Why do I keep missing confounds and sampling errors in my studies?

You know the mechanics — randomization, controls, power. But knowing the mechanics is not the same as catching the thing that ruins the study: the cause you failed to rule out, the group you failed to include. That gap is not a knowledge gap. It is a habit gap. Look closely at every study that went wrong. You designed it, then you trusted your design. You did not go back and ask what could still be false. The confound was sitting there before you ran a single trial. You simply did not look for it the right way. This will not fix itself with one more paper on methodology. It fixes itself when you build a check you run every time, before you spend a single resource — a short, deliberate attempt to break your own design before a reviewer, or reality, does it for you.