Lertap makes two or three "Stats" reports for a subtest.

 

Every time you run Lertap's Elmillon option, Stats1f and Stats1b reports are created. If the subtest is a cognitive one, a Stats1ul report is also usually created. (It is possible to stop Lertap from creating Stats1ul reports by using a setting in the System worksheet.)

 

The f in Stats1f stand for "full". These reports have the most detailed information for test items. Statsf reports go back to Lertap's birth in 1973; they have changed a bit over the years, but not by very much.

 

Some users often find Stats1f to be excessively detailed, full of numbers and tables which make sense to experienced test developers but, well, couldn't there be something easier to read? Yes: Stats1b.

 

Stats1b reports are b for briefer. They are designed to be easier to understand whilst retaining the most important information from Stats1f.

 

Stats1f and Stats1b are similar in that they're based on the use of correlation coefficients to reflect how items have performed.

 

When cognitive items are involved, another time-tested way to summarize item performance is to use "upper-lower" (U-L) methods, sometimes referred to as "high-low" (H-L) methods. These methods date back to pre-computer days when educators endeavored to assess item performance by hand, without the help of any electronic aide. The idea is a simple one: to see if an item is working well, we'll look at results from the best test takers, those with the highest test scores. We will compare these results from those gathered by looking at how the weakest (lowest) students did. If the top (or upper) students got the item right, while the bottom (lower) students did not, we say the item has "discriminated"; such items are ones we'll use again in the future when we want to have a test which we know can identify the strongest and weakest students.

 

The Stats1ul report employs the upper-lower method to index item quality.

 

When more than one subtest is involved in a Lertap analysis, there will be more reports, two or three for each subtest. Thus we will have Stats2f and Stats2b, and maybe Stats2ul, corresponding to results for the second subtest, and, on a really lucky day with three subtests, we'll be able to bask in the light of Stats3f, Stats3b, and, well, surely you get the idea. The number in the report name refers to subtest number.

 

You'll now be keen to page ahead, getting more detailed information for each of Lertap's Stats reports. But before you do, note: there's a top-flight manual about this stuff, a real printed book whose battery lasts forever and ever, something which can even be read in the brightest of sunshine, withstanding coffee and tea spills without crashing. You can even drop it, or run over it with your cycle or the baby's perambulator. Old technology, but rock solid. And it doesn't assume you're full bottle on statistics -- it holds your hand with loving humor, gently nudging you along ever so gently and respectfully. Read more about the manual here, and do try and read it; it covers Lertap's various statistics reports in more detail, and has several practical examples.