# Correlational and Survey Research by 7U2odi

VIEWS: 12 PAGES: 33

• pg 1
```									Correlational and Survey
Research
PSYC301: Week 4
Today’s Agenda

•   Hand in papers (no sharing, I promise!)
•   Correlations
•   Survey Research
•   Entering Data
Correlational Data

• How one variable changes in response to
another.
• Co Relates!
• Covarying, association
Why Use Coorelational Data?

• When we don’t have much control over the
data.
• When we can’t possibly assign groups.
– Impossible to assign…
– Unethical to assign…
What Do Correlations Show?

• A relationship between two variables
occurring in nature.
• How strong that relationship is.
• What the direction of that relationship is.
• But one important thing…
Okay, enough with the animation

• Correlation ≠ Causation
• Shows that there is a relationship
• Does not show that one causes the other.
Correlations

• A third variable could be causing them both:
– As ice cream sales go up, so does deaths relating
drowning.
– Is the Good Humor man a murderer?
The Good Humor Man Cannot Hurt
You
• Spurious relationship
– A 3rd variable is somehow affecting the 1st and 2nd
variables.
– Heat is related to both high ice cream sales and
more people to drown.
– But is heat causing drowning? I don’t think so.
Mediating Variable

• Something that comes between two things
that relate.
• Higher heat = more drowning.
– (We’ll leave the ice cream out of this).
• Fills the logical hole.
• Higher heat = more swimming = more
drowning.
Moderating Variable

• A third variable that directs the other two.
• Let’s assume there’s a negative correlation
between year and school and drinking.
• Freshman drink more than seniors.
– (Drink milk. Because the other stuff’s illegal).
Moderating Variable

• But what if we could make this a stronger
correlation?
• Students who live on campus drink more.
• Living on campus is a moderator.
• Younger students drink more, especially if
they live on campus.
Today’s Lecture is Brought to you by
the letter r
•   How do we measure this stuff?
•   Many ways.
•   Pearson’s correlation.
•   Abbreviated as r.
Pearson’s r
• Shows how much two variables correlate.
• Measured from -1.0 to 1.0

- 0.43
- and + indicate direction.    The number represents
percentage two variables relate.

Nothing is 100%!
Some Examples
• Visual representation
Correlation Between GPA and SAT   r = .62
4
3.8
3.6
3.4
3.2
3
2.8
2.6
2.4
700       900    1100   1300    1500
Some Examples
Correlation Between GPA and Average Drinks Per Week

4                                                               r =- .62
3.8

3.6

3.4

3.2

3

2.8

2.6

2.4
0      2         4        6        8        10       12   14
Some Examples

Jack Bauer’s Damnits and Donovon McNabb’s
6                  Touchdowns in a Week
5
4
3
2
1
0
0     50      100      150
Survey Research

• Correlational
• Helps us answer a question with things that
• Say we want to look at general satisfaction
with George Mason University and overall
course performance.
What makes up a survey?

• Predetermined set of questions.
– Operational definitions!
• Representative sample.
– Is going to an old folks home going to help with
our research?
– Our sample tells the story about the population.
– So what would be a good sample?
How do we sample?

• Simple Random Sample
• Everything has an equal chance of being
samples.
• Stratified Random Sample
• Split up into groups.
Convenience Sampling

• Non-probabilistic
• Participants are involved based on their
willingness to respond.
• Anyone do Sona-Systems for fun?
Avoiding Bias

• Polarized responses
– People who have a strong negative or positive
feeling are more willing to share than people who
are so-so.
Types of Biases

• Selection Bias: using a method of selection
that won’t fairly select a group that represents
– 85% of people disagree with teaching evolution in
the classroom.
– But this is your sample:
Many types of methods

•   Phone
•   Internet
•   Mailing
•   Public Places
Using Surveys in Research

• How does depression work throughout a
lifespan?
• Cross Sectional
• Longitudinal
Cross Sectional

• Many people are given surveys at one time.
• Survey many different ages to assess depression.
Longitudinal
– One group of people are given surveys over time
– Survey one group of people over a 20-year period.
Self-Report

• Potential Problems?
• Ways to assess self-report:
– Likert-type scale
• On a scale of 1 to 10…
– Yes or No
• Do you like coffee?
– Multiple Choice?
• Starbucks, Jazzman’s, etc.
How do we know it’s a good survey?

• Reliability
• Validity
Reliability

• Reliability
– How consistent your survey is
– If you gave the survey to the same group of
people twice, would the answers be the same?
Validity

• Validity
– Does your survey say what you want it to?
– Operational definition!
• Many Types
– Construct Validity
– Convergent Validity
• Does it agree with other measures?

• Life Satisfaction Survey