Transformations: What Shifting and Scaling Do
- Predict the effect of adding a constant on center, spread and shape
- Predict the effect of multiplying by a constant on center, spread and shape
- Explain why standardizing produces a distribution with mean 0 and standard deviation 1
Adding a constant moves, it does not stretch
Add the same number c to every value and the whole distribution slides along the axis. Every measure of center — mean, median, quartiles, percentiles — increases by c. Every measure of spread — standard deviation, IQR, range — is unchanged, because the gaps between observations are exactly what they were. The shape is unchanged too. This is intuitive once you picture the histogram physically sliding: the picture is identical, the axis labels have shifted.
Multiplying stretches everything
Multiply every value by a constant a and both center and spread are multiplied. The mean and median become a times what they were; the standard deviation and IQR become |a| times what they were — the absolute value, because a spread cannot be negative. The variance, being the square of the standard deviation, is multiplied by a². Shape is again unchanged: stretching a histogram horizontally does not create or remove skew.
Standardizing is a transformation, not a new idea
The z-score z = (x − μ)/σ is exactly a linear transformation with b = −μ/σ and a = 1/σ. Subtracting μ shifts the center to 0; dividing by σ scales the spread to 1. So a standardized distribution always has mean 0 and standard deviation 1 — by construction, not by coincidence. Crucially, standardizing does not make a distribution Normal. A skewed distribution standardized is a skewed distribution with mean 0, and treating a z-score as a Normal percentile without checking shape is a common and costly error.
A set of temperatures has mean 20 °C and standard deviation 4 °C. Converting to Fahrenheit uses F = 1.8C + 32. Find the mean, standard deviation and variance in Fahrenheit.
- 1.This is a linear transformation with a = 1.8 and b = 32.
- 2.Mean: 1.8(20) + 32 = 36 + 32 = 68 °F.
- 3.Standard deviation: |1.8|(4) = 7.2 °F. The +32 does not appear, because shifting every value leaves the gaps unchanged.
- 4.Variance: (1.8)²(16) = 3.24(16) = 51.84, which is also 7.2² as it must be.
Every score in a class is increased by 5 points. Which statistic is unchanged?
Standardizing does not normalize. A z-score tells you how many standard deviations from the mean an observation lies, which is meaningful for any distribution — but converting it to a percentile with a Normal table is only valid if the distribution was approximately Normal to begin with.
If every value in a data set is multiplied by 3, the variance is multiplied by —
Answer the 2 checkpoints as you read.
Sign in to save your progress