Build a table and plot that preserve units, individual observations, and the shape of variation before summarizing.
Field diagramPlot evidence with units
Estimated reading
8 min
Estimated practice
15 min
Equipment
Spreadsheet or graph paper, supplied representative dataset, and a data dictionary template.
Progress stays in this page session. Allow lesson storage in Cookie settings to remember it; browser settings may also prevent saving.
Objective
By the end, you can:
Design a tidy raw-data table with identity and units.
Choose a plot that shows individual observations and relevant comparison structure.
Calculate and interpret a descriptive summary without implying a causal or inferential result.
A table is part of the evidence
Start with one row per observational unit or planned measurement, not one row per conclusion. Include sample or replicate ID, group, date or run, raw value, unit, instrument or method version, and a status field for suspect, missing, or repeated values. A separate data dictionary explains each column. Keep calculations in derived columns so the original reading remains visible. This design makes it possible to find a transcription error without reverse-engineering a bar chart.
NIST’s statistical handbook treats plots as tools for understanding data structure. A plot should answer a specific question: Are groups separated? Is there drift over run order? Are values skewed? For a small dataset, show points. A mean with an error bar can hide the distribution, especially when the number of observations is small. Label axes with quantities and units; never rely on color alone to distinguish groups.
Describe before you explain
A descriptive summary can include count, minimum and maximum, mean, median, and standard deviation, when each is appropriate. The mean is sensitive to extreme values; the median is a positional summary. Standard deviation describes spread among observations; it does not state the uncertainty of the mean by itself. Describe what the data show, then state what they cannot establish. A visible difference is a prompt for reasoning, not automatic proof of a treatment effect.
Worked example
A modest result statement
Method A readings are 9.8, 10.0, and 10.1 mL; Method B readings are 10.4, 10.5, and 10.6 mL. The dataset is representative practice data.
01A mean: 9.97 mL; B mean: 10.50 mL; each n = 3.
02A point plot shows B higher in all three observations and similar within-group spread.
03Bounded statement: ‘In this small supplied dataset, B readings were higher than A; the exercise does not establish general accuracy or cause.’
Materials
Set out what you need.
Representative practice dataset labeled as such
Table template
Plotting tool
Data dictionary
Safety & stop conditions
Pause if the work no longer fits the plan.
Use supplied or fictional data; do not upload private or sensitive research data to an unapproved service.
Keep raw data read-only or versioned when possible.
Do not present representative values as Tall Pine experimental evidence.
Method
Work through the steps.
01
Define the row
Write what one row represents and keep independent units distinct from repeated readings of the same unit.
02
Preserve raw fields
Enter the original value, unit, ID, run, and status before calculating summaries.
03
Add derived fields
Calculate conversions or normalized values in separate, labeled columns with formulas visible.
04
Plot the question
Choose points, dot plots, scatterplots, or another suitable display that keeps the relevant structure visible and includes units.
05
Write a bounded result
Report count, center, spread, and pattern, then state the limitation of the design and data.
Checkpoints
Observe, record, investigate.
Every value has an ID and unit.
The plot can be read without a legend-only color cue.
Derived values can be traced back to raw fields.
The sentence describing the plot avoids causal language unless design supports it.
Common mistakes
Correct the process, preserve the record.
A bar chart hides the individual points.
Overlay points or choose a display that shows the observations and distribution.
The y-axis has no unit.
Write the measured quantity and unit in the axis label or table heading.
Repeated readings are counted as independent samples.
Keep the experimental unit separate from technical readings and summarize at the planned level.
A mean is reported with no count or spread.
Include n and a justified spread summary, plus the raw-data path.
Practice
From rows to a plot
Use the worked example below as the supplied representative dataset: Method A readings are 9.8, 10.0, and 10.1 mL; Method B readings are 10.4, 10.5, and 10.6 mL. Build a raw table, make a point plot, and write three descriptive sentences.
