Quantitative Methods Flashcards
When is an independent samples T-test used?
Used to determine if the mean differs significantly between 2 populations
When is a paired T-test used?
When samples are linked with each other e.g. heart rate before and after exercise
What does Mann-Whitney tell us?
(Non-parametric) Are two independent population medians different?
What does Wilcoxon signed ranks test tell us?
(Non-parametric) Are paired sample values different
What do ANOVA and linear regression both require to be used?
Normal residuals and equal variances (or sphericity for >2 levels)
What tests for normality are used?
Kolmogorov-Smirnov
Shapiro-Wilk
What is the difference between a One-way ANOVA and a Two-way ANOVA?
They either have one or two independent variable
How to test for sphericity?
Use Mauchly’s test of sphericity
When do you need to test for sphericity?
When doing a repeated measures ANOVA
What to do if sphericity is not assumed?
Use Greenhouse-Geisser or Huynh-Feldt
Assumptions of regression
- Residuals are normally distributed
- Error variance of y is equal across the range of predicted y-values
- The relationship is linear
- No relationship between the residuals and the X and Y variables
What is a Levene’s test?
Used to test for equality of variants before carrying out a T-test or ANOVA
What is Pearson’s correlation also known as?
Is it parametric or nonparametric?
Pearson’s product moment correlation coefficient (PARAMETRIC)
What is Pearson’s correlation?
Ranges from -1 (perfect negative correlation) to 1 (perfect positive correlation)
A value of 0 indicates that there is no association between variables
What is the non-parametric version of Pearson’s correlation coefficient?
Spearman rank correlation
What is the difference between a univariate and multivariate statistical test?
For a univariate test, you test only a single response variable but for a multivariate test, you examine two or more response variables
What is a One Sample T-test?
Tests whether or not the mean of a set of values you collected is significantly different to a fixed value
What is the non-parametric version of the One-Sample T-test?
Wilcoxon matched-pair signed-tank test
What is the non-parametric equivalent of ANOVA?
Kruskal-Wallis test
What is a Bonferroni correction?
Carried out when you do multiple tests with a chance of getting false positives. Multiply any p-values you obtain by the number of tests you perform
When to use linear regression?
If your DEPENDENT variable increases or decreases linearly with your INDEPENDENT variable
What to do if your dependent variable doesn’t vary linearly with your independent variable
Transform the data or do non-linear regression
Is the Chi squared test parametric?
No, so it does not make assumptions about normality or equal variances
What is the Chi squared test used for?
Used for looking at counts of discrete events
What does the Chi squared test require?
That all counts are independent
What are the two ways to use the Chi squared test?
- Goodness of fit
- Test of association
The difference is the way we calculate the expected values with which to compare our observations
Chi squared goodness of fit
Prior knowledge of what expected ratios will be (e.g. Mendelian inheritance has known ratios of offspring)
Chi squared test of association
No fixed idea about the rations of expected results - instead they are calculated based on the assumption that there is no association between the two variables
What is R Squared?
The proportion of variance measured which is due to the independent variable you are changing
What is binary logistic regression?
Deals with cases where you measure a binary outcome such as dead vs. alive and you want to see if a predictor influences the proportions of your two possible outcomes
What statistic do you look at for binary logistic regression?
Wald statistic
What is the non-parametric version of a repeated-measures ANOVA? What is its shortcoming?
Friedman test
Can’t deal with between-subjects factors
What is Poisson probability?
Used to calculate the probability of n occurrences of an event when you already have a measure of the average
What is homoscedasticity?
Equal variance over a range of y-values
Does binary logistic regression assume equal variances and normally distributed residuals?
No