Skip to main content Link Menu Expand (external link) Document Search Copy Copied
Biology Skills

Longitudinal, paired and repeated measures

Summary

Data collected over time or repeated under different conditions on the same subjects require special plotting. Time-course plots should preserve time order and mark missing time points. For paired data (same subject before/after), use connected line segments for each subject or plot the difference. Facetting by subject can reveal consistent patterns. Do not treat repeated measures as independent replicate counts: annotate them as repeated (e.g. with subject ID as color or facet).

Core rules

  • Time plots: Plot time on the x-axis and connect points chronologically. If data are irregularly spaced, mark axis carefully.
  • Paired connectivity: Use lines to connect measurements from the same subject across categories (e.g. pre/post) to show within-subject changes.
  • Faceting: For clarity with few subjects, create small multiples per subject. Large numbers of subjects may require summarizing with mean±error over time.
  • Variance captioning: When showing time-series, indicate if lines are individuals or means.

Required context

  • Subject or unit identifier for repeats.
  • Measurement times or condition labels (ordered categories).
  • Number of repeated measures per subject and intervals.

AI behaviour

  • Check structure: If aes(group=subject) is applicable, ensure lines are drawn or facets used.
  • Identify trends: If asked about time trends, verify that plotting style allows slope interpretation (e.g. use the same y-axis for all subjects).
  • Handle missing repeats: If some subjects miss time points, do not drop them without note; show gap or annotate “no measurement”.

Common failure modes

  • Disconnected points: Treating pre/post points as separate groups (e.g. two unlinked boxplots).
  • Spaghetti plot overload: Plotting hundreds of individual lines in one panel (becomes unreadable).
  • Chronology lost: Shuffling time points (plotting by category) incorrectly shows broken trend.

Authoritative standards

  • Time-series visualization: Standard practice is lines connecting repeated measures per subject (Long & Alcock guidelines).

Examples

Example: Treatment over time

  • Data: Blood sugar in patients at weeks 0, 4, 8.
  • Check: Plot each patient’s trajectory as a line. Add a bold line for population mean.
  • Good outcome: Thin lines (n=10) plus thick mean line, x-axis labeled weeks.

Example: Pre/post intervention

  • Data: Enzyme levels before/after in individuals.
  • Check: Scatter plot with arrows or lines linking each subject’s before and after points, or plot paired differences.
  • Good outcome: Each patient shown with different color; caption “lines connect paired measurements (n=8 patients)”.