📈 Describing data: graphs, center, spread

Shape, center, spread. Three things and you have described any data set.

⭐ The one idea

Raw numbers are noise. A picture shows the shape, one number gives the center, one number gives the spread. Report all three and you have described the data.

📊 Organizing data: tables and graphs

🧪 Our running example

Fasting glucose (mg/dL) for 8 clinic patients: 92, 98, 104, 104, 106, 110, 115, 126

Class (mg/dL)FrequencyRelative frequencyCumulative frequency
90–9922/8 = 0.252
100–10930.3755
110–11920.257
120–12910.1258
💡 Class width = (max − min) ÷ number of classes, rounded up. Classes never overlap and every value lands in exactly one. Relative frequencies add to 1.
Same 8 readings, binned01.5390s100s110s120sPatients
A histogram: bars touch, x-axis is numeric classes, height = frequency. Use it for quantitative data.
Stem-and-leaf: glucosestemleaf92 8104 4 6110 512611 | 0 5 means 110 and 115. Turnit sideways and it is ahistogram.
A stem-and-leaf plot keeps every actual value and still shows the shape.

📊 Bar graph

For categorical data. Bars have gaps; order does not matter. Patients per unit, admissions by diagnosis.

🥦 Pie chart

Parts of one whole; slices add to 100%. Fine for 3–6 categories, hopeless for 12.

📈 Histogram

For quantitative data. Bars touch; x-axis is a number line in classes.

📉 Time-series / line graph

One value over time. Vital-sign flowsheet, weekly infection counts.

Pie: 100 admissions by service48%27%15%10%Medical — 48Surgical — 27Obstetric — 15Pediatric — 10
Pie chart: each slice = category ÷ total.
Incident reports this quarterFalls14Med errors9Pressure injuries6CLABSI2
A Pareto chart is just a bar graph sorted tallest to shortest.

🏠 Shape

The four shapesUniformflat: notypical valueBell-shapedmean = medianSkewed righttail right,mean > medianSkewed lefttail left, mean< median
Symmetric, skewed right (long tail to the right, mean dragged up), skewed left, uniform. Hospital length of stay and costs are almost always skewed right.
🧠 The tail names the skew. Skewed right = tail to the right = a few very large values = mean > median. Income, length of stay, ER wait times. Skewed left: mean < median. Age at death from natural causes.

🎯 Center: mean, median, mode

⚖️ Mean x̄

x̄ = Σx / n = 855 / 8106.875 mg/dL

The balance point. Uses every value, so one outlier drags it.

📌 Median

Sort, take the middle. With 8 values, average the 4th and 5th: (104 + 106) / 2

105 mg/dL

Resistant to outliers. Use it for skewed data (income, length of stay).

🔁 Mode

The most frequent value.

104 mg/dL

The only “center” that works for categorical data (most common blood type: O+).

⚠️ Outlier test with one glance

Change the 126 to 226. Mean jumps to 119.375; median stays 105. When mean and median disagree a lot, the data are skewed or contain an outlier — report the median.

↔️ Spread: range, variance, standard deviation

📏 Range

max − min = 126 − 9234 mg/dL

Quick, but it only looks at two values.

📊 Variance s²

s² = Σ(x − x̄)² / (n − 1)108.4

Average squared distance from the mean. Units are mg/dL² — nobody thinks in those, so take the root.

⭐ Standard deviation s

s = √s²10.4 mg/dL

The typical distance of a value from the mean. Same units as the data. Bigger SD = more spread out.

🔧 The SD by hand, every step
  1. Mean: x̄ = 106.875
  2. Deviations squared: (92−106.875)² + (98−106.875)² + (104−106.875)² + (104−106.875)² + (106−106.875)² + (110−106.875)² + (115−106.875)² + (126−106.875)²
  3. Sum of squares = 758.875
  4. Divide by n − 1 = 7: s² = 758.875/7 = 108.41
  5. Square root: s = 10.41

Divide by n − 1 for a sample (that is what Desmos stdev does). Divide by N only for a whole population (stdevp). Exams almost always want the sample version.

💡 Empirical rule preview: for bell-shaped data about 68% of values sit within 1 SD of the mean. Here that is 106.875 ± 10 → roughly 96 to 117 mg/dL. Full story on the Normal page.

📦 Position: percentiles, quartiles, box plots

Percentile = % below you92981041041061101151265 of the 8 are below 110110 mg/dL sits at (5 ÷ 8) × 100= 62.5, so call it the 63rdpercentile.
Percentile of a value = (number of values below it ÷ n) × 100.

🧪 Second example: length of stay (days) for 10 patients

2, 3, 3, 4, 4, 5, 5, 6, 7, 12

Q1 (25th percentile) = 3   Median = 4.5   Q3 (75th) = 6.25   IQR = Q3 − Q1 = 3.25

Outlier fences: Q1 − 1.5·IQR = -1.875 and Q3 + 1.5·IQR = 11.125. The 12-day stay is above the upper fence → outlier.

Box plot of length of stay (days)min2Q13med4.5Q36.25max7out12
The box is the middle 50% (Q1 to Q3). The line inside is the median. Whiskers reach the last non-outlier; dots beyond the fences are outliers.

💡 Reading a box plot

  • Median line off-center → skewed.
  • Long right whisker → skewed right.
  • Wider box → bigger IQR → more spread.
  • Two box plots side by side = the fastest way to compare groups.

🧠 Five-number summary

min, Q1, median, Q3, max. Desmos gives all five with quartile(L,0)quartile(L,4), or just type boxplot(L) and hover.

⚠️ Different books (and Desmos) split quartiles slightly differently, so your Q1 may differ from a classmate’s by a little. Use the method your instructor showed; on the final, Desmos’s answer is the one that matches the key.

🧮 In Desmos

1L = [92, 98, 104, 104, 106, 110, 115, 126]
2mean(L)= 106.875
3median(L)= 105
4stdev(L)= 10.412
5var(L)= 108.411
6max(L) − min(L)= 34
7quartile(L, 1)= Q1
8histogram(L, 10)bin width 10
9boxplot(L)drawn on the graph

Full recipes on the Desmos page.

✅ Quick self-check

❓ Five patients’ pain scores: 2, 3, 3, 8, 9. Mean? Median? Which describes the typical patient better?
Mean = 25/5 = 5. Median = 3. The two 8-and-9 scores pull the mean up; the median (3) is the better “typical” value here.
❓ Unit A heart rates have SD 6 bpm; Unit B has SD 14 bpm. Which unit’s patients are more alike?
Unit A. Smaller SD = values cluster closer to the mean.
❓ A histogram of ER wait times has a long tail to the right. Which is larger, the mean or the median?
Mean. Right skew drags the mean toward the tail.
MAT 300 · built from your own course files (the statistics study guide, the Desmos guide, the formula sheet and the final-exam study questions) with nursing examples. Not a substitute for the textbook — check any number against your own notes before an exam.