QUANTITATIVE METHODS FOR POLITICAL SCIENCE – Final Exam
UNIT 1: Intro to Sta@s@cs
INTRODUCTION
Social scien+sts are interested in answering ques+ons of the sort: How prevalent is child
obesity in Mexico? Is ball possession in football associated with good results?
In order to answer these ques+ons, we need to:
• Gather data.
• Analyse it.
• Interpret the results.
What are STATISTICS? à The discipline that concerns the collec+on, organiza+on
analysis and interpreta+on of data.
DEFINITIONS
POPULATION à The complete set of all items that interest the inves+gator. Its size can
be very large (oCen assumed to be infinite). We need to know what the inves+gator
needs to answer.
SAMPLE à Observed subset (or por+on) of a popula+on. It is used to create a
HYPOTHESIS –you can never be 100% sure because you cannot have access to the en+re
popula+on–. Must be representa+ve à “Small copy of the popula=on”.
DESCRIPTIVE STATISTICS à Graphical and numerical procedures used to summarize and
process data à “EXPLORATIVE DATA ANALYSIS” (EDA). Consists in conver+ng data to
informa+on.
INFERENTIAL STATISTICS à Using informa+on contained in the sample to learn about
the popula+on. Consists in interpre+ng the informa+on in order to make conclusions.
SAMPLING à Process whereby the sample is selected from the popula+on. Its goal à
Finding a sample that represents the popula+on well.
Different procedures:
• SIMPLE RANDOM SAMPLING à Ideal representa+ve procedure through
random selec+on. Is the one used the most in the prac+ce.
o Procedure whereby n objects from the popula+on are selected purely by
chance.
o The choice of a given member of the popula+on has no effect on the
selec+on of any other member.
o Most desirable sampling methodology. However, not always possible to
carry out à Because it is the most expensive procedure.
• STRATIFIED SAMPLING
, o Used when the popula+on can be divided according to well defined,
observable characteris+cs.
o Example à Gender, na+onality, loca+on, etc.
o Method à Define the strata and randomly select individuals within each
stratum.
o Most oCen, the number of observa+ons selected from each stratum
mimics the weight of the stratum in the popula+on.
• CLUSTER SAMPLING à Not all groups are used.
o Methodology:
§ Group individuals in the popula+on into subgroups (CLUSTERS).
§ Randomly select a given number of clusters.
§ Use ALL individuals from the selected clusters to construct the
sample.
o OCen easier to implement than SIMPLE or STRATIFIED SAMPLING.
• NON-RANDOM SAMPLING à NOT RANDOM because the existence of biases
when selec+ng people.
o CONVENIENCE SAMPLING à The sample is drawn from individuals that
are easily available.
§ Example à Asking every other pedestrian at La Rambla about
their poli+cal views.
§ Advantages à Very low cost – good for pilot studies.
, § Disadvantages à Most likely, not a good representa+on of the
popula+on.
• WHY? à SELF-SELECTION BIAS (among mul+ple other
poten+al biases).
CLASSIFICATION OF DATASETS
Researchers oCen work with datasets à Usually SAMPLES.
• These contain:
o OBSERVATIONS/INDIVIDUALS à A case of the data being collected.
o VARIABLES à Adributes of each observa+on.
• There are three main TYPES OF DATASETS:
o CROSS–SECTIONAL DATA à Collec+on of data on many subjects gathered
at one point in +me.
, o TIME SERIES à A set of data points indexed in +me order. Only one object
of study.
§ Example à Unemployment rate in Spain (2017–2020).
o PANEL DATA à Data that contains informa+on on different cross-sec+ons
across +me. Many objects to study in many +me points.
CLASSIFICATION OF VARIABLES
VARIABLE à A specific characteris+c of an individual or object.
Examples: height, weight, wage, poli+cal orienta+on, number of siblings, na+onality,
favourite colour, etc.
Correctly classifying variables is essen+al to perform sound sta+s+cal analysis.
Criteria:
• CATEGORICAL/QUALITATIVE à Produce responses that belong to categories.
Are not numerical, cannot be counted. Do not have numerical informa+on.
UNIT 1: Intro to Sta@s@cs
INTRODUCTION
Social scien+sts are interested in answering ques+ons of the sort: How prevalent is child
obesity in Mexico? Is ball possession in football associated with good results?
In order to answer these ques+ons, we need to:
• Gather data.
• Analyse it.
• Interpret the results.
What are STATISTICS? à The discipline that concerns the collec+on, organiza+on
analysis and interpreta+on of data.
DEFINITIONS
POPULATION à The complete set of all items that interest the inves+gator. Its size can
be very large (oCen assumed to be infinite). We need to know what the inves+gator
needs to answer.
SAMPLE à Observed subset (or por+on) of a popula+on. It is used to create a
HYPOTHESIS –you can never be 100% sure because you cannot have access to the en+re
popula+on–. Must be representa+ve à “Small copy of the popula=on”.
DESCRIPTIVE STATISTICS à Graphical and numerical procedures used to summarize and
process data à “EXPLORATIVE DATA ANALYSIS” (EDA). Consists in conver+ng data to
informa+on.
INFERENTIAL STATISTICS à Using informa+on contained in the sample to learn about
the popula+on. Consists in interpre+ng the informa+on in order to make conclusions.
SAMPLING à Process whereby the sample is selected from the popula+on. Its goal à
Finding a sample that represents the popula+on well.
Different procedures:
• SIMPLE RANDOM SAMPLING à Ideal representa+ve procedure through
random selec+on. Is the one used the most in the prac+ce.
o Procedure whereby n objects from the popula+on are selected purely by
chance.
o The choice of a given member of the popula+on has no effect on the
selec+on of any other member.
o Most desirable sampling methodology. However, not always possible to
carry out à Because it is the most expensive procedure.
• STRATIFIED SAMPLING
, o Used when the popula+on can be divided according to well defined,
observable characteris+cs.
o Example à Gender, na+onality, loca+on, etc.
o Method à Define the strata and randomly select individuals within each
stratum.
o Most oCen, the number of observa+ons selected from each stratum
mimics the weight of the stratum in the popula+on.
• CLUSTER SAMPLING à Not all groups are used.
o Methodology:
§ Group individuals in the popula+on into subgroups (CLUSTERS).
§ Randomly select a given number of clusters.
§ Use ALL individuals from the selected clusters to construct the
sample.
o OCen easier to implement than SIMPLE or STRATIFIED SAMPLING.
• NON-RANDOM SAMPLING à NOT RANDOM because the existence of biases
when selec+ng people.
o CONVENIENCE SAMPLING à The sample is drawn from individuals that
are easily available.
§ Example à Asking every other pedestrian at La Rambla about
their poli+cal views.
§ Advantages à Very low cost – good for pilot studies.
, § Disadvantages à Most likely, not a good representa+on of the
popula+on.
• WHY? à SELF-SELECTION BIAS (among mul+ple other
poten+al biases).
CLASSIFICATION OF DATASETS
Researchers oCen work with datasets à Usually SAMPLES.
• These contain:
o OBSERVATIONS/INDIVIDUALS à A case of the data being collected.
o VARIABLES à Adributes of each observa+on.
• There are three main TYPES OF DATASETS:
o CROSS–SECTIONAL DATA à Collec+on of data on many subjects gathered
at one point in +me.
, o TIME SERIES à A set of data points indexed in +me order. Only one object
of study.
§ Example à Unemployment rate in Spain (2017–2020).
o PANEL DATA à Data that contains informa+on on different cross-sec+ons
across +me. Many objects to study in many +me points.
CLASSIFICATION OF VARIABLES
VARIABLE à A specific characteris+c of an individual or object.
Examples: height, weight, wage, poli+cal orienta+on, number of siblings, na+onality,
favourite colour, etc.
Correctly classifying variables is essen+al to perform sound sta+s+cal analysis.
Criteria:
• CATEGORICAL/QUALITATIVE à Produce responses that belong to categories.
Are not numerical, cannot be counted. Do not have numerical informa+on.
