Reading a z-table doesn't have to be confusing. If you've got raw data points and want to find their exact probabilities under a bell curve, the standard normal distribution table is your direct map. It translates raw scores into standardized values so you can instantly calculate percentiles, assess risk, or run hypothesis tests.
We'll skip the textbook filler and walk straight through how these tables work, how to compute the numbers yourself, and how to avoid common lookup errors that trip up analysts.
Anatomy of a Standard Normal Distribution Table
A standard normal distribution table—commonly known as a z-table—refers to a normal distribution with a mean ($\mu$) of $0$ and a standard deviation ($\sigma$) of $1$. When you look up a value in this table, you're finding the cumulative probability that a score falls at or below a specific $z$-score.
Most standard tables are split into two layout styles:
- Cumulative from mean: Shows the area between $z = 0$ and your target $z$-score.
- Cumulative from negative infinity: Shows the total area to the left of your $z$-score (ranging from $0.0000$ to nearly $1.0000$).
The rows typically represent the integer and first decimal place of your $z$-score (e.g., 1.9), while the columns represent the second decimal place (e.g., 0.06 for a total $z$-score of 1.96).
Calculating and Applying Z-Scores
Before you can use the table, you have to standardize your raw data. This step shifts your custom dataset onto the standard normal scale using the National Institute of Standards and Technology (NIST) formula guidelines for statistical control:
$$z = \frac{x - \mu}{\sigma}$$
Where:
- $x$ is your specific raw data point.
- $\mu$ is the population mean.
- $\sigma$ is the population standard deviation.
Imagine you manage customer service response times. Your team's average resolution time is $45$ minutes with a standard deviation of $8$ minutes. A customer's ticket took $61$ minutes to resolve.
Plug those figures into the formula: $$z = \frac{61 - 45}{8} = \frac{16}{8} = 2.0$$
Looking up $z = 2.00$ on a cumulative left-tail table gives you an area of $0.9772$. That means $97.72%$ of customer tickets are resolved faster than $61$ minutes, putting this specific ticket in the upper tail of wait times.
Step-by-Step Lookup Process
When you're working with real datasets, matching your raw numbers to table probabilities requires a repeatable process. Follow these steps to ensure accurate lookups every time:
- Verify your distribution shape: Check if your data is roughly bell-shaped. If it's heavily skewed, standard normal tables will give you misleading probabilities.
- Calculate the mean and standard deviation: Pull these parameters from your dataset before attempting any standardization.
- Compute the z-score: Subtract the mean from your target value and divide by the standard deviation. Keep two decimal places.
- Split the z-score digits: Use the integer and first decimal place to find the correct row, and the second decimal place to find the column.
- Interpret the table output: Identify whether your table measures left-tail area, right-tail area, or center-out area, and subtract from $1$ if you need the opposite tail.
Common Pitfalls and How to Avoid Them
Even experienced analysts run into trouble when misinterpreting table boundaries or ignoring underlying data assumptions.
- Confusing left-tail and right-tail tables: If your table shows the area to the left, looking up $z = 1.96$ gives you $0.975$. If you need the area in the right tail (the top $2.5%$), you must calculate $1 - 0.975 = 0.025$.
- Assuming normality blindly: Real-world data often has fat tails or extreme outliers. If you don't run a normal probability plot first, your table lookups will misestimate real probabilities.
- Mixing up sample and population formulas: Using sample standard deviation ($s$) in place of population standard deviation ($\sigma$) on small samples distorts your $z$-score. For small samples, Student's t-distribution is a safer bet.
For deeper context on probability distributions, the Probability and Statistics Cookbook offers great foundational notes on how these curves behave mathematically.
Frequently Asked Questions
What is the difference between a z-score and a t-score?
A z-score is used when you know the exact population standard deviation and have a large sample size. A t-score is used when the population standard deviation is unknown and you're estimating it from a smaller sample.
Why does a z-score of zero equal 0.5000 on the table?
Because a z-score of zero sits directly on the mean of a standard normal distribution. Exactly half of the data falls below the mean, and the other half falls above it, representing $50%$ or $0.5000$ of the total area.
Can a z-score be negative?
Yes. A negative z-score simply means your raw data point is below the mean. For example, a z-score of $-1.5$ sits one and a half standard deviations below the average.
How do I find the area between two different z-scores?
Look up the cumulative probabilities for both z-scores independently on your table. Then, subtract the smaller probability from the larger probability to find the proportion of data trapped between them.
