Test Statistic Formula Explained with Examples

When researchers want to know whether their findings are meaningful or simply the result of chance, they turn to a fundamental tool in statistics: the test statistic. At its core, a test statistic is a single number calculated from sample data that helps determine whether a hypothesis about a population holds up under scrutiny.

From clinical drug trials to marketing experiments, the test statistic formula serves as the backbone of data-driven decision-making. It translates raw numbers into a value that can be compared against known probability distributions, allowing analysts to draw conclusions with measurable confidence.

What Is a Test Statistic?

A test statistic is a numerical value computed from sample data during a hypothesis test. Its sole purpose is to measure how far your observed data strays from what you would expect if the null hypothesis were true. The greater the distance, the more reason you have to question that null hypothesis.

Think of it as a signal-to-noise ratio. The “signal” is the difference between your observed data and the expected value under the null hypothesis. The “noise” is the natural variability you’d expect from random sampling alone. When the signal is strong relative to the noise, the test statistic grows large — and your confidence that something real is happening grows with it.

Every hypothesis test produces a test statistic, though the exact formula varies depending on the type of data and the question being asked. A t-test, a chi-square test, and an F-test each generate their own version, but they all serve the same fundamental role: converting raw data into a single, comparable number that drives the final statistical decision.

General Test Statistic Formula

While specific tests use their own variations, most test statistics share a common underlying structure:

Test Statistic = (Observed Value − Expected Value) / Standard Error

Each component carries a distinct role. The observed value is the result you actually measured from your sample — a mean, proportion, or other summary figure. The expected value is what the null hypothesis predicts that measurement should be. The standard error captures how much variability you’d naturally expect in your estimate due to random sampling.

This structure is elegant in its logic. The numerator quantifies the gap between reality and the null hypothesis’s prediction. The denominator puts that gap in context — a difference of 10 points means something very different when your data is tightly clustered versus wildly spread.

The result is a standardized number, often called a test statistic value or t-score, z-score, or F-ratio depending on the test applied. This standardization is what makes the value comparable to a known probability distribution, ultimately allowing you to calculate a p-value and decide whether your results cross the threshold of statistical significance.

Types of Test Statistic Formulas

Different research questions call for different statistical tests, each with its own formula tailored to the data type and hypothesis at hand.

Z-Test

The z-test applies when you know the population standard deviation or are working with a large sample size. Its formula compares the sample mean to the population mean, scaled by the standard error. It produces a z-score measured against the standard normal distribution.

T-Test

When the population standard deviation is unknown — which is most of the time in real research — the t-test steps in. It follows the same general structure as the z-test but uses the sample standard deviation instead, and compares the result against a t-distribution that accounts for the added uncertainty of smaller samples.

Chi-Square Test

Rather than comparing means, the chi-square test works with categorical data. It measures how far observed frequencies in a dataset deviate from expected frequencies, making it the go-to test for analyzing counts, proportions, and relationships between categorical variables.

F-Test

The F-test compares variances across two or more groups. It is the engine behind Analysis of Variance (ANOVA), determining whether the variation between group means is large enough relative to the variation within groups to suggest a genuine difference rather than random fluctuation.

Summary Table

Test	Best Used When	Compared Against
Z-Test	Large sample, known population SD	Normal distribution
T-Test	Small sample, unknown population SD	T-distribution
Chi-Square	Categorical data	Chi-square distribution
F-Test	Comparing variances or multiple groups	F-distribution

How to Calculate a Test Statistic

Step 1: State Your Hypotheses

Begin by defining your null hypothesis (H₀) — the assumption of no effect or no difference — and your alternative hypothesis (H₁), which represents the outcome you’re testing for. These two statements frame everything that follows.

Step 2: Select the Appropriate Test

Choose your test based on your data type, sample size, and what you’re measuring. Comparing two means with a small sample? Use a t-test. Working with categorical counts? Reach for chi-square. The wrong test applied to the right data produces meaningless results.

Step 3: Gather and Summarize Your Data

Calculate the summary statistics your formula requires — typically the sample mean, sample size, and standard deviation. Accuracy here is critical, as every subsequent step builds on these figures.

Step 4: Apply the Formula

Plug your values into the appropriate formula. For example, using a one-sample t-test:

t = (x̄ − μ₀) / (s / √n)

Where:

• x̄ = sample mean
• μ₀ = hypothesized population mean
• s = sample standard deviation
• n = sample size

Step 5: Interpret the Result

Compare your calculated test statistic against the critical value from the relevant distribution table, or use it to derive a p-value. If the test statistic exceeds the critical value — or the p-value falls below your significance threshold (commonly 0.05) — you reject the null hypothesis.

Worked Example

A nutritionist claims the average daily sugar intake is 50g. You sample 30 people and find a mean of 54g with a standard deviation of 10g. Applying the t-test formula:

t = (54 − 50) / (10 / √30) = 4 / 1.83 ≈ 2.19

With 29 degrees of freedom at a 0.05 significance level, the critical value is approximately 2.045. Since 2.19 exceeds 2.045, you reject the null hypothesis — the evidence suggests average sugar intake differs meaningfully from 50g.

Example Calculations

Example 1: Z-Test

A manufacturer claims their light bulbs last an average of 1,000 hours. You test 50 bulbs and find a sample mean of 980 hours. The known population standard deviation is 60 hours.

z = (x̄ − μ) / (σ / √n) z = (980 − 1000) / (60 / √50) z = −20 / 8.49 ≈ −2.36

At a 0.05 significance level, the critical z-value is ±1.96. Since −2.36 falls beyond this threshold, you reject the null hypothesis. The bulbs are not lasting as long as claimed.

Example 2: T-Test

A school principal believes students average 7 hours of sleep per night. A sample of 20 students reveals a mean of 6.4 hours with a standard deviation of 1.2 hours.

t = (x̄ − μ₀) / (s / √n) t = (6.4 − 7) / (1.2 / √20) t = −0.6 / 0.268 ≈ −2.24

With 19 degrees of freedom at a 0.05 significance level, the critical t-value is ±2.093. Since −2.24 exceeds this threshold, you reject the null hypothesis. Students are sleeping significantly less than 7 hours.

Example 3: Chi-Square Test

A survey asks 200 people whether they prefer tea or coffee, split across two age groups. The observed and expected frequencies are:

Group	Prefers Tea	Prefers Coffee
Under 40 (Observed)	55	45
Under 40 (Expected)	50	50
Over 40 (Observed)	45	55
Over 40 (Expected)	50	50

χ² = Σ [(O − E)² / E] χ² = (5²/50) + (5²/50) + (5²/50) + (5²/50) χ² = 0.5 + 0.5 + 0.5 + 0.5 = 2.0

With 1 degree of freedom, the critical chi-square value at 0.05 significance is 3.841. Since 2.0 falls below this threshold, you fail to reject the null hypothesis. There is no statistically significant relationship between age group and beverage preference.

Test Statistic vs P-Value

FAQs