Experimental Design – Preliminaries

Preliminaries

Stage 1 – Formulate a clear research question

Start with the big questions – what is it you want to know? Write it down if you haven’t already. Seriously, it’ll help you focus to have the big overarching question there in front of you. That question will shape what you do next.

Normally it isn’t possible to answer your big question in a single experiment so next you should break it down smaller bite-sized questions that are small enough to answer with a single experiment. Now decide on the question you want to address.

Key point 1 – It’s better to answer 1 question well than 3 questions badly

Key point 2 – It’s better to answer 1 important question than 3 non-important ones

Write down what the question you plan to address. Again, I am serious here, one the biggest problems early researchers struggle with is being able to clearly articulate what it is they want to know. Having it clear in your mind from the beginning will make you better at discussing and understanding the data, better at writing it up and better able to spot issues within your design.

You can download a checklist for this and the rest of the decisions here Experimental Design Checklist

Stage 2 – Form hypotheses that answer the question

Make a clear statement that definitively predicts what you expect to see. What is the answer to the question you are posing. This is what your experiment will test.

Next form a series of predictions from the hypothesis;

Key point 3 – Predictions have to be testable

Key point 4Predictions have to be specific

Stage 3 – Answer these three key questions

Q. Do your predictions follow naturally from your hypothesis

If the answer is no, go back to stage 2

Q. Are the predictions testable

If the answer is no, go back to stage 2

Q. Is each prediction the strongest test of your hypothesis?

This last questions is the kicker. You want to walk away from the experiment with the knowledge that you have tested the hypothesis as far as you possibly can. If you are not convinced that you are truly testing it, you’ve guessed it, go back and make your predictions more robust.

Predictions in place? It’s time to move on to choosing your model system…

 

Advertisements
%d bloggers like this: