Ronald P. Carver

A Critical Review of Mathemagenic Behaviors and the Effect of Questions upon the Retention of Prose Materials1

The recent research on the effect of questions upon the retention of prose materials was critically reviewed with respect to four important determining variables—material difficulty, individual ability, learning time, and learning strategy. A detailed analysis of the literature suggested the following: (a) a failure to adequately control for the learning time and strategy variables has rendered the recent research ungeneralizable to important theoretical and applied situations, and (b) the concept of mathemagenic behavior is a barbarism that contributes more to confusion of thinking than to clarity of thinking. It was concluded that behavior should be relegated to a secondary role in future research on learning from prose, and that the investigation of learning strategies should be elevated to a primary role.

Measuring the Primary Effect of Reading: Reading-Storage Technique, Understanding Judgments, and Cloze*

The newly developed reading-storage (RS) type of test was evaluated by comparing it to a modified version of the cloze test in three experiments involving 48 college students reading passages at four levels of difficulty. The experiments involved the following conditions: reading versus non-reading of passages, varying degrees of passage deletion, and varying amounts of interpolated activity between passages and tests. The RS test seemed to be more reliable and equally valid as the cloze test in measuring gain in understanding during reading. The RS type of test appears to offer a better objective technique for measuring the primary effect of reading.

A Test of an Hypothesized Relationship between Learning Time and Amount Learned in School Learning

AbstractIt has been hypothesized that the degree of learning, other things being equal, is a simple function of the amount of time during which the pupil engages actively in learning. This hypothesis was tested with forty-eight students in an introductory psychology course. Classroom examinations were used to indicate amount learned; intelligence tests were used to indicate learning ability, and hours of study time reported by students were used to indicate learning time. When learning ability was controlled, statistically, the correlation between amount learned and learning time was statistically significant, r = .30, a result which provides support for the original hypothesis. Considering the attenuating effect of indicants, it was concluded that learning time is a highly important variable which should be taken into account in investigations of classroom learning.

Analysis of the Chunked Reading Test and Reading Comprehension.

The newly developed Chunked Reading Test was further analyzed by correlating the scores on this test with three other standardized reading tests—Davis Reading Test, Nelson-Denny Reading Test, and Tinker Speed of Reading Test. A rational analysis of the scores on all of the tests suggested that each score could be designated as measuring one of the following three types of variables: efficiency (E), accuracy (A), and rate (R). The tests were administered to 41 college students and the inter-correlations were factor analyzed. Two factors fit the data, and they were readily interpretable as an A factor and an R factor. A single factor fit forced upon the data was readily interpretable as an E factor. The results suggested that: (a) apparent differences among the variables measured by standardized reading tests for adults are more superficial than real, (b) all of the scores on these tests can be interpreted as being valid indicators of individual differences in the efficiency, accuracy, and rate at which thoughts are understood while reading, and (c) empirical support exists for the purported theoretical relationship, E=AR.

Factor Analysis of the Ability to Comprehend Time-Compressed Speech1

Sources of individual differences in the listening comprehension of speech presented at different rates were investigated by factor analyzing 11 variables, including measures of comprehension and field-independence. Twenty-two different tests, two for each variable, were administered to 50 college students. It was concluded that: (a) the cloze type of test includes a large component of variance unrelated to comprehension and quite specific to the technique itself, and (b) the comprehension of highly speeded speech probably involves a perceptual ability to be field-independent in addition to a separate ability involved in comprehending speech at normal rates.

Comments: Some Unbelievably Good Results and Their Implications

In answering questions on prose material, Natkin and Stahler (1969) hypothesized that high arousal would produce low immediate recall and high long-term recall, and that low arousal would produce high immediate recall and low delayed recall. The data they have reported seem to lend excellent support to this hypothesis. However, caution is advised in the acceptance of this evidence, even though Natkin and Stahler report statistical significance in support of their hypothesis. Lykken (1968), in a recent article on statistical significance in research, argues that replication is necessary for confirming hypotheses and that statistical significance is perhaps the least important attribute of a good experiment. Furthermore, Lykken contends that really large effects, differences, or relationships may even argue against the hypothesis being tested. It appears that such a case can be made against Natkin and Stahlers evidence. The low arousal condition used in their research consisted of building an expectancy of adjunct questions in a preliminary condition, and then a continuation of this condition in the succeeding experimental condition (Group B). The high arousal condition was simply the absence of adjunct questions in the preliminary reading condition and inclusion of the adjunct questions in the experimental condition (Group A) . Group B, the low arousal group, had a lower mean on the long-term recall test than they did on the short term. This was an expected result since learning theory would predict higher immediate recall than long-term recall. The most important result, however, is that

Comments: Note on a Schema for Proper Utilization of Multiple Comparisons in Research:

A listening test may be a better indicator of educational potential among the disadvantaged than traditional aptitude measures, and the newly developed listening test may have certain advantages over a traditional listening test. However, the uniqueness coefficient for the listening test with respect to a traditional aptitude test is probably somewhat lower than 50 per cent and the uniqueness coefficient with respect to a traditional listening test is estimated to be only .13. Therefore, it appears reasonable to question the conclusion that the newly developed listening test is highly unique.