PS295 Chapter Notes - Chapter 9: Internal Validity, Random Assignment, Demand Characteristics
Threats to Internal Validity:
-cofounding
-if experimental conditions are not equalized before participants receive IV then there is
biased assignment
-biased assignment is also called selection threat to internal validity
-can arise when efforts to randomly assign participants to conditions fail to create
experimental conditions that are equivalent prior to manipulation of IV
-random processes do not always produce random results but fortunately random
assignment works most of the time
Differential Attrition:
-attrition is the loss of participants during a study (if they die = subject mortality)
-when attrition occurs randomly and affects all experimental conditions equally, it is only a
minor threat to internal validity
-differential attrition: when the rate of attrition differs across the experimental conditions
-this weakens internal validity
Pretest Sensitization:
-pretests provide useful baseline data but may lead participants to react differently than they
would have if they had not been pretested
-when this is a concern, researchers may include conditions in their designs in which some
participants take the pretest whereas others do not
Document Summary
If experimental conditions are not equalized before participants receive iv then there is biased assignment. Biased assignment is also called selection threat to internal validity. Can arise when efforts to randomly assign participants to conditions fail to create experimental conditions that are equivalent prior to manipulation of iv. Random processes do not always produce random results but fortunately random assignment works most of the time. Attrition is the loss of participants during a study (if they die = subject mortality) When attrition occurs randomly and affects all experimental conditions equally, it is only a minor threat to internal validity. Differential attrition: when the rate of attrition differs across the experimental conditions. Pretests provide useful baseline data but may lead participants to react differently than they would have if they had not been pretested. When this is a concern, researchers may include conditions in their designs in which some participants take the pretest whereas others do not.
