Learning Efficiency Test-II
This is not meant to be a comprehensive review of the LET-II but
merely a summation of questions and concern.
The Learning Efficiency Test-II (1992 revision) is published by
Academic Therapy Publications and authored by Raymond Wheeler, a
Professor of Psychology at East Carolina University in Greenville,
North Carolina. It costs $60.00 for a manual (187 pages), stimulus
cards, 50 record forms, and a vinyl folder. Norms are provided for
ages 5 through 75 and older, with tables presented in one year
intervals between 5 and 16 years.
The catalog and manual claim that it is a quick and reliable
measure of visual and auditory memory characteristics and that it can
provide useful information about a person's preferred modality of
learning, as well as providing information about the impact of
interference on memory storage and retrieval.
The LET-II can be administered in 10-15 minutes. It consists of
strings of 2 to 9 non-rhyming letters presented either orally or
visually. The child/adult responds verbally and subtests raw scores
are based on string length. The raw scores are then converted to
scaled scores and percentile ranks. Memory on the LET-II is assessed
in two modes (visual and auditory) and in three recall conditions
(Immediate, short-term, and long-term). The six subtest scores (Mean
10, SD 3) can be converted into Modality scores as well as a Global
Memory score (Mean 100 SD 15).
Given the way that memory can affect a child's functioning in
school, and the need to assess preferred learning modality, the
LET-II is a test that we might be tempted to buy.
An examination of the LET-II tests material, manual, and catalog
raised the following concerns:
The manual states that the LET-II "has been empirically
demonstrated to be highly predictive of actual classroom levels of
performance in reading and mathematics for students with average
ability as well as handicapped students" (page 10). The support for
this claim is 4 citations. Examining the bibliography at the back of
the manual, these four empirical studies turn out to be 3 unpublished
Master's degree theses done at East Carolina University in
Greenville, North Carolina. The fourth citation is for the LET (1981)
The LET-II was normed on a sample of 1126 children and adults
between the ages of 5 years 0 months and 85 years, 4 months. Data is
provided for the age, sex and race of the sample but no
Socio-economic status data is provided. The manual states that the
"participants came from a broad range" of SES backgrounds. Sex
variables are fairly even across the sample with 46% male and 53%
females being included. This is very close to the U.S. Census
Bureau's 1990 estimates of 48% and 52% respectively. Race variables
are less comparable to the Census data. The LET-II reports
percentages of 66 versus 33 for Caucasians and Blacks in the total
sample. This in contrast to the 77 and 12 percent estimates for the
U.S. population. For the ages 5 to 16 the percentages are 58 versus
42 for Caucasians and Blacks. No geographical data is provided. It is
my assumption that most came from North Carolina.
For each of the one year age intervals, 5 through 16 (not 5
through 15 as stated on page 7 of the manual), the samples average
only 55 people, with a high of 84 at age 8 and a low of 40 at age 16.
These are well below the sample size of 100 recommended by Salvia and
Yssledyke (Assessment in Special and Remedial Education/Third
Edition, 1985) for the computation of standard scores.
No data about the IQs of the children in the normative sample is
provided. Adults in the sample were administered the Peabody Picture
Vocabulary Test and anyone scoring below 85 were excluded from the
sample. It is further noted that "no known cases of mental
retardation as defined by a general IQ score of less than 85 (on
either a group or individually administered intelligence test) were
included in the sample." (italics added). Two points. Was this the
PPVT or the PPVT-R and would the age of the norms have an effect on
the IQ? Who defined mental retardation as "a general IQ score of less
Reliability studies using the LET-II were not conducted.
Information based on 2 studies using the LET (1981) are included in
the manual as measures of the LET-II reliability. One unpublished,
"informal" study, involved 55 learning disabled student in grades 4
through 12. Coefficients for this group were found to be .71 to .86
(median .80). No breakdown by age, grade, IQ, or time interval is
provided in the manual. Without that information, and the interval
between test and retest, this data seems almost meaningless. A second
study was done which involved only 40 students identified as having
"learning and behavior problems." These students each had IQs above
89 and were test-retested between 1 to 6 weeks. Reliability
coefficients ranged from .81 to .97 and are generally considered
adequate. However, there are no specific information about age and or
grade of the subjects. All are simply labeled "secondary" students in
the manual. Since these reliability studies were carried out using
the LET and not the LET-II, no coefficients, and hence no Standard
Error of Measurements are provided for the two Modality scores or for
the Global Memory score. Without this information, there is no way to
calculate meaningful differences between the Modality and Global
memory scores. The reported coefficients also offer no information
from which to draw conclusions about the permanency of the scores
since the time limit between the two administrations of the test was
Validity in the manual is addressed in a number of ways: content
validity, diagnostic validity, and predictive validity.
Diagnostic validity was shown by examining patterns of
performance among 4 groups of special education students and
comparing these patterns to a group of "average" students in grades 4
though 7. The 197 Average students had average Verbal, Performance,
and Full Scale WISC-R IQs of 111, 110, and 112 respectively. This
raises some question about what might be expected when this 'average'
group is compared to a special education population. Average is
typically thought of as having a mean of 100, not 112! Won't
comparing a group labeled "LD" who have a mean IQ of 93 with a group
labeled "average" with a mean IQ of 111 create some confusion when
interpreting? My average and this sample's average are different.
One confusing aspect of the statistical tables presented in the
manual is the arrangement of the categories for the "learning
groups." In the tables listing Correlations, Mean recall
characteristics, Percent of information loss, Reading and Math
achievement, and Stepwise and Simultaneous regression, the order in
which the categories are presented change from page to page. The
first table lists the order as Average, LD, EH, EMH and SL; yet the
very next table lists the order as Average, EH, LD, EMH, and SL. In
the next 3 tables, the order changes again and the EMH group becomes
an EMR group? One may quibble that this is a minor point, but in a
manual that contains other errors, some fairly serious, this does
raise questions about the validity of such tables and statistics.
Table 2, which presents 4 pages of Means and Standard Deviations
for raw scores based on string length by age level grouping, contains
either a very curious developmental anomaly or a very serious error.
Page 163 is a table for Auditory Memory-Ordered. For the Immediate
and Long term trials the table shows the expected increasing lengths
of the string as the person's age increases. A child of 5 remembers
an average 2.91 letters while a person of 16 remembers 5.41. However,
for the column for the Short term recall trial we discover that all
ages had mean recalls of approximately 1 letter?? In fact, according
to this chart, at age 70-74, a person remembers 1.71 letters but the
standard deviation is larger than the mean (2.04)? This is almost
certainly a serious error in the table. How many of the other tables
are in error? I don't know for sure, but curiously, looking at the
conversion tables for raw score to scaled score for a 15 year old
(page 99) I found another! As you examine the table you note that the
higher the raw score one has, typically the higher the scaled score
will be. Not so for age 15, unordered, immediate recall. OOPS!
Table 6 (page 167) and Table 7 (page 168) display mean recall
characteristics and percent of information lost for each recall trial
for each special group. These pages are interesting for the way the
information is explained in the text of the manual. The manual
explains that special groups show distinct patterns of differences in
recall capacity. The tables do show this but one must ask about the
use of percentages as the measure of loss. For example, the average
child, with an IQ of 112, starts by remembering 4.7 letters and drops
to 3.7 on short term recall, a loss of 22.3%. The LD child starts at
3.5 letters and drops to 2.2 on short term recall. A drop of 37%. The
percentages are different by such large percentages only because the
LD child started at a lower number. The actual loss is about the
same, 1 letter!
One aspect of validity not addressed in the LET-II manual is that
of carry over to real life. The LET-II has items that are discrete
units of abstract symbols (letter) delivered in a non meaningful
manner. With regard to children in school, much of the material to be
learned requires the recognition and use of meaningful items. Krupski
(1985) demonstrated that for learning disabled children, memory
performance is approximately equal to that of average students if the
material being presented is meaningful in nature. If this is true,
the LET-II may be measuring some important aspect of non meaningful
memory, but tells us nothing about the child's performance in real
Since the LET-II proposes to measure and relate certain learning
styles to academic functioning, there should be some studies to show
the meaningful relationship of different learning styles to classroom
success. The manual does not offer any study to support any claim
that academic deficiencies can be attributed to a weaker modality
score on the LET-II.
Administration and Scoring:
Administration of the LET-II should be fairly straight forward
and trouble free. Unfortunately, it isn't. Although the manual
presents detailed instruction on how to administer and score the
LET-II, these instructions are at times unclear and worse, they
When administering the LET-II immediate recall subtests, the
examiner presents the visual stimulus items for 2 seconds while the
Auditory stimulus items are read 1 per second. After the child
repeats the items back to the examiner, s/he is asked to perform some
interference task ("Count from 2 to 12"). The manual states that some
children have difficulty with the counting interference task, but
examiners are not instructed what to do if the person can't perform
Recording responses is addressed on pages 36-37 of the manual.
Unfortunately the examples used on these pages contain errors in the
scoring. (In fact, there are two errors in this section). When
explaining how to score an item for which the child has extended the
string beyond the length, the examiner is told to disregard any
letter beyond the proper length. The example given is for a child
presented the string "Q-R-H-X"and who responds "Q-R-X-H-Y." Scoring
is for both ordered (correct placement of letter) and unordered
(correct letter, regardless of placement). The manual states:
"Scoring of items would end with the fourth item in the student's
response. Both Ordered and Unordered raw score would be four."
(italics added). Gee, I thought the score would be 2 and 4
The second error on the page is in the notation of the sequence.
Letters that a child gives that are not in the presented string are
circled to indicate such an error has occurred. The examples on page
37 have letters that were included in the string circled to indicate
that they were not included in the string!! But wait...the same
notation error is made again on page 72 for a different protocol! And
on page 76, another error, but this time in a different notation. Can
anyone get this right?
Okay, I'm done with errors, but I stopped looking.
How about just wanting to be average. Examining the norms tables
for ages 5 to 20, it is interesting to note that a child cannot
obtain an "average" scaled score of 10 a large percentage of the
time. For example, in the Visual modality tables, there is no raw
score to scaled score equivalent of 10 in 58% of the cases. In the
Auditory Modality tables the child will fare better. Here there is no
raw score to scaled score equivalent of 10 in 56% of the cases. What
do raw scores convert to then if not average? Often the converted
score is 8 or 11. A 3 point difference based on 1 raw score point.
Three points! One standard deviation!
So what's my point? I guess it's that you can't tell a test by
it's advertising. No matter how good a test looks; or what company
publishes it; or that it is published at all, we as examiners are
often left to critique the tests we buy. In fact, that is our ethical
responsibility. If we use it , we are responsible for defending the
reliability and validity of it's use. After spending lots of money on
a test, it's too bad how often we can be disappointed by the product.
As far as the LET-II is concerned, I'd follow Lennon and
McCartney's refrain "Let it be. Let it be." As for all others, Caveat
Emptor (Buyer beware).