-Experimental Group gets treatment -Control identical except no treatment not every study has a control group -Quasi-experimental: natural variability in IV -Read Estimating the reproducibility of psychological science (article) Critical thinking- search for issues in the following studies (format, data, concept): Study #1 -Listen in on conversation at a mall, record the sex of the person at whom people are laughing -RESULTS: In 30 conversations, this person was more often a man (60% of cases) -Therefore, men are funnier Study #2 -We measure laughter volume (db) in 200 people who watched clips of Youtube comics (2 M, 2 F) -RESULTS: When the comic was a woman, people laughed 50% louder -Therefore, women are funnier Potential problems: -the people in the lab study laughed harder or differently because they knew they were being watched? - some of the people at the mall were on drugs? - the YT clips were different lengths? or different jokes? (Substantially different content being presented/compared) Key factors in experiments: -Confounds: anything that affects the DV and may vary across conditions -Third variables and directionality -Representativeness of sample -Validity Selecting participants: Population: group you want to know about Sample: subset you actually test -Random sample: drawn at random from population (equally likely for everyone) -may try to ensure representativeness
-Convenience sample: drawn from easily accessible group (e.g.,
undergrads, MTurk) -culturally representative? Biased respondents? Randomisation -Random Assignment: use some random method to allocate participants to conditions reduce the influence of confounds or selection bias Getting Meta Meta-analyses aggregate findings from multiple studies to determine an overall average result/effect -control for the influence of study size -test differences between study characteristics Important Stats Concept -Statistical significance: result is unlikely to be found by chance (usually 5% cut-off) -Practical significance: does the result have any real-world value or meaning? -Effect sizes: standardized way of quantifying the impact of a result (e.g., differences between groups) Interactive demo on how p values are problematic http://www.fivethirthyeight.com/features/science-isnt-broken/ Statistics & quantitative methods -acknowledge your knowledge -consume relevant literature -train and learn Research Ethics -When might it be unethical to conduct experiments (vs correlational methods)? -IRBs (institutional review boards): safety, privacy, risk/benefit, & rights -Informed Consent -Deception & Debriefing Animal Research Ethics -Allow research models in areas where human experimentation would be unethical Review boards (e.g., IACUC) -ensure animals safe, well -evaluate impact of research Ethics Code
-Broader guidelines by which some psychologists are governed
-Ethics in clinical work -Ethics in research & academic pursuits -publication & collaboration http://www.apa.org/independent-review/APA-FINAL-Report-7.2.15.pdf