PSY2234: Introduction to Social Research (Part 2) PDF

PSY2234: Introduction to Social Research (part 2) Associate Professor Trevor Case Content warning 2 Essential Reading Gilovich et al. (2024) Ch.2 ― NOTE: The exam questions are taken from the lecture m...

PSY2234: Introduction to Social Research (part 2) Associate Professor Trevor Case Content warning 2 Essential Reading Gilovich et al. (2024) Ch.2 ― NOTE: The exam questions are taken from the lecture material, tutorials, and the readings 3 Lecture outline What is Social Psychology? Where do research questions come from? The Experimental approach ― Between subjects designs ― Within-subjects (aka repeated measures) designs Part 2 (next week) ― Factorial designs Quasi-experimental designs The Correlation approach Replication 4 Steps in the research process 1. Research question 2. Generate hypotheses (specific, directional predictions) 3. Operationalise ― Measure: What? How? (e.g., how would you operationalise superstition) Who? (representative sample, generalisation) 4. Design experiment/ correlation study 5. Collect data 6. Analyse data 7. Draw appropriate conclusions (overinterpreting; probs. with methods) 5 Example 3: Factorial Design Portion size and consumption (Wansink & Kim, 2005) Background ―People overeat food that they like. ―But does portion size influence consumption? 6 Example 3: Factorial Design 1. Research question ― Does container size have the same effect on consumption of palatable and less palatable food? 2. The hypothesis ― Large container size will increase intake of both fresh and stale popcorn compared to small container size 7 Example 3: Factorial Design 3. Operationalise (Measure what/how?) ― Actual consumption (weight) of fresh and stale popcorn ― Measure who? Moviegoers 4. Design experiment ― Randomly assign participants to receive free popcorn on entry to a cinema ― Two independent variables: ― 2 container sizes (medium vs. large) x 2 palatability (fresh vs. stale 14 days old) 8 Example 3: Factorial Design 5. Collect data ― give free popcorn at entry ― weight written on bottom of container ― both containers were large enough so that not all the popcorn would be consumed ― collect-up containers from moviegoers at end of movie and weigh ― Also, ratings of popcorn quality (stale was rated as stale) 6. Analyse data and look at the results… 9 Example 3: Factorial Design MAIN EFFECT OF CONTAINER SIZE 90 80 70 Grams eaten 60 50 40 30 20 10 0 Medium container Large container 10 Example 3: Factorial Design MAIN EFFECT OF POPCORN FRESHNESS 90 80 70 Grams eaten 60 50 40 30 20 10 0 Stale popcorn Fresh popcorn 11 Example 3: Factorial Design FACTORIAL DESIGN 2X2 12 Example 3: Factorial Design 7. Draw appropriate conclusion – Size does matter – Large container increased consumption of both fresh and stale popcorn; – Implications for increasing consumption of less preferred healthy food (e.g., children eating carrot sticks) Advantages of factorial design: – effects of IV's alone/together 13 Example 3: Factorial Design Factorial designs typically involve two or more categorical IVs These can be: ― Between-subjects factors (treatment vs. control) ― Within-subjects factors (the same participants are tested on two or more occasions) 14 Experimental Designs KEY POINTS Experimental designs typically involve a comparison of a treatment and a control group Experimental designs generally allow us to impute causality Common types of experimental design are: ― Between subjects design (random allocation) ― Within subjects design ― Factorial designs ― Quasi-experimental designs? 15 Example 4: Quasi-experimental designs Stress and magical thinking (Keinan, 1994) Background ―Magical thinking and superstition are a common feature of human cognition ―Stress affects cognitive functioning and might also increase magical thinking 16 Example 4: Quasi-experimental designs 1. Research question ― Is there a relationship between magical thinking and stress? 2. The hypothesis ― Those who are exposed to high stress will exhibit higher levels of magical thinking. 3. Operationalise (Measure what/how?) Magical thinking questionnaire e.g., ― If, during a missile attack, I had a photo Saddam Hussein with me, I would rip it to pieces. ― To be on the safe side, it is best to step into the sealed room right foot first. Residents of Israeli cities during the Gulf War (Feb 1991) ― Some cities were frequently bombed (e.g., Tel Aviv) ― Others were never attacked (e.g., Jerusalem) 17 Example 4: Quasi-experimental designs 4. Design experiment ― High stress: Those residing in cities that were frequently attacked were in the High stress condition ― Low stress: Those residing in cites that were never attacked 5. Collect data ― Interviewers knocked on doors; participants filled out the questionnaire 6. Analyse data and look at the results ― Higher rating means higher magical thinking 18 Example 4: Quasi-experimental designs RESULTS 3 Magical thinking 2 1 0 High stress Low stress 19 Example 4: Quasi-experimental designs 7. Draw appropriate conclusion ― Stress results in magical thinking? Advantages and disadvantages of quasi-experimental designs ― Investigate naturally occurring characteristics that could not be induced in the lab (e.g., depression, extreme stress). ― No random assignment, so it is difficult to impute causality (correlation not causation) 20 Example 5: Correlation Designs Pathogens, personality and culture (Schaller & Murray, 2008) Background ― There are cross cultural differences in personality, but the origins of these differences are unknown ― Regional differences in the prevalence of infectious disease might account for some of these cultural personality differences 21 Example 5: Correlation Designs 1. Research question – Is regional disease prevalence correlated with personality? 2. Hypothese – Where infectious disease is prevalent, personality styles that reduce sexual and social contact are functional – Specifically, there will be a negative relationship b/w regional disease prevalence and – Unrestricted sexual style (many casual sexual partners) – Extraversion – Openness to experience 22 Example 5: Correlation Designs 3. Operationalise Measure what? ― Regional disease prevalence ― Regional personality differences Measure how? Archival data ― Old medical atlas data of relative prevalence of infectious disease in 71 regions throughout the world ― 9 different infectious diseases (e.g., malaria, leprosy, tuberculosis etc) ― Cross cultural studies of Big 5 personality factors and socio-sexual orientation (SOI) Measure who? (varied) 23 Example 5: Correlation Designs 4. Design study (notice no GROUPS, so no treatment and control conditions) – Devise reliable coding scheme (e.g., 0= disease absent; 3= present at severe levels at least once) 5. Collate the data – Tabulate all the findings of these many cross-cultural studies using the Big Five and SOI 6. Analyse data and look at the results Disease prevalence was negatively correlated with: – a promiscuous sexual style (r = -.62, in women, -.27 in men) – Openness to experience (r = -.59) – Extraversion (r = -.59) 24 Example 5: Correlation Designs 7. Draw appropriate conclusion Can we conclude that infectious disease prevalence caused differences in personality? (a) Is there a relationship (correlation) between infectious disease prevalence and personality? YES (they have established this) (b) Is the relationship between infectious disease prevalence and personality spurious (false)? § DON'T KNOW FOR SURE (we have no treatment/control), but after removing the effects of many different variables, the relationships still survived § controlled for life expectancy, economic development, climate, Individualism/Collectivism (c) Does the 'cause' (infectious disease prevalence) precede the 'effect' (personality differences)? 25 Example 5: Correlation Designs Usually not possible to determine using correlation But current day personality was more strongly Cause: Effect: correlated with old (1940s) historical disease Disease Personality data than with recent disease data prevalence differences – This suggests disease prevalence precedes personality differences But it could be the reverse… If the reverse were true (i.e., current day personality caused the spread of disease) Effect: Cause: Disease Personality – Then current disease data (not old disease data) would be expected to be more strongly prevalence differences correlated with personality—it wasn’t 26 Correlation designs KEY POINTS Correlation designs have no random allocation to treatment and control groups With correlation designs it is difficult to impute causality ― i. Relationship ― ii. Non-spurious ― iii. Order (cause precedes the effect) Advantage of correlation designs: ― allows us to explore questions that would be difficult or impossible with experimental designs 27 Conclusion Variants of these basic designs are used in most of the social psychology research. Knowing the basic designs will help you to: a) determine whether the researchers have really demonstrated what they are claiming to demonstrate b) think about how you might be able to conduct your own study 28 Conclusion REPLICATION IN PSYCHOLOGY Paper published in Science in 2015 https://doi.org/10.1126/science.aac4716 Nosek and teams of others attempted to replicate 100 psychology studies published in 2008 in 3 top journals. Attempted to do direct replications ― same materials ― greater power Found 1/3 to 1/2 replicated This has led to improvements Increased N (reduces the chance of a fluke) Preregistration of studies (hypotheses; all measures) ― Increased transparency; sharing data 29 30