MARITINA LAMPROU • STATISTICS
WHICH STATISTICAL TEST
SHOULD I USE?
A practical decision guide for university students, theses and research projects
RESEARCH QUESTION VARIABLES DESIGN
What are you trying to compare, Are your variables continuous, ordinal How many groups? Independent or
associate or predict? or categorical? paired?
Includes: decision tree • t-tests • ANOVA • chi-square • correlation • regression •
non-parametric alternatives • assumptions • reporting examples • quick-reference
tables
Lamprou Maritina
Statistics • SPSS • Data Analysis • Thesis Support
Which Statistical Test Should I Use? | 1
, MARITINA LAMPROU • STATISTICS
1. How to Use This Guide
The core idea
Do not choose a statistical test because it is familiar or because you have seen it in another thesis. Start with
your research question, identify the variables involved, and then match the design to an appropriate family of
tests.
A statistical test is a tool for answering a specific question under a set of assumptions. The same dataset can
support different analyses depending on the research question. This guide is therefore organised around the
decisions you make before opening SPSS or R.
The five questions to answer first
• What is the goal? Compare groups, compare repeated measurements, examine association, or predict an
outcome?
• What is the outcome variable? Continuous, ordinal, binary, nominal, count, or another type?
• How many groups or conditions? One, two, or three or more?
• Are observations independent or paired? Different people in each group, or the same/matched people
measured more than once?
• Are key assumptions reasonable? Distributional shape, outliers, independence, linearity, equal variances and
expected cell counts may matter.
Important: This guide is an educational decision aid, not a substitute for checking the exact assumptions of
your method, your study design, and your supervisor's requirements.
A useful workflow
Step Question Output
1 Write the research question in one sentence. Clear analytic goal
2 Label each variable by role and measurement level. Outcome / predictor / grouping variables
3 Identify independent vs paired observations. Study design
4 Choose a test family. Comparison / association / prediction
5 Check assumptions and sample limitations. Final method choice
6 Run analysis, interpret effect + uncertainty, then report. Result
Which Statistical Test Should I Use? | 2
WHICH STATISTICAL TEST
SHOULD I USE?
A practical decision guide for university students, theses and research projects
RESEARCH QUESTION VARIABLES DESIGN
What are you trying to compare, Are your variables continuous, ordinal How many groups? Independent or
associate or predict? or categorical? paired?
Includes: decision tree • t-tests • ANOVA • chi-square • correlation • regression •
non-parametric alternatives • assumptions • reporting examples • quick-reference
tables
Lamprou Maritina
Statistics • SPSS • Data Analysis • Thesis Support
Which Statistical Test Should I Use? | 1
, MARITINA LAMPROU • STATISTICS
1. How to Use This Guide
The core idea
Do not choose a statistical test because it is familiar or because you have seen it in another thesis. Start with
your research question, identify the variables involved, and then match the design to an appropriate family of
tests.
A statistical test is a tool for answering a specific question under a set of assumptions. The same dataset can
support different analyses depending on the research question. This guide is therefore organised around the
decisions you make before opening SPSS or R.
The five questions to answer first
• What is the goal? Compare groups, compare repeated measurements, examine association, or predict an
outcome?
• What is the outcome variable? Continuous, ordinal, binary, nominal, count, or another type?
• How many groups or conditions? One, two, or three or more?
• Are observations independent or paired? Different people in each group, or the same/matched people
measured more than once?
• Are key assumptions reasonable? Distributional shape, outliers, independence, linearity, equal variances and
expected cell counts may matter.
Important: This guide is an educational decision aid, not a substitute for checking the exact assumptions of
your method, your study design, and your supervisor's requirements.
A useful workflow
Step Question Output
1 Write the research question in one sentence. Clear analytic goal
2 Label each variable by role and measurement level. Outcome / predictor / grouping variables
3 Identify independent vs paired observations. Study design
4 Choose a test family. Comparison / association / prediction
5 Check assumptions and sample limitations. Final method choice
6 Run analysis, interpret effect + uncertainty, then report. Result
Which Statistical Test Should I Use? | 2
