Psychology 2300 Chapter 9 Experimental Studies Elements of a true experiment aka things that allow us to begin making causal statements Manipulate one or more independent variables to create at least two conditions an experimental and control group Control all other variables so that conditions differ only for the independent variable Measure the effects on the dependent variable o The independent variable is usually something we manipulate Experimental group receives treatment variable Control group receives no treatment or placebo expectation to get well Only one thing is allowed to vary among the conditions o The dependent variable the effect of the manipulation Placebo a substance that has no effect actual medication but given to reinforce a patient s Is usually a measured or quantified variable at least one of the two Internal Validity the ability to confidently conclude that the independent variable caused the differences on the dependent variable within the experiment Requirements o Covariance dependent variable varies with independent variable when another variable covaries with the independent variable we call it a confound o Time order relationship independent variable before dependent variable o Elimination of plausible alternative causes done by manipulating independent variable Confound something that might vary along with the independent variable and could influence the dependent variable o How do we control for confounds Holding conditions constant treatment group has the same experience as the control group Balancing participants in each condition are similar random selection and assignment Independent Groups Designs Separate groups of people in each condition o Random groups design equate groups with random assignment o Matched groups Design equate groups by administering a pretest dependent variable form pairs based on pretest results then randomly assign Often used with small samples Matching maintains internal validity o Natural Groups Design groups based on individual difference Example o Natural Groups design Physical characteristics gender height weight Social characteristics ethnicity marital status Personality characteristics extroverted v introverted Mental health characteristics depression etc Threats to internal validity dependent variables o Subject loss differential loss of participants Matched groups design is the easiest to correct for o Intact groups assigning existing groups not individuals to conditions but using individual scores as Do the elements of a true experiment include the research setting o Most experiments are conducted in a lab which can help in the control of extraneous variables o However a true experiment doesn t require a lab setting as long as it has all the proper elements Repeated Measures Designs within subjects design Participants are in both the experimental and control conditions Advantages Disadvantages o Requires fewer participants which is more efficient o Don t need to worry about subject loss as much o Practice effects also called order effects doing task more than once affects dependent variable Balancing averaging practice effects across conditions Counterbalancing in repeated measures designs All possible orders o Determine all possible orders of conditions o Randomly assign participants to one of the possible orders o 4 or fewer conditions Block randomization o Presenting a random order but not accounting for all possible orders General rule each condition must appear in each ordinal position equally often External validity extent to which results can be generalized beyond the experiment People settings conditions Obtain external validity by Field experiments Clinical trials o Seeing if results replicate across different samples settings and operational definitions High control over independent variables random assignment of subjects Occurs in real life settings activities context is not artificial Can have methodological and ethical issues with control groups Strengths o Combines scientific rigor with clinical relevance Limitations o Difficult to do No experiment is perfect All experiments suffer from some lack of perfection o Often high internal validity is obtained at the cost of some external validity and vice versa Charles Gelso called this the Bubble Hypothesis Can have methodological and ethical issues with control groups o Combines scientific rigor with clinical relevance Strengths Limitations o Difficult to do No experiment is perfect All experiments suffer from some lack of perfection o Often high internal validity is obtained at the cost of some external validity and vice versa Charles Gelso called this the Bubble Hypothesis
View Full Document