2008 Unit 2 Methods

Download Report

Transcript 2008 Unit 2 Methods

II. Research Methods
College Board - “Acorn Book”
Course Description
6-8%
Scientific Bases of Psychology
3
Unit II. Methods
Summary Outline
 A. Experimental, Correlational, and
Clinical Research
 1. Correlational (e.g. observational,
survey, clinical
 2. Experimental
 B. Statistics
 1. Descriptive
 2. Inferential
 C. Ethics in Research
4
Unit II. Methods
A. Experimental, Correlational, and
Clinical Research
 Testable Hypotheses
 Operational Definitions
 Correlational Relationships
 “Correlation does not imply
causation”
 Causal Relationships
5
Unit II. Methods
A.1. Correlational
(e.g. observational, survey, clinical
Naturalistic Observation
Case Studies
Surveys
Correlational Research
6
Unit II. Methods
Natural
Observation
Professor
Wainwright’s
painstaking
field
research to
decode the
language of
bears comes
to a sudden
and horrific
end
7
Unit II. Methods
Natural Observation
 Naturalistic Observation Exercise
 by Alan Feldman
 Students work in teams to observe in public
places
 They observe alone, write up observations, and
compare observations later
 Natural Observation in the Real World
 by Marissa M. Sarabando
 Students work in teams to observe in public
places
 They are asked to make a hypothesis of what
they expect to observe
 Share the results in class
 Hypothesis supported or not?
Insight into Negative Correlation
9
Unit II. Methods
Experimental & Control Groups
10
Unit II. Methods
A.2. Experiments
Experimental Design
Selection of Participants
Population
Sample
Random sampling
Assignment of Participants to Groups
Experimental Group
Control Group
Random Assignment
11
Unit II. Methods
Key Ideas in Experiment Design
Treatment of Groups
Variables
Independent Variable (IV)
Dependent Variable (DV)
Placebo
Experimenter Bias (double-blind
design)
12
Unit II. Methods
Identifying Independent Variables and Dependent
Variables
Martin Anderson
 For the following statements create an
hypothesis (Your hypothesis should,
theoretically, be testable). Then identify the
IV and DV.
 Blondes have more fun.
 Hypothesis: Changing people’s hair color to blonde will
increase the amount of fun that they have.
 IV ____________________ DV __________________
 A rolling stone gathers no moss.
 Hypothesis:
 IV ____________________ DV __________________
 A researcher is interested in how the activity
level of 4-year-olds is affected by viewing
Teenage Mutant Ninja Turtles. He shows one
group a 30-minute video of Teenage Mutant
Ninja Turtles and another group a 30-minute
video of Barney.
 IV:______________DV:_______________
 Experimental group:__________________
 Control group:_______________________
An Exercise in Designing
Research
 Turn the saying “An apple a day keeps the
doctor away” into a hypothesis and design an
experiment to test its validity. (This chart
can be used to analyze other experiments)
 Other Ideas for Research Design
An Exercise in Designing Research
Step One: Form an hypothesis
Form an
Hypothesis
State the
relationship you
expect to find.
Hypothesis
The hypothesis needs to be
testable.
Think in terms of operational
definitions.
An Exercise in Designing Research
Step Two: Pick your subjects
Pick Subjects
Describe the
process you will
use to select
subjects for your
experiment.
Subject Selection
The population is the group you
are studying,
Your subjects are a sample from
the population.
An Exercise in Designing Research
Step Three: Assign your subjects to groups
Assign Subjects
Assignment to Group
Describe how you
will divide the
subjects into the
control group and
the experimental
group.
The control group and
experimental group
should be as similar as possible to
each other.
An Exercise in Designing Research
Step Four: Defining your independent variable
Independent
Variable
Describe your
Operational
Definition How are you
going to
measure the IV?
Experimental
Group
Control Group
This group “gets” This group
the IV.
doesn’t “get” the
IV.
(It might get a
placebo,
however).
An Exercise in Designing Research
Step Five: Defining your dependent variable
Dependent
Variable
Expected Result Expected Result
Describe your
Operational
Definition –
How are you
going to
measure the DV?
The DV is the same for both
groups.
You expect a measurable
difference between the groups.
An Exercise in Designing Research
Step Six: Statistical analysis
Analysis of Outcomes
How are you going to see
if the groups are
different?
(You do not need to
identify the exact
statistical procedures)
Comparison of Groups
The statistical analysis
gives you an idea if the
differences between
the two groups are big
enough to be greater than
chance.
Experimenter
Bias
Possible sources
of Experimenter
Bias
Ways to
control for
Experimenter
Bias.
The experimenter may
consciously or unconsciously do
things that can affect
the experiment so that it will
confirm the experimenter’s
hypothesis.
Confounding Variables
Possible Confounding Variables
Ways to
control for Confounding
Variables
Explain Blind and
Double Blind
Designs
Variables (other than the
variables that are being
studied)
that can have an affect on the
outcome of the study.
Confounding Variable
“When Dr.
Henderson
comes in,
everybody
play dead.”
Ethical Considerations
Describe how ethical
considerations will be
dealt with in the
research.
Basic Ethical Principles
•Informed Consent
(Use of Deception?)
•Protection from harm
and discomfort
•Confidentiality of
information about
participants
•Debriefing
participants after the
research
Flawed Experiment
(Source unknown)
 A psychologist wishes to study the effect of a reinforcement
of food on the performance of a fine motor skill involving eye
hand coordination. To accomplish this, he had his subjects
thread as many needles as possible in a five minute period.
 The subjects were divided into two groups:

 Group A
 Group B
Males
20
28
Females
30
22
Total
50
50
 The psychologist explained the tests to each group in the
same way. However, she offered Group A a voucher for a
free lunch for every 20 needles threaded. After the fiveminute time period had expired, he counted the number of
needles threaded by each group.
Flawed Experiment
 The results were as follows:
Total Number of Needles

Threaded
 Group A
80
 Group B
45

Average Number of Needles
Threaded per Person
1.8
0.9
 From these results, the psychologist concluded that the reward
of food caused Group A to thread more needles that Group B.
 What was the independent variable? What was the dependent
variable?
 Which of the two groups was the control group? Why?
 Which of the two groups was the experimental group? Why?
 How could the following variables negate the psychologist’s
conclusions?
 Age of the subjects?
 Sex of the subjects?
B.1. Descriptive Statistics
Measures of Central Tendency
Mode
Median
Mean
Measures of Variability
Range
Standard Deviation
28
Unit II. Methods
Normal Curve - Normal Distribution
68%
Skew – Skewed Distributions
Positive Skew
Negative Skew
29
Unit II. Methods
Creating A Living Frequency Distribution:
A Way to Introduce Key Statistical Terms and
Concepts
Martin Anderson
 Overview
 This exercise demonstrates basic statistical concepts. By






forming a “living frequency distribution” based on height,
students will gain direct, first-hand knowledge of the
following terms and the concepts they represent:
Variable
 Discrete, Continuous
Nominal Classification
 Dichotomy / Trichotomy
Continuum
Measures of Central Tendency
 Median, Mode, Arithmetic Mean
Measures of Variability
 Range
Distribution
 Histogram, Normal Curve, Skew, Outlier
Creating A Living Frequency Distribution (2)
 Materials
 Cardboard sheets for signs
 Marking pen
 30’ length of cord (extension cords connected together
work well)
 Create signs labeled with key terms and concepts
as follows to give to designated students:
 Median, Mode (best to make several of these), Range From
, Range To
 Create additional signs to indicate various heights
from 4’6” through 6’5”:
 Time Required
 This exercise is easily done in one class period.
Creating A Living Frequency Distribution (3)
 Step One: Creating a Dichotomy
 I ask students to divide into two groups - tall individuals on
one side of the room and short individuals on the other
 Step Two: Creating a Continuum
 Step Three: Identifying the Median
 Step Four: Identifying the Range
Creating A Living Frequency Distribution (4)
 Step Five: Identifying the Mode
 Step Five: Calculating Arithmetic Mean
 Step Six: Demonstrating Normal
Distribution
Creating A Living Frequency Distribution (5)
 Step Seven: Demonstrating Skew
 Step Seven:
Discussion
 Step Eight: Follow-up
A Three-Dimensional Model of the Normal Curve
William E. Addison & Kristine R. Hillman
 A common problem is conceptualizing relationships
among areas under the normal curve
 Students have access to a wooden model of the
normal curve.
 The model consists of a total of six pieces, two
each of three different sections.
 2 of 34%, 2 of 14%, 2 of 2%
 The pieces correspond in relative size and shape to
approximate areas delineated by standard
deviation units in an empirical normal distribution
 This helps them to visualize the concept of symmetry
 And how the symmetry of the normal curve relates to
concepts such as central tendency, variability,
relative standing
Normal Distribution Curve
Insight into Negative Correlation
37
Unit II. Methods
Correlation Coefficients
Positive Correlation
Negative Correlation
Discussion: Ways to explain and
demonstrate this to students.
38
Unit II. Methods
2. Inferential
Evaluation of chance
Probability that results are by
chance alone
Tests of Significance
Determine the likelihood that a
specific outcome was obtained by
chance alone
Importance of Random Assignment
39
Unit II. Methods
Is it Representative?
Can we Generalize?
Sample / Population
Replication
40
Unit II. Methods
What is “Significance?”
Common Usage
Important
Meaningful
etc.
Term of Art: Statistical Significance
Reliable
Repeatable
41
Unit II. Methods
Significance Testing
 Significant Result in Research
 Not necessarily large or important
 Not necessarily dramatic
 But probably did not occur by chance
 Null Hypothesis
 Opposite of Hypothesis
 Hypothesis: Anxiety reduces test
performance
 Null Hypothesis: Anxiety does not
effect test performance.
Significance Testing (continued)
 p-level (Probability Level)
 Probability that your null hypothesis is
correct
 Probability that the statistic is really zero
 Examples:
 Difference in means
 Correlation coefficient
 p < .05 (p < .01)
 Your null hypothesis has less than a 5%
chance of being right
 You have less than a 5% chance of being
wrong
Type I and Type II Errors
 Type I Error
 Deciding that one variable has an effect on (or a
relationship to) another variable when it doesn’t
 p-level gives the odds of making this kind of
error
 Type II Error
 Deciding that one variable does not have an
effect on (or a relationship to) another variable
when it does
 There is no easy way to estimate the odds of
this kind of error
C. Ethics in Research
Informed consent
Minimize risk and discomfort
Potential benefits must outweigh
risk to subjects
Confidentiality
Debriefing
Ethics of animal research
Approval of research committee
45
Unit II. Methods
How Psychologists Do Research
Spearman Rank Order
Correlation Coefficient
 We have seven subjects whose art projects are
being ranked by two judges, and we wish to know if
the two judges rankings are related to one
another. In other words, do the two judges tend to
rank the seven art show contestants in the same
way?
The formula for the Spearman Correlation
Coefficient is:
In which
r2 =
Spearman Correlation
Coefficient
∑ = Sum of
D2 = Difference squared
N =
number of subjects
Spearman Rho for Rankings of Two Judges of Art
Projects for Seven Subjects
subject
Judge
A
Ranking
Judge
B
Ranking
D
2
D
Alan
1
2
-1
1
Betty
2
1
1
1
Carlos
3
4
-1
1
Diana
4
3
1
1
Edward
5
6
-1
1
Fern
6
7
-1
1
Greg
7
5
2
4
∑
10
Thus we can see that there is a high
positive correlation, but not a perfect
correlation between the two judges
ratings.