A z score of -1.7 is approximately at the 4.46th percentile in a standard normal distribution. In other words, about 4.46% of observations fall below it, while about 95.54% fall above it, assuming the data follow a normal distribution.
What the Percentile Means
A percentile describes the proportion of values at or below a particular point. So the 4.46th percentile does not mean that a score is 4.46% of the average, or that it is “4.46 points” from anything. It means that, in the reference distribution, roughly 4.46 out of every 100 values are lower than the value represented by the z score.
Because -1.7 is negative, it lies below the distribution’s mean. Its distance from the mean is 1.7 standard deviations. For example, if a normally distributed test has a mean of 100 and a standard deviation of 15, a z score of -1.7 corresponds to a raw score of 74.5: 100 + (-1.7 × 15). That score is around the 4.46th percentile under the normal-model assumption.
How the Conversion Works
A z score tells how many standard deviations a value is from the mean. To convert a z score to a percentile, find the cumulative area under the standard normal curve to its left. This cumulative probability is commonly written as Φ(z). For z = -1.7:
- Φ(-1.7) ≈ 0.0446.
- Convert the probability to a percentage: 0.0446 × 100 ≈ 4.46%.
- Thus, the score is at approximately the 4.46th percentile.
The standard normal distribution has a mean of 0 and a standard deviation of 1. Its curve is symmetric, so the area below -1.7 is the same as the area above +1.7. A z table, statistical calculator, or software function for the normal cumulative distribution can provide the value. Results may differ slightly in the last decimal place because of rounding.
What a Z Score of -1.7 Says
The score is substantially below the mean, but it is not an extreme tail value in the way a score several standard deviations below the mean would be. Under the normal model, around 4 to 5% of observations are lower. The exact interpretation depends on what is being measured and how the reference population was defined.
For instance, if adult heights in a particular reference group were modeled as normally distributed, a person with a height z score of -1.7 would be shorter than about 95.5% of that group and taller than about 4.5%. This does not establish anything about health by itself; the population, measurement quality, and relevant clinical context matter. The same percentile calculation could describe a test score, measurement, or other variable, but the real-world meaning changes with the subject.
Also distinguish “below the mean” from “below average” in casual speech. The z score quantifies distance from a reference mean in standard-deviation units. It does not explain why a value is low, whether it is unusual for a particular subgroup, or whether the measurement is important.
Percentile, Percentage, and Rank
Percentile and percentage are easy to confuse. A percentile is a location in a distribution; a percentage is a proportion or rate. Being at the 4.46th percentile does not mean earning 4.46% on a test. A test score could be 74.5 points and still fall near that percentile if its mean and standard deviation were 100 and 15 and its scores followed the stated normal model.
Percentile rank can also be defined differently in practical datasets, especially when many people have the same score. The normal-curve answer is a model-based cumulative percentage, not necessarily the exact rank of one person in a finite class, exam cohort, or database. For an observed dataset, the percentile may instead be calculated directly from the number of values below or tied with the value.
Assumptions and Common Misreadings
The figure of 4.46% is accurate for a standard normal distribution, and it is the usual conversion when a z score is interpreted against a normal model. But standardizing data does not automatically make the data normally distributed. A z score can be calculated for any dataset with a meaningful mean and standard deviation; if the distribution is skewed, has heavy tails, or contains unusual clusters, the normal-table percentile may not match the actual proportion below that value.
- Do not reverse the tail: -1.7 is about the 4.46th percentile from the bottom, not the 95.54th. The 95.54% figure is the proportion above it.
- Do not treat it as a probability about an individual: the percentile describes a position relative to a distribution, not the likelihood that a person will experience an outcome.
- Check the reference group: a percentile is meaningful only in relation to the population or comparison distribution used.
- Check the direction of the measure: a low percentile can be desirable for some quantities, such as time to complete a task, and undesirable for others.
If the question concerns a real dataset, use its empirical distribution or a suitable statistical model rather than assuming normality without checking. A histogram or normal probability plot can help reveal whether a normal approximation is reasonable, though formal analysis may be needed for consequential decisions.
Related Z Scores for Comparison
Nearby values show how quickly percentiles change around this part of the curve. Approximate standard-normal percentiles include:
- z = -1.0: 15.87th percentile.
- z = -1.5: 6.68th percentile.
- z = -1.7: 4.46th percentile.
- z = -2.0: 2.28th percentile.
- z = +1.7: 95.54th percentile.
These comparisons are useful for checking the sign and scale of a table lookup. The positive and negative values are mirror images: their cumulative percentiles add to 100%, apart from rounding.
Frequently Asked Questions
Is a z score of -1.7 the 5th percentile?
It is close, but more precisely it is about the 4.46th percentile under the standard normal distribution. Calling it approximately the 4th or 5th percentile is reasonable when only a rough estimate is needed.
What percentage is above a z score of -1.7?
About 95.54% of values are above it in a standard normal distribution. This is the complement of the 4.46% below it.
How do I calculate the raw score from z = -1.7?
Use raw score = mean + (z × standard deviation). For a mean of 100 and standard deviation of 15, the raw score is 100 + (-1.7 × 15) = 74.5.
Does a z score of -1.7 always equal the 4.46th percentile?
No. That percentile follows from the standard normal curve. For a non-normal real-world distribution, the actual percentage below the score may be different.
Is a z score of -1.7 considered unusual?
It is below the mean and lies in the lower tail, with about 4.46% of normal-distribution values below it. Whether that is unusual or concerning depends on the measurement, comparison group, and purpose of the analysis.
