Scientific Methods II: Correlational Research Discuss correlational research Example: Research on television and aggression.

Scientific Methods II:Correlational Research

• Discuss correlational research

• Example: Research on television and aggression

Example of Correct Answer

• Hypothesis: Children who play violent video games will behave more aggressively.

• IV: Children play either a violent video game or a nonviolent video game

• DV: How long kids play with violent (e.g., guns, swords) or nonviolent (e.g., trucks, dolls) toys in a free play situation

Correct Answer (continued)

• Ensure internal validity by– Randomly assigning children to the violent vs. nonviolent

condition– Keeping experimenters unaware of the kind of game played by

the child (to reduce experimenter bias)– Making sure children are not aware of the purpose of the study

(to reduce participant bias)

• Advantage: Can detect cause and effect relationship• Disadvantage: May not generalize to situations outside

of the lab or to all kinds of aggressive behavior

Research Ethics

• Lisa Simpson conducts an experiment on Bart and her hamster. Is it ethical? Why or why not?

Ethical Issues in Social Psychology

• Guideline for Ethical ResearchInformed consent is an agreement to participate in an experiment, granted in full awareness of the nature of the experiment.

Deception is misleading participants about the true purpose of a study

Debriefing is explaining to participants, at the end of an experiment, the true purpose of the study

Scientific Methods II:Correlational Research

• EXAMPLES• "MARRIAGE SLOWS CANCER DEATHS“

Evidence that married people have a better chance of surviving cancer than do singles means that the unmarried might be good targets for cancer-prevention programs. Married people with cancer had a 23% higher overall survival rate than the unmarried.

Example 2

• Isolation increases with Internet use• Psychologist Robert Kraut and his

colleagues report that greater use of the Internet leads to shrinking social support and happiness, and increases in depression and loneliness…And the findings were unexpected, Kraut says, given that most people use the Internet for chat lines and e-mail, not just to isolate themselves in mounds of electronic information.

Example 3

“Revenge of the Beaker Bunch”..Researchers Find that Scientists Live Longer

If it were a movie, it would probably be called "Nerds: The Ultimate Revenge.“ A study spanning nearly 70

years suggests that, all else being equal, scientists live longer than non-scientists… According to Friedman'sstudy, nonscientists are 26 percent more likely to die atany given age than scientists. In a sample of 600 menborn around 1912, Friedman's group found that only 67% of nonscientists were still alive by age 70, compared to 72% of the scientists...”

Correlational Research

• Correlational research examines whether 2 factors are naturally associated. [Co-relation: the degree to which one variable is related to another.]

• Examples focus on naturally occurring relationships – marriage & cancer recovery– Internet use & depression– occupation & longevity

Correlations: The Basics

• Numerical range: -1.00 to +1.00

The Correlational Method

Correlation Coefficient

Positive correlations indicate that an increase in one variable is associated with an increase in the other.

Negative correlations indicate that an increase in one variable is associated with a decrease in the other.

The Correlational Method

The Correlational Method

• Surveys

The correlational method often relies on surveys, as well as on observational data. Surveys are used when the variable of interest is not easily observable.

How do we evaluate correlational findings?

• Amount use Internet is associated with greater depression. Why?

• 3 possible explanations• Using Internet depression • Depression greater Internet use (cannot

determine direction of causality)• A 3rd variable might be related to both

using the Internet and becoming depressed

3rd Variables

• People with few friends may be more likely to spend time on Internet and to become depressed

• People w/low self-esteem may use Internet (to avoid rejection) and are at risk for depression

Evaluating Correlational Research

• Correlational research cannot determine causality.

• Problems knowing direction of causality

• Problems knowing whether some 3rd variable (confounding variable) might explain the observed correlation

How do we evaluate correlational findings?

• Consider in context of research on TV exposure and aggression

• Start w/theoretical idea and to test specific hypotheses.

• Hypothesis: Children who watch more violent television shows behave more aggressively.

TV & Violent Content

• 8 out of 10 Sat. morning programs contain violence

• Sat. morning children’s shows: 20 violent acts per hour

• Average American watches 4 hrs of TV per day, or over 1000 hrs. per year.

• What is relationship between watching TV and children’s aggressive behavior?

TV and Aggression

• Many studies

• Sample correlational paradigm:– Parents record how much TV their kids watch– Teachers record aggressive behaviors

(hitting, pushing, shoving, insulting, etc.)

IV AND DV in Correlational Research

• Quasi-Independent Variable: how much TV kids watch (*not* manipulated, but acts like IV)

• Dependent Variable: amount of aggressive behavior at school

TV and Aggression: Correlational Studies

• Conclusion: Children who watch TV that includes violent content are more likely to behave aggressively.

Why?

• Theory: Watching TV causes aggression

• But: Maybe more aggressive kids are attracted to watching violence on TV (direction of causality)

• Or: A 3rd variable might account for the relationship

Third Variables

• What are some possible third variables that could explain the correlation between watching TV and aggressive behavior?

Marriage and Surviving Cancer

• Direction of causality is not as much of a problem (unlikely that surviving cancer caused marriage!)

• What are some 3rd variables that could account for the greater survival rate of married individuals?

Disadvantage

• Correlation cannot determine causality

• Even if we measure possible third variables (e.g., measure eating habits, measure health care, etc.) and use statistical techniques to “control” for them, we still cannot be sure about the cause-and-effect relationship.

Then…Why do correlational research?

• Allows for the study of problems that would be difficult to study in the lab

• Can’t manipulate who is married or not, who has a stressful job or not, gender, race/ethnicity

• Useful for generating hypotheses that can be tested later in a controlled experiment

• Video on correlational method

• See concepts on outline

Retrospective Study

• Observe some outcome (lung cancer) and then gather information from people with and without the outcome (e.g., about smoking, eating habits, exercise, parenting style)

Prospective Study

• A sample of individuals (preferably a representative sample) is followed over time to see whether certain experiences predict a later outcome (e.g., whether smoking predicts later lung cancer)

Common Cause

• Kids who eat more sugar might be more aggressive (during sugar high) and watch more TV when their blood sugar drops and they become lethargic.

• Common cause would explain why we see an association between TV and aggression

Confounding Factor

• Something that occurs (accidentally) that provides an alternative explanation for the correlation. (Video: speed limit lowered, but confound = gas crisis and less driving)

Coincidence

• Two variables are coincidentally associated, but doesn’t happen again.

Simpson’s Paradox

• Lurking variables can change the direction of an association.

• In video, more men than women applied to law school, and the law school had easier admission standards than the business school. Thus, the overall numbers (across both programs) appeared to show that more men were accepted to the professional programs.

Professional Schools Applied AcceptedMen 360 198 55%Women 200 88 44%*Looks like few women admittedBusiness SchoolMen 120 18 15%Women 120 24 20%*Greater % women admitted than menLaw SchoolMen 240 180 75%Women 80 64 80*Greater % women admitted than men

• Using a correlational design (not a controlled experiment), how would you investigate whether playing violent video games is ASSOCIATED with aggressive behavior in children?

• What would serve as your quasi-IV?• What would serve as your DV?• What procedure would you follow?