FROM MY PT SCHOOL NOTES · EVIDENCE-BASED PHYSICAL THERAPY · CHAPTER 6
How Should You Describe Clinical Data?
Before you interpret significance, describe the numbers. A mean, a median, a distribution and an effect size each tell you something different.
Describe the typical score
Descriptive statistics organize and summarize observed data. The lecture covers frequencies and relative or cumulative percentages, frequency tables, bar charts, histograms and graphical displays of distributions.
Central tendency: mean is the arithmetic average and is influenced by outliers; median is the middle ordered observation; mode is the most frequent value. Shape and skewness affect which summary best represents a sample.
Describe variability and distribution
Measures of spread include range, standard deviation, percentiles, interpercentile ranges and coefficient of variation when meaningful. The lecture also distinguishes standard error of measurement (error in an individual’s test score), standard error of the mean (sample-mean uncertainty) and standard error of estimate (spread about a regression line).
Describe the size of a difference
Effect size describes the size of a difference or relationship, often in raw clinical units or standardized units such as Cohen’s d. The lecture introduces normal distributions and skewed distributions; any statistical interpretation should reflect the actual dataset.
Check your reasoning: Why might the median better represent a highly skewed sample?
The median is less affected by extreme values than the arithmetic mean, so it may better describe the typical observation in a skewed distribution.
Updated-standard note
One original slide says effect size should not be calculated when results are statistically nonsignificant. That is not a general rule. Magnitude and uncertainty remain worth reporting even when a p-value does not cross the chosen threshold. Fixed size labels such as ‘small’ or ‘large’ depend on context.
Put it into practice
Scenario: A small dataset contains walking-test times of 5, 5, 6, 7 and 50 seconds for a hypothetical task.
What should you conclude?
Reasoning: The median is 6 seconds; the mean is 14.6 seconds. The single extreme observation moves the mean sharply. Describe the distribution and investigate whether the value is real or a data issue rather than silently removing it.
Common trap: Using the mean as ‘typical’ without checking skew or outliers. Remember: A numerical summary needs a distribution and units.
Course basis: my Evidence-Based PT school notes, source lecture Chapter 9. New questions and headings are teaching aids, not official exam items. For full definitions and examples, use the class notes and assigned readings.
