This is massive topic - people write whole books on this. Normally to find something out about the world, you'd do something, and see what happened. Then you'd change just the thing you were studying and see what happens then. You'd do this lots of times and hopefully get some consistent results.

For example, I dangle a weight off a spring and measure how long the spring is. I dangle another weight off the same spring and measure how long the spring is. I do this lots of times and discover that the heavier the weight is, the more the spring stretches. I maybe even get to a point where if you tell me the weight, I can predict how much the spring will stretch.

When you try and apply this type of approach to people (controlled experiments - not dangling weights off them), things get really tricky. Firstly it's really hard to change just one thing with people. If I change something as fundamental as my delivery style, I can't help but change a load of other things - the length of the session, my body language, the amount of activity to name but three.

Even if I get a reliable correlation between two things that doesn't mean that one causes the other. For example, in children of primary school age, there's a strong correlation between ability to do sums and shoe size. That doesn't mean we should include foot stretching as part of the school syllabus, it just means that as kids get older they get better at sums and their feet grow.

There are also all sorts of odd effects when experiments involve people, for example Rosenthal effects where the experimenter knowing what answer they expect to find, unconsciously changes what they observe or Hawthorne effects where the novelty of changing to doing things a new way produces a short improvement whether or not the new way of doing things is intrinsically better.

So now let's try and invent a way of testing coaching styles.

Attempt 1: I get a load of students and I decide I'm going to teach them about forward paddling whilst delivering using different VAK modes. I have someone who looks confused when I talk about forward paddling, but when I do a demo, they suddenly get it. So they're a visual learner and VAK works, right?

Well maybe, but what we just saw doesn't prove that. Maybe they were going to get it anyway and just needed an extra minute to think about it. Maybe it was something they picked up from my demo that I could equally well have verbalised, but I didn't.

Attempt 2: This time I get a load of students and split them into two groups. One group I teach in an auditory style, one I teach in a visual style and I see which group does best.

Now we're getting somewhere, but what if one group was naturally better at forward paddling than the other? What if I'm just better at coaching visually than I am verbally?

You can go round and round in circles trying to design experiments to test coaching, but it's really hard to come up with a system that you can't pick holes in. One design that has been suggested is;(11)

1. Test a load of learners to find out what their learning preferences actually are (but don't tell them).
2. Randomly assign them a learning preference (which might or might not be the same as their actual preference).
3. Tell the coach what their randomly assigned preference is, (but not what their actual learning preference is).
4. Measure how good the learners are at something.
5. Get the coach to coach that thing to suit their randomly assigned style.
6. Measure how how much the learners have improved.

If there's anything to the idea of learning styles, you'd expect that those who've been coached in a style that matches their real learning style would do better than those who have been coached in a way that doesn't mesh with their learning style. In practice, very few experiments have been done with this kind of structure (and I don't know of any in sport), and those that have find that there's little or no evidence that coaching to match the particpants learning style makes a difference. (10)

In psychotherapy, where there's a similar situation with several competing models of how people's minds work. Studies that have been done (The Great Psychotherapy Debate, Bruce E. Wampold), find little differences in the efficacy of treatment provided by therapists who support very contradictory models. What they found was significant was the length of time the therapist had been practicing and their own personal level of belief in the model they subscribe to. Perhaps for us as coaches this is telling us that it doesn't matter what we believe in, as long as we believe in something and that the way to get good at coaching is to do it lots!