Correlation & Regression
1) Pearson correlation
relationship between 2 continuous variables
types of relationships
r Pearson product moment correlation coefficient
t-test for r
2) single variable line
Running Head: HORMONES ON ATHLETIC PERFORMANCE
The Effect of Hormones on Athletic Performance
University of Western Ontario
1
HORMONES ON ATHLETIC PERFORMANCE
2
In hopes to create a super athlete, our
Week 1: Introduction
What am I doing here?
Why is my evil professor making
me take statistics?
Week 1: Intro
1
Some Basic Concepts:
Getting Your Head Back into it
a database
final grades (%) from rece
Repeated-Measures Designs
also called within-subjects designs
can be any # of factors
we'll stick to 1 factor today
> 2 levels (conditions)
or
else, would be a paired-samples t-test
Repeated Meas
One-way
Independent-Groups Anova
1) when is a one-way Anova used?
2) sources of variation
3) an example of a one-way analysis
of variance
4) assumptions underlying F-distribution
Week 4: One-way Anova
Monte Carlo Studies
Switch gears
Be statisticians, rather than researchers
Compare empirical vs. theoretical distributions of statistics
Look at the effect that violating assumptions has on the t-test
Week 2: t-tests
3 types of t-tests:
1. Single sample
2. 2-sample test:
Independent groups
3. 2-sample test:
Paired groups
(correlated, dependent)
What are they and when do
you use them?
Examples of in
Completely Randomized Factorial Design
any number of factors (IVs)
we'll stick to 2 today:
e.g., effects of alcohol (1 vs. 6 oz)
and alcohol tolerance (low vs. high)
on a reaction time vigilance bu
Split-Plot Analysis of Variance
1) intro to split-plot (mixed) ANOVA
2) a short example
3) a medium length example
4) a long example
5) choosing error terms for simple main effect analyses
Split-Plot
High-level Review
1) Background stuff
assumptions of t and F distributions
effects of violating them
Monte Carlo investigations
Type I error, Type II error
, , power
Review
1
2)
Hypothesis Testing
Dif
Factor Analysis
NOT used for testing the effect of a treatment variable on some
dependent measure, as in t-test or ANOVA
in fact, no IV/DV distinction
*no significance testing involved*: no treatment
Multiple Correlation and Regression
1 dependent variable, a bunch of independent variables
> 1, at least
What is the best way to predict the DV from this set of IV's?
e.g., have a bunch of persona
Chi-square (2)
1) some background on 2
2) 2 test of an association between 2 nominal
variables
- basic 2 x 2 (2 variables, 2 levels each)
3) 2 test of association with variables with
more than 2 level
1
Examining Power and Type II Error in Monte Carlo Simulations
The University of Western Ontario
In the present study, researchers used the MONTE program to perform Monte Carlo
investigations of the t