Introduction to Experiments

• Correlation


  A relationship between two variables where they co-vary, but does not necessarily imply causation
• Causation


  A relationship where one variable directly causes changes in another variable
• Requirements for a causal claim

  • Covariation: cause and effect are related
  • Temporal Precedence: cause comes before effect
  • Elimination of confounds: no alternative causes
• Correlation does not equal causation
• Number of bars in New York City


  Number of churches
• Spurious correlation: A correlation where variables co-vary but are not causally related
• Things that can cause spurious relationships: chance, third variable problem
• Internal validity is sometimes called the third variable problem
• Without an experiment, you cannot tell whether two variables covary due to a causal relationship or spurious correlation
• Program evaluation


  A type of research that aims to assess whether a program or intervention successfully changes behaviour or works as intended
• Program evaluation examples


  • PSA: seatbelt campaigns
  • Workplace programs: employee retention
  • Behavior modification: token economies
  • Learning and teaching outcomes: LSAT prep classes, training, tutoring, driving school

  
• To determine whether a driving prep course helps people pass the driving exam, you could compare the driving test scores or prep rates of those who took the prep course with those who did not take the course

  
• Internal Validity


  The degree to which changes in the dependent variable can be attributed to the independent variable, rather than to another variable (i.e., confounding variables)
• Confounding variable


  An extraneous variable that offers an alternative explanation for the observed differences
• Confounding design: When the confound varies systematically with the independent variable (true confound)
• Possible confounds in driving prep course example


  • Motivation
  • Money
  • Education
  • Awareness of course
• Experimental designs


  • Involve the (1) manipulation of an independent variable (IV)
  • Participants are (2) randomly assigned to different levels (conditions) of the IV
  • The researcher then measures the dependent variable (DV) and (3) compares the different levels of the IV to see if they differ on the DV
• Independent Variable (IV)


  What the researcher directly manipulates to determine its influence on people's behaviour (the DV)
• Dependent Variable (DV)


  A response or behaviour that is measured
• Levels


  Values that an independent variable can take
• Treatment group


  The participants in an experiment who are exposed to the independent variable level that involves the medication, treatment, or intervention
• Control group


  The participants in the experiment who serve as a comparison to the treatment group. They receive a neutral or "no treatment"
• Random assignment


  A process by which participants are placed into different levels of the IV, and each participant has an equal chance of being in any condition
• Selection effects


  When participants in the control group differ from the treatment group before any treatments or manipulations
• Random assignment eliminates group differences and is the only way to ensure groups are equal
• Matched-groups design


  Participants are matched with other participants based on some important variable, and the individuals in those pairs are then randomly assigned into groups
• Quasi experiment


  Where one grouping variable cannot be randomly assigned
• Without full random assignment, it is impossible to make a full causal claim
• Random assignment works; it always works (long term)
• After collecting the data, you need to determine if the levels of the IV differ from each other on the DV by comparing the means of the different levels
• The three requirements for a causal claim (covariation, temporal precedence, eliminating alternative explanations) are all checked in an experiment
• Options for driving prep course experiment


  • Option 1: Two levels of IV (take course, don't take course)
  • Option 2: Three levels of IV (take course, no course, study alone)
• Hypotheses for the driving prep course experiment