Langdon Adult Intelligence Test

STATISTICAL REPORT

LAIT NORMING #2, 7/15/79
Copyright © 1979 by Kevin Langdon
Permission to republish granted to Darryl Miyaguchi

    This report provides an overview of the norming of the Langdon Adult Intelligence Test completed in July 1979, including 553 testees.  Only a handful of the earliest responses to the test's appearance in the April 1979 issue of Omni are included.  A further norming will be completed after the bulk of the Omni response has been analyzed.

    The norming sample included 207 persons tested on Form A and 346 tested on Form B.  Form A was an early version of the test, now out of print, differing from Form B only in a few items.  Thus, one form cannot be used to obtain an independent measure of intelligence for an individual tested using the other.

    Raw scores of all testees were computed using the appropriate formula for form A or Form B.  Additionally, a score for items unchanged between the two forms, and scores for each of two matched sets of items containing one half of the items on each part of the test, were computed for each testee.

    Correlations between halves of a test (split-test correlations) are generally lower than those which would be obtained if the tests were full length because any chance variation is a larger percentage of the half test.  To compensate for this effect, it is usual to apply the formula where r₁ is the uncorrected correlation coefficient and r₂ is the corrected correlation.  For Form A of the LAIT r₁ = .822 and r₂ = .902 and for Form B r₁ = .815 and r₂ = .898.

    Scores on other tests reported by testees were entered into the computer with other data from the answer sheets and paired with LAIT scores.  A table of LAIT-previous score pairs for LAIT total score and each subscore was constructed for each test which was used in the norming (see Table 1) and arranged in LAIT score order (lowest to highest).

    Correlations between the LAIT and previous score distributions for LAIT-previous score pairs from the lowest LAIT score through each LAIT score were calculated and printed out and a cutoff point was determined to maximize r²N, where r is the correlation coefficient and N is the number of score pairs included.  Table 2 shows the total number of previous scores reported, the total falling below the cutoff, the overall correlations, and the correlations for the truncated distributions for each test used.  Scores on each test were weighted by this correlation figure, representing the relationship between the LAIT and the test concerned without the effects of the generally lower effective ceiling of most other tests, in calculating the overall means, standard deviations, and correlations for LAIT and previous score distributions for all tests included in the norming.

    A scatter diagram of LAIT scores against all reported scores on other tests was produced for LAIT total scores and subscores.  Outlying points were identified and were not used in calculating distribution means and standard deviations and correlations between LAIT and previous scores.  Table 3 summarizes the values obtained.

    A correction for the very tight distribution of previous scores reported due to the bulk of the norming population having been preselected by these scores was applied to the previous score standard deviations by the formula , where is the uncorrected standard deviation, is the corrected standard deviation, and r is the correlation between LAIT and previous score distributions.

    At this point, the LAIT and previous total score means and standard deviations were equated and I.Q.'s were calculated.  Total and part score means and standard deviations for the entire score distributions were equated to yield subscore I.Q.'s.  General population percentiles were looked up in an internal table and tested population percentiles were calculated directly.