Module 2 – Types Of Studies + Types Of Data

Statistics and Data

  • Statistics are procedures and rules to organize and interpret data.

  • Psychologists convert behaviors into numbers for statistical analysis.

Variables

  • A variable is a property that can take on different values.

    • Continuous variables: Take on any value (quantitative or numerical data).

    • Categorical variables: Take on a small set of possible values (frequency or count data).

  • Example: Food dish visits by mice can be a continuous variable (any number of visits).

  • Sex of a mouse is a categorical variable (male or female).

  • Continuous variables can be "chopped up" into discrete categories.

  • Variable type is important because statistical tests require specific variable types.

Dependent vs. Independent Variables

  • Dependent variables: Measured variables (also known as response variables).

    • In the hungry mice example, food dish visits are dependent variables.

  • Independent variables: Usually controlled by the researcher (also known as explanatory or predictor variables).

    • Researchers may observe or manipulate these variables.

Types of Studies

Observational Study
  • Researchers observe ongoing behavior without intervening.

  • Example: Facebook analyzed user posts to study laughter types.

    • Dependent variable: Frequency or count of laughter types (categorical).

    • Independent variable: Sex (categorical).

    • Independent variable: Age (continuous).

Experiment
  • Researchers manipulate a variable to see if that variable have any effects.

  • Example: Giron et al. (2015) studied whether "anger inoculation" decreases road rage.

    • 40 college students used a driving simulator.

    • Half argued against aggressive driving before the simulator.

    • Dependent variable: Accident or no accident (categorical).

      • If the total number of accidents compared, then the dependent variable would be continuous

    • Independent variable: Anger inoculation or no anger inoculation (categorical).

      • 50% of participants who argued against aggressive driving had an accident, versus 70% of those who didn’t.

Causality

  • Observational studies cannot infer causality.

  • There might be a third variable affecting both independent and dependent variables.

  • Example: Ice cream sales and shark attacks.

    • Association does not equal causation.

    • A third variable like heat could cause both.

    • Using emoji or LOL won’t change a person’s age.

    • There is an association between online laughter and age.

Experimental Research & Eliminating Third Variables
  • Control: Scientists manipulate the independent variable and try to keep everything else constant.

  • Random assignment: Each participant has an equal chance of being in a condition.

Population vs. Sample

  • A population is the entire collection of events of interest.

    • Facebook was able to analyze the entire population of Facebook users.

  • A sample is data collected from a subset of the population.

  • Inferential statistics are used to infer things about the population based on a sample.

  • Samples must be random.

    • Every member of the population has an equal chance of being included.

    • Psychological findings may only relate to Westernized, Educated, Industrialized, Rich, Democracies because most research subjects are American college students.

Observational Study Example

  • A graduate student determined the efficacy of an educational program that used the environment for teaching.

    • Fourth-grade teachers took students to a nature reserve.

    • The idea was that being outside would increase learner engagement and improve learning.

  • The student chose an observational study.

    • Compared students who went to the site to those who didn’t.

    • Independent variable: number of visits to the site (categorical - never, 1, 3, or 4 times).

  • Dependent variable: Commonwealth Accountability Test (CATs) scores (science and reading), which are two continuous dependent variables.

  • ANOVA test: A statistically significant relationship between program participation and reading [F(3,651)=3.002,p=.030][F(3,651) = 3.002, p = .030] and science [F(3,651)=4.192,p=.006][F(3,651) = 4.192, p = .006] scores.

  • Cannot infer causality without an experiment.

  • The study has the third variable problem.

    • Example: Only the best teachers take their students, and these teachers might try harder every day than teachers that didn’t go to the site.

Experiment: Program Increases Scores?

To show the program increases scores:

  • Select a random sample of the population.

  • Randomly assign half of classrooms to visit the site and half to do something else.

  • Then any differences in CATS performance could be attributed to the program.