Gifted Issues Discussion homepage Forums Identification, Testing & Assessment MAP 2-5 vs 6+

Hello All
Does anyone know if/why the percentile calibration for these are the same? My DD is in 6th grade and took the 6+, and she dropped from 95th percentile to the 89th (based on 2011 norms). She said the test was harder than the 2-5.
I'm worried because our school district uses the fall Map score for advanced placement. So I wonder if she will get kicked out of the program.
Interestingly enough, I'm seeing online that this drop is not unusual. If that is indeed the case, wouldn't it make sense for the percentiles to be different?

One of the phenomena that occurs on norm-referenced testing is that, near the top of the range of the test, the spread becomes less reliable. Her report that the test was more challenging supports this, as the existence of higher-level problems indicates a greater likelihood of capturing her full instructional range. On the 2-5 version, all the high performing students in the grade 5 norms were bunched up, because there weren't enough hard problems, so that it was difficult to precisely distinguish between, say, a 79th %ile performer and a 97th %ile performer. The reliability of a 95th %ile score at the top of the MAP 2-5 is much less than the reliability in the middle percentiles, because it is so close to the ceiling of the test that minor differences in raw score (say one or two raw score points), which may occur for reasons other than "true" skill levels, can skew the %ile quite a bit one way or the other. Now she's testing nearer the bottom of the item pool for grade 6+, with plenty of headroom above her. Consequently, the likelihood of a noticeable change in %iles between the 2-5 and the 6+ is fairly great.

Another factor is the adaptive testing system, which starts everyone from the same place when testing for the first time (roughly one grade level below nominal grade placement), and gives them a limited number of items to work up (or down) to their true level. It is possible that a high functioning student might run out of the number of problems administered before the system reaches her instructional level. If that is the case, you might see a bump in her percentile, at the winter MAP.

In addition, regression to the mean would predict greater drops in percentile on re-test for the upper extremes than for percentiles closer to the 50th %ile.

In the big picture, 95th to 89th is a noticeable, but not necessarily dramatic, drop in performance. It's only about half an SD, some of which can be explained purely by standard error of measurement. High scores usually have higher SEMs, according to the publisher's documentation.

Whether she will continue to be eligible for the program depends, of course, on your local district criteria.

Thanks for the detailed reply, aeh. I can understand it better now. For this school year, she will stay in the advanced class because she was in the 99 percentile last fall. I have to wait and see what the school decides to do for next year.

You are correct that a drop is not unusual and aeh already provided a really good explanation of why. At the same time, neither rising scores nor stable scores are unusual either and that is why the "percentile calibration" are the same. DS & DD, who took the 6+ version for the first time last year, are instances of stable and/or rising scores, but they also scored at the top of the charts. For what it is worth, DS/DD concur that the 6+ tests were harder, especially for DS, who scored substantially above the 99 percentile threshold.

2015 normative data is available, if that helps you - https://www.nwea.org/resources/2015-normative-data/

My son just took it and his scores all went WAY up compared to last year, so I don't know if things will be adjusted in the future or not.

I am not clear on how to use this normative data. Unless it is as straight forward as it seems. Find the mean and figure the SD and compare growth. I didn't see a chart for percentile like the NWEA_2011_RIT_Scale_Norms.

Eh...these tables don't have the growth norms that you may be thinking of. You can use the status norms to find a child's approximate percentile, based on RIT score. You can also use the growth to describe the expected growth of a population of students, but individual students don't necessarily have the same curve, with known differences in growth for students who are above and below the mean at the beginning of the year. What you really want for comparing growth is growth norms that are broken out by percentile tiers. E.g., the mean beginning-to-end of year growth for the child who performed in the 85th %ile in fall was xx points.

To do that, you might enjoy playing with this little Excel tool that NWEA developed for teachers:

https://legacysupport.nwea.org/support/article/2015-achievement-status-and-growth-calculator

Input the beginning and end RIT scores by content area and grade (if you don't know the exact week of school in which testing was conducted, use the defaults), and it pops out your student's growth percentile, which, one hopes, will fall at or above the 25th %ile (average and up). Though in some schools, 16th %ile is the cutoff for average.

The 2011 RIT Scale Norms chart had the percentile in one column 1 thru 99 and 8 columns for grades 3 thru 8 with corrisponding scores in those columns to match the percentile column. You could see where 99% started. Growth has been stale for 2 years as far as RIT score. I was just wondering how much has changed since 2011 as far as the 99% group (where it starts.)

It seems the median dropped a few points, at least as far as 7th grade. I was wondering where the ends of the curve were. Maybe they will provide something when they get more data.

Thanks for the link.

Some percentiles/RIT scores. The highest listed here is 95.

https://www.nwea.org/content/uploads/2015/06/2015-MAP-Comparative-Data-One-Sheet-AUG15.pdf

The 2015 norm data are not available to the general public at the level of granularity you are likely seeking:

https://legacysupport.nwea.org/support/article/norms-study-resources

They do give you the mean and the SD in the One Sheet Report for each test by grade. With a little statitisitcs and a spreadsheet you can easily generate the same data given in the 2011 report. Completely dumb they don't just print it out via PDF...probably want $$$ for it.