Control of socioeconomic status

It would have been desirable for the two communities to have differed only in respect to the variable being investigated: the degree of structure in teaching method.
The structured schools were in an industrial city, with three-family tenement houses typical of the residential areas, but with one rather sizable section of middle-class homes.
The unstructured schools were in a large suburban community, predominantly middle- to upper-middle class, but fringed by an industrial area.
In order to equate the samples on socioeconomic status, we chose schools in both cities on the basis of socioeconomic status of the neighborhoods.
School principals and guidance workers made ratings of the various neighborhoods and the research team made independent observations of houses and dwelling areas.
An objective scale was developed for rating school neighborhoods from these data.
Equal proportions of children in each city were drawn from upper-lower and lower-middle class neighborhoods.
Subjects

Individual differences in maturation and the development of readiness for learning to read indicate that not until the third grade have most children had ample opportunity to demonstrate their capacity for school achievement.
Therefore, third-grade children were chosen as subjects for this study.

For purposes of sample selection only (individual tests were given later) we obtained group test scores of reading achievement and intelligence from school records of the entire third-grade population in each school system.
The subjects for this study were randomly selected from stratified areas of the distribution, one-third as underachievers, one-third medium, and one-third over-achievers.
Children whose reading scores were at least one standard deviation below the regression line of each total third-grade school population were considered under-achievers for the purposes of sample selection.
Over-achievers were at least one standard deviation above the regression line in their school system.
The final sample was not significantly different from a normal distribution in regard to reading achievement or intelligence test scores.
Twenty-four classrooms in twelve unstructured schools furnished 156 cases, 87 boys and 69 girls.
Eight classrooms in three structured schools furnished 72 cases, 36 boys and 36 girls.
Administrative restrictions necessitated the smaller sample size in the structured schools.

It was assumed that the sampling procedure was purely random with respect to the personality variables under investigation.
Rating scale of compulsivity

An interview schedule of open-ended questions and a multiple-choice questionnaire were prepared, and one parent of each of the sample children was seen in the home.
The parent was asked to describe the child's typical behavior in certain standard situations in which there was an opportunity to observe tendencies toward perfectionism in demands upon self and others, irrational conformity to rules, orderliness, punctuality, and need for certainty.
The interviewers were instructed not to suggest answers and, as much as possible, to record the parents' actual words as they described the child's behavior in home situations.

The rating scale of compulsivity was constructed by first perusing the interview records, categorizing all evidence related to compulsivity, then arranging a distribution of such information apart from the case records.
Final ratings were made on the basis of a point system which was developed after studying the distributions of actual behaviors recorded and assigning weight values to each type of behavior that was deviant from the discovered norms.
Children scoring high in compulsivity were those who gave evidence of tension or emotionality in situations where there was lack of organization or conformity to standards and expectations, or who made exaggerated efforts to achieve these goals.
The low compulsive child was one who appeared relatively unconcerned about such matters.
For instance, the following statement was rated low in compulsivity, "She's naturally quite neat about things, but it doesn't bother her at all if her room gets messy.
But she cleans it up very well when I remind her".
Measurement of anxiety

Castaneda, et al revised the Taylor Anxiety Scale for use with children.
The Taylor Scale was adapted from the Minnesota Multiphastic Personality Inventory, with item selection based upon clinical definitions of anxiety.
There is much research evidence to validate the use of the instrument in differentiating individuals who are likely to manifest anxiety in varying degrees.
Reliability and validation work with the Children's Anxiety Scale by Castaneda, et al demonstrated results closely similar to the findings with the adult scale.
Although the Taylor Scale was designed as a group testing device, in this study it was individually administered by psychologically trained workers who established rapport and assisted the children in reading the items.
Relationship of Anxiety to compulsivity

The question may be raised whether or not we are dealing with a common factor in anxiety and compulsivity.
The two ratings yield a correlation of, which is not significantly different from zero;
therefore, we have measured two different characteristics.
In theory, compulsive behavior is a way of diminishing anxiety, and one might expect a negative association except for the possibility that for many children the obsessive-compulsive defenses are not sufficient to quell the amount of anxiety they suffer.
The issue of interaction between anxiety and compulsivity will be taken up later.
Criterion measurement

In the primary grades, reading permeates almost every aspect of school progress, and the children's early experiences of success or failure in learning to read often set a pattern of total achievement that is relatively enduring throughout the following years.
In establishing criterion measurements, it was therefore thought best to broaden the scope beyond the reading act itself.
The predicted interaction effect should, if potent, extend its influence over all academic achievement.

The Stanford Achievement Test, Form J, was administered by classroom teachers, consisting of a battery of six sub-tests: Paragraph Meaning, Word Meaning, Spelling, Language, Arithmetic Computation, and Arithmetic Reasoning.
All of these sub-tests involve reading except Arithmetic Computation.
Scores are stated in grade-equivalents on a national norm.
The battery median grade-equivalent was used in data analysis in this study.

The Wechsler Intelligence Scale For Children was administered to each sample third-grade child by a clinical worker.
The relationship of intelligence test scores to school achievement is a well-established fact (in this case, Af);
therefore, in the investigation of the present hypothesis, it was necessary to control this factor.

The criterion score used in the statistical analysis is an index of over- or under-achievement.
It is the discrepancy between the actual attained achievement test score and the score that would be predicted by the I.Q.
For example, on the basis of the regression equation, a child with an I.Q. of 120 in this sample would be expected to earn an achievement test score of 4.8 (grade equivalent).
If a child with an I.Q. of 120 scored 5.5 in achievement, his discrepancy score would be, representing of one year of over-achievement.
A child with an I.Q. of 98 would be expected to earn an achievement test score of 3.5.
If such a child scored 3.0, his discrepancy score would be -.5, representing of one year of under-achievement.
In this manner, the factors measured by the intelligence test were controlled, allowing discovered differences in achievement to be interpreted as resulting from other variables.

Results

test of interaction of compulsivity and teaching methods

Tables 1 and 2 present the results of the statistical analysis of the data when compulsivity is used as the descriptive variable.
Figure 1 portrays the mean achievement scores of each sub-group graphically.
First of all, as we had surmised, the highly compulsive children in the structured setting score significantly better (Af) on achievement than do similar children in the unstructured schools.
It can be seen too that when we contrast levels of compulsivity within the structured schools, the high compulsive children do better (Af).
No significant difference was found in achievement between high and low compulsive children within the unstructured school.
The hypothesis of there being an interaction between compulsivity and teaching method was supported, in this case, at the level.

While we had expected that compulsive children in the unstructured school setting would have difficulty when compared to those in the structured, we were surprised to find that the achievement of the high compulsives within the schools where the whole-word method is used in beginning reading compares favorably with that of the low compulsives.
Indeed their achievement scores were somewhat better on an absolute basis although the difference was not significant.
We speculate that compulsives in the unstructured schools are under greater strain because of the lack of systemization in their school setting, but that their need to organize (for comfort) is so intense that they struggle to induce the phonic rules and achieve in spite of the lack of direction from the environment.

It is interesting to note that medium compulsives in the unstructured schools made the lowest achievement scores (although not significantly lower).
Possibly their compulsivity was not strong enough to cause them to build their own structure.

Our conjecture is, then, that regardless of the manner in which school lessons are taught, the compulsive child accentuates those elements of each lesson that aid him in systematizing his work.
When helped by a high degree of structure in lesson presentation, then, and only then, does such a child attain unusual success.
Test of interaction of anxiety and teaching methods

The statistical analyses of achievement in relation to anxiety and teaching methods and the interactions of the two are presented in Tables 3 and 4.
Figure 2 is a graph of the mean achievement scores of each group.
As predicted, the highly anxious children in the unstructured schools score more poorly (Af) than those in the structured schools.
The interaction effect, which is significant at the level, can be seen best in the contrast of mean scores.
While high anxiety children achieve significantly less well (Af) in the unstructured school than do low anxiety children, they appear to do at least as well as the average in the structured classroom.

The most striking aspect of the interaction demonstrated is the marked decrement in performance suffered by the highly anxious children in unstructured schools.
According to the theory proposed, this is a consequence of the severe condition of perceived threat that persists unabated for the anxious child in an ambiguous sort of school environment.
The fact that such threat is potent in the beginning reading lessons is thought to be a vital factor in the continued pattern of failure or under-achievement these children exhibit.
The child with high anxiety may first direct his anxiety-released energy toward achievement, but because his distress severely reduces the abilities of discrimination and memorization of complex symbols, the child may fail in his initial attempts to master the problem.
Failure confirms the threat, and the intensity of anxiety is increased as the required learning becomes more difficult, so that by the time the child reaches the third grade the decrement in performance is pronounced.

The individual with high anxiety in the structured classroom may approach the learning task with the same increased energy and lowered powers of discrimination.
But the symbols he is asked to learn are simple.
As shown earlier, the highly anxious individual may be superior in his memorizing of simple elements.
Success reduces the prospect of threat and his powers of discrimination are improved.
By the time the child first attacks the actual problem of reading, he is completely familiar and at ease with all of the elements of words.
Apparently academic challenge in the structured setting creates an optimum of stress so that the child with high anxiety is able to achieve because he is aroused to an energetic state without becoming confused or panicked.

Sarason et al present evidence that the anxious child will suffer in the test-like situation, and that his performance will be impaired unless he receives supporting and accepting treatment from the teacher.
Although the present study was not a direct replication of their investigations, the results do not confirm their conclusion.
Observers, in the two school systems studied here, judged the teachers in the structured schools to be more impersonal and demanding, while the atmosphere in the unstructured schools was judged to be more supporting and accepting.
Yet the highly anxious child suffered a tremendous disadvantage only in the unstructured school, and performed as well or better than average in the structured setting.