4

I'm doing a stationary sport like golf, so the details matter.

I'm recording each of my scores with detailed information about the equipment I'm using, and the weather conditions. So 1 score has 2-3 pieces of equipment attached to it.

I can plot averages for may main piece of equipment, but then how do I isolate the effect of the others(performance accessories/safety equipment)?

mdewey
  • 16,541
  • 22
  • 30
  • 57
Andrei Erdoss
  • 141
  • 3

2 Answers2

4

You can apply multiple factor ANOVA with your score as dependant variable and equipment and weather conditions as factors. You would need to meet certain assumptions:

1) scores should be normally distributed under each factor

2) ratios of variance between groups should not exceed 3 or be less then 1/3

3) most importantly: all scores must be independent among and within groups

Germaniawerks
  • 1,027
  • 1
  • 10
  • 15
  • 1
    +1 The second item isn't actually an assumption of anova. The *assumption* is equality of variance. What you have there is advice about how much heteroskedasticity can be tolerated. I'm curious about how they arrived at those particular numbers. Is there a paper from which this advice comes? – Glen_b Jan 27 '14 at 08:39
1

Depending on your response variable which might be skewed (e.g. distance from tee, distance from hole), or counts (frequency of some event, like 'on the green'), you'd likely use some form of GLM. In some cases, the response might be nearly normal, in which case you might use some form of regression/ANOVA.

But before you worry about analyzing data you need to design your experiment; a good experiment will mix and match your various types of equipment in an appropriate manner (to your assumptions and hypotheses - if you aren't looking for interactions, it's relatively simple), and will likely randomize these combinations to different holes, preferably more than one per combination of equipment. If you want to estimate interactions, you could be in for a lot of holes of golf.

Glen_b
  • 257,508
  • 32
  • 553
  • 939