Experimental Psychology PSY 433

1
Experimental PsychologyPSY 433

• Chapter 2
• Observation and Correlation

2
Does Country Music Cause Suicide?

• http//www.youtube.com/watch?v-Xu71i89xvs
• Stack Gundlach found that metro areas that
  played more country music had higher suicide
  rates, concluding that country music causes
  suicide.
• Maybe depressed people seek out sad music?
• http//www.jstor.org/stable/2580303

3
Non-Experimental Research

• Variable -- a characteristic that can have
  different values (height, weight).
• Value -- usually a single, specific number (6
  feet tall, 140 pounds).
• Measurement the process of assigning numbers to
  entities in the world.
• In non-experimental research, variables may be
  measured, but nothing is being manipulated by the
  experimenter.
• No independent variable, just DVs.

4
Naturalistic Observation

• Methods for observing behavior in its natural
  environment.
• Behavior is complex and humans have limited
  attention span, so we delimit, or narrow, the
  range of behaviors we plan to observe.
• Reactivity -- subjects may behave differently
  than usual when they know they are being observed.

5
Unobtrusive Observation

• Unobtrusive observation -- subject is unaware of
  being observed in presence of observer
• Example chivalry study, bathroom study.
• Unobtrusive measurement -- observer collects
  evidence in absence of subject and infers
  behavior of subject.
• Example graffiti study, collecting scat
• There is a danger in anthropomorphizing or
  incorrectly interpreting what is observed.
• Participant observation going native

6
Survey Techniques

• Gives a picture of peoples attitudes, beliefs,
  behaviors, and feelings about a topic.
• Sample from a population, then infer based on
  sample -- only as good as the sample.
• Return rate may produce sampling problem.
• Collect large amounts of data from large number
  of people quickly.
• Does not show causality among variables.
• Can also be used to provide data for the
  correlational method.

7
Relational Research

• Contingency Research
• Variables are presented in a contingency table.
• A Chi Square statistic is computed to determine
  whether relationships among variables exist.
• Values in the tables are counts or frequencies
  for categories, not measurements.
• Data is ex post facto

8
Correlational methods

• Correlation a statistical technique that
  expresses the degree of linear relationship
  between 2 variables.
• If the correlation is high, a strong linear
  relationship exists.
• If the correlation is low, a weak relationship
  exists.
• If the correlation is zero, there is no
  relationship.

9
Correlation Coefficient (r)

• r is a numerical index of the degree of linear
  relationship between 2 variables.
• r is computed by taking into account pairs of
  scores one score from one variable and the
  other score from another variable.
• Correlation coefficient (r) has a strength (0-1)
  and a direction ( or -).
• r allows us to more precisely compare different
  sets of variables
• SAT GPA vs IQ GPA

10
Using Coefficient (r)

• Does income level predict reading level?
• Measure income level at grade 1
• Measure reading level at grade 1
• Compute correlation between reading level
  income
• What if r 1.0? What if r -1.0?
• What if r .88? What if r -.88?
• What if r .15? What if r -.19?
• What if r 0.0?

11
Values of r for Multiple Variables
12
Causality and Correlation

• The directionality problem -- for any correlation
  between X and Y
• X may cause Y
• Y may cause X
• Z may cause both X and Y
• Classic examples
• Hours watching violent TV violent behaviors ()
• Grades in physics grades in statistics ()

13
Spurious Correlations

• Spurious correlation -- a correlation exists but
  no causal relationship exists.
• Spurious correlations generally occur because
  both variables are mediated by a third variable.
• Classic examples
• Number of churches in a town and number of
  murders.
• Number of toasters owned in a household and
  number of teen pregnancies.
• Kids in private schools get higher test scores.

14
Biases

• Selection bias a kind of spurious correlation
• Students in Mississippi had a higher average SAT
  than in California even though California spent
  more per pupil than Mississippi.
• In Mississippi only top students took the SAT
  whereas in California nearly all took it.
• Restriction of range -- a correlation may
  underestimate the relationship between two
  variables if the range of either variable is
  restricted.

15
An Example of Restricted Range