Behavioural Research Methods 2

Quantitative research cycle (6 steps)

1. Problem formulation

2. Hypotheses

3. Research design

4. Data collection

5. Analysis

6. Conclusions

Categorical or nominal variable

Distinct groups with no inherent order

Ordinal variable

Ordered categories, uneven intervals (never/always)

Interval or continuous variables

Numerical with equal intervals

H0

Assumes no effect or difference

H1

Assumes a difference or effect

When p value is below 0.05, we (…) H0

Reject

Type I or alpha error

False positive (Reject a true H0)

Type II or beta error

False negative (Failing to reject a false H0)

Central limit theorem

The distribution of sample means approximates a normal distribution for large enough samples

Test for Categorical by Categorical with cell count over 5

Chi squared

Test for Categorical by Categorical with cell count under 5

Fisher’s exact

Test for Categorical by Interval with normal distribution

T test

Test for Categorical by Interval with no violated normality

Mann-Whitney U / Wilcoxon Ranksum

Test for more than two categories by interval

Anova

Test for Interval by Interval with normal distribution

Pearson Correlation

Test for Interval by Interval with violated normality

Spearman Corelation

Test for multiple variables by Interval

Multiple regression

Test for Continuous or Categorical by Binary

Logistic Regression

Empirical vs. theoretical distributions

Observed data vs. idealized data used to estimate probabilities

Confidence interval

Range of plausible population values for a parameter

Low degree of peakedness

Platykurtic distribution

High degree of peakedness

Leptokurtic distribution

What is a chi squared test

The chi-squared test measures whether the observed counts in categorical data differ significantly from expected counts under independence or a specified distribution.

What does shapiro test do

Calculate average distance between dots and line

Procedure Wilcoxon Ranksum test

Orders data from lowest to highest, ranks the data and sum numbers per group and compare the ranksums

What does Cohen’s D calculate

How many standard deviations the difference between group means is

P value answers: (…), effect size answers: (…)

Is there evidence for a difference, how big is the difference

Regression

Compares means of continuous outcomes (Y) between 2 or more groups (X)

Anova

Models how continuous outcome (Y) changes with 1 or more predictors (X)