09.12.2012 Views

I__. - International Military Testing Association

I__. - International Military Testing Association

I__. - International Military Testing Association

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

tally significant. Hotelling's T2 was 5.1998 with a probability<br />

of .027 at 1 and 2 degrees of freedom. It should be noted, however,<br />

that the test,;was made on a group to group basis and there<br />

was no indication which individual pairs may have changed the<br />

most. It was in fact possible that no pairs would significantly<br />

vary even though the full model rejected the null hypothesis.<br />

Also, since the test was non-directional, it was not possible to<br />

identify which group contained higher RGLs than the other although<br />

they appear to have been higher more recently. I<br />

.<br />

An observation related to this, however, was the significant'<br />

correlation of the RGLS of the two samples (r = .4709, p =<br />

.OOl). This raised an interesting situation in which samples<br />

taken in 1982 and 1990,: although significantly different,. were<br />

none the less related. The relatedness, however, was not developmental<br />

over time. One possible solution to this ambiguity was<br />

that RGL was varying with frequent but intermittent corrections<br />

using a current "clearly written text" standard.<br />

1990 RGLs and First Time Examination Failures. RGLs were not<br />

significantly related to first time examination failure rates (r<br />

= .0741 and probability = .327).<br />

1990 RGL and Overall Course Failures. ~11 students failing the<br />

first final examination were provided a retest. Course failure<br />

required failure of both the first examination and the<br />

reexamination. As was the case with first time exam failures,<br />

course failures were not significantly related to RGL in the<br />

1990 sample (r = .0404, probability = .403).<br />

FORCAST RGL and AFHRL Targets. The correlation between FORCAST<br />

RGLs of course materials in the 1990 sample and AFHRL targets of<br />

actual student reading ability (50th percentile reading ability)<br />

was .0695 and was not significant (p = .333). A reduced target<br />

at the 15th percentile also failed to be significantly related<br />

to the obtained FORCAST RGL (r = .0249 with probability = .439).<br />

The data failed to demonstrate that the variation within the<br />

reading ability of personnel was linearly related to FORCAST<br />

RGLs of the CDC material.<br />

FORCAST RGL and Flesch-Kincade RGL Comparison. Due to resource<br />

limitations on the Flesch-Kincade RGL side, comparisons were<br />

made on only one CDC consisting of four volumes. The means and<br />

SDS were 11.2725 and . 5187 for FORCAST and 9.0800 and .1619 for<br />

Flesch-Kincade. The obvious difference between the averages was<br />

significant with Hotelling's T2 of 92.8489 and probability equal<br />

.004 '(df = 1 and 3). Flesch-Kincade generated significantly<br />

lower RGL estimates than did FORCAST. The correlation, although<br />

129

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!