Descriptive Statistics
Outline of lecture
Purpose of descriptive statistics
Describing categorical variables
Frequency analysis
Describing continuous variables
Summary statistics
Measures of central tendency, variability and normality
Normal distribution
Relevant SPSS commands:
Descriptives, Compare means, Histograms, Explore
Next lab session
lbic.navitas.com
navitas.com
Objectives for today
Understand the importance of exploring the characteristics of your data before conducting the final analyses
Appreciate why categorical and continuous data require different kinds of descriptive analysis Identify and define key measures of:
Central tendency, Variability, Normality
Explore these data characteristics using various SPSS commands (next lab session) lbic.navitas.com navitas.com
Population and Sample
Population
A group of people: e.g. all students at Brunel, all employees in
Company A, all customers
A set of objects: e.g. all cars produced by Company A in 2012
Census
When data is gathered from the whole population
Sample
Subset of the population
Constraints
Time
Money
Availability of researchers lbic.navitas.com 4
navitas.com
Why do descriptive statistics?
What do we need to do before we run statistical tests that will provide conclusive answers to our research questions?
We need to get to know our data better
A summary or overview
lbic.navitas.com
What do descriptive statistics allow us to do?
Describe the general characteristics of the sample from which the data was collected (sample/group sizes, mean age etc.)
Check variables for violations in the assumptions underlying the statistical tests you intend navitas.com to use
Categorical variables
Examples of categorical variables?
Gender, nationality, marital status, grade, education level
Level of measurement?
Nominal and ordinal
It is usually valuable to know how many cases belong to each group as defined by your categorical variables
How many cases are there in each category group?
Most statistical tests require
roughly equal numbers of cases in each group at least 10 cases (but more = better) in all groups
...in order to produce reliable results lbic.navitas.com navitas.com
Categorical variables
Example research questions:
“Is the Product easy to use, attractive, etc?”, “Is website A better than website B?”, “Does working from home decrease productivity”, “Does stress increase customer defection?”
What was the relative proportion of males and females in your sample?
Demographic information is normally reported in the method section of your report
For e.g. If this is not 50/50 then is the sample representative of the population of interest?
You may need to collect more data
If the sample does not reflect the demographic structure of the target population
Or is otherwise unbalanced or group sizes are too small for analysis lbic.navitas.com
navitas.com
Categorical variables
In SPSS to obtain descriptive statistics for categorical variables, select:
Analyse > Descriptive
Statistics > Frequencies
Examining the frequency
table for the Gender category variable.
Comparing the groups may be
problematic in this case. Why?
lbic.navitas.com
navitas.com
Categorical variables
Frequency tables may also be sufficient to fully answer simple questions
e.g. Running a count on responses to a question:
“Which is your favourite social media platform?”
Could also provide a rank order of popularity
But no test of statistical ‘significance’
(simplistic) example: in a sample of 200, 99 males and
101 females said they use Twitter.
So, can we conclude that females use Twitter more than males? We’ll look at significance testing next session lbic.navitas.com navitas.com
Continuous variables
Continuous variables are scale level measures
e.g. Age, weight, height, time, test scores,