Complex Experimental Design and Multi-Step Methods (SIN 303, 402, 501)

Passages with three experiments are rarely three copies of the same test. Usually Experiment 1 holds one variable fixed while varying another, Experiment 2 flips which one is fixed and which varies, and Experiment 3 lets both vary at once. The fastest way through this is to build a grid: one row per experiment, one column per variable, and fill in what was fixed, what was varied, and over what range. Once the grid is built, you can answer almost any question just by reading across a row or down a column, instead of rereading paragraphs.

The payoff for crossing variables in Experiment 3 is seeing whether they interact. If temperature alone changes an outcome and pH alone changes it too, that still doesn't tell you whether temperature's effect depends on which pH you're at. Only a design that runs every temperature at every pH can show that. This is also why a complex design is not just a simple design with extra numbers: the whole point is the structure of what's combined with what, not the count of trials. Treating it as bolted-on arithmetic misses the actual question being tested.

This structure connects directly to control and instrumentation. If one experiment in the chain switches to a different tool, like a titration instead of a probe, any difference in readings is now tangled up with the method change itself. You can't tell whether a gap in results reflects the real variable or just the new instrument's calibration, so the design loses its ability to isolate cause from measurement.