Because the data do not fit
theoretical null distributions, attempts to infer natural selection from polymorphism data will require genome - wide surveys of polymorphism in order to identify anomalous regions.
The null distribution looks an awful lot like our actual results: it is almost flat in the middle and flares at the ends (see Figure 2).
The estimates hewed very close to
the null distribution, suggesting that little but estimation error was present.
So the tight fit of
the null distribution suggests that the rankings are, if not entirely random, then darn close.
Remember,
the null distribution shows what program rankings would look like if they were entirely random.
To highlight the role of random error, we calculated the «
null distribution,» or what the distribution of program rankings would look like if all the programs were actually identical and nothing but random estimation error were present.
In fact, when we lay
the null distribution over the Texas results, the fit is almost perfect (see Figure 3).