research methods exam 3

factorial designs

measures 2 independent variables (often working together) at the same time

interaction

The effect of one independent variable depends on the level of another independent variable

main effect

overall effect of one IV, averaging across all levels of the other IVs

parallel lines in a factorial design

no moderator

What is the difference between an interaction and a moderator effect?

• The interaction effect is the pattern in the data (lines crossing or changing slope).

• The moderator is the variable responsible for that pattern.

between subjects

• treatments/manipulations are administered to different groups of people

• one independent variable

within subjects

single group of participants exposed to ALL treatments/manipulations

how to handle error variance

• make conditions as uniform as possible

• increase effectiveness of your IV

• random assignment

between subjects design limitation

level of effect—> need multi-level design

matched group design

An experimental research technique where participants are paired based on shared, relevant characteristics

advantages of matched groups design

control a key variable to reduce confounds

limitation of matched groups design

time consuming and difficult

within subjects design is also known as

repeated measures design

advantages of within subjects design

• requires fewer participants

• minimizes person confounds or error due to individual differences

disadvantages of within subjects designs

• demand on participants

• attrition

• increased likelihood of figuring out hypothesis

• carryover effects

attrition

drop out

carryover effects

lingering impact of a previous treatment, action, or experience on a subsequent, different condition or time period

sources of carryover

learning, fatigue, habituation, sensitization, contrast

counterbalancing

vary the order in which participant experience different conditions

complete counterbalancing

use every possible order of all experimental treatment conditions

partial counterbalancing

taking a limited number of orders from a random pool of all possible orders

descriptive statistics

Describe a data set in terms of location and variability (graphs, frequency counts, measures of central tendency, measures of variability)

bar graph

The length of the bar represents the value of the dependent variable

histogram

frequency distribution across categories/ levels/ responses

normal distribution

symmetric distribution

skew

unsymmetrical distribution

line graph

usually for continuous data as opposed to categorical

scatterplot

plot each pair of observations on a graph to notice trends and identify outliers

frequency distributions

used for categorical data

central tendency

indicator of where the center tends to be— what’s a typical person in your sample like

skewed distribution use

median

normal distribution

mean

variability

• tells us how similar the scores are and how bunched up or spread out they are

• 2 types

standard deviation

square root of the deviations (variance)

deviation

measure of the distance of all points from the mean

quasi-experimental design

estimates the causal impact of an intervention on a population without using random assignment 9could be confounds)

quasi experimental design advantages

generalizability, lower resources

quasi experimental disadvantages

no random assignment

posttest-only design

intervention is applied and dependent variable is measured only after the treatment

pretest-posttest design

measures participants before and after an intervention to assess change

ecological momentary assessment designs

repeated assessments over the course of time while people are functioning within their natural settings

ecological momentary assessment designs strengths

reduces retrospective recall bias, ecological validity, and ability to test temporal relations

ecological momentary assessment designs disadvantages

costly, measurement reactivity, attrition

longitudinal study advantages

controls confounds and can view trends

longitudinal study disadvantages

time-consuming, not generalizable, attrition, history effects

inferential statistics

uses data analysis from a small, representative sample to make predictions, generalizations, or inferences about a larger population

inferential statistics is based on

probability theory

null hypothesis significance testing

take a set of observations and compare them with what we would expect to observe if there were no difference/relationship

null hypothesis

There is no relationship in the population, and any relationship in the sample is sampling error

alternate hypothesis

There is a relationship in the population, or if you are testing for differences, the 2 conditions are essentially drawn from DIFFERENT populations

p value

measures the probability that observed data occurred by random chance, assuming the null hypothesis is true

alpha level

threshold at which you’ll decide tghst the observed relarionship is probably ISN’T due to chance; usually .05

type 1 error

occurs in hypothesis testing when the null hypothesis is true but is incorrectly rejected

cohen’s d

strength of the effect

type 2 error

when you fail to reject a null hypothesis that is actually false, essentially missing a real effect or difference

power

the probability that a test will reject the null hypothesis when the null hypothesis IS false (usually .80)

powerful designs uses

enable us to observe differences or changes that really exist

ways to increase power

sensitive measures, increase EFFECT SIZE, increase sample size, decrease alpha level, control extraneuous variables

correlation

when you want to find a RELATIONSHIP between 2 CONTINUOUS VARIABLES

t-test

when you want to find a DIFFERENCE between 2 conditions (movie with sound/vs none)

ANOVA

when you want to find a DIFFERENCE between MORE THAN 2 conditions (birth order)

correlation coefficient (r )

index of the relationship between the 2 variables

key features in interpreting r

direction of relationship (+/-) and strength of relationship (-1 to +1)

independent samples t-test

between subjects design

paired samples t-test (dependent samples t-test)

within-subjects desgin

t-test

differences between 2 groups

ANOVA

3 or more groups

one way ANOVA (one independent variable)

oen independent variable wirh 3 or more groups

factorial ANOVA

two or more independent variables

if categorial IV…

Factorial ANOVA

if continuous independent variables…

typically in multiple regressions

p does not effect…

effect size

observer bias

when observers see what they want to see

observer effects

when the observer’s presence changes participant behavior

reactivity

wgen participants react to being watched

ways to prevent observer bias

detailed operationalization, observer training, interrater reliability