Measure twice, cut once

While “measure twice, cut once” is an old proverb that’s usually associated with carpenters, it applies very well to a lot of situations that arise in grad school, especially when getting ready to run an experiment.

The best thing you can do, is take the time to make sure you’ve sorted out all possible (foreseeable) problems before you run your experiment. While this may mean it takes longer to actually run your experiment, it will also save you a lot of time. It will decrease the number of participants you need to find (you can’t use the same ones twice) and you won’t have to analyze data twice. All in all, it’ll save you a big headache when you try to find out what went wrong.

So what are some areas you can “measure twice”?

1. The wording of any questions you ask participants. If you have a survey, or a form that participants will fill out, have people who won’t be involved in your experiment look it over. And then listen to them if they say anything is unclear.
2. It’s likely you’ll be giving your participants instructions, possibly both written and verbal. Do what you can to make them short. Participants don’t want to sit through a long lecture, or read a long page on what they need to do. Most likely they’ll miss an important detail, and that may make their results unusable. Try to write your instructions in short sentences, and use either a numbered list or bullet points.
3. If you’re writing code, or making a system, test it out. And when testing it, don’t just do what you want the participants to do, but try your hardest to break it. In my experience, participants will find the oddest ways to break your system.
4. What information are you trying to collect from your experiment? Make sure you know exactly what information you need before hand, and then make sure you’re actually collecting it. It’s very difficult to try to get useful information out of data that doesn’t answer your question.
5. Where is your experiment taking place? And/or what equipment is needed for it? Once you know those, test your experiment using the same equipment or location you will use during your experiment.

Even after you have done your best to check everything over, do a dry run of your experiment. Get friends, family, co-workers, fellow students, etc, to come and run through your experiment. Have them be brutally honest about anything they find unclear, confusing, odd, etc, and then take the time to fix it. If they uncover any big problems, you’ll need to do a second dry run (with different people). It’s a lot of work (but all of grad school is), but it’s worth it, when you finally get around to running your experiment and everything goes well.