PredictionYour expectations for what will happen after operationalizing your hypothesis.InductionObserve then generalize. Making many observations and then trying to find a general principle that would explain or connect those to observationDeductionStart with an assumption and test it. We know that X and Y are true-they should lead to Z.External ValidityExtent to which results from the study can be generalized to a broader set of circumstancesReplicationWould the results replicate if we conducted the study again using the same methodsConditionsWhich part of the study participating inMain EffectThe effect of one IV on the DV, collapsing across the levels of the other variables. One m.e. per IVMarginal MeanThe average level of one level of your IV, compared to the average of the other level of your IV . You need as many marginal means as levels.InteractionThe effect of one IV on the DV depends on the level of the other IVModerator VariableVariable that changes the strength of the relationship between the IV/Predictor and the DV/outcomeFailure of Random AssignmentAlthough you do use proper methods to randomly assign participants, the two groups just differ because of change in some way that is relevant to the study
-height/weight in a study of appetite
Prevention: really big sample, control for it using statsFailure to manipulate your IVYour manipulation did not manipulate what you meant it to because of poor operationalization
Prevention: operationalize carefully, manipulation chart, ask what am I really manipulatingProcedural ConfoundWhen you manipulate more than one thing. (didn't hold everything else constant)
Prevention: don't be too ambitions, think about it, play it safeInternal ValidityConfidence that the experimental manipulation (IV) actually caused the change in the outcome measure (DV)Relationships between confounds and internal validityProcedural confounds compromise the internal validity of a studyPoor Operationalization of your DVWhen your DV is not really measuring what you want it to
Prevention: Look at previous research, multiple measures of the same DV, think about itWithin Subject StudyWhen a group of people see all levels of one of your IV'sOrder EffectsWhen the order you give the IV's in effects the DVPractice EffectsWhen people get better at your DV over time because of practiceFatigue EffectsWhen participants get habituated to your DV or tired over the course of the experiment
-mentally, physically, or boredCounterbalancingPresenting your within subject IV in all possible ordersLatin SquareA square with every possible order. Used with lots of levels of IVAdvantages of within subject design-less participants
-less chance that status variables are influencing results b/c participants act as their own control
-no chance of failure of randomizationDisadvantages of within subject design-order effects
-practice effects
-fatigue
-increase suspicionCase StudyIn depth study of single case or small group of people- Phineas Gage
Pros: Yield exploratory or descriptive data, great way to study unique cases
Cons: Cannot know about external validity( do results generalize) cannot establish causalityRandomized Controlled Trial Major components- how is different from regular experiments-The ultimate gold standard experiment
-They should follow the consort statement guidelines
-very strict set of rules about recruitment
-reporting your methods
-controlling the confounds (e.g. Double blind designs)
-used for clinical trialsCorrelational Studies-Non experimental studies where you do not randomly assign participants to conditions
-Instead, just looking to see whether two or more naturally occurring groups differ( men+women, young+old )
-give us important insightsLongitudinal StudiesStudy following the same group of individuals over time in multiple assessments
Pros: look at change over timeCross Sectional StudyGroups of people are compared to at the same time
Pros: much faster and cheaper than longitudinalWhy don't correlational studies help you assess causality?There are 3 different options that can explain it. There could be a cause but you have to play an experiment to make sure there is a causalityMeta-AnalysisA study of studies
-your subjects are previous experiments instead of individual people
Pros: Helps us answer broad questions, give us average effect size across studies
Cons: Have to be a lot of studies before you can do thisWhat are 3 logical explanations that could explain a correlational relationship found?1. Playing violent video games causes aggressive behavior
2. Aggressive individuals tend to play violent video games
3.Something else is causing individuals BOTH to play violent video games and to be aggressiveMundane Realism"Real World" realism studies that look like the real world
-studying gambling behavior by turning lab into casinoExperimental Realism"Psychological" realism: Studies that are psychologically meaningful to the subject or feel like the real-world
-Asch's conformity studiesField StudyCollection of data in natural settings
Correlational study in "natural settings" could be:
-Shopping mall
-Elevator
-Forest
-BathroomField ExperimentIntroducing a manipulation (IV) into setting outside the lab and observing the effects on people's behavior (DV)ReliabilityReliability is an index of how consistently your measure assigns the same number to the same observation.
Ex. Ruler measures the same thing everytimeValidityValidity is an index of how well your measure actually measures the thing it is supposed to measure.
Ex. Ruler says its 8 inches then it is validNominalDependent measure that has a categorical outcome
Data have no numeric relationship to each other
Examples: DV = Race -> White vs. HispanicOrdinalVariables that involve ordered ranking.
1st place, 2nd place, 3rd place, 4th place, etc.
ExamplesIntervalMeasurement scales that make use of real numbers that reflect relative differences in magnitude. 1, 2, 3, 4, 5, 6, 7, 8, etc.
Examples: DV = aggression
- How much aggression are you feeling right now on a scale from 1 to 10?Self ReportSubjects answer questions that they are asked on paper, orally, or over the internet
Examples:Measuring sexual behavior and condom use
Pros: Cheap, directness, ask a lot of questions in a short time
Cons: May be treated casually, try to make themselves look good, may not be honest with themselvesLikert ScaleA likert scale is when you have people rate how much they endorse a statement on a scaleBehavioral MeasuresGive subjects an opportunity to perform a behavior and see if they do it, how much they do it, how long they do it for, or how many times they do it.
Pros: measure a behavior without the subject knowing what you are measuring, may be closer to construct of interest than responding to a questionnaire
Cons: Harder to do, isn't always possible to measure the behavior you care aboutImplicit MeasuresMeasures of people's unconscious, automatic attitudes that we get without asking them directly.
Often used when assessing "touchy" social issues: prejudice
Examples:Priming
Pros: not influenced by self-presentation concerns, do not require subjects to have accurate self-knowledge
Cons: not always clear what you are actually measuring, can be sensitive to contextPhysiological MeasuresSample some kind of physiological output
Can these also be implicit measures of stress, arousal, neural responses?
Examples: Heart rate / blood pressure (arousal, effort), Cortisol (stress), Neural responses (measuring blood flow to different parts of the brain)
Pros: Subjects usually don't have control over these variables, Can be fairly easy to measure (cortisol)
Cons: Can scare people (blood samples),
Can be expensive, Can be difficult to interpret-if cortisol or heart rate increases is the person stressed or excited or do they have a condition?Double Jeopardy TheoryProposes that there is an additive effect of stigmatized identities or lower status traits on discrimination.
For each stigmatized identity or lower status trait that you have... the more prejudice and discrimination you will face.
-High Status/Non Stigmatized: Male, white, Heterosexual, High income, Firefighter, Christian
-Low Status/Stigmatized: Gay/Lesbian, Black, Hispanic, Transgender, Poor, Janitor, MuslimSocial Dominance TheoryConflict between groups is fundamentally male on male, with males of the dominant group maintaining dominance over males of the subordinate groupInteraction/ModerationWhen the effect of one independent variable on the dependent variable differ depending on the level of a second independent variable
Ex:Does the effect of gender on car price differ depending on what race you are?How do you tell if there is moderation from a bar graph?Draw lines to bars that match. If lines are not parallel then there is an interactionHow do you tell if there is moderation from a line graph?Are the lines parallel? Want the lines to cross that means there is an interaction a difference in pricesHow do you tell if there is moderation from a table?Check to see if there is a difference between the two columns
Ex. Women vs. women, men vs. men, or race vs. race
MARGINAL MEANS ARE COMPLETED DIFFERENT- THEY ARE AN AVERAGE
