Tutorials › Introduction to Statistics › Comparing Relative Positions Across Different Scales

Home  /  Tutorials  /  Center, Spread and Position

Center, Spread and Position · Tutorial 168 of 996

Comparing Relative Positions Across Different Scales

Use z scores to compare how far observations lie from their own group means, even when the measurements use different scales.

Beginner 9 min read

What You'll Learn

  • Calculate z scores using the appropriate group mean and standard deviation.
  • Compare relative positions when measurements use different units or scales.
  • Distinguish being above a mean from being far from a mean.
  • Interpret negative z scores and compare distances from group means.
  • Explain what a z-score comparison does not establish about percentiles or real-world importance.

Why Compare With Z Scores?

A raw score is difficult to compare across different scales. A score of 82 points on one test and 740 points on another cannot be compared just by subtracting or ordering the numbers: the tests may use different units, ranges, and amounts of variability. A more useful question is how each score compares with the mean of the group to which it belongs.

A z score expresses an observation’s signed distance from a mean in standard-deviation units. When two observations come from differently scaled measurements, calculating a z score for each using its own group’s mean and standard deviation puts both positions on a common scale. The comparison is about relative distance from each group mean, not about which raw score is numerically larger.

Definition: For an observation \(x\), its z score relative to a mean \(\mu\) and standard deviation \(\sigma\) is \(z=(x-\mu)/\sigma\). If the reference values are sample summaries, use \(z=(x-\bar{x})/s\). The z score is unitless: it gives the observation’s signed distance from the reference mean in standard-deviation units. A positive z score is above the mean; a negative z score is below it.

How to Make a Relative-Position Comparison

For each observation, identify the relevant group mean and standard deviation, subtract the mean from the observation, and divide by that group’s standard deviation. Use the summaries for the observation’s own group; using one group’s standard deviation for both observations would not standardize their positions correctly.

Once both z scores are calculated, their signs and magnitudes answer different questions. If both are positive, the larger z score is farther above its group mean. If both are negative, neither observation is above its mean; the z score closer to zero is less far below its mean, while the one with the larger absolute value is farther from its mean. If one is positive and the other negative, only the positive-scoring observation is above its mean. Absolute values compare distance from the mean regardless of direction; signed values retain the direction.

1
Match each observation to its reference group.
Record the mean and standard deviation appropriate for each measurement.
2
Calculate each z score.
Subtract that group’s mean from its observation, then divide by that group’s standard deviation.
3
Compare direction and distance.
Use the signs to tell whether observations are above or below their means; compare signed values or absolute values according to the question.
4
State the conclusion in context.
Describe the relative position in standard-deviation units, not as a difference in raw measurement units.

Worked Example: Comparing Two Positive Z Scores

Suppose two hypothetical students take different exams. On Exam A, a student scores 84 points; the class mean is 72 points and the standard deviation is 6 points. On Exam B, another student scores 91 points; the class mean is 80 points and the standard deviation is 5 points. Which student is farther above the mean of their own class?

For the Exam A score, the difference from its mean is \(84-72=12\) points. Therefore,

$$ z_A=\frac{84-72}{6}=\frac{12}{6}=2 $$

The score is 2 standard deviations above the Exam A class mean. For Exam B,

$$ z_B=\frac{91-80}{5}=\frac{11}{5}=2.2 $$

The score is 2.2 standard deviations above the Exam B class mean. Both z scores are positive, and \(2.2>2\). Thus, the Exam B student is farther above their own class mean in standardized terms. This does not mean that 91 is simply “better by more points” than 84: the exams use different scales, and the comparison concerns distance from each exam’s class mean relative to its class’s standard deviation.

Why Standardizing Allows a Comparison

The calculation divides a difference from the mean by the standard deviation of the same group. The numerator and denominator use the same measurement units, so the units cancel. The result is not a number of points, kilograms, or minutes; it is a number of standard deviations.

This cancellation explains why z scores are useful when measurements are reported in different units. For instance, converting a measurement from centimeters to meters changes both the observation’s difference from the mean and its standard deviation by the same positive conversion factor. Their ratio stays the same.

Justification: Suppose a measurement is converted by \(y=a+bx\), where \(b>0\), and the observation, mean, and standard deviation are all converted consistently. The transformed difference from the mean is \(y-\mu_y=b(x-\mu_x)\), while the transformed standard deviation is \(\sigma_y=b\sigma_x\). Therefore,
$$ z_y=\frac{y-\mu_y}{\sigma_y} =\frac{b(x-\mu_x)}{b\sigma_x} =\frac{x-\mu_x}{\sigma_x} =z_x. $$

The same cancellation applies when sample means and sample standard deviations are used consistently. A change in units alone therefore does not change the standardized position. Standardization also accounts for differences in spread: a raw difference of 10 units is relatively larger in a group with a standard deviation of 2 than in a group with a standard deviation of 20.

Worked Example: Comparing Negative Z Scores

Consider two hypothetical measurements. A reading of 45 is from a group with mean 50 and standard deviation 5. A reading of 68 is from a different group with mean 80 and standard deviation 4. Which reading is farther below its own group mean?

For the first reading,

$$ z_1=\frac{45-50}{5}=\frac{-5}{5}=-1 $$

It is 1 standard deviation below its group mean. For the second reading,

$$ z_2=\frac{68-80}{4}=\frac{-12}{4}=-3 $$

It is 3 standard deviations below its group mean. Both scores are negative, so both readings are below their respective means. Since \(|-3|=3\) is greater than \(|-1|=1\), the second reading is farther from its mean and farther below it. In signed terms, \(-1>-3\), so the first reading is less far below its mean than the second; it is not farther above its mean. Neither reading is above its own mean.

Relative Distance Is Not the Same as Raw Difference

A z-score comparison can reverse the impression given by raw differences. An observation may be many raw units from its mean in a highly variable group, yet only a few standard deviations away. Another observation may differ from its mean by fewer raw units but be more unusual relative to a group with little variability.

Worked Example: The Larger Raw Difference Is Not the Larger Relative Distance

Suppose a hypothetical athlete’s jump distance is 2.4 meters in a group with mean 2.0 meters and standard deviation 0.5 meters. A second athlete’s sprint time is 12 seconds in a group with mean 14 seconds and standard deviation 1 second. Compare how far each observation is from its own group mean.

For the jump distance, the raw difference is \(2.4-2.0=0.4\) meters, giving

$$ z_{\text{jump}}=\frac{2.4-2.0}{0.5}=\frac{0.4}{0.5}=0.8. $$

The jump distance is 0.8 standard deviations above its group mean. For the sprint time, the raw difference is \(12-14=-2\) seconds, giving

$$ z_{\text{sprint}}=\frac{12-14}{1}=\frac{-2}{1}=-2. $$

The sprint time is 2 standard deviations below its group mean. The raw differences, 0.4 meters and \(-2\) seconds, cannot be compared directly because they measure different things. The absolute z scores show that the sprint-time observation is farther from its group mean in standardized terms: \(2>0.8\). It is below its mean, while the jump distance is above its mean.

What the Comparison Does and Does Not Tell You

A z score describes position relative to a chosen mean and standard deviation. It does not, by itself, give the observation’s percentile, the probability of observing a value at least that extreme, or the shape of the group’s distribution. Two groups can have the same z score for particular observations while having different distribution shapes. In that case, those observations are the same number of standard deviations from their respective means, but the z scores alone do not establish equal percentile ranks.

A z score also does not tell whether an observation is desirable or important. Being far above the mean could be favorable for one measurement and unfavorable for another. The interpretation depends on what is being measured and the purpose of the comparison. It is safest to report both the numerical z score and its direction in context.

For the calculations here, a normal distribution is not required: subtracting the mean and dividing by the standard deviation gives a standardized position whether or not the data are normally distributed. However, the comparison is meaningful only when the group means and standard deviations are suitable reference summaries and the measured quantities are understood. If a standard deviation is zero, the formula cannot be used because division by zero is undefined.

Common Mistakes

  • Comparing raw values from different scales. A larger numerical value need not be farther above its own group mean. Calculate each z score using that group’s mean and standard deviation.
  • Calling a negative z score “above the mean.” A negative z score means the observation is below the mean. If two scores are negative, compare their absolute values to identify which is farther from its mean.
  • Ignoring direction when the question asks “above.” A larger absolute z score means farther from the mean, not necessarily farther above it. Check the sign before describing direction.
  • Treating z scores as percentiles. A z score states a distance in standard-deviation units; it does not alone specify a percentile or probability.
  • Mixing groups’ summaries. Use each observation’s own reference mean and standard deviation. Otherwise the scores do not represent the intended within-group positions.
Key takeaway: To compare observations across differently scaled measurements, calculate each one’s z score from its own group’s mean and standard deviation. Read the sign for direction and the magnitude for distance; when both scores are negative, neither observation is above its mean.

Check Your Understanding

For each comparison, distinguish direction from distance and state what the z scores do and do not establish.

  1. A reading has \(z=1.4\) and another has \(z=0.9\). Which is farther above its group mean?
  2. Two observations have z scores \(-0.6\) and \(-2.1\). Which is farther below its mean, and which is closer to its mean?
  3. One observation has \(z=1.2\) and another has \(z=-1.8\). Which is above its group mean? Which is farther from its mean?
  4. Why can’t a raw difference of 8 points on one test be directly compared with a raw difference of 3 seconds on another measurement?
  5. Does a z score of 2 by itself tell you the observation’s percentile rank? Explain briefly.

Previous

Next

Back to Tutorials