STATISTICS
A small study may fail to show a statistically significant difference from the target, but the confidence interval may be too wide to support confidence in acceptability. In that situation, ‘not statistically different from the target’ should not be treated as ‘acceptable’.
Paired and independent t-tests The distinction between paired and independent data is critical. In a paired design, each observation in one condition is linked to a specific observation in the other condition. Split-sample studies are a common laboratory example. If 20 specimens are tested immediately and again after 24 hours of storage, the analysis should focus on the 20 within- specimen differences. Each specimen acts partly as its own control. In an independent design,
observations in one group are not linked to observations in the other group. For example, if turnaround times are compared between outpatient and inpatient samples collected from different patients on different days, the groups are usually independent. Confusing paired and independent
data is not a minor technical error. It changes the analysis. A paired t-test asks whether the mean within-pair difference differs from zero. An unpaired t-test asks whether two independent group means differ. These are different questions.
Welch’s t-test The classical Student t-test assumes equal variances in the two independent groups. That assumption is often unrealistic in laboratory data, where group sizes and variability may differ. Welch’s t-test does not require the equal-variance assumption and is often a safer default for comparing two independent means. This should not be interpreted as
meaning Welch’s t-test is always the answer. It still compares means and still requires the question to be appropriate for a mean-based comparison. However, it avoids one fragile assumption of the classical Student t-test.
Fig 1. The maths behind t-test variations. Ignore the symbols and complicated terminology – the top of the equation (numerator) measures the differences (within or between group[s]) depending on the test. The denominator (below the equation line) is a measure of the variability of the data and/or groups. So we are generating a statistic (t statistic) that is a ratio of the observed difference divided by the variability. Bigger number = bigger difference = bigger than can be described by data variability = significant.
18
WWW.PATHOLOGYINPRACTICE.COM June 2026
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56
