The SAGES-2 reasoning test is a single task-type, whereas the CogAT is a few types, so it's not entirely surprising that, in some children, you might obtain noticeably different scores on them. Remember that SAGES-2 is a screening instrument, not intended to produce a true measure of cognition. (And, for that matter, the CogAT also does not claim to be a true cognitive assessment (aka, IQ test), but an aptitude test measuring learned abilities.)
And, as always, I have to mention that very young children can be unstable test-takers for many developmental reasons not associated with the test per se. (SAGES-2 was normed on kindergartners, so it does include them, test behavior and all, in the scoring, but the range of "norm"-al is very broad in five and six year-olds.)
