[Editor's note: Thanks to the author for submitting this manuscript in HTML format]

Distinguishing Science-Related Variations in the Causal Universal of College Students' Worldviews
by
William W. Cobern, Ph.D.
Associate Professor of Science Education
Education and Professional Development
Western Michigan University
2112 Sangren Hall
Kalamazoo, MI 49008-5192
Voice: +616.387.2255
FAX: +616.387.2882
Internet: bill.cobern@wmich.edu
         http://www.wmich.edu/slcsp/slcsp.htm
         (Scientific Literacy and Cultural Studies Project)
 

Abstract

An investigation derived from the logico-structural theory of worldview was conducted for the purpose of examining the relationship between science interest and variations in the Causal universal within college students' worldviews. This required the development of a special pen-and-paper instrument for detecting worldview variations in the Causal universal. The instrument was based on the assumption that when a student is faced with an unfamiliar phenomenon, he or she is more likely to accept an explanation that is more consistent with his or her worldview than an explanation of the phenomenon that is less consistent. The test involved making a choice between explanations that were scientifically-more and scientifically-less compatible. The test along with a measure of science interest was given to a 120 first year college students. The test alone was given to a group of professional scientists. The results suggested that there was considerable worldview  variation among the students and that students who favored scientifically-more compatible explanations showed higher levels of science interest. It was also found, however, that even the students with science interest were less likely to choose a scientifically-more compatible explanation than were the professional scientists. It was concluded that the investigation lends corroboration to the logico-structural theory of worldview and provides a further rationale for pursuit of research in this area.

Introduction

        There is a long worthy tradition of worldview studies among anthropologists and of recently a growing interest among science education researchers (e.g., Allen, 1995; Lassiter, 1993; Lawrenz & Gray, 1995; Ogunniyi et al., 1995). Traditionally worldview researchers have attempted to identify and compare worldviews by locating within a worldview a single theme considered representative of the worldview (e.g., Pepper, 1942). Although very interesting, the thematic approach leads to oversimplification and thus has not precipitated the research one would expect from a good theory (Jones, 1972). In 1984 the cultural anthropologist Michael Kearney published a theoretical book in which he presented his logico-structural model of world view. Subsequently this model was adapted for research purposes in science education by Cobern (1991). The sensitivity and richness of the logico-structural model stands in marked contrast to the earlier thematic approaches to worldview research. The logico-structural model is a multifaceted representation of worldview thus allowing rational justification for the expectation of intra-worldview variation in the typical school classroom, quite opposite of what one would expect working under a thematic worldview framework (e.g., Proper, Wideen & Ivany, 1988). Of course such an assertion about intra-worldview variation raises the question of empirical evidence. And if the evidence is forthcoming the question then becomes, do these variations actually exert a significant influence on (say ) science attitude as predicted by the theory? The purpose of the study reported on here was, first, to test for the existence of theory-predicted intra-worldview variation and, second, to examine the relationship between worldview variation and science interest.

Theoretical Framework
        Worldview refers to the culturally-dependent, generally subconscious, fundamental organization of the mind. This organization manifests itself as a set of presuppositions or assumptions which predispose one to feel, think, and act in predictable patterns. Kearney (1984, p. 1) refers to worldview as, "culturally organized macrothought: those dynamically inter-related basic assumptions of a people that determine much of their behavior and decision making, as well as organizing much of their body of symbolic creations ... and ethnophilosophy in general." To be rational means to think and act with reason, or in other words to have an explanation or justification for thought and action. Such explanations and justifications ultimately rest upon one's world view, one's presuppositions about the world. Or in other words, a worldview inclines one to a particular way of thinking. According to Kearney (1984, p. 41) a worldview "consists o f basic assumptions and images that provide a more or less coherent, though not necessarily accurate, way of thinking about the world." Specifically, a worldview defines the Self. It sets the boundaries of who and what I am. It also defines everything that is not me, including my relationships to the human and non-human environments. It shapes my view of the universe, my conception of time and of space. It influences my norms and values.

        Experience is useless unless interpreted. Therefore, beginning in childhood, each person interacts with his or her physical and social environment, and through this myriad of environmental interactions, worldview presuppositions are unconsciously constructed. The process occurs over a long period of time, with the formative, childhood years being of most importance. Through the years of schooling, formal education contributes to worldview development; and in turn, a worldview provides a foundation upon which cognitive frameworks are built during the learning process.

        At some point of maturity (e.g., as an adult) the malleability of a worldview begins to decrease. It becomes resilient in the face of change providing an adult with cognitive stability; and yet a worldview has an adaptive function which allows even adults to adjust to new environments. Thus, while worldview presuppositions are strongly held, they are not immutable. The strength with which a mature worldview is held appears to be inversely related to the degree of heterogeneity in a culture. The more heterogeneity, the less strongly a worldview is apt to be held. This process of worldview development and change is what Kearney (1984, p. 3) calls "dialectical constructionism" and it shares much with Piaget's genetic epistemology (1971) as well as with Ausubel's constructionist theory of learning (Ausubel, Novak & Hanesian, 1978). In human mental architecture, worldview is the foundation upon which cognitive and perceptual frameworks are built.

        Kearney notes that the general paradigm used by American anthropologists doing worldview research has been that of theme. The same has been true of science education researchers who have employed the theoretical work of Pepper (1942). The mono-thematic or configurationalist approach is an

"attempt to discover and describe the underlying pattern, configuration, basic personality, ethos, or world view of a society. What all of these concepts have in common is that they refer to an hypothesized mental principle that organizes in a distinctive way nonmaterial elements...of a given society. These mental constructs are assumed to shape social and cultural behavior and the material and nonmaterial results of this behavior." (Kearney, 1984, p. 23)

The difficulty with this approach is that it greatly over-simplifies and thus distorts the concept of world view. For example, using the Pepper scheme some have concluded that Westerners tend to have a mechanistic world view. The tendency may well be factual, but to call the Western worldview mechanistic is to gloss over significant aspects of Westerners that are decidedly non-mechanistic. A similar problem is encountered when one speaks of a scientific worldview (e.g., AAAS, 1993). Most who advocate the development of a scientific worldview as an important goal in science education are essentially advocating a philosophy of scientism (Duschl, 1988). To advocate that all of experience should be interpreted by 20th Century scientific methods and theories is surely to advocate an impoverished view of the world, devoid of aesthetics and theology. What is more, scientism does not fit with reality. Scientism is an monolithic philosophy that glosses over substantial differences among today's scientists, such as the differences between B. F. Skinner and Fritjof Capra, or Owen Gingerich and Carl Sagan. It would be more reasonable to abandon the monothematics, and to instead ask, What aspects of a worldview are relevant to science? Thus, it is more appropriate to speak of scientifically-more and scientifically-less compatible worldviews.

The Logico-Structural model of World View
        In contrast to the thematic approach, Kearney's logico-structural model begins with the idea that a worldview is an organized set of fundamental, cognitive presuppositions about reality. He assumes that this organization is shaped by the,

"internal equilibrium dynamics among [the worldview assumptions]. This means that some of these assumptions and resultant ideas, beliefs, and actions predicated on them are logically and structurally more compatible than others, and that the entire worldview will `strive' toward maximum logical and structural consistency. The second and main force giving coherence and shape to a worldview is the necessity of having to relate to the external environment." (Kearney, 1984, p. 52)

In other words, a worldview tends to be internally consistent, in that presuppositions are logically integrated and universals are structurally integrated; hence, the model is termed logico-structural. A worldview is externally valid in that the human need to relate to the external environment fosters coherence.

        Kearney suggests that all worldviews are a structural composite of seven, basic cognitive categories or universals: Self, NonSelf, Relationship, Classification, Causality, Space, and Time (for a complete discussion of these categories see Cobern, 1991 and Kearney, 1984). These universals he likens to the diagnostic categories used by physicians. "Although the doctor is confronted with a variety of patients, he can presumably describe the most significant medical facts about them in terms of...features common to all patients, e.g., blood pressure, pulse, respiration" (Kearney, 1984, p. 65). In principle groups of people and even individuals can be identified by worldview variations which result from the content variation in worldview universals. Logically consistent presuppositions about reality are the content. Each universal is composed of a hierarchically arranged set (or sets) of assumptions, or presuppositions, at the end o f which is a final absolute presupposition or first order presupposition, an ultimate presupposition beyond which there are no others. One might think of a 1st order presupposition as akin to Aristotle's final cause. At the opposite end, these hierarchies blend into the cognitive frameworks with which educators are more familiar.

        In constructivist theory cognitive frameworks are a crucial factor in the assimilation of new knowledge (Tobin, 1993) and in forming attitudes and interests. It follows then that worldview as the foundation for cognition and perception has a significant influence on learning and attitude development. Where students are of diverse cultural backgrounds there is a prima facie case for considering worldview as a factor in the processes of education. However the logico-structural theory of worldview suggests that even in situations usually considered culturally uniform there is likely to be some worldview variation precisely because worldview is a composite of seven universals. In other words, in any given educational setting many presuppositions within these seven universals will be shared by most students; however, there will likely exist some presuppositions shared by only a few. This intra-worldview variation is potentially of significant influence in science achievement and attitude development.

Limitations and Hypotheses
        In order for one to test for a relationship between intra-worldview variation and science interest the investigator must have some knowledge of how the presuppositions and attributes in various universals correspond to the nature of science. The investigator, in other words, must have some knowledge of what constitutes a scientifically compatible worldview. The status of worldview research unfortunately is not such that reliable information is available for all seven universals. However, it is generally agreed that a scientific explanation must meet certain criteria, therefore the focus of this investigation was limited to the Causal universal. The research questions are modified accordingly. Is there evidence for worldview variation within the Causal universal among culturally uniform college students? If so, is this variation in the Causal universal significantly related to interest in science?

        This paper thus reports on the development of the Test of Preferred Explanations (TOPE) which was designed to test these questions, i.e., for distinguishing science related variations in the Causal universal of college students' worldviews. Collected data was used in testing three null hypotheses:

Instrument Development

        Undoubtedly, there are many ways one could use to distinguish worldview variations among students. For qualitative approaches see Cobern et al. (1995). In this investigation the approach was to develop a paper-and-pen instrument that could be given to a large number of students in a short period of time. The instrument is intended to be a preliminary discriminating device used prior to more incisive, investigative techniques, probably techniques of the ethnographic type, perhaps using an Interview Vee (Ault, Novak & Gowin, 1984). The content of the instrument derives from the contention that a scientifically compatible worldview must include presuppositions in the Causal universal that are appropriate to scientific explanation.

        The primary problematic feature of any instrument designed to discriminate among students according to worldview variations is that the instrument itself must not be a test of scientific knowledge. As explained in earlier work on worldview theory in science education research (Cobern, 1991), being ignorant of scientific concepts does not necessarily indicate a worldview variation. With regard to distinguishing variations in the Causal universal, this can be avoided by making the following assumption:

When a student is faced with an unfamiliar phenomenon, he or she is more likely to accept an explanation that is more consistent with his or her worldview than an explanation of the phenomenon that is less consistent.

If one presents a student with an unfamiliar phenomenon and two explanations, one cast in a scientific style and the other not, one would expect students with scientifically compatible worldviews to choose the first explanation more frequently than students with variant worldviews. This suggests that an effective instrument could be constructed with unfamiliar phenomena as items.

        The construction of TOPE began with the identification of unfamiliar phenomena to be used in the items. Unfortunately, one can never be sure who is familiar with what. An alternative procedure that avoids this problem is to create descriptions of fictitious or quasi-fictitious phenomena. For the current study 28 such descriptions were created. The original instrument used these plus three more descriptions based on factual, but obscure phenomena. The test instructions indicated that the items did not necessarily contain factual information, and therefore the test was not a test of knowledge.

        The investigator assumed that presuppositions amenable to scientific explanation are present in a student's worldview if a student frequently chooses explanations that are scientifically compatible. Thus a scientifically compatible explanation was needed for each item. Obviously the explanations for the fictitious phenomena would be fictitious. The explanations for the obscure phenomena items also needed to be fictitious in order to avoid confounding affects of students who might happen to be knowledgeable about the obscure phenomena. The criteria for designing a fictitious, but scientifically compatible explanation came primarily from Braithwaite's book Scientific Explanations and to a lesser extent Aicken's The Nature of Science. It is important to note that (especially) Braithwaite's positivist view of scientific explanation is not unproblematic. His view nonetheless represents the view of many scientists and is a view prominent in science education (Smolicz & Nunan, 1975).

        According to Braithwaite, an explanation and hypothesis are virtually the same thing. To be acceptable in science they must be empirical and above all, testable. A scientific explanation or hypothesis always involves natural causes and tends to be mechanistic and reductionistic. Thus the scientifically-more compatible explanations were designed to be:

 
Any scientific explanation is also a part of a theoretical structure or system composed of many explanations, generally on different levels of explanatory power. Scientific explanations are not given in isolation (Martin, 1972), however with the exception of one TOPE item that relates experimentation to theory, all of the items contain ad hoc explanations. The notion that explanations should be related to other explanations in an explanatory system is not unique to scientific thinking however, and thus was not included as a criterion for the items in this instrument.

        The foil in each item was a rational explanation designed to be scientifically-less compatible or simply scientifically unacceptable. The criteria for composing such explanations were basically the opposite of the criteria for designing the scientifically-more compatible explanations. An attempt was made to write rational explanations that were holistic rather than reductionistic, plausible though non-testable. The 31 items in the original instrument were primarily written by the investigator. Two physicists and a mathematician offered expert advice for revising items and suggestions for original items. The adjectives scientifically-more and scientifically-less were chosen with purpose. None of the explanations is scientifically compatible because all are fabrications. However, it can be argued that when the explanations are considered hypothetically some are more compatible with science than others.

The Selection of Items
        Having compiled and edited 31 items, the next step was to test their discriminating power. This was done by giving the instrument to participants identified as having a strong or weak scientifically-oriented world view, and retaining only the items that discriminated between the two groups. Scientists and engineers comprised the former group. The group identified as having a weak, scientifically oriented worldview was comprised primarily of non-science students at the University of Sokoto, Nigeria. This identification was deemed sound because these were students raised in a non-scientific, non-technological society who at the university level still had professed little interest in science and had no recent science instruction. The second group also included female secretaries at two American colleges who professed little science interest and who had no recent science instruction. The demographics of this group were such tha t one would expect them to have much less of a scientific orientation than a group of scientists (Vetter & Babco, 1987).

        The test was constructed in three, 31-item formats. In Format A, each item contains a phenomenon description followed by two explanations of opposing style. In Format B each phenomenon description is followed by one explanation randomly chosen to be scientifically-more or less compatible. Format C is identical to Format B except that each item has the opposite explanation. Each item in Formats B and C ends with a five-point scale (1 to 5) of acceptability, where a five indicates a high degree of acceptability.

        The instructions to participants called the instrument a survey, rather than a test, and indicated that the instrument did not call for technically correct responses. At no point was any indication given that this instrument was a part of research in science education. In Format A, the subjects were to choose the one explanation of the two that they found more acceptable. Though the instructions discouraged the use of it, there was also an undecided response. In Formats B and C, participants were to indicate on the scale how acceptable they found the single explanation. Format A was scored by giving one point for each scientifically-more compatible explanation chosen, a half point for an undecided response, and a zero for each scientifically-less compatible explanation chosen. The same basic scoring scheme was used with Formats B and C. On the response scale, four's and five's were scored as one, three as a half point, and one's and two's as a zero.

        An a priori selection criteria for inclusion of an item in the final instrument was set at a 0.4 minimum percentage difference between the science and non-science groups. For example, if on an item 80 percent of the science professors chose the scientifically-more compatible explanation but only 40 percent of the second group chose this explanation, then the item was retained. The decision to use a 0.4 difference was based on Hopkins and Stanley's (1981) discussion of item discriminating power.

        Following this guideline, the initial analysis of the data led to the retention of 14 items. Four other items were very close to the 0.4 difference. Subsequently, these four items were given to a professional scientist who had not participated in the initial study. Based on his comments one of the four items was dropped as too confusing, and the other three were retained after modification to better fit the scientific model of explanation. The final instrument (TOPE) contained seventeen items. Fifteen of these came from Format A, and two from Formats B and C. The Appendix contains the complete instrument (also see Cobern, 1988).

        The process of establishing item validity was concurrent with item development and selection. To be explicit, validity in the study was approached first as a matter of content. The items were judged to have content validity because they were based on the work of experts, i.e., Braithwaite and Aicken. Furthermore, content validity was affirmed by having the items reviewed by a panel of scientists and science educators. In their opinion the items contained accurate examples of scientifically-more and scientifically-less acceptable explanations. Construct validity was affirmed by choosing only those items that discriminated between known groups.

Research Method and Results

        The investigation was conducted with college freshmen instead of younger students both as a matter of logistical convenience, and to minimize confounding factors such as reading ability. In the fall of 1987, 120 first year students at a southern liberal arts college completed TOPE. These students were enrolled in a required first year, general education course and represented just under half of the first year class. As an indicator of science interest the students were asked to examine a list of college disciplines and check the one or two disciplines that most interested them as possible majors, and the one or two of least interest. The two questions were combined and scored as 1 for any science area being checked on the first list but not the second, 0.5 for science not being checked on either list, or 0 for science being checked on the second list but not the first. The students were thus divided into three groups: those with explicitly stated science interest, those with no inidcation of science interest (interpreted as some interest in science), and those who explicitly state no science interest.

        Copies of TOPE were also sent to 200 scientists randomly selected from a directory of American scientists. Of these 88 usable, completed tests were returned (44%). The theory predicted that the mean TOPE scores of the student groups and of the professional scientists would be significantly different, and that TOPE scores would increase with interest. The null hypotheses were tested using a one-way ANOVA procedure and T-tests between cell means (Walonick, 1986). The p value for rejection was set at 0.01. The results are given in Tables 1, 2, and 3.

 
 

Discussion of Results

        Based on the student scores, the test-retest reliability was calculated to be 0.81 (a small random sample of students agreed to re take the test three weeks later). The three null hypotheses were rejected at p =< 0.01. The expected order of results was confirmed. The mean score for professional scientists was highest followed by the mean score of students with high science interest. The students with low science interest had the lowest mean score on TOPE. Given the cursory gauge of science interest the positive results are encouraging. One would expect to find even greater differences with science interest measured by a more sophisticated instrument.

        The group mean for scientists is lower than what the researcher expected. One factor to consider is that the scientists in the study were largely from four-year liberal arts colleges. Working in more interdisciplinary environments and with less emphasis on research, many could be more open to different explanation styles than (say) scientists at research universities. Another possibility is that the lower than anticipated scores indicate a lack of instrument sensitivity. Redoing the study using a group of research scientists at research universities would help to answer the question. It must be remembered however that the theory does not demand that scientists always choose scientifically compatible explanations. The Causal universal is influenced by the presuppositions that form the other universals, particularly those in Classification and the NonSelf. In other words, even scientists are likely to have more than one notion of causality.

Conclusion

        The objectives of the investigation were achieved. The very fact that TOPE was successfully developed based on differences among people corroborates the worldview theory inference of worldview variation. Secondly, the analysis of TOPE scores by science interest among students and by the comparison of student TOPE scores with the TOPE scores of professional scientists corroborates the inference that worldview variation can significantly influence science interest. The results raise interesting questions specifically about origin and direction of effect. The results indicate that students who dislike science tend to prefer nonscientific explanation styles and vice versa. Given the dialectical constructivist approach to worldview development one would want to investigate family, socio-cultural, and educational factors that give rise to these preferences. Furthermore, one is prompted to ask whether our conception of science and method of teaching science is too inflexible to attract the interest of students who have very different ways of understanding the world around them. Thus, the investigation both lends corroboration to the logico-structural theory of worldview and provides further research questions.

References