Latticework — Master the skills engineers actually use

Explanation

Statistics' hypothesis-testing module used a z-test (statistics.NormalDist) specifically because Python's standard library has no t-distribution implementation — noted at the time as a deliberate simplification, accurate for large samples but not the textbook-standard choice for small ones. scipy.stats closes that exact gap: a real Student's t-distribution, and a one-line function for the most common test.

from scipy import stats

def one_sample_ttest(sample, popmean):
    t_statistic, p_value = stats.ttest_1samp(sample, popmean)
    return (round(t_statistic, 4), round(p_value, 4))

Try it

This is the real, textbook-standard t-test — the exact test Statistics' own course had to approximate with a z-test due to a genuine stdlib limitation, now available in one function call once scipy is in play.

Loading editor…

Exercise

Write `one_sample_ttest(sample, popmean)`: use `scipy.stats.ttest_1samp` to test whether `sample`'s mean differs from `popmean`. Return `(t_statistic, p_value)`, each rounded to 4 decimal places.

Quiz

Statistics' own hypothesis-testing module had to use a z-test instead of a t-test, and explicitly noted why. What was that reason?

Checkpoint

You can run a real one-sample t-test with scipy.stats, closing the exact gap the pure-stdlib Statistics course had to work around.

A Real T-Test with SciPy