Introduction to Statistics and SPSS
Methodology of Conflict and Democracy Studies
December 3

Aim of this lecture
•Variables and their categories
•
•Population and sample
•
•Hypotheses and null hypotheses
•
•Statistical significance
•
•Introduction to SPSS
•
•How to make your own variables

Logic of Statistics
•Deductive logic of research
•
•What we do:
•Derive hypotheses from the theory
•Define variables and operationalize our concepts
•Collect the data
•Test the hypotheses using statistical models
•Provide interpretation and decide whether our hypotheses hold or not
•
•This all requires more than just a few cases

Variables
•Measurable items that change their values
•
•Number of cars on highways, maximum daily temperature, local turnout in elections
•
•Independent (predictor) and dependent (outcome) variables
•
•Main tool for testing hypotheses
•
•

Levels of Measurement
•Completely different categorization of variables than IV and DV
•
•Categorical:
•Nominal
•Ordinal
•
•Continuous:
•Interval
•Ratio

Nominal Variables
•Their values cannot be ordered in a logical way
•
•Names of towns, names of streets, telephone numbers, colors, species of animals, numbers of
players
•
•Binary variables – nominal variables with just two values
•Someone is employed or is not employed
•Citizen either voted in election or did not vote
•You either attend this lecture or you do not

Ordinal Variables
•Their values can be ranked in a logical way however we cannot tell the exact differences between
the values
•
•School grades, Olympic medals, military ranks, age groups
•
•Ordinal variables tell us more than nominal variables (ordering values) but less than interval and
ratio variables

Interval and Ratio Variables
•Interval:
•We can order the values and we know the differences
•Equal intervals on a scale represent equal differences
•Temperature in Celsius
•
•Ratio:
•Same as interval but ratios of values are meaningful
•They contain a true zero
•Distance in kilometers, time in seconds, number of books in a library
•
•In SPSS interval and ratio variables are under the same label (scale)
•

Continuous or Discrete?
•Scale (interval, ratio) variables can be either:
•Continuous
•Discrete
•
•Depends on whether the values can take any values on a scale
•
•Success rate in a test (in %)
•Number of kids in families

Nominal
Ordinal
Scale
(interval, ratio)
Can we logically order the values?
No
Yes
Yes
Do we know differences between the values?
No
No
Yes
Continuous or discrete
Discrete
Discrete
Continuous / discrete

Population and Sample
•Population:
•Includes all possible subjects of a dataset
•All towns of a country, all university students
•
•Sample:
•Includes only part of the cases and it is a subset of the population
•Important feature – representativeness
•1,000 people in a survey
•Many ways of selection – random and non-random
•
•

Population and Sample
•When your work with population data:
•You have data for the whole population
•Your findings apply to the whole population
•
•When you work with sample data:
•You have data for the sample only
•Your aim is to generalize the findings to the whole population
•
•Nobody cares if 53 per cent of 1,000 survey respondents support Brexit but whether 53 per cent of
UK population has this opinion
•
•

Hypotheses
•Logical conjecture about the nature of relationships between two or more variables expressed in
the form of a testable statement (O’Leary 2004)
•
•“Higher unemployment leads to higher frustration of the society”
•
•Null hypothesis:
•Statement about absence of any relationship between independent and dependent variable
•Every hypothesis has its null hypothesis
•
•In statistics, all operations test the null hypotheses
•
•After testing the null hypotheses either hold or they are dismissed (what gives support to our
hypotheses)

Statistical Significance
•Working with samples is always connected with some sampling error
•
•Statistical significance allows to estimate whether the found effects are not only random and so
they can apply to the whole population
•
•Levels of significance: 95 %, 99 %, 99.9 %
•
•Significance and hypotheses testing:
•If a result is significant, we reject the null hypothesis and we gain confidence in our own
hypothesis
•If a result is not significant, we keep the null hypothesis and we thus we have no support for our
own hypothesis




•A statistically significant effect does not necessarily mean that it is also important and
meaningful
•
•A finding that a new medicine reduces body temperature of the patient by 0.01 °C (significant at
99.9 %)
•
•A finding that a new medicine reduces body temperature of the patient by 1 °C (significant at 99.9
%)
•
•
Statistical Significance