Scientific Research Expertise 
from Middle School to Professional Practice

 

Wolff-Michael Roth (MROTH@UVIC.CA)

Lansdowne Professor
Applied Cognitive Science
Faculty of Education, University of Victoria, Victoria, BC, V8W 3N4 Canada
Symposium presentation at the 1999 annual meeting of the American Educational Research Association, Montréal, Québec.

Abstract

In this paper, I report studies of experimenting among three different populations: middle school students (Grade 7 and 8), university science graduates in a science teacher education program, and scientists. The middle school students and scientists were observed in `natural' situations where individuals and small groups designed and conducted investigations, the outcome of which they reported to a larger community. The university science graduates were asked to design and conduct an investigation for which they had 7 hours spread over two days. Unlike the other two population, the science graduates did not engage in research as a regular part of their daily activities. Our results show that the science graduates designed experiments, operationalized variables, and represented scientific results in ways that were not much different from what the much younger middle school students had done. Because of the considerable competencies of the Grade 8 students who had learned to conduct their own research, we therefore have some indications that what we learned from our studies of scientists can be transported into middle school classrooms.


Introduction

Ethnographic research in scientific laboratories and scientific field work showed that designing investigations, collecting data, transforming data, and interpreting the resulting representations are some of the quintessential scientific practices (Latour, 1993; Roth & Bowen, 1999)--though, unlike previously theorized, these practices cannot be thought separately from each other and are highly contingent structured activities that are correlated with a person's familiarity in the field of inquiry. Recent reform documents have increasingly called for such "authentic" practices in mathematics and science education which would allow students to engage in these subjects in ways that correspond to everyday practices in these fields (AAAS, 1993; NCTM, 1989). (I am including mathematics, for a considerable number of scienfic practices are related to the representations they develop and ultimately include in their publications.) For example, mathematics curricula in Grades 5-8 should enable students to (NCTM, 1989):
* describe and represent relationships with tables, graphs, and rules; (p. 98)
* analyze functional relationships to explain how a change in one quantity results in a change in another; (p. 98)
* systematically collect, organize, and describe data; (p. 105)
* estimate, make, and use measurements to describe an dcompare phenomena; (p. 116)
* construct, read, and interpret tables, charts, and graphs; (p. 105)
* make inferences and convincing arguments that are based on data analysis; (p. 105)
* evaluate arguments that are based on data analysis; (p. 105)
* represent situations and number patterns with tables, graphs, verbal rules, and equations and explore the interrelationships of these representations; (p. 102) and
* analyze tables and graphs to identify properties and relationships. (p. 102)
Because these competencies focus on data collection, analysis, and presentation, they are easily integrated with science curriculum reform at the same grade levels. In fact, the integration of mathematics and science school activities may not only be interesting because children collect their own data, but they may be essential for developing a thick layer of experiential knowledge that underlies much of scientists' understandings (Roth, 1996; Roth & Bowen, in press; Roth, Masciotra, & Bowen, 1998). Such integration of rich experiences with physical phenomena and subsequent transformation and analysis of the data appears to lead to robust mathematical and scientific understandings of phenomena (Greeno, 1988; Roth, Masciotra, & Bowen, 1998; Roth & McGinn, 1998).
To date, many science and mathematics teachers have not yet realized the potential that lies in situating science and mathematics education in students' self-directed inquiries about natural environments as a way to implement standards such as those of NCTM. More so, there is some evidence that science teachers may not enact competent data interpretation themselves (Roth, McGinn, & Bowen, 1998) making it difficult for them to begin such practices in their own classroom and even less to scaffold students into these practices. In the past, all of our research studies in which students developed competent research practices always included science teachers who, at a minimum, had obtained a masters degree in a natural science. But we were cognizant of the fact that our own curriculum innovations may not scale to the regular classroom because of teachers' lack of experience in scientific practices. My presentation today is therefore fundamentally concerned with the question, "What are the practices enacted by Grade 8 students, B.Sc. graduates in a secondary science teacher education program, and scientists during open-ended inquiry?"

Lab Science: In Schools and in Praxis

Traditional Science Teaching

When students do laboratory work, science comes out short: students do learn something, but that often has more to do with turning a set of instructions into an accounted-for course of action involving laboratory equipment (Amerine & Bilmes, 1990; Lynch, Livingston, & Garfinkel, 1983) than it does with actually learning any science. In fact, students often find themselves in a double bind situation which prevents many of them to learn: To know whether they have followed the laboratory instructions they need to know that what they observed is what they should have observed; however, to know whether what they observed is what they should have observed, they need to know that what they have done is what they should have done (Roth, McRobbie, Lucase, & Boutonné, 1997a). The practices tapped and developed by such laboratory tasks are far less theoretical scientific than practical (and possibly routoine) in nature. A vivid description of current science teaching was provided by ethnographic studies of high school science in Australia and the United States (Gallagher & Tobin, 1987; Tobin & Gallagher, 1987).
Several problematic elements were identified with laboratory teaching. First, laboratory investigations embody a "cookbook" approach. Students (attempt to) follow recipes, gathering and recording data without a clear sense of the purposes, the procedures, or the interconnections between the two. Second, investigations provide low cognitive demands that preclude reflective action by means of which students might learn to express themselves about phenomena of scientific interest. Third, students engage in activities not intended by the curriculum planners, spending much of their laboratory time in off-task activity with short periods of attention to complete the work. Time off-task is used for non-science related socialization with peers. To make up for the time lost in these activities, teachers set a high pace in regular classroom periods to transmit even more factual information. Fourth, in common practice, laboratory activities are treated as if students could see the conceptual structure of the discipline, that they could discover the nature of nature; this despite the generally accepted understanding that all observation is theory-laden and that therefore students will, without knowing the science they are to learn, perceive phenomena in the ways necessary to construct the concepts in the first place. (We documented elsewhere the differences between students' perception of a science demonstration and students' efforts in making their perception consistent with teacher talk about the phenomena that appeared incongruent [Roth, McRobbie, Lucas, & Boutonné, 1997b].) Thus, laboratory activities in high school are all too often ritualized verification exercises in which the myths of science as fact and sets of rational scientific process skills are perpetuated.
The results of such teaching are held accountable for the many students that drop science in their early high school years: High school science years have been described as "rites of passage" with the sole purpose of selecting students for science programs at the university (Brookhart Costa, 1993). At the same time, those who are not successful, learn that science is for a special breed of smart people, "geeks," and "geniuses." While recent research in science and technology studies has largely debunked the myth of scientists as a special breed, many ordinary people uncritically accept the products of science and technology (Postman, 1992). However, we would expect members of a scientifically and technologically advanced society to make choices informed by a critical view of scientific and technological knowledge. As part of this science- and technology-related literacy, one would expect students to experience and learn how scientists and engineers produce new (arti)facts. That is, students should experience not only ready-made science (and technology) but science-in-the-making, to use the Latourian (1987) notions of science as both product and process. We know from our own work in school science that if such an experience is complemented with critical reflection on the process of knowledge construction, students are in a much better position to relativize scientific laboratories as a special place, an enhanced environment in which natural order is improved (Knorr-Cetina, 1992; Latour, 1983).

Science is Everyday Practice

The ontology and epistemology portrayed by traditional science teaching are clearly at odds with recent field studies in scientific laboratories (Knorr-Cetina, 1981; Latour & Woolgar, 1986; Law, 1994; Lynch, 1985; Traweek, 1988). These studies show the tenuousness of the assumptions underlying the conceptions of teaching as outlined in the previous section. Little has changed in classrooms over the past two decades since the emergence of an increasing number of studies that show the socially constructed nature of scientific knowledge at various levels of analysis. Rather than being pure, science is highly meshed with other aspects of daily life, politics, economics, social relations, power, etc. Knowledge construction in scientific laboratories therefore has a strong local, indexical, and serendipitous character; at the same time, knowledge construction depends on relationships with funding agencies, other researchers, those in charge at research sites, and the public, thus illustrating its macrosocial nature.
One would expect that with a better understanding of the nature of science, scientific facts and theories, and scientists' everyday work, public policy makers and even just plain folks would be able to make more informed decisions about the priorities for spending public moneys. Schools are places where such better understandings can be fostered. To bring about change, the above-cited aspects of everyday practice in science and engineering should be shared with children. However, "telling children how scientists do science does not necessarily lead to far-reaching changes in how children do science; indeed, it cannot, as long as the school curriculum is based on verbally expressed formal knowledge" (Papert, 1991, pp. 10-11). Suzuki (1989) makes an explicit link between the quality of science education and being able to function as responsible citizens in an increasingly sociotechnical world. He thus suggests changes in science education: we do not need to inculcate children with facts, but foster in them an excitement through activities that engage them in exploring, discovering, and connecting.

Background

Over the past decade, I conducted (sometimes together with my teacher colleague and later graduate student Michael Bowen, others with university colleagues) a series of studies on the experimental and representational practices of middle and high school students (e.g., Roth, 1994, 1995; Roth & Bowen, 1993, 1994, 1995; Roth, McRobbie, Lucas, & Boutonné, 1997; Roth & Roychoudhury, 1993), B.Sc. graduates who were enrolled in a fifth-year secondary science teacher preparation program (Bowen & Roth, 1999; Roth, McGinn, & Bowen, 1998), and scientists (doctoral students, postdocs) (Bowen & Roth, 1998; Roth & Bowen, 1998; Roth, Masciotra, & Bowen, 1998). Here, I want to focus on those studies involving:
* Grade 8 students involved in an innovative science program that allowed them to construct their own research questions, and design the studies to answer these questions. Their only requirements were that the reports they constructed about and from their research should be convincing as judged by their peers who conducted similar or other research on the 50 acre property of the school. The students attended a private school in its first year of transition to a coeducational institution. By Grade 8, students of this school did not yet academically distinguish themselves from those who attended the surrounding publish schools (other than that, on the average, they came from more well-to-do homes). These students conducted their own research for a 10 week period in the course of which they became increasingly competent in the different aspects of doing research. The data from this part involves a continuous video record of the 10-week session from 2 Grade 8 sections, all written work, the results of experimental design research to test research hypotheses about scientific representation practices students had developed, and interviews.
* Preservice secondary science teachers all of whom had previously obtained undergraduate or graduate degrees, most of them in the biological sciences. As part of a "methods" course on the teaching of secondary science, these students conducted, over the course of 10 days, a research of their own design in a designated 1-acre area of mixed vegetation on the university campus. The data collected in this case consists of extensive written reports submitted at the end of the project. Furthermore, additional data were obtained in several sessions asking students to interpret data sets, and transform data, and to construct scientific representations.
* Together with Michael Bowen, I conducted a 18-month ethnography among ecologists beginning one spring (preparation), covering two summers, and into the second fall. As research assistants, we helped in capturing reptiles (lizards, skinks, rubber boas, garter snakes) and thereby apprenticed to the research of a herpetologist. We also attended local, national and international conferences with our participants, seminar presentations, conducted formal interviews with one participant (10 hours), and kept field notes on the many informal meetings with members of the ecology community. Videotapes, copies of papers, audiotapes, copies of field notes, etc. are part of our data base.

Grade 8 Students Enacting Science

Learning Process

Traditional forms of science education are based on the assumption that language, tools, instructions for an experiment, textbook problems, and other items relevant to teaching have singular meanings. Consequently, instructions should be converted into actions, tools used in canonical ways, unitary meanings learned by definitions: if there is trouble, students are usually blamed for not trying hard enough, malvolition, or cognitive deficiencies. From the same perspective, it is assumed that the laws of nature, concepts, and rules can be read from relevant documents (e.g., lab experiments). Our investigations in open-inquiry settings revealed the problematic nature of these traditional assumptions. (The settings include open ended inquiry, among others, in Grade 11 and 12 physics, Grade 4-5 engineering, Grade 6 and 7 physical science. For book length reports see Roth [1995, 1998].)
a. the events students observed underdetermined the laws of nature to such an extent that the convergence of their language (games) with accepted science discourse was only one of several possibilities;
b. tools did not embed unique cultural meanings that could be unproblematically appropriated by students;
c. textbook and teacher "problems" had multiple meanings that gave rise to different solution-related activities; and
d. the interpretive flexibility of objects, events, and semiotic representations was the very phenomenon which allowed shifts in language games, and thus learning, to occur.
Over a period of time dealing with emergent problems, students learned to cope with the ill-structured nature of the tasks, and in fact, they began to capitalize on interpretive flexibility that is, the fact that language, tools, materials could be interpreted in multiple rather than singular ways--students, because they have not yet been sufficiently disciplined, appear to attribute more different forms of meaning. This resulted in what we described as their ability to tinker, to convert objects, materials, or tools in ways to bring them closer to their answers; objects, materials, or tools were often used in ways for which they were not designed. In fact, one could say that students in these classrooms learned to live with greater indeterminacy and developed greater flexibility in terms of meaning. Therefore, boxes with tools and a steady supply of glue, various types of adhesive tape, cardboard, thermometers, soil corers, soil acidity meters, soil testing kits, rulers, measuring tape, and materials to repair equipment "on the fly" became central to the learning environment.
One of our descriptive metaphors for students was that of tinkerers or bricoleurs whose practices are characterized by their indexical logic, situationally contingent and circumstantial nature, and implied opportunism (Knorr-Cetina, 1981; Papert, 1991). Tinkerers flexibly interpret their settings to constitute problems and solutions as part of making things work. This leads to a conception of problems and problem solving that differs from traditional problem solving research in education and psychology (e.g., Roth & Bowen, 1993). Rather than being prefabricated with a definite structure, problems unfold and emerge from engaged activity and are structured by the dialectic between individuals, setting, and social context. Solution processes cannot be divided into separate means and ends: the formulation of problems and solutions are inseparable and may occur in any order. Because goals are endogenous to the constitution of problems, problems are owned by students rather than given in a normative, decontextualized form by external agents. Students' most significant learning in this context may have been to cope with the unstable and ever shifting characteristics of their problems, solutions, and the setting in which they worked. Rather than expecting a stable and structured world in which they seek "right" answers, students learned to structure and make the most of available physical and social resources, to find answers from a range of possible answers, depending on how they solved the problem they framed. These students learned to make sense of a world in flux rather than expecting rigid bodies of knowledge.
In traditional laboratory exercises, students often go through the activities without knowing what they are doing or why they are doing it; they leave the interpretation until after they have finished and when they have opportunities to "fix" data so they fit the theory.[1] The students in the participating classes, on the other hand, engaged continuously in interpretation and sense-making activities. Their ways of talking emerged from the continuous interaction of measurement procedures, instruments, and current language games in a "mangle of practice" (e.g., Pickering, 1995). With new language games, new ideas and topics became so important that they overrode earlier knowledge claims from their laboratory and field reports. Subsequently, individual claims were also modified to keep pace with changes of the discourse. Our data collection procedures allowed us to track some of these changes as they emerged in the course of students' activities. For example, we showed how Miles, a Grade 8 student doing field research, evovled his description of factors that influenced plant growth in his research site. During a 10-day period and in the course of various conversations with his teacher, research partner, peers, and a laboratory assistant, we noted how a nearby lake first entered his discourse, then progressively became a central referent in his effort to geographically order his sampling sites, to finally becoming a factor that influenced plant growth. At the end of this period of time, he constructed a graph to support his contention that there was a correlation between distance to the lake and plant growth. The following vignette (I am drawing on a case published in Roth & McGinn, 1998) further underscores the way students operated in this learning environment.
Jamie and Miles are two grade 8 students participating in an innovative 10-week science curriculum which asks them to formulate and answer research questions about biotic and abiotic aspects of their own 40 square meter ecozone on the school grounds. Jamie and Miles have chosen their ecozone in a forested area. Earlier in the day they decided to investigate the soil composition in three 1 square meter subplots that would later constitute comparative sites for a series of studies on plant growth. From each of the subplots, they collect samples of top soil. As Jamie and Miles return to the laboratory, they decide to follow the recommendations in one of their resource texts to float each soil sample in a large beaker. After the soil has settled, Jamie and Miles observe three distinct layers of materials; Miles instructs Jamie to make a drawing of the beaker and the layers so that they have a record that will show the relative composition of their samples in the three subplots. Miles measures the height of each layer and Jamie records each measure next to his drawing. Prompted by their teacher's comment to present their data in a way that permits comparison with others who might investigate soil composition, the two boys decide to find the relative contribution ("%-amount") of each layer to the total sample. They encounter some difficulty finding relative amounts so they ask the teacher who engages them in a conversation over and about little sketches they drew in Jamie's field notebook. Jamie then measures the total height of all layers allowing Miles to calculate the relative contribution of each category. They find a chart tin their resource materials that classifies soil according to the different contributions of three composite materials. Miles suggests copying the chart and then marking the compositions of the three samples onto the chart. [Roth & McGinn, 1998, p. xx]
Here, Jamie and Miles' research moved along a trajectory which began with digging up soil and ended with points on a soil chart. Between these beginning and endpoints, Jamie and Miles engaged in a number of practices that transformed real soil into dots on a chart via drawings, height measurements, and percentage calculations. They generated what Latour (1987, 1993) called a cascade of inscriptions of increasingly general nature. Their initial drawing and height measurements were very much a function of the amount of soil they had collected. From this early representation in which the signs were still soil-like, they literally abstracted (Lat. ab-, away and strahere pull) a part which was then embodied in their drawings, and, through continued abstraction ended in a series of calculated numbers and ultimately as singular points on a grid. Their transformation of the layer thickness into a relative height constituted a scaling of raw into normalized measures. Once they had relative amounts of soil types, their results could easily be compared with those of other students and they could be mapped onto the published chart.
Science teachers and parents are often afraid that pupils in our learning environments do not "cover the content" and will not compare well with others on the basis of achievement tests. Our studies show that the diversity of the projects allowed classes as a whole to cover more than the required curriculum. As students began to talk about and defend their products (design artifacts, research and results), they had to elaborate and therefore clarify many ideas (as embodied in the conversations about their work and objects therein) for themselves before they were actually ready to teach others; because peers act as teachers, the distance between the language of teaching and the language of listening is considerably smaller than if the same content is taught by the regular teacher. Furthermore, we observed that during discussions with other research teams in their classes, students learned tremendously; those discursive and tool-related practices which students found relevant rapidly spread throughout, and changed the classroom communities.[2]

Investigative Practices of Future Science Teachers with Science Degrees

To find out what research practices future science teachers with previous science degrees enact, we asked a group of such students to design their own research questions and answer them by collecting data on the university grounds. When the pre-service secondary teachers first entered the research area (located in "undeveloped" mixed forest at the edge of the university property) there was considerable discussion in the student pairs about how they were going to ask "do-able" questions and what those questions would be. As students continued to work on identifying the area in which they were going to conduct their research work, staking out boundaries, and drawing a map of the zone, they started formulating specific questions to address as they noticed more and more specific details of the zone and reflected about the equipment which had been made available to them.
The final questions addressed by the preservice teachers had many similarities to those framed by the Grade 8 students when they first started their outdoor research many being so conceptually "broad" that it would be difficult to address them in a single outdoor session. However, in the final analysis, these students cannot be faulted for not being more competent in the practices, for they had not been familiar with these processes or with the site in which they conducted their research. Nevertheless, let us take a look at the reports which these "experienced" students submitted after their investigations. Because of the wholistic nature of scientific research (e.g., Roth & Roychoudhury, 1993), each of the aspects of research discussed in the following interacts with all of the others. Any problems are likely to ripple through the entire project.

Research Questions and Research Design

The 25 preservice teachers had considerable difficulty in structuring focus questions that were appropriate for the type of activity they were doing in their class. For the particular project these students engaged in, with few exceptions, causal questions would be quite difficult if not impossible. Despite this limitation, of the twenty-four questions addressed by the students fourteen were causal in structure. Causal questions included, for example, "How does the moisture level affect the distribution and height of horsetails in our investigative site?" and "Do the exhaust gases from the cars parking in Lot C directly effect concentration of field flowers in front of the lot?" This type of question is more suitably addressed in an experimental regimen whereby moisture level is manipulated and distribution or growth measured. Similarlily, the second question requires an experiment in which hypotheses about the effects of exhaust gases are explicitly tested. This question is further confounded by the "busy road" which paralleled the parking lot on the other side of the plot on which the students were working. Even a correlational study maybe problematic because of the free movement of "exhaust gases" from the two sources back and forth across the plot of land where the plants were measured. What we can see in these examples is a common practice in everyday reasoning to use correlations, even spurious or contingent (case), as proof for a cause and effect relationship. Here, lacking experience in framing cause-effect type research questions, these students drew on their everyday practice to ask cause and effect questions when the context allowed them only to collect correlational data.
Students also addressed issues that had could be addressed only if the meaning relations differed from those in science. For example, students used signifiers such as "competition", "biodiversity", "growth", and "productivity" all of which are embedded in specific networks of meaning in biology that do not equate to "distribution", counts of limited numbers of organisms, or "height" as they were used by the students. Much like the middle school students in the initial phases of previous studies (Roth & Bowen, 1993), and despite their previous science degrees, these preservice teachers experienced difficulties constructing productive questions to direct their inquiries. Our ongoing research suggests that university science courses do not seem to assisted students in developing the kind of embodied knowledge which allows scientists to make sense of their world (e.g., Bowen & Roth 1998).

Operationalizing Variables

An important step in scientific research is the construction of defensible claims on the basis of the measurements completed. Here, the operationalization of variables mediates the relationships between original questions and claims. Our analyses show that several studies had operationalized their variables such that the results of the investigation could not be used to provide answers to the research questions. In part, this was the result of addressing questions involving biological factors such as "competion" and "biodiversity" and in asking causal questions in contexts where it was not possible to address them. However, problematic operationalization of variables also occurred in situations where these conditions were not present in such a way that they would interfere with that procedure. Two types of problems with operationalization became aparent:
i. measured variable ineffectively reflects conceptual intent of initial question, and
ii. insufficient replication or an inappropriate sampling regime.
The following example illustrates both of these breakdowns. In this example, students asked the research questions, " How does the side of a fallen log affect it's biodiversity?" and "How do the burned portions and the recent ad older exposure of new wood affect the snags biodiversity?" To address both questions one measure of "biodiversity" was the "frequency/quantity" of different types of organisms--lichen, moss; small plant growth (non-lichen, moss); spiders; beetles, larvae; and insects. This provides a constraint on the perspective on biodiversity and gives rise to a conceptual breakdown relating to the combining of dissimilar species within the same cells. There are also difficulties with how the "frequency/quantity" was determined. For instance, a count of "one" of moss/lichen represents a "patch" of indeterminate size, not a measure of individual bodies or surface area of coverage. In another example, a count of "2" insects in a section represents large (macroscopic) insects visible at the surface, not those beneath the surface of the soil or under plants. In these cases, insufficient operationalization and sampling (apart from other problems with that study) meant that even correlational claims would be inappropriate.

Constructing Tables

Field ecologists frequently structure their data analyses by constructing tables which assists them in representing data and defining variables (Roth & Bowen, 1998). Our ongoing ethnography of ecologists suggests that by using tables scientists ensure that they are collecting all relevant information--the tables act as memory aid in the data collection phase of the research. Few of the reports submitted by the preservice teachers made use of this representation device. Of twelve reports, eight used tables for representing their data. Of these eight, four were were structured such that patterns did not emerge. For the four reports that did not use a table, it was our determination that use of a table would have been appropriate for the data collected and would have aided interpretation.

Transforming Data

Supported by the heuristic used to scaffold students' inquiries, 10 (of 12) projects contained transformed data. However, graphs were frequently used in non-standard ways to depict the collected data. For example, there was frequent use of bar graphs rather than the scatter plots scientists would have used. Scatter plots (5) rarely included best-fit lines (2) which we found consistently among Grade 8 students. In other cases, further insights might have been gained if different representations (such as X-Y-Z plots or 3-D bar graphs) had been used instead of the bar, scatter, or point-to-point graphs students actually employed. Many graphs were labeled or structured in ways that later led to confounding interpretations of the graphs. Four graphs did not relate to questions that were being addressed, and in some instances there appeared no reason to construct one of these graphs (such as plotting a bar graph of averages of measures across a slope). In total, six of the ten reports which used graphs had some problems with how they used graphical representations to depict the collected data and this subsequently affected the claims that could be drawn from those representations.

Claims

In their claims, scientists draw implications from the data collected and discusses the data in the context of the original question(s). Several of the reports had conclusions that consistently extended from the data collected and its representations and transformations. However, many other of the reports made claims which either did not relate to the original question(s) or which did not logically extend from the data collected/depicted. Of these two types of problems, the latter is the more problematic and was found in ten of the reports (in some reports with regards to one claim, in others with regard to all of the claims made). In one example, students concluded that "intraspecific and interspecific competition affects the growth, density, and distribution of plants." However, they drew this causal conclusion from a dataset without measures of either "competition" and "growth." In a scientific context, the claim was therefore unwarranted. In five of the reports, claims were made which were not related to the original question, although in only two cases was this done and the original question not addressed (in another two cases, no claims were made related to a question posed in the study at all). Overall, problems with the claims' sections arose more frequently from claims made which did not extend from the collected data, a quite frequent problem, rather than from claims which did not address the original question.[3]

An Ecologist at Work

Over an 18-month period, we recorded and documented the construction of facts in field ecology. In this discipline, many members understand themselves as being engaged in an observational science operating without the grand theories one might find in physics. Although much of what members know is derived from naturalistic observation, the facts which they report in academic settings (posters, presentations, articles) are purely based on measurements. However, these observational data are diverse in their origin and error size and are collected over a considerable geographical area (here a 20 km stretch) and over long periods of time (3 years). Considerable work is required to coordinate these diverse data in time and space before individual data can be converted into population statistics. Along the trajectory from data to reported fact, ecologists are not always certain whether they have observed something. The following quotes from different stages in the construction of one lizard species[4] illustrate such uncertainties.
(Sam, herpetologist, seeking to catch lizards in one of her field sites:) `I usually find about five a day. I sort of am getting this feeling that they are more active later in the day. They can't tolerate, I think preferred temperature is about 20, mid 20's or maybe high 20's. Probably mid. So in the real heat of the day I don't look for the animals `cause they're buried down too deep and then I go out again in the 4 to 6 kind of range and lately I've noticed I've had better luck'.
(Sam in the field laboratory, timing lizards as she chases them along a race track:) `I don't know if I will be able to use these speed measures, but I do it anyway. Maybe there is something, maybe not'.
(Sam presenting the results of her work in a colloquium:) `And it turns out the longer the lizards are kept in the lab, the slower they run. Which is kind of interesting, but I can statistically control for this effect and go on to look to see if there are other things that are important. And it turns out there are. One of the things that's important is what sex you are. Adult males are typically shorter than adult females, their body lengths are shorter. And adult males also have relatively longer back legs than adult females. And it turns out that this body length and back leg length is important for predicting how fast it runs'.
On the one hand, we see self-doubts in the field and field laboratory where Sam talked about the uncertainties of finding lizards and whether the sprint trials with the animals she captured and returned to the field laboratory were any good. On the other hand, far away from the field in a more formal academic setting, we observe the matter-of-factness of propositional knowledge about lizards such as the statistical significance of correlations between sprint speed (dependent variable) and body length and back leg length (independent variables). On the one hand we see factual statements and hard inscriptions, on the other hand there are uncertainties related to objects, instruments, and measurement processes in the field. From our participant observer perspective, there existed a sharp and seemingly irreconcilable contrast. We must wonder how such firm statements and claims are possible when they have emerged from myriads of decisions and uncertainties that are evident during the ecologist's field work.
A (lay) sociologist of science and student of Science in Action (Latour, 1987) working backwards and tracing (authoritative) statements and inscriptions, would first find printouts from statistics software that had operated on a large database, which itself was imported from another, spreadsheet software package into which numbers had been entered during the ecologist's past field seasons. From here on, our `science lover' would find a dizzying heap of proliferating inscriptions. There would be field notebooks; tables partially filled with records of widely varying origins; forms; printouts containing codes and coding schemes; numbered metal tags for field use; labelled and code-bearing vials, socks, plastic holding boxes and wooden enclosures all identifiable by a one- or two-digit painted number. Linked to these `first' inscriptions she would notice an array of diverse instruments and associated measurement practices. Surprisingly, she would find little in terms of concepts, laws, and theories but, as her participants would tell her repeatedly, `a lot of conceptual mayhem that lies beneath all of that'. Because the ecologists understand their practice as an observational science and because this particular project is concerned with correlations of phenotypic aspects of the lizards, much of the ecological fieldwork appears to be driven by what is do-able in terms of measurement. However, there are also considerable variations in measurement practices, scales of measurement, and measurement error. That is, our scientists' activities and the resulting data on which their claims are built arise from a variegated observational topology associated with considerable co-ordination work which allow the statistical correlation of quite unlike aspects of the study object. Thus, the distillation of the field work, the lizard as it becomes visible to the scientists' audiences, is not just a natural object, not just an individual construction, and not just a social construction but arises out of the interaction of nature, individual, and culture.

Professional Vision

Lizards are literally visible to those in the field and field laboratory who catch and `process' the animals. In this sense, Sam, our ecologist, has a relationship with the lizards as living beings that move around, are difficult or easy to catch, and show specific behaviours allowing her to identify each individual--some of whom she gives names. She describes changes in observations on female lizards by taking the perspective of the animals. Her depiction of lizards shares a lot in common with naturalist portrayals of animal life.[5] But, from her perspective, all she can collect by means of literal seeing is `anecdotal' information; in the parlance of her field, even the colour descriptions one might find in field guides are anecdotal (Law & Lynch, 1990). In the context of her domain, behavioural ecology, Sam has to make lizards visible to others by means of a different perceptual machinery that allows her to arrive at `scientific' descriptions of the animals of interest. This process requires animals to go through the process of digitisation, a conversion that involves `hard' numbers and electronic bits, before other ecologists can `believe their eyes'. But as part of this process of converting nature to electronic bits, Sam's field activities are open to an indeterminable horizon of contingencies which arise in the course of the scientific work that results in `knowledge about the lizard'. Through Sam's work, the lizard is constructed and thereby made visible, involving a proliferation of inscriptions, conversions of nature into numbers, and electronic digits. Lizards, as they are seen by our `science lover', are reflexive of endlessly awkward processes and objects left behind in the field laboratory and terrain. That is, literal seeing while important to the individual ecologist is not as important as the `observations' possible once the animals and their environment have undergone multiple transformations. Instruments, field laboratory, inscriptions, and associated practices therefore constitute pieces of an observational machinery that does not have the even, spherical surface of the human retina but has its own multi-dimensional, heterogeneous, and heteromaterial topology. This article is about the topology of this observational machinery which, not unlike biological retinas, turns nature into (afferent) digital signals that are the data for subsequent processing in centres of computing. These signals are not pure in any sense but are always and inevitably formed by the enacted disciplinary practices. That is, any relevant aspect about the lizard's life history or natural history emerges from the interplay between the domain of inquiry and existing discursive practices and material configurations (Coulter & Parsons, 1991; Goodwin, 1994). In the present study, these discursive practices were more related to instruments and measurement and less to unifying concepts and theories regarding ecology and evolution.
My central claim therefore is that observing lizards is more than a matter of watching a few animals in the field--though naturalists, amateurs, and our participating scientist may have done this in the past and still do--but lizards have to be observed using the whole observational machinery available to a field ecologist, including analogue and digital scales, analogue and digital length-measuring devices (e.g., ruler, tape, map measurer, paces), Munsell charts, and digital stop watches. As a result of physically bringing together natural phenomena and the instrumental topology, numbers composed of digits begin to fill rows and columns of a spreadsheet. The entire laboratory machinery is used to produce digits which are then summarised, compared, correlated and otherwise processed by means of statistical software producing tables and graphs (i.e., series of non-trivial and arbitrary transformations of subsequent stages). These tables and graphs, in the context of some form of text, originate, support, and legitimize scientific propositions about the lizard such as `sprint speed is inversely related to lizard length' or `maternal tail length predicts number of offspring'. In the process of the scientific work, lizards as biting, writhing, running, feeding, and defecating creatures are turned into series of numbers composed of one or more digits--first on paper and subsequently into the electronic digits of spreadsheet and statistical software.

Mathematization of Nature

Ethnomethodologically specified, measurement is a hopelessly vulgar competence. Our interest lays, therefore, in the local, in situ practices by means of which ecologists conduct their measurement-related activities which includes producing local judgements related to the practical adequacy, accuracy, and appropriate correspondence between measuring device and measured phenomena. Despite the potential dangers of `going native' arising from our own training as natural scientists, and following an ethnomethodological maxim (Lynch, 1991), I did not treat scientific measurement as a homogeneous set of methods and standards but as an ecology of heterogeneous techniques that allow scientists to attain locally recognisable and locally adequate measures. In my studies, I provided documentary evidence that measurement, though a familiar term of the ecology trade, is neither a coherent interdisciplinary practice nor coherent for the same scientist across situations, even along physically similar dimensions (e.g., `length' and `distance'). Although the measurements generated by our ecologists may have had referents in the practices of the discipline, what we observed was how each of the scientists locally elaborated and enacted for him/herself the meaning of the cultural referents in this setting. Each participant also elaborated a sociology of science in the sense that they enacted their practices such as to be accountable to the field at large.
Mathematisation has been described as one aspect of the process of scientific seeing (Lynch, 1990). I provide here some indication about just how mathematisation is achieved in the context of ecological field work and describe some of the discovering practices in ecological fieldwork to disclose the order of the local contingencies of the day's work in an ecology field research camp. What I show is how the scientists in the field made their work accountable in, of, and as instances of measurement practices. However, what mathematisation achieves is more than just seeing, for it is no longer individuals that are seen but a conglomerate of indistinguishable replaceable individuals. It is the move from `this lizard is doing X' to `lizards do X', from the psychology of the individual to the sociology of the masses. Because sweeping and `fuzzy' generalisations describes no one individual in specific, statements such as `leg length determines speed' can stand as a factual statement about this animal. This epistemological rupture between the description of individual animals to statements about all the animals as a class is associated with scientists' shift from the field laboratory practices to data processing in computers, and a similar shift from filling rows of a spreadsheet associated with individual animals to processing columns of numbers that summarise individuals into classes.
Tables and graphs, which abound in scientific publications, are visual documents that integrate the substantive, mathematical, and literary resources of scientific investigation, and create the impression that the objects or relations they represent are inherently mathematical. At the same time, the real mystery is the adequation of mathematics with the empirical world not the superimposition of one mathematical form with another. This isomorphic relationship between mathematics and the empirical world, which is an a priori given for many scientists, was topicalised in the form of the couplet {Fundamental structure <--> Mathematical form} (Lynch, 1991). However, rather than being a simple relation or adequation, the double arrow has to be understood as a possibly infinite chain of inscriptions linked together by the embodied practices of their users. Reference is therefore a quality of a chain of representations that mediates between the purer elements of empirical world and mathematical form toward the extremities of the chain; there is no longer an adequatio rei et intellectus, just a chain that can be extended infinitely at both ends (Latour, 1993).

Discussion & Conclusion

Enculturation has been theorized in terms of the habitus that forms when people--physically and socially situated in the world--participate with others in activities and, as a matter of course, adopt an unthematized practical sense for doing things in particular ways (Bourdieu, 1997). My work goes beyond the work of earlier research on research practices (e.g., Latour, Woolgar, Knorr-Cetina, and Lynch) both in its scope--from grade 8 to professional practice--and in terms of the work observed--in work settings (outside) and on specially designed interpretation tasks (inside). In these studies our work is beginning to disclose interesting discontinuities and contrasts. First, there were considerable discontinuities between the investigative activities of practicing scientists and those of university students and recent graduates with bachelors and masters degrees in science. At the same time, surprising competencies were exhibited by pairs of Grade 8 students that, at a minimum, matched those of the university graduates in science despite the vast differences in the educational backgrounds. A comparative reading of the transcripts of group interpretations made by Grade 8 students (who engaged in innovative science curricula for their second year) and university students in a fifth year teacher education course (who had obtained science degrees but who had experienced the traditional fare of science courses) shows very little difference in the resources and practices enacted during the tasks and very similar breakdowns.
One major breakdown for most university students occurred when they had to deal with actual, scattered data that did not fall onto a line graph (linear, quadratic); the same situation was dealt with much more gracefully by the Grade 8 students experienced at preparing convincing representations of real data. A second breakdown was experienced by the university students because they took graphical models as representing real data so that they sought for ways how "impossible" data could have ever been collected. That is, my research indicates that present schooling at the elementary, secondary, and undergraduate levels does little to introduce students to the authentic scientific representation practices. My ongoing work among doctoral students and postdoctoral researchers suggests that it is during these years that individuals begin to participate in those experimental practices which we documented for practicing scientists. On the other hand, my experience with a special curriculum in a Grade 8 ecology class suggests that even at this early age, students can effectively engage and participate in scientific representation practices.
Traditional pedagogy--associated with the literature on general problem solving skills that can be transferred across settings (e.g., Anderson 1985)--also assumes that learning proceeds from the general to the particulars. Yet my work shows that without knowledge of particulars that can serve as referents, students have difficulties in making sense of the representations (as signs). On the other hand, those students with great familiarity of particulars and how to transform particulars into graphical representations also showed high levels of inscription-related competencies. The discontinuity between the graphing-related practices of most students (other than those in our Grade 8) and those enacted by practicing professionals may therefore arise from these differences in knowledge of particulars and familiarity in and with transformation practices. My ethnographic work among graduate students and postdocs in the field of ecology shows that the transitions from school to research practices do not come easy.
Ultimately, my work on scientific practices (mostly relating to representations) from Grade 8 to professional practice constitutes only a small slice of the kind of work that needs to be done to understand trajectories of scientific competence from elementary school to everyday life (professional science, activism, lay involvement). I am convinced that there is even a definite place for members of the science (and technology) studies community to cooperate with science educators both in terms of the research to be conducted as well as in policy making and curriculum design. It is to be hoped that this kind of work constitutes a beginning in dealing with our own conceptual blind spots so that we can investigate the discontinuities in the practices at the interface of schooling and everyday activity outside thereafter.

Acknowledgments

This symposium paper draws on a number of studies conducted over the past decade. It was made possible, in part, by a grant of the Social Science and Humanities Research Council of Canada (410-93-1127). My thanks go to Michael Bowen and Sylvie Boutonné who assisted in the collection and transcription of the databases on which I have drawn.

References

American Association for the Advancement of Science (1993). Benchmarks for science literacy. New York: Oxford University Press.
Amerine, R., & Bilmes, J. (1990). Following instructions. In M. Lynch & S. Woolgar (Eds.), Representation in scientific practice (pp. 323-335). Cambridge, MA: MIT Press.
Anderson, J. R. (1985). Cognitive psychology and its implications. San Francisco, CA: Freeman.
Bourdieu, P. (1997). Méditations pascaliennes [Pascalian meditations]. Paris: Seuil.
Bowen, G. M., & Roth, W.-M. (1998, April). Isolation of variables and enculturation to a reductionist epistemology during ecology lectures. Paper presented at the annual conference of the American Educational Research Association, San Diego, CA.
Bowen, G. M., & Roth, W.-M. (1998, October). Natural worlds and graphical representations: On the difficulties of learning ecology from lectures. Paper presented at the annual meeting of the Society for Social Studies of Science, Halifax, N.S.
Bowen, G. M., & Roth, W.-M. (1999, March). "Do-able" questions, covariation, and graphical representation: Do we adequately prepare reservice science teachers to teach inquiry? Paper presented at the annual conference of the National Association for Research in Science Teaching, Boston, Mass.
Brookhart Costa, V. (1993). School science as a rite of passage: A new frame for familiar problems. Journal of Research in Science Teaching, 30, 649-668.
Coulter, J., & Parsons, E. D. (1991). The praxiology of perception: Visual orientations and practical actions. Inquiry, 33, 251-272.
Crist, E. (1996) `Naturalists' portrayals of animal life: Engaging the verstehen approach. Social Studies of Science, 26, 799-838.
Gallagher, J. J., & Tobin, K. (1987). Teacher management and student engagement in high school science. Science Education, 71, 535-555.
Goodwin, C. (1994). Professional vision. American Anthropologist, 96, 606-633.
Greeno, J. G. (1988). Situated activities of learning and knowing in mathematics. In M. Behr, C. Lacampagne, & M. M. Wheeler (Eds.), Proceedings of the 10th Annual Meeting of the PME-NA (pp. 481-521). DeKalb, IL: IGPME.
Knorr-Cetina, K. D. (1981). The manufacture of knowledge: An essay on the constructivist and contextual nature of science. Oxford: Pergamon Press.
Knorr-Cetina, K. D. (1992). The couch, the cathedral, and the laboratory: On the relationship between experiment and laboratory in science. In A. Pickering (Ed.), Science as practice and culture (pp. 113-138). Chicago, IL: The University of Chicago Press.
Latour, B. (1983). Give me a laboratory and I will raise the world. In K. D. Knorr-Cetina & M. Mulkay (Eds.), Science observed: Perspectives on the social study of science (pp. 141-170). London: Sage Publications.
Latour, B. (1987). Science in action: How to follow scientists and engineers through society. Milton Keynes: Open University Press.
Latour, B. (1993). La clef de Berlin et autres leçons d'un amateur de sciences [The key to Berlin and other lessons of a science lover]. Paris: Éditions la Découverte.
Law, J. (1994). Organizing modernity. Oxford, UK: Blackwell.
Law, J., & Lynch, M. (1990). Lists, field guides, and the descriptive organization of seeing: Birdwatching as an exemplary observational activity. In M. Lynch & S. Woolgar (Eds.), Representation in scientific practice (pp. 267-299). Cambridge, MA: MIT Press.
Lynch, M. (1985). Art and artifact in laboratory science: A study of shop work and shop talk in a laboratory. London: Routledge and Kegan Paul.
Lynch, M. (1988). Sacrifice and the transformation of the animal body into a scientific object: Laboratory culture and ritual practice in the neurosciences. Social Studies of Science, 18, 265-289.
Lynch, M. (1990). The externalized retina: Selection and mathematization in the visual documentation of objects in the life sciences. In M. Lynch & S. Woolgar (Eds.), Representation in scientific practice (pp. 153-186). Cambridge, MA: MIT Press.
Lynch, M. (1991). Method: measurement--ordinary and scientific measurement as ethnomethodological phenomena. In G. Button (Ed.), Ethnomethodology and the human sciences (pp. 77-108). Cambridge: Cambridge University Press.
Lynch, M., Livingston, E., & Garfinkel, H. (1983). Temporal order in laboratory work. In K. D. Knorr-Cetina & M. Mulkay (Eds.), Science observed: Perspectives on the social study of science (pp. 205-238). London: Sage Publications.
National Council of Teachers of Mathematics. (1989). Curriculum and Evaluation Standards for school mathematics. Reston, VA: NCTM.
Papert, S. (1991). Situating constructionism. In I. Harel & S. Papert, Constructionism: Research reports and essays, 1985 - 1990 (pp. 1-11). Norwood, NJ: Ablex.
Pickering, A. (1995). The mangle of practice: Time, agency, & science. Chicago, IL: University of Chicago.
Postman, N. (1992). Technopoly: The surrender of culture to technology. New York: Knopf.
Roth, W.-M. (1994). Experimenting in a constructivist high school physics laboratory. Journal of Research in Science Teaching, 31, 197-223.
Roth, W.-M. (1995). Authentic school science: Knowing and learning in open-inquiry laboratories. Dordrecht, Netherlands: Kluwer Academic Publishing.
Roth, W.-M. (1996). Where is the context in contextual word problems?: Mathematical practices and products in Grade 8 students' answers to story problems. Cognition and Instruction, 14, 487-527.
Roth, W.-M. (1998). Designing communities. Dordrecht, Netherlands: Kluwer Academic Publishing.
Roth, W.-M. , & Bowen, G. M. (in press). Complexities of graphical representations during lectures: A phenomenological approach. Learning and Instruction.
Roth, W.-M., & Bowen, G. M. (1993). An investigation of problem solving in the context of a Grade 8 open-inquiry science program. The Journal of the Learning Sciences, 3, 165-204.
Roth, W.-M., & Bowen, G. M. (1994). Mathematization of experience in a Grade 8 open-inquiry environment: An introduction to the representational practices of science. Journal of Research in Science Teaching, 31, 293-318.
Roth, W.-M., & Bowen, G. M. (1995). Knowing and interacting: A study of culture, practices, and resources in a Grade 8 open-inquiry science classroom guided by a cognitive apprenticeship metaphor. Cognition and Instruction, 13, 73-128.
Roth, W.-M., & Bowen, G. M. (1998, October). Perceptual topology of and mathematization in ecology fieldwork. Paper presented at the annual meeting of the Society for Social Studies of Science, Halifax, N.S.
Roth, W.-M., & Bowen, G. M. (1999). Of cannibals, missionaries, and converts: graphing competencies from grade 8 to professional science inside (classrooms) and outside (field/laboratory). Science, Technology, & Human Values, 24, 179-212.
Roth, W.-M., Masciotra, D., & Bowen, G. M. (1998, October). From thing to sign and `natural object': Toward a genetic phenomenology of graph interpretation. Paper presented at the annual meeting of the Society for Social Studies of Science, Halifax, N.S.
Roth, W.-M., & McGinn, M. K. (1998). Inscriptions: a social practice approach to "representations." Review of Educational Research, 68, 35-59.
Roth, W.-M., McGinn, M. K., & Bowen, G. M. (1996). Applications of science and technology studies: Effecting change in science education. Science, Technology, & Human Values, 21, 454-484.
Roth, W.-M., McGinn, M. K., & Bowen, G. M. (1998). How prepared are preservice teachers to teach scientific inquiry? Levels of performance in scientific representation practices. Journal of Science Teacher Education, 9, 25-48.
Roth, W.-M., McRobbie, C., Lucas, K. B., & Boutonné, S. (1997a). The local production of order in traditional science laboratories: A phenomenological analysis. Learning and Instruction, 7, 107-136.
Roth, W.-M., McRobbie, C., Lucas, K. B., & Boutonné, S. (1997b). Why do students fail to learn from demonstrations? A social practice perspective on learning in physics. Journal of Research in Science Teaching, 34, 509-533.
Roth, W.-M., & Roychoudhury, A. (1993). The development of science process skills in authentic contexts. Journal of Research in Science Teaching, 30, 127-152.
Suzuki, D. (1989). Inventing the future: Reflections on science, technology, and nature. Toronto: Stoddart.
Tobin, K., & Gallagher, J. J. (1987). What happens in high school science classrooms? Journal of Curriculum Studies, 19, 549-560.
Traweek, S. (1988). Beamtimes and lifetimes: The world of high energy physicists. Cambridge, MA: MIT Press.

[1] In our experience, fixing the data is a rather common phenomenon in traditionally-conducted high school and university laboratories. We observed or discovered situations in which students with scholarships to the top Canadian and American universities "fudged" their data by one order of magnitude to make their measurements fit existing theory. Little did they know that in most high school and undergraduate laboratories, the closest one can come with this experiment is about one order of magnitude.
[2] Here, I draw on an extensive report of science practices presented in Bowen and Roth (1999).
[3] Here, I rely on a presentation given to sociologists of science (Roth & Bowen, 1998).
[4] There are different lizard species. To protect the identity of our informants as much as possible, we refer to the animals as `lizards' rather than the specific species.
[5] Eileen Crist (1996) provides an interesting analysis of the discourse used by naturalists in describing animal life. She argues that naturalists' (thick) descriptions of animal activity are in stark contrast to scientific writing of generic individuals and typical cases which makes animal behaviour appear automated. In Lynch's (1988) descriptions of laboratory rats, the transition from naturalistic rats to analytic animals was accomplished by means of the `sacrifice'.