The mean and (standard deviations)—calculated as an
average over the 3 years—are 2.5 (1.6) and 4.0 (2.2) m
FMK, for the non-electrified and electrified track units,
respectively. In our sample of 278 track units, 154 are nonelectrified
and 124 electrified. Estimating the full model (i.e.
Eq. (3)) it is possible, again using the F-test, to test for
whether a restricted model (now with 35 restrictions) is
better than the full model. We get F ¼ 0:538 (Pvalue
¼ 0.985) and—in the same way as for the Swedish
material—we cannot reject the restrictions. The final model
used for analyzing the Finnish data is