Controlling extraneous variables

1
Controlling extraneous variables

• How to recognize em,
• and
• how to fix em

2
Extraneous variables

• Extraneous variables are all variables other than
  the Ivs and the DVs.
• Some extraneous variables threaten the internal
  validity of a study by varying systematically
  along with the independent variable(s). These
  co-varying EVs are called confounding, because
  they confuse, mix up, or confound the explanation.

3
Control of extraneous variables

• Potentially confounding variables may be
  controlled by procedures which ensure that they
  do not co-vary with the IV.
• 1. Keep extraneous variables constant.
• 2. Keep the influence of evars constant.
• 3. Randomly distribute the influence of evars.

4
Theoretical threats

• Campbell and Stanleys list of categories of
  extraneous variables Threats to internal
  validity
• History Extraneous events which occur during the
  course of an experiment. Not just in repeated
  measures designs
• Maturation Changes within the individual which
  occur during the course of the study, as a
  function of the passage of time.

5
More theoretical threats

• Testing carryover In repeated measures studies,
  having been measured once with a test affects how
  one responds at the posttest.
• Instrumentation Changes in the measurement of
  the dependent variable during the course of the
  study, as a function of the passage of time.
• Statistical regression Similar to testing
  carryover, except that regression is the
  phenomenon that extreme scores change more from
  pretest to posttest than do average scores.

6
Still more threats

• Selection bias Non-random assignment of
  participants to groups
• Ultimately, this is an ex post facto design
• Interaction of selection bias with other threats
• Mortality Loss of group equivalence through the
  unequal loss of participants from groups.
• Thus, Campbell and Stanley list eight threats to
  internal validity.

7
Participant and experimenter variables

• Participants are active, not passive
• Participant demand characteristics
• Positive self-presentation motive
• Greatest when assessing the participants true
  intentions, beliefs, or feelings
• Minimal if attributions are global and external
• Behavior systems analysis is necessary

8
Positive self-presentation

• Intertreatment interaction What to do to look
  good depends on the experimental condition.
• Intratreatment interaction What to do to look
  good depends on the individual participant, or
  even on the phase of the experiment, within an
  experimental condition.

9
Experimenter variables

• Motives and knowledge
• Experimenter attributes
• Biosocial attributes Age, gender, race, religion
• Psychosocial attributes Anxiety, need for
  approval, dominance, intelligence
• Situational attributes Experience with research,
  reaction to participant attributes, laboratory
  cues
• Are these variables biasing?

10
Experimenter expectancies

• Effects on researcher
• Data collection
• Data recording and analysis
• Data interpretation Practically any set of data
  can be interpreted in different ways, depending
  on the orientation of the person doing the
  interpreting (Christensen, 1997, p. 250). Cf.
  our archival research.

11
Experimenter expectancies

• Effects on participants
• Interaction with positive self-presentation
  motive
• Most obvious in animal studies
• Means of influencing participants
• Recording errors -- Nonverbal cues
• Intentional biases -- Social influence
• Expectancy effects can exceed treatment effects

12
Sequencing effects

• Within-participant sequencing effects
• Carry-over effects
• Practice effects Learning
• Between-participant sequencing effects
• Order effects
• Counterbalancing