Effect size - cohen's d

#1
Hi everybody
Who can help
I got involved in an experiment at my college working on heart rates in horses before and after training.
I think the experiment is quite ordinary, some horses get trained on a special course and others don't (about 25 in each group). I got saddled with the statistic analysis but I don't know much about stats.
I thought I'd just run the results through an Anova and get a p value and all that but somebody told me to do an Ancova instead.
Why? What's the difference? And what's this got to do with non parametric tests?
I was also told to do an effect test using "for example" a Cohen's d.
Why? What's the use?
Can somebody please help and or tell me where I can find explanations for dummies and online software to do it for me - cuz man, am I stuck.
Sorry if I seem a bit confused - I am.
Thanks for all your help
Sophie
:shakehead
 

JohnM

TS Contributor
#2
You should run a 1-way Analysis of Variance, with "training course" as the independent variable or factor, "post-training heart rate" as the dependent variable, and "pre-training heart rate" as a covariate.

Basically you are comparing the effect of training course on heart rate, but controlling for the baseline heart rates of the horses before the training started.

In other words, the covariate will eliminate any potential confusion: let's say that you discover that heart rates are faster after the special course, but what if that particular group of horses had faster initial heart rates? Would the faster heart rates be due to training course, or just because they were faster to begin with? The covariate will take care of controlling for this, and allow a "clean" comparison of the training courses.
 
#3
Thanks so much

Gee John
I don't know what to say - your advice seems very clear and that's great.
Any idea of some books I can buy to learn more? To learn everything basically: I never did stats and it looks like I'll be needing to use them more and more.
Thanks again
Sophie