PSYCH 312 1st Edition Lecture 9Outline of Last Lecture I. evaluation of research hypothesis II. variance inferential statsIII. types of variationa. chanceb. IV (systematic)c. Confound (systematic)Outline of Current Lecture I. Statistical hypothesis testing II. Threats to internal validity Current Lecture oStatistical hypothesis testing oAlpha level Probability of committing a type I error Set at .05 (p< .05)These notes represent a detailed interpretation of the professor’s lecture. GradeBuddy is best used as a supplement to your own notes, not as a substitute.5 in 100 chance of committing type I error 95% of time we wont make type I erroroBeta level Probability of committing a type II error Represents inverse power (probability of avoiding a type II error)Power= 1-Beta**want power to be highoWhy do we avoid type I errors?Eliminate confounds oWhat do we do to decrease type II errors?Increase sample size (increases power)]Design study to minimize factors that reduces size of F ratioIncrease within grp Var and/orDecrease between grp VaroThreats to internal validity Major confounding variables Primarily affect bt-grp varMaturation confound Problems of interpretation resulting from participant maturation either between tests or over time Growing older, stronger, healthier, more tired, more bored, ectFatigue effectsCurrent exp: 3 year olds get tired fasterSolutionRandomize the order of conditionsTesting/practice confound Problems that result from repeated measurement of same individualSolution?Randomize order of conditions Use bt-subjects design rather than within-subjectWithin: same participants in all conditions Between: different participants in different conditions (left, right, original)History confoundEvents that take place between measurements in pretest-posttest design (not related to IV)Ex: record mood at pretest, administer drug, then test again 6 mo laterPretest during winter Posttest during summerSelection confoundParticipants vary on one or more variable (other than that IV) across the conditionsLikely when using an experimenter selected IVRegression toward the meanExtreme scores (DV) move toward the mean with repeated testing over time (unrelated to effect of IV)Diffusion or imitation of treatments Individuals in an experiment may communicate with each other, can reduce differences between conditions due to diffusion Solutions?Make participants blind to hypothesis Ask them not to talk about study Test over short periodInstrumentation Instrument (human or machine) used to measure DV is unreliable (gives different reading over time)Solutions?Training and practice Use multiple raters (check for inter-rater reliability)Counterbalance order of conditions across ratersMortality Loss participants due to drop out or refusal to participate Solution?Removing similar participant from other conditions Noting problem in results and
View Full Document