Social Psych Test 1

4 ways of knowing

intuition, authority, logic, observation

Diedrek Stapel 2011 Research Fraud

Caught fabricating research in 2011 in their lab, leading to many of their papers getting pulled from journals. Helped prompt the open science movement

Publish or perish

Good researchers publish their work. If your experiment isn’t significant or doesn't align with your expectations, you could lose grant funding.

File drawer Problem

Research that is not successful or significant is just stored away, meaning it cannot contribute to future developments.

Simmons et al 2011 False Positives Study

Showed how easy it was to find a significant correlation between two nonsense variables.

Researcher Degrees of Freedom

number of decisions or practices that a researcher makes that might not be reported

2015 reproducibility crisis

Tried to replicate 100 studies, and only 30-40% replicated

Three ways to find a research question

Observe the world around you, Look inward, Read past research

3 Criteria for Concluding Causality

Covariance / Correlation

Temporal Precedence

Non-spuriousness (only one cause)

why randomization is important for social psych

Lots of variability and statistical noise

Random Error

Variability among answers

systematic error

Gets in the way of making causal claims (confounds)

Random selection v random assignment

The selection of participants should be random and representative of the population. Once selected, participants should then be sorted into groups.

Between Subjects Design

two different groups where each one receives one level of the independent variable

Factorial Design

Looks at multiple IVs on a DV (interaction)

Within Subjects (repeated measures)

Each participant is in all levels of the IV. Completely removes random error, but order effects need to be accounted for

Mixed Design

Contains at lest 1 between subjects IV and one within subjects IV.

Carry over effects for within subjects designs

Habituation, sensitization, learning, fatigue, adaptation

Confound

something that systematically varies with our variable, and you dont realize it

ways to avoid experimenter bias

double blind study, lie detecting questions, anonymity

ways to avoid confounds

experimental conditions need to be exactly the same for each participant

quasi experimental design

Something prevents it from being a true experiment, often in social psych because of preassigned groups or examining a continuous moderator

Moderation

A continuous or categorical variable that explains “who, where, when” in an experiment. It is manipulated or measured, and changes effect of IV on the DV

Mediation

Looks at a variable that explains a relationship between two other variables, “how? why?'“ Typically continuous, but can be categorical, and explains the effect of IV on DV

total effect

relationship between predictor and outcome

Indirect effect

Amount of relationship explained by process variable (mediator)

Direct effect

amount of relationship not explained by process variable

Operationalization

Creating a way to measure the research question

Control Group

Must have as similar as possible of an experience to the expiremental group

Cover Story

The simplest form of deception in an experiment. Must be relatively simple and believable

Ways to administer an IV

Personal Administration, Written Instructions, or some combination there of

statistical power

significance of results, achieved by strength of manipulation and sample size (N)

Pilot testing

Initial testing of operationalization to see how well it works before the experiment is started. Done with real participants or those naive to the study. Also can check scale reliability

Manipulation Check

Checks if participants are picking up on the operationalization. Can be during the study (internal) or during a pilot test (external)

Operationilizing the DV

how you will measure the variable

Behavioral DV

measures behavior, typically the truist dv

Quantitative behavioral DVs

Frequency, Duration, Proximity/distance, amount, speed, all or nothing

Behavioral Trace

the result of a behavior that can be measured

Behavioroid

Measuring the intention of behavior / past behavior

Verbal DVs

measuring teh response of a participant, can be closed ended or open ended

Close ended verbal response

response is confined often by a scale or comparison

open ended verbal DV

no limit to response. could be recall, free response, or a structured interview

matching procedure

match people with similar charatersitics, and then randomly put one in the expiremnetal group and one in the control

ways to eliminate counterbalancing

counterbalancing (every possible combo), latin square (each variable in each ordered position) , random order

two main types of confounds

participant / expirementer, design

observational research

watching people in their natural element. much easier to conduct but cannot determine causality