Extraneous Variables and Controls

Extraneous Variables

  • An extraneous variable is any variable that you're not investigating that can potentially affect the causal relationship of your research study.
  • Types of Extraneous Variables:
    • Physical Variables
    • Social Variables
    • Personality Variables
    • Context Variables

I Physical Variables

  • Physical variables are aspects of the testing situation that need to be controlled:
    • day of the week
    • experimental room
    • Lighting
  • Ways to Eliminate Physical Variables:
    • Elimination: Removing extraneous physical variables prevents them from operating differently across different treatment conditions.
    • Constancy of Conditions: The constancy of conditions controls extraneous physical variables by keeping all treatment conditions identical except for the independent variable.
    • Balancing: Balancing controls extraneous physical variables by equally distributing their effects across treatment conditions.

II Social Variables

  • Social variables are aspects of the relationships between subjects and experimenters that can influence experimental results.
  • These include demand characteristics and experimenter bias.

Demand Characteristics

  • Demand characteristics are cues within the experimental situation that demand or elicit specific participant responses.
    • Example: Students cue professors to wrap up their lectures by packing their binders, books, and water bottles and looking at the door.
  • Demand characteristics can confound an experiment if they vary across experimental conditions.
    • Subjects may act to confirm what they think is the experimental hypothesis.

Single-blind Experiment

  • In a single-blind experiment, subjects are not told about their treatment condition.
    • Example: In a single-blind drug study, the experimental and control groups might receive capsules that look and taste identical.
  • How do single-blind experiments control demand characteristics?
    • When subjects are not told about their treatment condition, this eliminates cues that might alter their behavior.

Placebo Effect

  • The placebo effect is when a subject receives a passive treatment and improves because of positive expectancies.
    • Example: If a patient receives a fake pain medication in a healing environment and the healthcare provider tells the patient that they are receiving a powerful pain treatment, many patients would report positive results.
  • How do cover stories control demand characteristics?
    • A cover story is a false plausible explanation of the experimental procedures to disguise the research hypothesis from the subjects.
    • They should be used cautiously since they are a form of deception.

Experimenter Bias

  • Experimenter bias is any behavior by the experimenter that can confound the experiment.
    • Example: An experimenter might provide more attention to subjects in one condition than another.
  • What is the Rosenthal effect?
    • The Rosenthal effect is the phenomenon in which experimenters treat subjects differently based on their expectations, and their resulting actions influence subject performance.
    • Also called the Pygmalion effect - a self-fulfilling prophecy.
  • Why is a double-blind design superior to a single-blind design in controlling experimenter bias?
    • Single-blind experiments only control demand characteristics, since subjects are blinded to their condition.
    • Double-blind experiments control both demand characteristics and experimenter bias since both the experimenter and subjects are blinded. Hindi alam ng subjects if anong group sila Parehas na experimenters and participants hindi alam.

III Personality Variables

  • How might an experimenter's personality affect experimental results?
    • Research on experimenter personality shows that when experimenters are warm and friendly, subjects learn more, talk more, earn better test scores, and are eager to please.
    • Hostile or authoritarian experimenters obtain inferior subject performance.
  • How can experimenters control personality variables?
    • Treat the "experimenter" as an independent variable in statistical analysis. If an interaction is found, then the experiment is confounded.
    • Employ multiple experimenters to run an equal number of subjects in each of the experimental conditions (balancing).
    • When there is a single experimenter, minimize face-to-face contact and closely follow the script.
    • Videotape sessions to confirm consistent performance.

IV Context Variables

  • Context variables are extraneous variables produced by experimental procedures created by the research setting environment, like the assignment of participants to conditions.
  • When might subjects select the experiment?
    • When we allow subjects to sign up for experiments whose titles differ in their appeal:
      • "The Memory Test Experiment"
      • "The Heavy Metal Music Experiment"
  • However, this could result in a biased sample threatening the causal relationship.
  • Why shouldn't you run your friends in your experiment?
    • Selecting your friends might bias your sample, threatening internal validity. Both you and your friends might act differently in your experiment than strangers.