PROCEDURES OF THE STUDY
The Experimental Plan
This study utilizes an experimental plan known as
a complex factorial design.1 The design is presented
schematically in Figure I. Referring to
the front face of Figure I, it is noted that there are three variables; therapy,
age and education. Therapy is varied in three ways, age is varied in two
ways, and education is varied in two ways. Thus, there are twelve separate
conditions (or cells) which vary systematically from those (in the first
cell) who are characterized as being in therapy group one and age group one
and educational group one, to those (in the last cell) who represent the
combination of therapy group three and age group two and educational group
two.
Within each of these twelve cells (or sets of conditions)
there are three numbers representing three persons—all of whom meet
the conditions for the particular cell.
Figure I represents three
dimensions. The depth in the design depicts simultaneous measurement
in the areas of intelligence, personality, and mathematics. The same
subjects, then, are measured in each of these areas.
In any area, the score for an individual is entered into
its appropriate place within the particular cell, which characterizes him.
The score for each individual indicates the quantity
of change having occurred through the interval between testings.
Selection of Subjects
The dianetic center publicized rather widely (in
newspapers and correspondence) the advent of a new series of sessions of
dianetic therapy, and called for a meeting of all those interested in
participating. At this meeting, the director first talked generally
about dianetic therapy. He pointed out that this next series was planned
for the following two-month period. He requested that only those apply
who could definitely set aside a number of hours each week during this period.
The director then discussed the cost of this series. He asked
all those who could fulfill the obligation of time and money to come to the
secretary at the end of the meeting for the purpose of recording names,
addresses, and free times.
After the meeting, letters were sent to the first twenty-four
applicants, notifying them that another meeting would be held for the purpose
of routine psychological testing. No other selective device was
utilized.
The Test Materials
Interview
The examiner brought
to each personal interview a prepared sheet, which called for the name, amount
of previous exposure to dianetic therapy, date of birth, and educational
history for each subject. Although this statistical information might
have been obtained with less trouble by including a specially prepared form
with the regular test materials, the interview served as a vehicle for another
purpose. The situation provided an opportunity to stimulate motivation.
This was attempted by impressing upon the subjects the idea that maximal
effort would result in test results that the dianetic center would be able
to use to plan his therapeutic procedure to give him greater benefit.
The results of the interview
and tests were not made available to the center until after the completion
of the study.
Tests of Intellectual Functioning
There is a high intercorrelation among most of the
standard tests of intellectual functioning. Because of their variability
of content, however, it is desirable to have more than one measure so that
the mean result will be more valid and reliable in terms of internal ecological
considerations.2 That is to say, the combined
score is a more representative measure than either of its components.
The first test in this area was the SRA Non-Verbal
Form.3 The alternate form for this test is
the SRA Verbal Form.4 The forms are highly
correlated and both show significant validity and
reliability.5,6 The Non-Verbal
Form was given in the first testing situation and the Verbal Form
in the second.
The second test in this area was the Revised Alpha
Examination Form.7 The alternate
form for this test is the Revised Alpha Examination
Form.8 Both are highly correlated and
show significant validity and
reliability.9,10,11
Form 5 was given in the first testing situation, and Form 7
in the second.
Since both of these types of tests must be taken into
account for better representativeness in the area of intellectual functioning,
some combination of their scores is necessary. The SRA
Manual12 gives enough data for the calculation
of normative standard deviations, as does the Wells’
Manual.13 Since both of these error terms
reflect the variation of a normal population, the difference between them
is due mostly to differences in test construction. Then, the standard
scores are comparable—being corrected for differences in test construction.
The raw scores were converted, by means of the appropriate standard
deviation, into standard scores, and these were combined for each subject
to represent his performances in the area of intellectual functioning.
Tests of Arithmetical Ability
In the area of mathematical ability, test constructors
have taken cognizance of the factors of manipulation of fundamentals, and
special reasoning processes.
This kind of reasoning was measured by the Arithmetical
Reasoning
Test.14,15 This
test has alternate forms (A and B). The forms are highly correlated
and both show significant validity and
reliability.16 Form A was administered
in the first testing situation and Form B in the second.
The manipulation of fundamentals was measured by the
Schorling-Clark-Potter Hundred Problem Arithmetic
Test.17,18 This
test has alternate forms (V and W). The forms are highly correlated
and both show significant validity and
reliability.19,20 Form
V was administered in the first testing situation and Form W in the
second.
Since both of these factors must enter into any consideration
of arithmetical ability, some combination of them would best represent
performance in this area. Thus, it was necessary to find some means
of equating the two tests. The Schorling
Manual21 presents normative standard deviations
while the Cardall Manual22 gives enough data
for these to be calculated. Since these error terms both reflect the
variation of normal population, the difference between them is due mostly
to differences in test construction. Then, the standard scores are
comparable—being corrected for differences in test construction. The
raw scores were converted, by means of the appropriate standard deviation,
into standard scores and these were then combined for each subject to represent
his performance in the area of arithmetical ability.
Test of Personality Conflicts
To measure personality conflicts, the test chosen
was Rotter’s Incomplete Sentence Blank — Adult
Form.23 This provided a valid and
reliable score which indicated the effect of the intensity of conflicts in
personality.24
The area, which subsumes personality conflicts, is probably
the least clearly delineated in psychology. This emphasizes the need
for representativeness in measurement. However, a study of the literature
of available tests did not reveal any two-group tests whose scores were
comparable. Thus, the choice was narrowed in a single measure.
The advantages of Rotter’s form were that it was
specifically designed to measure personality conflicts, and that it presented
more difficulty than the other tests to the subjects in anticipating what
was being scored.
General Remarks
It will be noted that the tests chosen have met the
criteria of being practical for group administration, having equivalent alternate
forms, and being both valid and reliable measures. Group tests were
used because the time involved in the administration of individual measures
constituted an undue interference with the dianetic center’s schedules.
Alternate forms were used because, in the retest situation, it was
desirable to avoid the complications that arise with increasing familiarity
with the test material. The insistence upon the criterion of a high
degree of validity and reliability was pronounced beyond the levels, which
are usually set. This was desirable in that it provided a finer degree
of measurement so that subtle variations in change, if present, would be
isolated in the extremely refined statistical analysis.
Each of the tests in this study has a manual with specific
directions for administration. These were followed exactly.
The Testing Situations
The dianetic center offered the use of a large auditorium
for the test sessions. The arms of the chairs were equipped with writing
surfaces. There was enough room for the subjects to be seated both
a row and a seat apart to forestall collaboration.
The subjects were tested simultaneously at one uninterrupted
session.
Tabulation of Data
The tests were first scored by the experimenter and
then checked independently by two graduate students in psychology who were
enlisted for this purpose.
The same procedure was followed with other data.
The number of therapeutic hours for each subject during
the experimental period was crosschecked. First, this information was
kept as a continuous record by the dianetic center. Second, this
information was obtained directly from the subjects during the second testing
session, after the therapeutic interval, during the un-timed Rotter test.
Statistical Treatment of the Data
The method for the statistical treatment of the data – the analysis of variance of a complex factorial design – was chosen for three reasons: (1) It affords the maximum surety of the result with the smallest number of cases, (2) It enables an analysis of the interactions of the variables with maximum surety of the result because of its simultaneous nature, and (3) It is a refined technique which is sensitive to slight changes.25
Difference Scores
For each subject, in each area of measurement, there
was a first testing session score and a second testing session score. The
first score was subtracted from the second. Thus, a positive difference
indicated a greater numerical performance in the second test. A zero
difference indicated that the performance of the first test was the same
as the performance on the second. A negative difference indicated a
lesser numerical performance on the second test.
Scanning the array of difference scores for all subjects in each area,
the greatest negative value was noted. Then, the value of plus one
was added to the real value (disregarding sign) of this greatest negative
and the resultant number was taken as a constant to add to each difference
score in the array for each area. Thus, these final coded values preserved
the relative amounts of change among the subjects for each area. The
coding also took away all negative values; a condition necessary for the
statistical analysis.26
The coded scores were then entered into a table of analysis
for each area similar to the front face of the design represented in
Figure I.
Prerequisite Test of Homogeneity
Within each area of measurement, the main extractable
variables (age, education, therapy and random sequence) were each subjected
to the test for homogeneity of variance.27
This is a necessary condition, which must obtain in the data before
the extraction and analysis of variances.28
In all of the test of homogeneity, except for one, the
hypothesis was upheld. In this case, the data was transformed in
scale29 until homogeneity was found a tenable
hypothesis.
Extraction of Variances
For each of the areas of measurement, the variances
were extracted and tabled.
When this was completed, a summary of the variances in
each area of measurement was tabled.
Test of Variances
The choice of an appropriate error term with which
to test the mean variances of the summary tables depended upon the possible
of hypotheses, which might derive from the experimental design.
The first possible error term is that of the highest
order interaction.30 However, this makes
the assumption that the categories within each of the controlling variables
is a random selection.31 This assumption
had not been met in this study, and that error term was discarded. The
other possible error term is that of the residual mean
variance.32 The use of this error term
confines speculation to these particular age categories, these particular
educational categories, and these particular therapy categories. It
provides no test for speculations beyond the limits actually incorporated
in the raw data.33
The mean variances for the variables and their interactions
(for each summary table) could be tested against the appropriate residual
variance. However, since this error term was able to be broken down
into two components (variance due to random variation and residual error
variance), a finer test of the difference is afforded by using the residual
error after the extraction of the sampling error. This was done and
the results incorporated into the summary tables.
A Brief Note
This study was designed to afford an objective test of the claims for dianetic therapy, and to do this with definitiveness. It provided for adequate information without anticipating the direction of the effects of dianetic therapy. The data derived permitted an extensive analysis of the therapy because of the range of the measured controlling factors. Since dianetic claims only specifically emphasize the areas of mathematical ability, intellectual functioning, and personality conflicts, this study utilized standardized tests, which were especially designed to measure these areas. The total design is somewhat complex, but an attempt was made to clarify it by representing it diagrammatically (see Figure I).