RESEARCH METHODS FOR
POSTGRADUATE STUDY
Analysing Groups and Relationships
Aims for today
Understand what statistical tests are and why use them
Choose the right test for the data
Identify the preparatory steps
Examine relationships
Calculate Pearson’s r on SPSS
lbic.navitas.com
navitas.com
Research questions
Researchers often want to know if there is a significant relationship between two variables
Is there a relationship between waiting time and customer satisfaction? Is there a relationship between TV ads and number of sales?
Is there a relationship between employees’ levels of stress and customer defection?
Does gender predict early technology adoption?
lbic.navitas.com
navitas.com
Research questions
Or if there is a difference between one or more scores/conditions/groups
lbic.navitas.com
Is product A more attractive than product B?
Does working in pairs improve performance?
Does working from home decrease productivity?
Has the new manufacturing process increased productivity?
Does gender affect the frequency and size of online purchases? navitas.com
Statistical tests: Why we need them?
Does marital status affect life satisfaction?
‘In a scale from 1 to 8, how satisfied are you with your life?’
Great! I found a difference!!
How confident are you? Is it an accident (due to chance)?
We need to have a statistical test! lbic.navitas.com navitas.com
Choosing the right test
If you browse any introductory statistics text book you’ll find a bewildering array of different statistical tests
Each has
a specific purpose (i.e. exploring relationships, comparing groups) Specific data requirements (categorical, ordinal or continuous data, normal distribution)
SPSS makes it easy to run most of the well known tests
However, it is important to be able to
Select the most appropriate test given your research question
Understand conceptually what the test is computing
Effectively interpret the output (SPSS often gives more than one result per test) lbic.navitas.com navitas.com
Preparatory steps
1. Choose the right test depending on your research question
2. Select your variables
3. Describe your data
lbic.navitas.com
navitas.com
Step 1: What is your question?
Remember, when conducting research it is important to be clear about the questions you are trying to answer
Ideally before you begin data collection
The questions...
Is there a relationship between waiting time and customer satisfaction? Is there a relationship between TV ads and number of sales?
Is there a relationship between employees’ levels of stress and customer defection?
Does gender predict early technology adoption?
...require quite different statistical tests to questions like:
lbic.navitas.com
Is product A more attractive than product B?
Does working in pairs improve performance?
Does working from home decrease productivity?
Has the new manufacturing process increased productivity?
Does gender affect the frequency and size of online purchases? navitas.com Tests of relationship vs. tests of difference
Questions of the first kind are about relationships. Common techniques include:
Correlation – are variable X and variable Y related?
E.g. Head size and IQ, salary and job satisfaction
Questions of the second kind are about differences between groups/categories:
Chi-square shows whether there is an association between two categorical variables (e.g. Is gender associated with smoking?)
T-tests measure the difference between two groups (e.g. males, females) according to some continuous variable (e.g. Height, sales, job satisfaction)
Analysis of variance (ANOVA) measure differences when there are more than two groups and/or more than two independent variables
(e.g. marital status as in slide 4)
Today we will look at Pearson’s r correlation
lbic.navitas.com
navitas.com
Step 2: Select your data
Which variables will you be