A lot of people on here seem to be assuming that a test age 11 is a good predictor of ability.
They're not. Look at computer generated GCSE target grades and how inaccurate they can be. If everyone took the 11+, using the best predictive tests we have at the moment and the top 25% went to a grammar, about 1 in 5 students would be allocated the wrong school.
From the Sutton Trust report into the effectiveness of grammar schools:
"All tests are unreliable to some extent, so a person’s score is partly a matter of chance. This means that for some, the decision to offer a grammar school place or not will be something of a lottery.
One way in which the adequacy of a selection test might be judged is in terms of its predictive validity. If a test at age 11 could accurately predict academic achievement at, say, age 16, then we might argue that such a test would be a good way of discriminating between those children at age 11 who were ‘academic’ and those who were not. The correlation between test scores at age 11 and achievement scores at age 16 is a measure of the validity of the prediction, a correlation of 1 indicating perfect prediction, and a correlation of 0 indicating no predictive ability at all. If the correlation was poor (close to 0) and a large number of those who ‘failed’ the test at 11 went on to achieve good academic results at 16, we might be less convinced that the test was really appropriate for the purpose of selection.
Fortunately, data exist for a number of different tests taken at age 11 and the corresponding performance at GCSE of the same pupils (e.g. Cognitive Abilities Test (CAT), MidYIS, London Reading Test (LRT). In none of these cases is the correlation much above about 0.7. Whether this is high enough to show adequate predictive validity is a matter of opinion.
Figure 1 illustrates how, with a correlation of 0.7 and a cut-off pass mark that selects 25% of 11 year olds, children can be wrongly selected or not selected. If, as a crude generalization, those 16 year olds who achieve in the top 25% are taken to be those who should have gone to a grammar school, we can see that about 78% go to the appropriate school for their ability, leaving around 22% wrongly allocated."
The graph associated with this data is on page 21 (electronic page 36) of this document: www.suttontrust.com/wp-content/uploads/2008/10/SuttonTrustFullReportFinal.pdf