l stoodialinza
← All topics

Step 4 · Complex designs · Topic 17 of 17 · 7 min

Data preparation and a reproducible report

Transformations, normalization, exclusions and what belongs in the Methods section.

In short#

  • A transformation changes the scale and the hypothesis: a t-test on logarithms compares geometric means, that is, their ratio.
  • Normalizing to a reference without accounting for its uncertainty can understate the final uncertainty.
  • Excluding observations is acceptable under a predefined, documented rule.
  • Report: design, observational unit, n per group, exclusions, method, corrections, software, estimates with intervals.

In plain words#

To let another researcher inspect and reproduce an analysis, keep the source data, record each step, and document exclusions instead of erasing inconvenient values.

A logarithmic scale is useful when multiplicative differences matter, for example when concentrations vary several-fold. Equal ratios have equal distances on the scale, so x to 2x spans the same distance as 2x to 4x.

Before publication, check that the figure caption states what the points and error bars represent, how many observations are included, and which tests and comparisons each significance marker denotes.

In Linza#

  1. Transform data creates a new table and keeps the source table together with its operation history.
  2. Exclusion (×) keeps the value visible and records the decision in the project.
  3. Research JSON preserves the project, source data, settings and preparation history. Methods TXT contains the selected analysis, its assumptions and calculation details.
Try it in Linza Opens in a new tab with synthetic data. Work in the original tab remains unchanged.

Common mistakes#

  • Removing inconvenient observations without documenting why they were excluded.
  • Leaving the caption unclear about what error bars represent and how many observations are included.

Check yourself#

You compared groups with a t-test after taking logarithms. What is the conclusion really about?

Optional. Your answer is not saved.

Sources#

  • Bland J.M., Altman D.G. (1996) Transforming data. BMJ, 312. Search for publication Opens in a new tab
  • Lang T.A., Altman D.G. (2015) Basic statistical reporting for articles published in biomedical journals: the SAMPL guidelines. International Journal of Nursing Studies, 52(1). Search for publication Opens in a new tab
  • Wilkinson M.D., Dumontier M., Aalbersberg I.J. et al. (2016) The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data, 3. Open publication Opens in a new tab

References are for further reading. A method’s suitability depends on the study design and its assumptions.