We have discussed the sampling distribution of the sample mean when the population standard deviation, σ, is known. However, in practice, we rarely know the population standard deviation. In the past, when the sample size was large, this did not present a problem to statisticians. They used the sample standard deviation, s, as an estimate for σ and proceeded as before, calculating a confidence interval with close enough results. However, statisticians ran into problems when the sample size was small. A small sample size can cause inaccuracies in the confidence interval.

Student’s t-Distribution

William S. Gosset (1876–1937) of the Guinness brewery in Dublin, Ireland, ran into this problem. His experiments with hops and barley produced very few samples. Just replacing σ with s did not always produce accurate results when he tried to use existing inference techniques. He realized that he could not use a normal distribution for the calculation since finding that the actual distribution depends on the sample size. This is because s is a more reliable estimate of σ as samples get bigger. This problem led him to “discover” what is called Student's t-distribution (after Gosset’s pen name, Student).

Until the mid-1970s, some statisticians used the normal distribution approximation for large sample sizes and used the Student’s t-distribution only for sample sizes of at most 30. In our current age of technology, the oft-accepted practice now is to simply use the Student’s t-distribution whenever s is used as an estimate for σ.

In summary, if you draw a simple random sample of size n from a population that has an approximately normal distribution with mean μ and unknown population standard deviation σ and calculate the t-score, t = , then the t-scores follow a Student’s t-distribution with n – 1 degrees of freedom. The t-score has the same interpretation as the z-score. It measures how far is from its mean, μ. For each sample size n, there is a different Student’s t-distribution.

The following images compare the z (standard normal) and t (Student’s t). What differences do you notice?

Degrees of Freedom

The degrees of freedom (df), come from the calculation of the sample standard deviation, s. Remember when we calculated a sample standard deviation, we divided the sum of the squared deviations by n − 1, but we used n deviations (x – ) to calculate s. Because the sum of the deviations is zero, we can find the last deviation once we know the other n – 1 deviations. The other n – 1 deviations can change or vary freely. We call the number n – 1 the degrees of freedom.

For example, if we have a sample of size n = 20 items, then we calculate the degrees of freedom as df = n – 1 = 20 – 1 = 19, and we write the distribution as T ~ t₁₉.

The following image shows what happens to the t-distribution as you change the degrees of freedom. What happens as the df increases? What happens once n reaches around 30, and how does that relate to what you already know about the CLT?

Properties of the Student’s t-Distribution

To summarize the properties of the t-distribution:

• The graph for the Student’s t-distribution is similar to the standard normal curve, in that it is symmetric about a mean of zero.
• The Student’s t-distribution has more probability in its tails than the standard normal distribution because the spread of the t-distribution is greater than the spread of the standard normal. So the graph of the Student’s t-distribution will be thicker in the tails and shorter in the center than the graph of the standard normal distribution.
• The exact shape of the Student’s t-distribution depends on the degrees of freedom. As the degrees of freedom increases, the graph of Student’s t-distribution becomes more like the graph of the standard normal distribution.
• The underlying population of individual observations is assumed to be normally distributed with unknown population mean μ and unknown population standard deviation σ. The size of the underlying population is generally not relevant unless it is very small. If it is bell-shaped (normal), then the assumption is met and doesn’t need discussion. Random sampling is assumed, but that is a completely separate assumption from normality.
• The notation for the Student’s t-distribution (using T as the random variable) is T ~ t_df where df = n – 1.

Finding t-Distribution Probabilities

A probability table for the Student’s t-distribution can also be used. The table gives t-scores that correspond to the confidence level (column) and degrees of freedom (row). When using a t-table, note that some tables are formatted to show the confidence level in the column headings, while the column headings in some tables may show only corresponding area in one or both tails. Notice that most t-tables gives t-scores given the degrees of freedom and the right-tailed probability.

You’ll find that t-tables are adequate for finding critical values but are very limited when trying to find p-values. Calculators and computers can easily calculate any Student’s t-probabilities.

Figure References

Figure 6.1: Phillip Glickman (2019). clear glass cup close-up photography. Unsplash license. https://unsplash.com/photos/4wnZbnW9Bv0

Figure 6.2: Kindred Grey (2021). Comparing the standard normal distribution and Student’s t-distribution. CC BY-SA 4.0.

Figure 6.3: Kindred Grey (2021). t-distribution with different degrees of freedom. CC BY-SA 4.0.

Figure Descriptions

Figure 6.1: A Guinness Draught beer in a glass next to a candle in an English pub.

Figure 6.2: Two lines on one x,y plot that both follow the bell curve. X axis ranges from -3 to 3 by 1. Density is on the Y axis and goes from 0 to 0.4 by .1. Normal distribution is taller at the maximum and more narrow on the sides. t distribution is shorter at the maximum point and wider on both sides.

Figure 6.3: Four lines on one x,y plot. X axis ranges from -3 to 3 by 1. Density is on the Y axis and goes from 0 to 0.4 by .1. All 4 lines follow bell curve and are very similar. From most density at the maximum point to lowest: Normal, df=30, df=10, df=5.