Our lives involve a wide range of decisions in which we attempt to anticipate the optimal course of action, even though the consequences cannot be known with certainty. Examples of such decisions include how to respond during a global pandemic, whether or not to administer a life-saving medication, choosing an enrichment program, or selecting a particular diet. Decision-making under conditions of uncertainty is often essential in everyday life. However, research indicates that decisions made under uncertainty, even when they are highly consequential, are often not based on knowledge grounded in data analysis that explicitly and thoughtfully accounts for uncertainty. Instead, such important decisions are frequently based on anecdotes—cases encountered personally or heard about incidentally.
Decision-making under conditions of uncertainty is central to scientific practice and scientific thinking. Scientists develop scientific theories based on data while assessing the degree of certainty in the hypotheses they propose. To do so, they must integrate knowledge about uncertainty related to scientific knowledge and to the nature of science with knowledge about uncertainty related to statistical reasoning. Uncertainty related to scientific knowledge refers to what is still unknown about the phenomenon under investigation (for example: What do we still not know about the phenomenon?). Uncertainty related to the nature of science refers to knowledge about how research can be designed in ways that enable reliable and valid decision-making (for example: What kind of scientific study would make this possible?). Uncertainty related to statistical thinking refers to knowledge about data, such as its type, amount, and reliability (for example: What scientific knowledge do we need? What data are required?). An example that integrates these three domains is deciding whether to take a specific vitamin supplement. Such a decision requires understanding how to conduct reliable and valid research (knowledge about the nature of science), learning what is currently known about the vitamin (scientific knowledge), and investigating data and drawing inferences while assessing the plausibility of the vitamin’s effects (statistical knowledge).
This way of thinking is not intuitive. In fact, students tend to perceive scientific knowledge as certain and absolute. In addition, they often approach uncertainty arising from data through one of two extremes: a deterministic view, in which data are seen as predicting future outcomes with complete certainty, or a relativistic view, in which no conclusions about future outcomes can be drawn from data at all (complete uncertainty). Engaging students in scientific work, while explicitly addressing the nature of scientific practice, helps them understand that certainty is not absolute, and how scientists act to improve what they can know.
Citizen science projects provide fertile ground for cultivating deliberation about these forms of uncertainty, particularly projects in which students are involved in both statistical and scientific inquiry and are exposed to the scientific practices commonly used by project scientists. An example can be found in the "Radon Gas" Project, in which students are exposed to the scientific work of project scientists, for example through visits to radon laboratories and learning about different measurement instruments. At the same time, students learn about what is known regarding radon gas, its properties, and the risks associated with prolonged exposure. They collect data from their homes and investigate these data. These activities raise questions among students that reflect different types of uncertainty, for example regarding the validity of different measurement tools (knowledge about the nature of science), factors that may be associated with increases or decreases in gas levels (scientific knowledge), and the ability to generalize from measured gas levels to the population of residential buildings more broadly (statistical knowledge). The need to make consequential decisions within the project leads learners to connect these different forms of uncertainty and to construct knowledge under conditions of uncertainty.
Deepening and Expansion ▼
Statistical uncertainty
Ben-Zvi, Aridor, Makar, and Bakker (2012) explain that investigating a phenomenon under conditions of uncertainty involves data analysis that includes evaluating the data and its limitations from a statistical perspective. However, statistical reasoning that accounts for uncertainty runs counter to the human tendency to base claims primarily on anecdotes. To develop statistical reasoning that incorporates uncertainty, learners must learn to think simultaneously about the story emerging from the data (the signal) and about evaluating and quantifying statistical uncertainty with respect to the characteristics of the phenomenon. This uncertainty raises questions such as: Does the story emerging from the data apply to all the data? To part of it? What is the likelihood that this story indeed characterizes the phenomenon under investigation?
To evaluate and quantify statistical uncertainty, learners must consider data from a perspective that accounts for two opposing ideas: sample variability and sample representativeness. That is, they must assess the extent to which a data sample represents the studied phenomenon, while also considering the degree to which random samples of the same size drawn from the same population may differ from one another. In other words, when drawing conclusions based on data, one must assess how far these conclusions can be generalized to the phenomenon, taking into account the probabilities associated with their occurrence.
This process is particularly challenging for novice learners, who tend to express opposing conceptions of uncertainty. One conception is deterministic, in which a sample is seen as fully representing the population from which it was drawn (complete certainty). The second is relativistic, in which a sample is viewed as incapable of representing the population due to sample variability (complete uncertainty). Distinguishing between sources of uncertainty may encourage the development of a middle ground, in which generalizations are qualified by the probabilities.
Scientific uncertainty
According to Gasparatou (2017), the construction of scientific knowledge involves repeated reinterpretations of theories proposed in previous research. New hypotheses often lead to the collection of new data that are added to existing data. Examining these new data may lead to validation of an existing theory, but also to its refinement or refutation. This process requires coordinating newly collected data with existing or emerging theories, that is, understanding the relationship between data and theory and the dual and opposing roles of data (as supporting a theory or as challenging it). Such understanding is fundamental to the process of scientific knowledge construction and is therefore central to scientific thinking. According to Chalmers (2013), this process is driven and advanced by recognition of scientific uncertainty and its sources: the complexity of the world that scientific research seeks to unravel, and the limited resources of human beings (both in knowledge and in tools).
According to Popper (1963), a dogmatic conception of theory treats data as conclusively confirming a theory, leaving little room for scientific uncertainty. An opposing early conception emphasizes the role of data in falsifying theories, expressing absolute uncertainty regarding scientific theories.
Uncertainty in thinking about the nature of science
Developing understandings of the nature of science (what science is, what scientific knowledge is, how scientists conduct research, and so forth) is fundamental in science education. The nature of science is considered a gateway into scientific culture and a catalyst for critical examination of scientific practice and research. Hofer and Pintrich (2001) identified three conceptions of scientific knowledge:
(a) Knowledge originates in the world around us; scientific knowledge can be known with certainty and is not subject to challenge or refinement. Once scientific knowledge is constructed, it is permanent, rigid, and immune to change, skepticism, or critique.
(b) Knowledge originates in the knowing subject rather than in the external world; therefore, scientific knowledge is subjective and uncertain, always provisional and open to refutation. Focusing exclusively on the uncertainty and falsifiability of scientific knowledge prevents the exercise of objective judgment in evaluating claims.
(c) An integrative conception combining the first two, according to which scientific knowledge may change over time, yet its uncertainty can be evaluated and quantified.
Integrating scientific thinking, statistical thinking, and thinking about the nature of science in the context of uncertainty
According to Aridor, Dvir, Tsybulsky, and Ben-Zvi (2023), citizen science activities enable engagement in authentic scientific and statistical practices. As such, they may elicit conflicts between intuitive modes of thinking and extreme conceptions of uncertainty in statistical thinking, scientific thinking, and thinking about the nature of science. For example, viewing scientific knowledge as provisional (uncertainty regarding the nature of science) alongside rejection of the possibility of inference from samples (absolute statistical uncertainty) may conflict with a scientific conception that views data as confirming theories (absolute certainty in scientific thinking). Such conflict may lead to connections among these modes of thinking, leading to more balanced views of uncertainty across scientific and statistical thinking, and ideas about the nature of science.
References ▼
- Aridor, K., Dvir, M.,Tsybulsky, D., & Ben-Zvi, D. (2023). Living the DReaM: The interrelations between Statistical, Scientific and Nature of Science uncertainty articulations through Citizen Science. Instructional Science, 51, 729–762. https://doi.org/10.1007/s11251-023-09626-8
- Ben-Zvi, D., Aridor, K., Makar, K., & Bakker, A. (2012). Students’ emergent articulations of uncertainty while making informal statistical inferences. ZDM - The International Journal on Mathematics Education, 44(7), 913-925.
- Gasparatou, R. (2017). Scientism and scientific thinking. Science & Education, 26(7-9), 799-812.
- Kali, Y., (2006). Collaborative knowledge-building using the Design Principles Database. International Journal of Computer-Supported Collaborative Learning, 1(2), 187-201.
- Popper, K. (1963). Conjecture and Refutations: The growth of scientific knowledge. London and New York: Routledge and Kegan Paul.