A good experimental design should always: include more than one independent variable have predictable outcomes be repeatable more than one of these