Table 1.
Pair compared | Sample
sizes
|
Mean growth rates,
1/h
|
df† | t′(B−A) | Opportunity
for selection,
×100
|
df† | t′(B−A) | |||
---|---|---|---|---|---|---|---|---|---|---|
A* vs. B* | A | B | A | B | A | B | ||||
F30 vs. M30 | 127 | 229 | 0.418 | 0.417 | 254.1 | −0.083 | 0.380 | 0.714 | 353.9 | 2.211 |
F38 vs. M38 | 143 | 266 | 0.247 | 0.200 | 244.9 | −7.448** | 5.266 | 9.926 | 386.7 | 3.884** |
F30 vs. F38 | 127 | 143 | 0.418 | 0.247 | 210.1 | −31.84** | 0.380 | 5.266 | 145.9 | 7.238** |
M30 vs. M38 | 229 | 266 | 0.417 | 0.200 | 433.9 | −47.74** | 0.714 | 9.926 | 275.7 | 9.178** |
indicates probability of the type I error lower than 0.001. Critical values of t′ were calculated for four planned comparisons.
A and B serve as variables that should be substituted within each row for appropriate F or M assayed at 30°C or 38°C.
Degrees of freedom are not natural numbers because they were calculated according to Welsh's procedure to compensate for unequal sizes and variances of the compared samples (19).