Transcript Introduction

Introductory Topics
PSY 211
8-29-07
Scientific Method
Why Statistics?
• Common sense is often mistaken
• Cognitive biases:
– Hindsight bias*
– Overconfidence bias*
– Positive illusions*
– Confirmation bias
– Fundamental attribution error
– Errors in emotional prediction
– Me-too fallacy
Science
• Common sense and past findings can
serve as theory for new studies
– Working model designed to organize or
predict behaviors and events
– A narrative (story) explaining why things are
what they are
• Test theories using scientific method
Scientific Method*
1.
2.
3.
4.
5.
6.
7.
Theory
Hypotheses made
Design study
Run study
Analyze results
Draw conclusions
(Make new, better theories)
Types of Studies
Correlational Studies
• Examines whether two variables are
associated (related)
• Correlation coefficient (r): -1 to +1
• Sign indicates direction of relationship
• Magnitude indicates strength of relationship
• Examples: Surveys, ratings, performance,
physical measurements
• Is depression correlated with anxiety?
PSY 100… Correlation ≠ causation
PSY 211… Correlation = causation sometimes
Experimental Studies
• Rather than observing two variables, we
manipulate one and observe the other
• Independent variable: experimentally
manipulated
– Control condition vs. Experimental condition
– Participants randomly assigned to condition
• Dependent variables: outcome variable,
depends on the independent variable
• Examples: drug trials, interventions
Does Prozac decrease depressive symptoms?
IV:
Treatment
Control
DV:
Of course, experiments have flaws too…
Reporting Results
Statistics
• Definition: using numbers to summarize or
interpret information
• Population: large group of interest
– “Adults” “Americans” “Depressed children”
“Europeans over 65 at risk for heart disease”
– Studies are designed to answer questions
about a population
– Problem: unless the population is very small,
not all will be able to participate in the study
• Sample: small group similar to the
population; participates in the study
– Stratified sample
– Random sample
– Convenience sample
• Use numbers to describe groups
– For populations, numbers called “parameters”
– For samples, called “statistics”
PARAMETERS
STATISTICS
=M
= SD
Sampling Error
• Samples usually do not perfectly represent
the population
• Sample statistics usually differ from the
population parameters
– Sampling error
– “chance” “fluctuations” “bad draw”
• Sometimes sampling error is big;
sometimes small… it’s all probability
Reporting Results
• Descriptive statistics: used to describe the
sample
– M, median, mode, SD, r
– Early focus in this class
• Inferential statistics: used to draw
conclusions about the population
– Significance tests, confidence intervals
– Late focus in this class
Homework #1
• Continuous variables:
– Ordinal scale: rankings (1st, 2nd, 3rd, etc.)
– Interval scale: ranked, and space between
scores are the same size
– Ratio, ranked, space between scores are the
same size, have absolute zero point
• Categorical variables:
– Nominal scale: labeled groups
• Follow the guidelines for easy points
• Brainstorm ways your questions could be
misinterpreted or leave people out
• Ask a friend/classmate to try to find flaws in
your questions, and throw out bad ones
• Use available references
• If stumped, brainstorm by checking this site
(no copying):
http://ipip.ori.org/newIndexofScaleLabels.htm
There have been times in life when I wished
everything was over:
1
2
3
4
5
Completely
False
Mostly
False
Neutral
Mostly
True
Completely
True
Suicidality is associated with a number of
important variables, such as stress, depression,
hopelessness, and maladjustment (Myers,
2001). This item measures suicidality because
suicidal people want to give up on life.
Because people may be defensive in
responding, I tried to soften the item by
avoiding the word “suicide”.