Midterm Flashcards
What is social psychology
The study of individuals (thoughts, feelings, actions) in environments. Often empirical and experimental approach.
Aristotle on social psych
To what extent is beauty “in the eye of the beholder”
1900 - Triplett experiment
Noticed that cyclists were faster in the presence of others. Designed experiment to test this. Showed that context shapes our behaviours without our knowledge (even when trying our hardest).
Allport 1924 on social psych
In particular stressed interactions between individuals and social context and focus on experiments.
1930s-40s: Hitler
Social psych is very North American, because people fled here during WWII, and then started studying why these things happen - humans trying to determine how humans could do such things to other humans.
The goal of any science
To slowly accumulate evidence in support of (or refuting) theories about the world. Impossible to “prove” a theory, the evidence in support of it just becomes overwhelming. The “evidence” is studies that people do testing various parts or predictions of theories.
Replication
For evidence to be considered evidence, it is important that independent labs can run the same experiment, and get similar results, over and over again.
Initial Spark experiment
100 participants were showed erotic and non-erotic photos. They guess behind which door was the erotic stimulus. Actual location of stimulus wasn’t there until AFTER participants made their guess. Work was generally considered well done. but no one believed that psi was a thing. So he followed all the rules, and published what many considered to be rubbish. Must mean that the rules allow rubbish… what else is rubbish?
Ingredients that contribute to less replicable science
- Institutional Pressures: “Publish or perish”
- Flashy and significant effects needed to be published: Publication bias
- Lots of ways to analyze data: Garden of Forking paths, p-hacking, intentional and unintentional.
Solutions for replication crisis
Establishing best (statistical and methodological) practices to avoid p-hacking. Revisiting established effects and support for replicating what we though of as real things. Registered reports. Pre-registration of hypotheses. Open data, open code, methods.
Examples of effects proven to be wrong
Power posing: Posing with expansive posture will activate testosterone and lead to more confident behaviour.
Ego-depletion: self-control or willpower draws upon a limited pool of mental resources that can be used up.
Stereotype Threat
The concern about confirming a stereotype leads to confirmation of that stereotype.
Replication crisis on science in general
The replication crisis generally illustrates that science is working as it is supposed to. Weaker effects, not real effects, don’t replicate, and get dropped from the literature over time.
Experimenter Bias
The experimenter having an effect on the study.
Ultimate goal of science
To understand why things happen. If we have a good understanding, we can predict the future. We want to know to what extend X causes Y.
Experiments
Key is manipulation. Two conditions are exactly the same, except for one thing. If outcome is different across conditions, we can have some certainty that manipulation caused that difference.
Observation vs experiment
There are tons of stuff we can’t manipulate: age, race, tidal waves/earthquakes. So use observation then.
Random sampling
Want to generalize our results to all humans. To do this we have to randomly sample from all humans. It maximizes odds that the we don’t have a “cohert” effect and that our results generalize.
Weird problem of random sampling
Most research comes from western, educated, industrialized, rich, democratic places. Most data comes from psych undergrads and rats. This is because of convenience samples.
Random assignment
We want the only thing different between different conditions to be the experimental manipulation. Humans vary in a lot of ways. So we randomly assign people to a condition, and trust that the random participant characteristics even out/is non-systematic enough that it doesn’t influence results.
Control group.
Important for determining direction of effect, and if the effect is actually different from the baseline.
Single-blind
Participants don’t know what condition they are in. Eg. real drug vs. placebo
Double-blind
Participants and researchers don’t know what condition participants are in.
Questions to ask when evaluating claims
- Based on any data?
- Questionable motives of people doing the research?
- How many participants?
- Who were participants?