Yes I know it is a different test (called middle-level) but the basic logic does not change at all because it is taken by far smaller population designed with tighter range in scaled score (710-440=270 middle vs 800-500=300 upper), meaning the fluctuation in % per incremental raw score change for middle-level will be much bigger.
According to the 2011-2012 Interpretive Guide to the SSAT, from August 2008 to July 2011, the specific numbers of students taking the SSAT per grade were:
Grade 11: 2,546
Grade 10: 10,232
Grade 9: 23,481
Grade 8: 99,080
Grade 7: 17,342
Grade 6: 14,603
Grade 5: 13,005
Notice the big jump in 8th graders’ taking the SSAT. Also notice how small (and select) these groups are (cumulative numbers for 3 years!).
If you get 14 questions wrong out of 60 in verbal for middle, you would get around 73%, but if you get 15 questions wrong out of 60 in verbal for upper, you will still get around 84%. Of course, the middle-level questions are easier but your child is younger too. The percentile drop or gain is greater - each mistake will be more punishing and each correct answer more rewarding for middle-level. So it is not to be taken lightly at all, simply because it is “only” middle - mistakes could do greater harm.
BTW, it is exactly same number of questions, format and time between middle and upper (only elementary level is significantly different); the only differences are difficulty level and writing prompts. So she could actually do much better at upper after some practice and strategy. And 58% is an impressive score for testing cold turkey - it means she can hold her own against the most select group.
As I said, where you are on the scoring scale, a small difference in raw scores can result in a large difference in % ranking. For example, a student who gets a raw score of 25 (out of 60) on the verbal section of the upper level SSAT will increase by nearly 20% points by getting only 5 more correct answers. For middle-level, if you leave 10 blank 97%, if you write wrong answers to 14, it drops to 73%. Only 4 more wrongs plus penalties result in such a big difference in percentile. So the important things are “guess management” and taking the test multiple times.