10. Choosing the Right Test

A Practical Guide to Parametric and Non-Parametric Options

Author

A. J. Smit

Published

2026/03/19

In This Chapter

How to connect the biological question to the statistical method
Choosing among common parametric tests
When to consider non-parametric alternatives
Why data type, design, and assumptions matter together

1 Introduction

Many students learn statistics as a menu of named tests. In practice, test selection should follow from a small set of prior questions:

What is the biological question?
What type of variables do I have?
How many groups or predictors are involved?
Are the observations independent?
Are the assumptions of the preferred method plausible?

2 Key Concepts

The practical rule is to choose a method by question, data structure, and design together.

Test selection starts with the biological question, not the software function.
Variable type determines which classes of method are even plausible.
Study design and independence constrain valid inference.
Assumptions help decide between parametric, transformed, or alternative methods.
Named tests are easier to remember when seen as answers to recurring data situations.

3 A Practical Decision Sequence

To compare two groups, start with a t-test or its non-parametric analogue.
To compare three or more groups, consider ANOVA or its non-parametric analogue.
To quantify association, use correlation.
To model a response as a function of predictors, use regression.

4 Parametric and Non-Parametric Options

Parametric methods are usually more powerful when their assumptions are reasonable. Non-parametric methods are useful when assumptions fail badly or when the data are ordinal or rank-based by nature.

The key is not to memorise separate lists, but to understand which method matches:

the question,
the data type,
the design, and
the inferential goal.

5 Links

This chapter replaces the earlier split material:

11-decision_guide.qmd
12-glance.qmd

Reuse

CC BY-NC-SA 4.0

Citation

BibTeX citation:

@online{smit,_a._j.2026,
  author = {Smit, A. J., and J. Smit, A.},
  title = {10. {Choosing} the {Right} {Test}},
  date = {2026-03-19},
  url = {http://tangledbank.netlify.app/BCB744/basic_stats/10-test-selection.html},
  langid = {en}
}

For attribution, please cite this work as:

Smit, A. J., J. Smit A (2026) 10. Choosing the Right Test. http://tangledbank.netlify.app/BCB744/basic_stats/10-test-selection.html.

--- title: "10. Choosing the Right Test" subtitle: "A Practical Guide to Parametric and Non-Parametric Options" author: "A. J. Smit" date: last-modified date-format: "YYYY/MM/DD" --- ```{r code-brewing-opts, echo=FALSE} knitr::opts_chunk$set( comment = "R>", warning = FALSE, message = FALSE, fig.width = 6.5, fig.height = 4.5, out.width = "88%", fig.asp = NULL, fig.align = "center", fig.retina = 2, dpi = 300 ) ggplot2::theme_set( ggplot2::theme_minimal(base_size = 8) ) ggplot2::theme_set( ggplot2::theme_bw(base_size = 8) ) ``` ::: {.callout-note appearance="simple"} ## In This Chapter - How to connect the biological question to the statistical method - Choosing among common parametric tests - When to consider non-parametric alternatives - Why data type, design, and assumptions matter together ::: # Introduction Many students learn statistics as a menu of named tests. In practice, test selection should follow from a small set of prior questions: 1. What is the biological question? 2. What type of variables do I have? 3. How many groups or predictors are involved? 4. Are the observations independent? 5. Are the assumptions of the preferred method plausible? # Key Concepts The practical rule is to choose a method by question, data structure, and design together. - **Test selection** starts with the biological question, not the software function. - **Variable type** determines which classes of method are even plausible. - **Study design and independence** constrain valid inference. - **Assumptions** help decide between parametric, transformed, or alternative methods. - **Named tests** are easier to remember when seen as answers to recurring data situations. # A Practical Decision Sequence - To compare **two groups**, start with a `t`-test or its non-parametric analogue. - To compare **three or more groups**, consider ANOVA or its non-parametric analogue. - To quantify **association**, use correlation. - To model a **response as a function of predictors**, use regression. # Parametric and Non-Parametric Options Parametric methods are usually more powerful when their assumptions are reasonable. Non-parametric methods are useful when assumptions fail badly or when the data are ordinal or rank-based by nature. The key is not to memorise separate lists, but to understand which method matches: - the question, - the data type, - the design, and - the inferential goal. # Links This chapter replaces the earlier split material: - `11-decision_guide.qmd` - `12-glance.qmd`