Study Notes: Module 0.4 Correlation and Experimentation

Psychological Research Methods

• Psychologists utilize various methods to describe, predict, and explain human thoughts, feelings, and actions.

• Correlational research: A non-experimental method used to describe the relationship between two or more variables.

• Experiments: Research methods that attempt to establish a cause-and-effect connection between variables.

Understanding Correlation

• Correlation: A measure of the extent to which two factors vary together, indicating how well either factor can predict the other. Describing behavior serves as the first step toward predicting it.

• Variable: Anything that can vary and is feasible and ethical to measure.

• Correlation Coefficient: A statistical index representing the relationship between two variables, ranging from $-1.00$ to $+1.00$.

• Scatterplot: A graphed cluster of dots where each dot represents the values of two variables.

  • The slope of the points suggests the direction of the relationship.

  • The amount of scatter suggests the strength of the correlation; specifically, less scatter indicates a higher correlation.

Identifying Correlation Directions and Strengths

• Positive Correlation: Two sets of scores tend to rise or fall together (e.g., height and weight).

  • A perfect positive correlation is represented as $r = +1.00$.

• Negative Correlation: Two sets of scores relate inversely; as one goes up, the other goes down (e.g., the distance from a person's head to the ceiling as they grow taller).

  • A perfect negative correlation is represented as $r = -1.00$.

  • A negative correlation does not indicate "weakness" or "badness"; it only describes the inverse direction.

• No Relationship: When there is no discernible pattern or predictive power between variables, represented as $r = 0.00$.

• The strength of a correlation is determined by the absolute value of the coefficient, while the sign ($+$ or $-$) indicates direction. For example, $-0.85$ indicates a stronger relationship than $+0.75$.

Statistical Illumination and Observation

• Statistics are necessary because casual observation often misses patterns that data reveals.

• Czech and Slovakian Volunteer Study (Polák et al., 2020): 2,291 volunteers rated fear and disgust for 24 animals on a scale of 1 to 7.

  • Animal data averages include:

    • Cockroach: Fear $3.10$, Disgust $4.16$

    • Maggot: Fear $2.90$, Disgust $4.49$

    • Viper: Fear $4.34$, Disgust $2.83$

    • Spider: Fear $4.39$, Disgust $4.47$

    • Panda: Fear $1.57$, Disgust $1.17$

  • Despite the raw data being difficult to interpret by eye, the scatterplot reveals a positive correlation of $r = +0.72$.

• Pattern Recognition: People notice discriminatory patterns (such as gender bias in salary and seniority) more easily when data is statistically summarized than when it is presented case-by-case.

• Examples of Correlational Findings:

  • Positive: College students who sleep more tend to have better academic performance (Okano et al., 2019).

  • Positive: More time spent on social media correlates with higher risk for depression in teen girls (Kelly et al., 2019).

  • Positive: Longer breastfeeding duration correlates with higher later academic achievement (Horwood & Ferguson, 1998).

  • Negative: Higher consumption of leafy vegetables in older adults correlates with less mental decline over 5 years (Morris et al., 2018).

The Limits of Correlation: Correlation and Causation

• Fundamental Principle: Correlation does not equal causation.

• Directionality Problem: Even if two variables are correlated, it is impossible to know which variable causes the other based on correlation alone.

  • Example: Mental illness correlates with smoking. Does smoking cause mental illness, or does mental illness lead to smoking?

• Third Variable Problem: An underlying third factor might influence both variables.

  • Example: Both smoking and mental illness could be triggered by a third variable such as a stressful home life.

  • Example: Marriage length correlates positively with baldness in men. Marriage does not cause baldness; perhaps aging (third variable) causes both.

• Predictions: Correlation predicts the risk of certain outcomes but does not prove the cause (e.g., low self-esteem predicts depression risk, but may not cause it directly).

Illusory Correlations and Regression Toward the Mean

• Illusory Correlation: Perceiving a relationship where none exists, or perceiving a relationship as stronger than it actually is. This occurs when we notice and recall instances that confirm our beliefs while ignoring those that do not (e.g., believing dreams forecast events).

• Illusion of Control: The false belief that chance events can be personally influenced (e.g., a gambler throwing dice harder for high numbers).

• Regression Toward the Mean: The tendency for extreme or unusual scores or events to fall back toward the average.

  • Extraordinary happenings are usually followed by more ordinary ones.

  • Examples: An outlier grade usually regresses toward a student's average; a team's unusually bad game is usually followed by a better one.

• Superstitious Thinking: Failure to recognize regression can lead to false explanations. A coach might believe a scolding "worked" because the team's performance returned to normal after an unusually bad half, when they were actually just regressing to the mean.

Characteristics of Experimentation

• Experiment: A research method where an investigator manipulates one or more independent variables to observe the effect on a dependent variable.

• Goal: To isolate cause and effect by manipulating specific factors and controlling others.

• Structure:

  • Experimental Group: The group exposed to the treatment or one version of the independent variable.

  • Control Group: The group not exposed to the treatment; used as a comparison to evaluate the effect of the treatment.

• Random Assignment: Assigning participants to groups by chance (e.g., coin flip) to minimize preexisting differences between the groups. This equalizes groups in terms of age, attitudes, and other characteristics.

• Difference from Random Sampling: Random sampling creates a representative sample for a survey; random assignment equalizes groups within an experiment.

Experimental Procedures and the Placebo Effect

• Placebo Effect: Experimental results caused by expectations alone; behavior change caused by the administration of an inert substance that the recipient assumes is active.

  • Expensive placebos ($2.50$) often work better than cheap ones ($10$ cents).

  • Examples: Decaf coffee boosting alertness in people who think it's caffeinated; athletes running faster on "supposed" performance enhancers.

• Single-blind Procedure: Participants are unaware of whether they are receiving the treatment or a placebo; this controls for social desirability bias.

• Double-blind Procedure: Neither the participants nor the research staff knowing who is receiving the treatment or the placebo. This is standard in drug evaluation to avoid experimenter bias and the placebo effect.

Independent and Dependent Variables

• Independent Variable ($IV$): The factor that is manipulated and whose effect is being studied.

• Dependent Variable ($DV$): The outcome that is measured; the variable that may change in response to manipulations of the $IV$.

• Confounding Variable: A factor other than the independent variable that might influence the results (e.g., age, intelligence, temperament).

• Experimenter Bias: Unintentional influence by researchers to confirm their own beliefs.

• Operational Definitions: Precise specifications of procedures used to manipulate the $IV$ and measure the $DV$. These allow for replication of the study.

• Example (Self-Testing Experiment by Lee et al., 2014):

  • $IV$: Study procedure (reviewing answers vs. self-testing).

  • $DV$: Final exam performance.

  • Result: Students who self-tested scored $75\%$ compared to $51\%$ for those who only reviewed material.

Validity and Real-World Application in Research

• Validity: The extent to which a test or experiment measures or predicts what it is supposed to measure.

• Reliability: The extent to which findings can be replicated.

• Rental Housing Experiment (Carpusor & Loges, 2006):

  • Objective: Test the effect of perceived ethnicity on housing availability.

  • Method: 1,115 landlords in Los Angeles received identical emails with names suggesting different ethnicities.

  • Results (Positive Replies):

    • "Patrick McDougall": $89\%$

    • "Said Al-Rahman": $66\%$

    • "Tyrell Jackson": $56\%$

  • $IV$: Ethnic connotation of the name.

  • $DV$: Percentage of positive replies.

• Facebook Deactivation Experiment (Allcott et al., 2020):

  • Objective: Study social media's effect on mental health.

  • Method: 1,700 people deactivated Facebook for 4 weeks.

  • Results: The experimental group reported lower depression, higher happiness, and greater life satisfaction than the control group.

Questions & Discussion

• Question: Which of the following news reports is a negative correlation? (1) More exercise/less depression, (2) Lower IQ/lower grades, (3) Less study/lower grades, (4) Shoe size/grades.

  • Answer: (a) People who spend more time exercising tend to experience less depression.

• Question: In an experiment to test room temperature on performance, what is the $IV$?

  • Answer: (d) The temperature of the room.

• Question: Income levels and sleep hours are both lower in certain individuals. What is the correlation?

  • Answer: (a) Positive correlation (both factors move in the same downward direction).

• Question: Which coefficient represents the strongest relationship? (a) $+0.75$, (b) $+1.3$, (c) $-0.85$, (d) $-0.05$.

  • Answer: (c) $-0.85$ (coefficients cannot exceed $1.0$; the highest absolute value determines strength).

• Question: What is the purpose of random assignment?

  • Answer: (c) To reduce potential confounding variables.

• Question: What kind of study ensures neither researcher nor participant knows who gets the drug?

  • Answer: (c) Double-blind.

• Question: What is the purpose of a control group?

  • Answer: (b) To determine a cause-and-effect relation between the independent and dependent variables.

• Question: If a team's performance improves after a coach yells following an unusually bad half, what is the alternate explanation?

  • Answer: Regression toward the mean; after an unusually poor performance, the team is statistically likely to return to their average performance regardless of the yelling.

• Question: How would you interpret a correlation coefficient of $-0.87$?

  • Answer: It represents a strong negative correlation, meaning as one variable increases, the other decreases significantly.

• You Try It: Interpret the finding that teens who feel loved by parents behave in healthier ways.

  • Interpretation 1: Parental love causes healthy behavior.

  • Interpretation 2: Healthy, well-behaved children elicit more parental love.

  • Interpretation 3: A third variable, such as high-functioning family environment or genetics, causes both high parental love and healthy behavior.