L4 Flashcards
What type of test do we do if we want to know if there is a difference in the means when there are more than two means?
ANOVA
(Analysis of Variance)
What are the assumptions of a one-way analysis of variance (independent)?
What are boxplots useful for?
Spotting outliers
Useful for spotting impossible scores and observations
What are density/histogram’s useful for?
Probe the normality assumption and the normality of variance assumption
What does the distribution look like in our data?
What are dot plots good for?
They are goot at spotting outliers and for looking at the distribution.
- Raw data points help represent the distributional data while it is overlayed with the boxplot*
- Best of both worlds*
What does Levene’s test do?
Tests for the homogeneity of variances
What is a contrast test?
Statistical method for testing theoretical predictions about differences between group means against empirical data
a way of testing more general hypotheses about population means.
What is a treatment contrast test?
One level is special and acts as a “baseline” category (i.e. placebo or control condition) against which the other level are defined
Only makes sense to use this type of contrast when we are comparing against a true baseline category
What is a Helmert Contrast?
i.e. when we are interested in an additive effective - e.g. looking at the diffeence in the increase in vitamin C for instance
What are Sum to Zero Contrasts?
A paired comparison, where we are comparing two entitites to a baseline rate.
When can we use Sum to Zero Contrasts
- We have one specific level of our variable that we are interested in comparing each other level of our variable to
- IV of interest is discrete or qualitative
* I.e. happiness levels in states, no principled order of our variables of interest (state in this case)*
* Handy when we are dealing with unordered, discrete variables*
What type of contrasts do we use when the IV is ordered?
Polynomial Contrasts
What is a Polynomial Contrast?
A special set of orthogonal contrasts that test polynomial patterns in data with more than two means
(e.g. linear, quadatic, cubic, quartic, etc)
When should we use a polynomial contrast?
- Used for describing the trend across the different in the variables of interest
- Trying to capture the overall trend across different levels of our independent variable
- Used when dealing with an IV that has an order to it
What type of contast is a trend analysis?
Polynomial contrast
What are trend analysis useful for?
Further articulating the trend or pattern that we are observing
What are the 3 types of trend analysis?
- Linear Trend
- Quadtratic trend
- Cubic Trend
What is a Linear Trend?
No bends, difference between each variable is roughly the same.
What is a Quadratic trend?
There is a bend in the data across the levels of IV
What is a Cubic Trend?
There is at least two bend in the data
What type of analysis should we use if we are dealing with multiple means and a within-subjects design?
One-Way Analysis of Variance (Paired)
What are the assumptions of a one-way analysis of variance (paired)?
What is meant by “we want sphericity in our data” for one way ANOVA’s (paired)
What happens if this is violated?
We want to make sure that different scores are at roughly similar variances across each of our conditions.
If violated in within-subjects design can increase the probability of a type 1 error
What test do we use to test for sphericity?
Mauchly’s sphericity test
How do we know if sphericity has been violated using the results of Mauchly’s test?
What is the test statistic?
If p < .05
Test stat: W
If Mortley’s test tells us spherisity has been violated, what test can we use to correct it?
Greenhouse-Geisser Correction (more conservative and common)
or
Heinfelt Correction
What type of analysis only has only one variable being analysed?
Univariate
What is a bivariate analysis?
Used to find out if there is a relationship between two different variables
What is a multivariate design
The analysis of three or more variables
What is a 2x2 within-subjects design?
There are two variables and there are two levels to each variable
Every participant contributes to each of the two levels of the two IV’s
What is a mixed design?
Some of the variables are within-subjects and some are between-subjects
What is sphericity?
It is the condition where the variances of the differences between all possible pairs of within-subject conditions (i.e., levels of the independent variable) are equal
