1/18
Looks like no tags are added yet.
Descriptive Statistics
Means, frequencies, standard deviation, etc.
Inferential Statistics
Assess whether or not there is a significant effect and the size
degrees of freedom
number of values in dataset that are free to vary
N - 1
because once all values but one are set, the last value must be the correct value to produce the observed mean
p-values
probability of observing the results IF the null hypothesis were true
p > .05 means there is a greater than 5% chance that we would get these results if the null hypothesis is true - thus results are nonsignificant
important characteristics of p-values
direction - which group had higher mean
magnitude - small or large effect
prediction - result is aligned with previous literature?
explanations for nonsignificant p-values
expected effect is absent
insufficient statistical power
low sample size
p-values can be highly variable
Basc assumptions of statistical tests
Normality of distribution (residuals, not necessarily raw data)
homogeneity of variance (or homoscedasticity for continuous IVs)
independent samples t-test
splits sample into two independent groups and compares
IV = categorical, 2 levels
DV = continuous
calculate effect size with Cohen’s d (group difference in SDs)
one way ANOVA
compares 2+ groups
IV = categorical, 2+ levels (ex. liberal, moderate, conservative)
DV = continuous
calculate effect size with eta squared
a priori analyses
pre-specified extra tests to illuminate results beyond primary analysis
ex. running a follow up t-test to specifically compare two levels of an IV (if there are more than 2, one would run an ANOVA as the primary test)
Post hoc analyses
unplanned, exploratory follow up tests
cannot use these to draw conclusions, only to bring up possibilities for future research
Factorial ANOVA
2+ Independent Variables
1 Dependent Variable
indicates whether there is a main effect of both IVs and if there is an interaction
Paired T-tests
samples in both levels of the IV are ‘paired’ in some way
often it’s two measurements of the same participant which significantly reduced the noise resulting of individual differences between groups
repeated measures ANOVA
Multiple DVs (ex. performance on multiple tasks)
requires assumption of sphericity
Assumption of Sphericity
assumption that different DVs in a repeated measures ANOVA show similar variances
counterbalancing
for within-subjects designs
present conditions of an IV or different DVS in varied orders to prevent order effects, learning effects, carry over effects, etc.
Correlations
estimate the linear association between two continuous variables
r = correlation coefficient
ranges from -1 to 1 (perfectly negative to perfectly positive)
extreme values/outliers pose serious problems by inflating/deflating correlations
carefully remove outliers
possible to use non-parametric correlations
Multiple Regression
using 2+ independent variables (categorical or continuous) to predict a continuous DV
interactions among predictors (IVs) may exist (moderation)
General Linear Model
between subjects tests follow the General Linear Model:
outcome = model + error