The correlation is not there
Across the research, reading rate and comprehension separate almost immediately once vocabulary is controlled for. Slow readers score the same as quick ones. Everybody has met the child who reads a page in ninety seconds and cannot say what was on it.
The test that works
Stop a student at the end of a paragraph and ask what it *did* — not what it said. *It gave an example.* *It changed its mind.* *It set up the objection it is about to answer.* Students who cannot answer that are the ones losing marks, at any speed.
Teach the moves
There are maybe eight things a paragraph can do in an argumentative text. Name them, then go and find them in something two pages long. It is a fortnight of work and it moves scores that a year of reading practice did not.
What speed is actually for
Rate matters when a test is running out of time, and then it is a timing problem rather than a reading one — a different fix, on a different section of the report.