Statistical Tests (Parametric & Non) Flashcards
Statistical Tests
Parametric vs Non-parametric tests
A parametric test is a statistical test that assumes the sample data comes from a population that follows a normal distribution.
Non-parametric tests are used when your data is not normally distributed. Every parametric test has a non-parametric equivalent. We have positively & Negatively Skewed.
- Something went wrong
- Parametric Test is simply inapplicable
What are the parametric tests and their non parametric counterpart?
Test of Difference (Is there a significant difference?):
* Parametric: T-test (2 groups), (IV nominal; DV interval/ratio) & ANOVA (> 2 groups), (IV nominal; DV interval/ratio)
* Non-parametric: Mann-Whitney Test (2 groups), (IV nominal; DV ordinal) & Kruskal-Wallis Test (> 2 groups), (IV nominal; DV ordinal)
Tests of Relation (Is there a significant relation?):
* Parametric: Pearson Correlation (r)
* Non-Parametric: Chi-square test of association (IV & DV nominal) & Spearman Correlation (IV & DV ordinal)
Tests of Prediction/Effect:
* Parametric: Linear Regression (IV & DV interval/ratio)
* Non-parametric: Logistic Regression (DV nominal) & Ordinal Regression (IV & DV ordinal)
Pearson Correlation Coefficient
What is a correlation?
Describes the association/relation between the independent and the dependent variable in terms of strength and direction. A scatter plot visually represents a correlation.
Positive: The independent and dependent variables are changing in the same direction (both increase or decrease).
Negative: The independent and dependent variables are changing in opposite directions (as one increases the other decreases).
Beware of spurious correlation- a mathematical relationship between two variables that appear to be causal but are not.
What are the assumptions of the Pearson Correlation Coefficient?
- Randomization
- Normality
- Interval or ratio level of measurement for the independent and dependent variables
- Linearity (straight line)
Table R in statistical table
What are the non-parametric tests of relation?
Spearman Correlation (IV & DV Ordinal) and Chi-square test of association (IV & DV nominal)
Correlation vs. Causation
Correlation only establishes that a relationship exists (it discuses what both variables have in common) whereas Causation means that the two variables share a causal relationship (the independent impacts the dependent).
Analytical difference between Correlation (r) and Linear regression
Pearson correlation measures the strength and direction of a correlation. Linear regression measures the effect of X (predictor variable) on Y (outcome variable).
If you know something about X, this knowledge helps you predict something about Y.
Test of prediction/effect
What is the purpose of linear regression?
- To determine the amount of variance (change) in the dependent variable is being accounted for or explained by the independent variable
- To determine the effect/ impact of an independent variable on the dependent variable
- To predict a score on the dependent variable from a score on the independent variable
Test of prediction/effect
What are the assumptions of linear regression?
same as correlation
- Randomization
- Normality
- Interval or ratio level of measurement for the independent and dependent variables
- Linearity (straight line)
same as correlation
What is the coefficient of determination (R2)?
A measure of how well a model predicts outcomes or tests hypotheses.
What are the non-parametrical tests of prediction/effect?
Logistic (DV nominal) and Ordinal (IV & DV ordinal) Regression
Test of difference
What is the t-test?
The t-test is a parametric statistical test which assesses whether the means of two groups are statistically different from each other.
What are the assumptions of the t-test?
- Normality
- Randomization
- Equal variance of both samples (homogeneity of variance)
- What level of measurement?
1. Independent variable?
2. Dependent variable?
What are the types of t-tests?
Independent samples t-test: used to determine if there is a significant difference between the means of two groups/samples. This uses a between-subjects study design.
Dependent samples t-test: used to determine if there is a significant mean difference within the same group of participants at two points in time. This uses a within-subjects study design.
