PowerPoint PresentationSlide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8One approach: Test correlationsProblem: With correlations, causality is ambiguousSlide 11Slide 12Slide 13Slide 14Slide 15Slide 16Slide 17Slide 18Slide 19Slide 20Slide 21Slide 22Slide 23Slide 24Slide 25Slide 26Slide 27Slide 28Slide 29Slide 30Slide 31Slide 32Slide 33Slide 34Slide 35Slide 36Slide 37Slide 38Slide 39Slide 40Slide 41Slide 42Today: Research MethodsIn Social Psychology (Chpt 1)But first: How to study!1. Don’t use laptops in class – take notes by hand (on PP hard copies). Why? You have to actively listen for “important nuggets,” since you can’t write everything downYou will be more engaged and less distracted (research shows 40% of time, students have something else up)You will have more “deep conceptual learning” which is what I try to test forSee this new research: http://pss.sagepub.com/content/early/2014/05/21/0956797614524581But first: How to study!2. Review your notes three times: the next day. Then the next week. Then the night before the test. Why? Because it capitalizes on what we know about how new memories are formed. Don’t wait till the night before!But first: How to study!3. Consider using CourseSmart.http://learnsmartadvantage.com/course-books/?product-course=social-psychologyWhy? Because it is highly interactive which makes it easier to form memories.The system pays attention to what you don’t know, and feeds you relevant information.Recall: Social Psychology is…“The scientific study of how people think about, influence, and relate to one another.”One Ad: “Rose Quartz Power Bead Bracelet, Rose Quartz Buddha Beads, Bead Bracelet. Stone of Gentle Love. Enhances Warmth and Affection.”How can we test controversial claims and ideas? (i.e., the “pink crystal” effect)Through careful research designOne approach: Test correlationsOne approach: Test correlationsAre couples who own more pink crystals more loving?Does degree of belief in the power of crystals correlate with longer-lasting marriages?Problem: With correlations, causality is Problem: With correlations, causality is ambiguousambiguousNeeded to Test Causality: A Controlled ExperimentNeeded to Test Causality: A Controlled ExperimentBasic Logic:1. In an experiment, only one thing is different between two conditions. 2. If the two conditions differ, it has to be because of that one thing.Important Steps in Designing an Experiment:1. Formulate a testable hypothesis, which can be compared to a “null hypothesis.”????Important Steps in Designing an Experiment:Muddy thinking here makes everything else meaningless.Important Steps in Designing an Experiment:2. Decide who to study (sample selection).Important Steps in Designing an Experiment:Bad decisions here limit the generalizability of your results.Important Steps in Designing an Experiment:3. Decide how to manipulate the “independent variable.”Important Steps in Designing an Experiment:Bad decisions here limit the meaningfulness of results.Important Steps in Designing an Experiment:4. Decide how to measure the “dependent variable” (the outcome)Important Steps in Designing an Experiment:Bad decisions here limit the meaningfulness of results.Important Steps in Designing an Experiment:5. Decide how to assign people to condition.Important Steps in Designing an Experiment:Random assignment, which equalizes people across conditions, is usually best.Random assignment assures all types get all “treatments.”Important Steps in Designing an Experiment:5. Make sure Ps are ignorant of their condition and the study hypotheses.Important Steps in Designing an Experiment:Or else, “demand” or “placebo” effects may explain the results.Important Steps in Designing an Experiment:Demand effect: Ps “give you what they think you want.”Placebo effect: It works only because Ps believe it will work.Important Steps in Designing an Experiment:6. Decide what variables to control or equalize. Problem: “confounds” may explain the results.Important Steps in Designing an Experiment:Confounds: factors that vary with the IV (like time of day, time of semester) that make it impossible to know which is causal: The IV, or the confound.Important Steps in Designing an Experiment:7. Analyze the data and draw conclusions.Important Steps in Designing an Experiment:Suppose: 4.5 “affectionate behaviors” in the crystal condition (on average), versus 3.2 behaviors in the control condition. IS THIS SIGNIFICANT?Important Steps in Designing an Experiment:It depends. A “statistically significant difference” (p < .05) is big enough to be very unlikely to happen by chance alone (only 5 times out of 100 by chance).Factors affecting statistical significance: 1. The size of the sampleFactors affecting statistical significance: 1. The size of the sample Larger sample = harder to keep getting a chance result = more likely to converge on the “true” effect that exists within the populationFactors affecting statistical significance: 2. The size of the effect (here, 4.5 vs. 3.2 behaviors)Factors affecting statistical significance: 2. The size of the effect (here, 4.5 vs. 3.2 behaviors)Larger effect = harder to get by chance alone = more likely to be realFactors affecting statistical significance: 3. The background variability or “noise.”Factors affecting statistical significance: 3. The background variability or “noise.”Do the scores “clump together” within the 2 conditions, or are they very spread out?Factors affecting statistical significance: Suppose: Crystal condition scores = 4,3,5,4,5, M=4.2Vs.Control condition scores = 3,2,4,3,3, M=3.0Tight clumping around each mean = little background variability = more reliable.Factors affecting statistical significance: Suppose: Crystal condition scores = 1,7,2,6,5, M=4.2Vs.Control condition scores = 5,1,2,6,1, M=3.0Much spread around each mean = much background variability = less reliable.My predictions for Crystal study:The null hypothesis is true: no significant difference will emerge.Point of today’s lecture: thinking critically about truth claims. Show me the data!If it sounds too good to be true, it probably isn’t. Consumers beware!Put it to work! With 2 or 3 other students, design an experiment to test whether playing violent video games is harmful. Consider hypothesis, sample selection, IV, DV, possible confounds, and what kind of difference you’d