Mathematics
Z-Scores (Standard Scores)
Quick fact
In a normal distribution, approximately 68% of all data points have a z-score between -1 and 1, 95% between -2 and 2, and 99.7% between -3 and 3.
Why this is interesting
Imagine you scored 85 on a math test and 70 on a physics test. Which subject did you perform better in? Without knowing the difficulty of each test, how can you fairly compare?
Read the full explanation
Understanding Z-Scores (Standard Scores)
A z-score repositions every data point according to its distance from the average, measured in units of spread (standard deviation). Think of the average score as the 'home base.' A z-score tells you how many steps (standard deviations) you are away from home base. If your math test had a high average and small spread, your 85 might only be half a step above average (z = 0.5). But if physics had a low average and large spread, your 70 could be two steps above average (z = 2.0). That means you performed relatively much better in physics. The formula is simple: subtract the mean from your score, then divide by the standard deviation. Z-scores are also called standard scores because they standardize different distributions to a common scale with a mean of 0 and standard deviation of 1.
A deeper explanation
The z-score works because it uses the standard deviation as a ruler. By converting raw scores into a dimensionless measure (standard deviation units), it eliminates the influence of units and scales. This is especially powerful when comparing data from different tests, measurements, or populations. Z-scores are fundamental in identifying outliers (typically |z| 3) and in the empirical rule for normal distributions. They also form the basis for z-tables, which give probabilities of observing values within certain ranges. In practice, z-scores allow researchers to combine data from different sources, detect unusual observations, and standardize inputs for machine learning algorithms. Understanding z-scores reveals the core idea of normalization: transforming data to a common reference frame without losing relative relationships.