Research Methods Lecture 2 PDF

Lecture 2 + Research Methods Research Methods Toolbox 1. Case study → focus on 1-2 people to give proof that stuff exists - Case studies usually don't apply to the rest of the world making it have low external validity (since it is an experience and every human is differen...

Lecture 2 + Research Methods Research Methods Toolbox 1. Case study → focus on 1-2 people to give proof that stuff exists - Case studies usually don't apply to the rest of the world making it have low external validity (since it is an experience and every human is different) - Rich information, low generalization - Pros → provides rich information by observing their experiences - Cons → low external validity - Validity → does the study measure what we want it to measure? - External validity → do the observations apply to real life/people? 2. Naturalistic Observation → observing w/out interfering with others and simply just watching them in their natural context - High generalizability but lacks control - Reactivity → when people know they are being watched they will most likely change how they act (eg. when u slow down when u see a cop, you’re more likely to slow down) - Pros → high external validity than case studies because they’re more relatable from personal observations - Cons → possible low internal validity (do the study procedures measure what we want them to?) 3. Archival research (“big data”) → doing research on existing databases/records (patient records, etc) - Less invasive, could have quality issues - Pro → non invasive/doesn’t interfere w/ your life - Cons → there’s a lack of quality control and internal validity because you’re not actively collecting the data on your own - Lack Internal validity → eg. when u wanna measure hand washing after toilet however as u leave bathroom u notice the door handle sticky so now u dont know whether people washing hands due to sticky door or after toilet use 4. Surveys → self-report measures usually used because they voluntarily report thoughts and feelings (very common method) - Convenient, but prone to bias and errors - Pros → easy to administer (quick to set up, print, etc), you can also use random selection to control generalizations - Cons → response error/bias Response error and bias - Anonymity helps provide more truthful answers - When conducting surveys, there may be errors in judgment from responders - Malingering → faking answers (ie. faking/overstating at the doctors) - Social desirability bias → acting in a certain way to please others - Operational definition → define variable that can be measured/quantified to avoid ambiguity/vagueness in measurement (eg. How many times do you smile/laugh a day? Reliability and validity - Reliability → when test results are consistent over time - How consistent a test is over time - Internal consistency → do survey responses agree? - Test-retest reliability → are test results stable? - Inter-rater reliability → do two people agree w/ the results? - Validity → when tests measure what it’s supposed to - A test measures what i suppose to be measured - Face validity → does it appear to measure what it says it measures? - Convergent validity → does it agree with others measuring the same thing? - Divergent validity → does it diverge from others measuring different things? Correlational Studies - Show the relationship b/w two variables - Pros → helps show a trend for predictions - Cons → can’t infer or make claims for why - Strength and direction is defined w/ the correlation coefficient (r value) - Strength and direction → -1.0 to +1.0 - -1.0 perfect negative correlation - +1.0 positive correlation - Illusory correlations → correlation isn’t there - Correlation doesn’t equal causation!! Experimental Designs - Confounds → rival hypotheses/variables that could explain your effects - Random assignment → distributes unknown confounding variables evenly b/w groups (cancels out differences) - Placebo effect → when you feel real effects from ineffective manipulation - Participant demands → behaving the way you think the researcher wants you to - Experimenter effects → when researchers bias the study - Manipulation → independent variables are designed by the researchers as the variable of interest - Quasi-experimental designs → no random assignments are allowed, but you can still manipulate variables because of existing group memberships - Ie. marital status, ethnicity, childhood experience, ability/disability, difficulty w/ causal inferences