Intro to Experimentation Ch 9 10 Lecture Thurs 2 21 Topic Outline for Exam 2 The Experimental Ideal Create a situation in which two groups are perfectly equal at baseline then introduce a single treatment a single change a single manipulation to one of the groups and take a measurement Why it falls short when studying people The everything equal at baseline ideal is currently impossible when we re studying humans Independent Variable IV Variable in which you manipulate to show a change in your dependent variable ex how sports events affect mood IV sports game Dependent Variable DV Variable in which you test for changes ex how sports events affect mood DV mood levels due to the effects of sports game Operationalizing the IV Researchers always need to make trade offs between practicality and experimental realism when operationalizing Independent Variables the ability to test non objective things such as fear love jealousy self esteem low social status injustice etc Manipulate the IV Changing the independent variable to see the different effects it has on the dependent variable ex changing sports events both negatively and positively to see the effects in which it has on people Random Assignment to Condition Random assignment refers to the use of chance procedures in experiments to ensure that each participant has the same opportunity to be assigned to any given group Experimental Group The group of participants who receive the drug or treatment being studied Control Group The control group is composed of participants who do not receive the experimental treatment Placebo Group Some patients in a study may be administered a placebo fake treatment while other participants receive the actual treatment The purpose of doing this is to determine whether or not the treatment has an actual effect Randomized minimizes the differences among groups by equally distributing people with particular characteristics among each of the trials Double Blind even the experimenter doesn t know what condition treatment the participant is given Placebo control Experiment Some patients in a study may be administered a placebo fake treatment while other participants receive the actual treatment controlled group The purpose of doing this is to determine whether or not the treatment has an actual effect External Validity Can the results of the study be generalized to the rest of the population Internal Validity Refers to how confidently one can conclude that the observed effect s were produced solely by the independent variable and not extraneous ones Why we might have poor Internal Validity Random error Aspects of the testing environment that affect both groups equally and that create noise in our data ex a flickering light anything that would distract us from detecting and effect Accounted for by controlled testing NULL HYPOTHESIS TESTING Systematic Error a k a Confounds Groups differ on a dimension that makes them unequal at baseline that may influence dependent variable ex having two conditions always run in two separate rooms or by two separate experimenters or at two times of day Selection bias Our two groups may be predetermined by a characteristic other than our IV ex personalities of people who sit at the front of the room verses people in the back of the room Needs random assignment Maturation effects Sometimes participants change over time and it has nothing to do with your manipulation short term factors that cause maturation boredom fatigue practice Counterbalance Switching the order in which you present to participants to eliminate the factor of practice maturation ex ordering 1 2 3 then 3 2 1 then 2 1 3 etc Internal Validity Threats History effects Sometimes participants change over time and it has nothing to do with your manipulation ex testing participants before and after a tragic event such as Sept 11 2001 controlled group would require the participants to go through the same conditions history Attrition Losing subjects over time in a way that may be systematically related to the IV or DV ex testing marital happiness with 10 couples over a span of 5 years loosing participants throughout those 5 years throw off your data prevention requires careful planning permission into databases for contact info and imputations Instrumentation Experimenter Effects If the testing apparatus changes over the course of the experiment it introduces error unrelated to your DV Demand Characteristics Between Groups Randomly assign subjects to experimental vs control condition Pro s Compares separate groups of individuals allows one score per participant participants score in not influenced by other factors fatigue practice etc Con s Individual differences high variability confounding variables large number of participants Within Groups Designs Repeated Measures Measure each subject twice in the control experimental conditions Pro s Reduces in error variance power Con s Participation in one condition may effect performance in other conditions carryover effect Null Hypothesis Testing T Tests Supplemental Reading Lectures Tues 2 26 Thurs 2 28 The Logic of Null Hypothesis Testing Testing whether the effects of our IV has an effect on our DV or if it is simply caused by random error Indirect Proof Null Hypothesis testing We have to prove that random chance random error is such an unlikely explanation for our why group averages differ that the only likely explanation is the IV did it Random Chance Random error we know 2 groups won t be perfectly equal at baseline Effect of IV as explanations for group differences If the difference between two group means is big enough it s unlikely that random error is the sole cause of group differences in which case the IV is more likely to be the cause Alpha To prove the null IV didn t effect DV wrong our group means need to be different enough that this difference would occur by chance alone less than 5 of the time to rule out random chance as the better explanation alpha 05 or 5 P values tells us the probability that we DID get this outcome just by chance alone Statistical Significance p value is less than alpha 05 large N Practical Significance it s too small to have much real world consequence Two types of error Type 1 error Rejecting the null hypothesis when in fact it is true False positive Minimized by replication Type 2 error Accepting the null hypothesis when in fact it is false False Minimized by Increasing N reducing random error increasing strength negative of IV Steps
View Full Document
Unlocking...