Psych 611 Advanced Statistics and Research Methods in Psychology

1
Psych 611Advanced Statistics and Research
Methods in Psychology

• Christopher Kello

2
(No Transcript)
3
Scale-Free Networks

• Internet connectivity
• Social ties
• Corporate holdings
• Semantic associations
• Biological / ecological networks

Previous picture taken from http//www.smu.edu/mat
h/netw.html
4
A simple law holds amidst all the complexities
of scale-free networks
5
Human Behavior

• Inherently complex and messy
• Often appears as noisy in data
• Yet often yields signals or patterns that are
  informative about
• Cognitive and social processes
• Clinical treatments, educational practices
• Opinions, attitudes, trends, etc.

6
Psychological Data

• Research Methods Ways of measuring behavior in
  order to yield signals/patterns
• Statistics Tools for finding signals/ patterns
  in behavioral measurements

7
Who We Are

• Chris Kello
• Richard Hermida
• David Kidd
• Joe Luchman

8
Course Structure

• http//archlab.gmu.edu/people/ckello/psyc611fall07
  .htm
• Textbook
• Lectures
• Literature review
• Labs
• Exams
• Presentations

9
First Lecture

• Research methods and statistics are ALWAYS
  fundamentally about variability
• Within- and between-subjects variance
• Observed, expected, residual variance
• True variance vs. error variance

10
Methods and Statistics

• Are useless without well-formulated questions
  and hypotheses
• Questions spring from desires to know, help, fix,
  improve, etc.
• Questions MUST be posed so that observations can
  answer them
• Hypotheses are possible outcomes of pursuing a
  research question
• Often Hnull versus Hresearch
• Usually should have a basis in theory

11
Descriptive Statistics

• Are exploratory, they help to paint a picture
  but not answer research questions by testing
  hypotheses
• Histograms, scatterplots, line and bar graphs,
  many other types of graphs
• Measures of central tendency, variance, relation,
  deviations from normality
• Dont underestimate the value of using
  descriptive statistics, but use with caution

12

Frequency
Lateral Deviation
Lateral Deviation
Ordered Free Throws
13
Inferential Statistics

• Are designed to answer questions by testing
  hypotheses
• Inferences based on samples drawn from some
  population
• Different types come with different assumptions
  that must satisfied in order to use them

14
Basic Design Types

• Experiments
• Independent vs. dependent variables
• Random sampling / random assignment
• Minimize error variance / confounds
• Test causal hypotheses
• Observational a.k.a. correlational studies
• Explore and examine relations among vars

15
Need to Know

• Formulas and concepts behind
• Mean, median, variance, standard deviation,
  standard error of the mean, z-score, correlation
  (z-score formula), covariance, linear regression
  (z-score formula), coefficient of determination
• Concepts behind
• Skewness, kurtosis, central limit theorem,
  quartiles, outliers, difference between
  correlation coefficient and covariance
• Be able to interpret
• Histograms, box plots, scatter plots (positive,
  negative, no linear relation, non-linear
  relations)

16
www.ojp.usdoj.gov/.../documents/documentff.html
17
www.omninerd.com/2006/04/21/articles/50
18
Standardized Observed (Actual) Mean Reaction Time
Standardized Model (Expected) Mean Reaction Time
19
data.princeton.edu/stata/Introduction.html
20
(No Transcript)
21
Homework 1

• Collect at least 50 pairs of observations for an
  empirical phenomenon you find interesting
• Can be any kind of data, not just behavioral
• Observations must be integers or reals

22

• Write a brief report (500-700 words)
• Introduction
• Lead reader into your research question and
  hypotheses with motivation / background, striving
  to achieve the aims listed on Methods and
  Statistics.
• Methods
• How observations were collected, design params in
  the terms listed on First Lecture, Inferential
  Statistics, and Basic Design Types overheads.
• Results
• Report / graph / address everything listed on
  Need to Know overhead (use z-score scatter plots)
• Discussion
• What were the most interesting / least boring
  aspects of the results? What can one take away
  from these results?