time. So what are differences between them? Thanks!
 A: It may be easier to answer if you gave a reference or example of where you are getting your ideas from.
The split plot design in general looks at multiple factors, but works with all of them at the same time.  For example if you have several plots for growing wheat (the term originally comes from agricultural experiments) and you want to look at the factors fertilizer (yes/no) and water (high/low) and it is possible to apply the fertilizer to only half a plot, but the entire plot will receive the same amount of water, then you can have a split plot design.  You would split every plot into 2 pieces and apply the fertilizer to one half (but not the other) randomly choosing which half to treat.  You would then randomize the plots to get either the high or low amount of water.  All combinations of high/low water and yes/no fertilizer are present in the same experiment and you can test/estimate both main effects and the interaction (but the denominator and precision will be different for split-plot vs. whole-plot factors).
In the one-factor-at-a-time method you might do one experiment where you randomly fertilized half the plots and not the others and gave them all the current watering treatment.  Then decide whether the fertilizer helps or not.  Then do a new experiment using the level of fertilizer you chose and randomizing between high and low watering.  This does not give all combinations and only allows the estimation/testing of conditional main effects and not anything about the interaction.
