EDHD306 FINAL EXAM STUDY GUIDE Fall 2014 Chapters 8 9 10 11 12 13 14 Experimental Design Experimental control and randomization o Ways to eliminate confounding influence from other variables o Experimental control Known as the control group Similar to population between tested but no change is implemented o Randomization Randomly organizing groups Confounding variables o Variables that vary along with the independent variable Need to be controlled or eliminated o Other variables that can effect the result o The experiment must be designed such that the independent variable can be the only cause of the results Internal validity Internal validity o How well a study is done o Avoids confounding variables o High internal validity increases confidence in researchers explanation of problem Basic experiments o Posttest only design 1 Obtain two equivalent groups of participants 2 Introduce the independent variable experimental group vs control group OR condition 1 vs condition 2 3 Measure the dependent variable Ex test driving ability before variable manipulated then again after variable is manipulated o Pretest posttest design Almost identical to posttest only A pretest is given prior to any treatment or experimental manipulation Ex test driving ability before variable manipulated then again Advantages and disadvantages of having a pretest after variable is manipulated o Advantages Possible to check for non equivalent groups Assesses equivalency of groups with small sample size Assigning participants to conditions o Independent groups Can use to select participants for study Mortality Helpful when there is a risk of participants will drop out Experimental mortality attrition If we have pretest scores we can assess whether the people who dropped out were different from the people who completed the study Sensitize participants to what is being studied Can be time consuming Reduce external validity May possibly give away the hypothesis of the experiment and therefore bias participant behavior o Disadvantages o Repeated measures randomly assign to ONLY one level of the independent variable random assignment is sued to assign participants to conditions participants are only in one conditions experimental or control random assignment prevents systematic biases assign to ALL levels of the independent variable the SAME individuals participate in all conditions at different points in time Advantages and disadvantages Advantages o fewer research participants are needed everyone participates in both conditions so they are equivalent Disadvantages o the order of presenting the treatments may affect the dependent variable Types of order effects o Practice effect o Fatigue effect an improvement in deterioration in performance as performance as a result of repeated practice with a task the participant becomes tired bored or distracted treatment carries over to influence the response to the second treatment the effect of the first o Carryover effect Counterbalancing all possible orders of presentation are included in the experiment o Matched pairs first match on a participant characteristic and randomly assign to ONLY one level of the independent variable people are matched on a participant characteristic e g cognitive ability and then each person from the pair group is assigned to either one condition or the other Goal to achieve the same equivalency within groups that you obtain in a repeated measures design without actually having the same participants in both conditions particularly helpful when you have small sample sizes randomization might not be 100 effective at producing equivalent groups Manipulations independent variables o Straightforward manipulations Researchers manipulate the independent variable using written verbal or visual material Presented directly to participants Ex manipulating mood by showing one group a sad movie or another group a happy movie o Staged manipulations Events that occur during the experiment are staged in order to manipulate the independent variable Create a desired psychological state Stimulate a real world circumstance Frequently employ confederate Someone who acts like a participant or bystander but is actually working with the researcher Ex manipulating mood by putting participants into a situation where they are insulted instead or complimented by a confederate who they think is another participant in the study o Strength of manipulation The differences between the groups should be maximized Makes it more likely that one will observe an effect But we also need to consider levels of the variable in the real world ethics Dependent variables o Types of Measures Self report participants provide information about themselves Behavioral measures observations of participants behaviors Physiological measures recordings of bodily responses o Sensitivity o Ceiling Effects Relates to the test s ability to identify a condition correctly The level at which an independent variable no longer has an effect on a dependent variable o Floor Effects The level above which variance in an independent variable is no longer measured or estimated the floor effect is a statistical phenomenon in which most data points fall in the very low range of possible values bottom out on the floor of the measure the floor effect is often seen in assessment when a test is too challenging for a given target population Participant expectations o Demand characteristics Can be controlled for minimized by Using deception Using filler items in a questionnaire Simply asking participants about their perceptions of the purpose of the research Might inform of the purpose Experimenter expectations o Expectancy effects experimenter bias May occur whenever the experimenter knows which condition the participants are in Experimenter might unintentionally treat participants differently in various conditions of the study Bias can occur when experimenters record the behaviors of the participants there may be subtle differences in the way the experimenter interprets and records the behaviors Clever Hans a horse whose brilliance was revealed by Pfungst to be an illusion if a clever horse can respond to subtle cues it is reasonable to suppose that clever humans can too Research has shown that experimenter expectancies can be communicated to humans by both verbal and nonverbal means Teacher expectancy self fulfilling prophecy Ones expectations about another person eventually lead the other person to behave in a way