College aantekeningen Practice of empirical research (6474PEOY_2526_S1)

LECTURE 1 FROM RESEARCH
PROBLEM TO ANALYSIS
GENERAL INFORMATION COURSE

Course goal

 Bring prior information on research methods together, and expand.
 Prepare you for:
o Master thesis
o Future career
o Not just for end paper

Course components

 Lectures, tutorials, computer labs
o Linked, but not completely overlapping.
 Brightspace
o All info
o Week-by-week
o Preparations  before sessions
 Coordinators
o Sarah Plukaard:
o Jasper Maas:

To pass the course…

 Participate in all tutorials and computer labs
 Review assignment (no grade)
 End paper (grade 5.5 of higher)

STATISTICS TECHNIQUES

Statistical techniques from the bachelor program

 Z-test & t-test
 Correlation & regression
 ANOVA
 Non-parametric tests

THIS IS ALL THE TECHNIQUES FOR YOUR END PAPER!!!!!

,What’s new in PEO?




FROM RESEARCH PROBLEM TO ANALYSIS

Multivariate analysis: the basics

 Bivariate analysis: the relationship between 2 variables
o X (horizontal)  Y (vertical)
o Purpose
 To identify correlations or associations.
 To test hypotheses about how variables relate.
o Common methods
 For 2 numeric variables: correlations (Pearsons,
Spearman), scatterplots, regression.
 For 1 numeric & 1 categorical variable: t-test,
ANOVA
 For 2 categorical variables: chi-square test, cross-
tabulation
 Multivariate analysis: relations between multiple variables at a
time.
o Variables
 X1 & X2…?  Y
 X  Y1 & Y2…?
 Both
o Purpose
 To see how several variables together influence an
outcome.
 To control for confounding variables  variables that
might distort the main relationship.
 To identify patterns or groupings in complex data.
o Types/methods

,  Multiple regression: examines how several
independent variables affect a single dependent
variable.
 MANOVA (multivariate ANOVA): compares groups on
multiple dependent variables simultaneously.
 Factor analysis : reduces many variables into fewer
underlying factors.
 Cluster analysis: groups observations based on
multiple variables.
 Canonical correlation: studies the relationship
between two sets of multiple variables.

Complexity of research questions

 Univariate descriptive: one variable at a time and describes its
main features.
o Purpose
 To summarize & describe a single variable
 To understand its distribution, central tendency and
spread.
o Common descriptive measures
 For numeric (continuous) variables
 Mean
 Median
 Mode
 Range, variance, standard deviation
 For categorical variables
 Frequency counts
 Percentages/proportions
 Bivariate: a descriptive way to summarize how two variables relate.
o Purpose
 To summarize how two variables vary together.
 To detect patterns, trends or associations.
o Common descriptive tools
 Cross-tabulation
 Scatterplots
 Correlation coefficients
 Grouped summaries
o Symmetric: the relationship between 2 variables is mutual 
no distinction between ‘cause’ (predictor) and ‘effect’
(outcome).
 No direction implied.
o Asymmetric, non-causal relationship: one variable is
considered the independent or predictor variable and the other

, is the dependent variable, but this relationship does not imply
causation. It just describes how 1 variable varies with another.
 Directional but not causal
o Asymmetric, causal relationship: one variable causes of
influences the other. This is the typical focus in experimental
research or causal modeling.
 Directional and causal.

Causali
Type Direction Example
ty

Symmetric None No Height vs. weight

Asymmetric non- One → Study hours → exam scores
No
causal other (descriptive)

One →
Asymmetric causal Yes Fertilizer → plant growth
other

 Multivariate: summarize and describe more than 2 variables at the
same time. Looks at patterns, relationships and distributions across
multiple variables simultaneously.
o Purpose: to understand complex interactions, patterns or
structures among several variables at once.
o Examples of multivariate descriptives
 Means and standard deviations for multiple variables
together.
 Correlations matrices
 Covariance matrices
 Cross-tabulation
 Multivariate plots
o When uses: often before performing advanced analyses like
factor analysis, multiple regression or cluster analysis to
understand the structure of the date.

Matching problem with technique