Interesting. So there's a linkage study used to discuss the predictive of value of STAR360 for STAAR, which suggests about .75 or so correlation, with about 80+% accuracy for predicting classifications (performance levels).
http://doc.renlearn.com/KMNet/R004573515GK64DE.pdfI'll have to do some more digging, but at a guess, there may be differences in the proportion of literary to informational text on the various tests, or whether reading comprehension is assessed by responding to questions or cloze (fill-in-the-blanks), or whether the text in question is sentences vs paragraphs. The SAT probably has more inferential comprehension on it, vs an elementary/middle school level text. Also, there are often differences between tests with a single grade-level item set (as many state tests have), and those with adaptive testing (as many progress-monitoring tests such as STAR360 do). (I haven't checked to see if either of those tests is adaptive.) Those are many of the questions I'd ask, anyway.