Nature of science in students’ discussions on disagreement between scientists following a narrative about health effects of the Fukushima Daiichi accident

ABSTRACT

We explored the potential for addressing nature of science through a historic narrative about disagreement between researchers concerning a socio-scientific issue, incidence of juvenile thyroid cancer following the Fukushima Daiichi nuclear accident. The narrative was developed from authentic sources and tested in two cycles. Eight groups of three to four high-school students were audio recorded. Transcripts were analyzed regarding what nature of science emerged in the discussions and what understanding about NOS could be discerned, using three complementary NOS-frameworks (Consensus-NOS, Whole Science-NOS, FRA-NOS). Together, the student groups touched upon 19 different NOS-themes as they tried to make sense of the disagreement related in the narrative. All groups addressed a common core of NOS-themes, most of which were central to the narrative itself, although some themes that were not part of the narrative also emerged. Students displayed a basic understanding of the tentative, empirical, and subjective nature of science together with the role of evidential relevance and completeness of evidence related to the choice of scientific methods. On the other hand, students did not reckon with peer review as a means for establishing knowledge and resolving disagreement. Moreover, although students readily accepted disagreement as a basic property of science, they had difficulty handling this disagreement when coping with the SSI in the narrative. We discuss how the combination of history of science-in-the-making and SSI in narrative form offers opportunities to teach NOS without risking simplified messages of how scientific knowledge develops or how science can be used to address socio-scientific issues.

KEYWORDS:

• Risk
• socio-scientific issues
• nature of science
• history of science
• Fukushima

Introduction

In this study we explore how students deal with a nonfiction narrative about conflicting evidence and disagreement between researchers concerning a socio-scientific issue (SSI), incidence of juvenile thyroid cancer following the Fukushima Daiichi nuclear accident. We are interested in the potential use of this kind of narrative involving disagreement between scientists, as starting points for teaching the Nature of Science (NOS).

Nature of Science has been conceptualised through different frameworks in science education research. The dominant one is the Consensus-NOS framework consisting of seven NOS-aspects with little scholarly disagreement and which have been empirically shown to be possible to teach to K-12 students (Lederman, 2006). The aspects are framed as statements, for instance that scientific knowledge is empirically based, tentative (i.e. subject to change), and subjective (i.e. involves interpretation of individuals or groups). The Consensus-NOS framework has contributed to the realisation that simply engaging students in scientific inquiry will not automatically lead to understanding of NOS and that NOS needs to be taught explicitly (Lederman & Lederman, 2014). However, others have critiqued the idea of teaching statements or views about NOS to students. For instance, Allchin (2011) argued that NOS-understanding needs to be functional, not declarative. It should enable students to interpret the degree of reliability of scientific knowledge for them to make informed judgments. Allchin (2011) suggested the Whole Science-NOS framework which consists of dimensions of reliability used by the scientific community to assess scientific knowledge. Examples of such dimensions are the degree of evidential relevance or completeness of evidence pertaining to a certain knowledge claim (Allchin, 2011, p. 525). Recently yet another framework, FRA-NOS, was presented as a more comprehensive and holistic approach to teaching the nature of science than previous ones (Erduran et al., 2019). The FRA-NOS framework conceptualises science as two systems, one cognitive-epistemic and one social-institutional, with each system containing categories that are described in a set of interrelated statements (Erduran et al., 2019, p. 316). Examples of categories are Methods and methodological rules (cognitive-epistemic) and Social certification and dissemination (social-institutional).

A common feature of most narratives aiming to teach NOS to students is that they build off of classic science controversies taken from the history of science (HOS) (Lederman, 2006). For instance, Schiffer and Guerra (2015) developed a fictional science story based on the Galvani-Volta controversy over animal electric fluid. The NOS that was targeted concerned primarily the empirical and cultural nature of science. Dai et al. (2021) created a non-fiction science story around the discovery of the structure of DNA. In this case, the targeted NOS concerned the role of imagination and creativity as well as cultural and social influences (further examples are, e.g. Forato et al., 2012; Hadzigeorgiou et al., 2012). However, Allchin (2003) criticised the use of science narratives for having simplified plots that linearly lead up to today’s accepted science, claiming that they may give a distorted picture of NOS. Instead of creating ‘just-so stories’ (Allchin, 2003, p. 346) which explain a certain outcome by tracking its history backwards from an already established endpoint, Allchin (2003) suggested that narratives building on historic controversies should focus on uncertainty and conflicting results and embrace the complexity of the issues at the time they were debated.

Conflicting evidence and controversy constitute crucial elements of many socio-scientific issues, and there is a close relationship between SSIs and NOS (Zeidler, 2014). Studies confirm this relationship both ways (Karisan & Zeidler, 2017). Thus, Zeidler et al. (2002) demonstrated that grade 9–12 students’ responses to statements following the presentation of a socio-scientific issue on animal rights differed depending on their understanding of NOS. The other way around, Eastwood et al. (2012) showed that subjecting grade 11–12 students to teaching including contemporary SSIs such as alcohol, diet and obesity, or stem cell research, led to increases in their understanding of NOS. Sadler et al. (2004) subjected high-school students to two arranged reports on global warming presenting conflicting conclusions drawn from partly exclusive, partly identical data. The study confirmed that the tentative, empirical, and social aspects of NOS were important for how students reasoned. At the same time, the authors stated that there is a need for further studies on ‘how science education topics such as NOS and SSI-decision making interact with one another’ (p. 405). Moreover, just as Allchin (2003) raised concerns about the way that NOS is coming across in simplified HOS-narratives, so Oulton et al. (2004) raised concerns about students too often being presented the view that SSIs can be addressed by listing and evaluating pros and cons as a basis for a personal decision. The authors argued that focus be on understanding the SSI as such, its nature and the many perspectives that are possible to take.

Obviously, both historical science narratives and socio-scientific issues have a role to play to increase students' understanding of NOS, as both address controversy over conflicting evidence and interpretations of empirical data. Moreover, both approaches have been challenged as ways to teach NOS (Allchin, 2003; Oulton et al., 2004). However, they have rarely been combined. This is what we aim at in the present study, to look at the ways that students address the nature of science as they encounter a narrative on conflicting evidence and disagreement from contemporary history of science within the context of a socio-scientific issue. Our overarching research question is: what possibilities for addressing the nature of science emerge as students discuss a narrative about conflicting evidence and disagreement between researchers regarding the incidence of thyroid cancers among young inhabitants of the Fukushima prefecture following the Fukushima Daiichi nuclear accident in 2011?

We employed the three frameworks Consensus-NOS (Cons), Whole Science-NOS (WS), and Family Resemblance Approach to NOS (FRA-NOS) to identify NOS in our data. The Consensus-NOS framework was chosen because it is still the dominant one and because it forms the basis for a considerable number of studies on the relation between NOS and SSI. We included the Whole Science-NOS framework because it was specifically developed to support students’ interpretation of the reliability of scientific knowledge claims, which was a central feature of the narrative about disagreement that the students encountered. The FRA-NOS framework, finally, was included since it is claimed to subsume previous frameworks (Erduran et al., 2019, p. 316) and, thus, could potentially work as a complement to the other two. The term NOS-theme is used for referring to the aspects (Cons), dimensions (WS) and categories (FRA) in the three frameworks.

Based on this conceptualisation of NOS, the following two specific research questions were addressed:

1. Which NOS-themes appear in students’ discussions about disagreement between scientists as related in the narrative?

2. What understanding of the identified NOS-themes can be discerned in students’ discussions about disagreement between scientists as related in the narrative?

The exploratory orientation of these two research questions enabled us to get insight into the kind of NOS that may potentially emerge from students’ discussions around the narrative. In this way, the study addressed an acknowledged need for careful analyses of well-chosen case studies of a historical or contemporary nature, to guide the selection of cases that are possible to use for the purpose of teaching NOS (Osborne et al., 2003).

Study design, setting, data collection and analysis

This study was conducted within a practice-oriented and collaborative (Feldman, 1993; Ruthven, 2005) research project where science education researchers, practicing science teachers, and researchers with specialisations in risk and risk analysis collaborated around how to include questions of risk in SSI-teaching in secondary science education (Schenk et al., 2019; Wojcik et al., 2019). The research design is best described as a combination of features from different but closely related methodologies of educational design research (McKenney & Reeves, 2012), with special emphasis on the following elements: (1) design, implementation, analysis, and re-design of teaching interventions (Brown, 1992; The Design-Based Research Collective, 2003), (2) designing the interventions from a combination of educational research, subject-matter expertise, and existing classroom practice (Ingerman & Wickman, 2015), and (3) conducting joint analyses between researchers and participating teachers in order to increase the relevance of results for classroom practice (Joffredo-Le Brun et al., 2018; Hamza et al., 2018). The research presented in this particular study reflects an in depth, exploratory part of the development of interventions, focusing on rich conversational data from student discussions, thereby reflecting our interest of looking in detail into what NOS emerged, and what understanding of NOS that was possible to observe, as students engaged the narrative (Erickson, 2012).

As an overview, the study design consisted in an initial development phase in which an intervention in the form of a narrative and its presentation was designed jointly in the research group based on a combination of expert knowledge of the particular science lying behind the narrative, educational knowledge of NOS and SSI, and practical knowledge of how to present information to students in ways that engage them in discussion. On special initiative from the teachers, a series of questions was also developed as prompts for the students’ discussions on the narrative. In the second phase, the intervention (narrative + questions) was implemented in two subsequent iterations in two different schools: seven grade 11–12 students in the first iteration, and 18 grade 12 students in the second iteration. All interventions were audio recorded. Small modifications were made between the iterations, based mainly on the teachers’ interpretations of the first iteration. In the last phase, we analysed the narrative and the student discussions for presence of NOS and students’ understanding of the NOS emerging in the discussions.

The narrative

During spring 2019, we developed a narrative about some events following the Fukushima Daiichi nuclear accident. In focus of the narrative were researchers disagreeing on the interpretation of data from a large screening survey of thyroid cancer in the exposed population following the accident, as well as disagreement on methods for establishing whether there was an increased risk of thyroid cancer in the Fukushima Daiichi area (Suzuki, 2016). Overall, it may be characterised as a non-fiction narrative (Aune et al., 2018) about a recent historical event including an SSI at its core. Below, we present a short outline of the narrative employing terminology drawn from Froese Klassen (2014) and Klassen (2009).

The setting for the narrative is the Fukushima Daiichi prefecture between 2011 and 2016. The narrator is Wojcik, an expert in radiation biology, who presents the story orally to the students. The first central event is the reactor core meltdown and release of radioactive elements – mainly ¹³¹I and ¹³⁷Cs – into the environment. A problematic situation appears as there is both uncertainty and public anxiety over increased risks of thyroid cancer in the exposed population. One of the agents of the narrative, the Japanese government, then makes a first critical decision to carry out evacuation measures but without sufficient dosimetric measurements of the evacuated and remaining people that would allow an estimation of the possible health effects of radioactive contamination (Ohba et al., 2020; Tokonami et al., 2012). This decision is taken on the grounds that there will be no measurable increase in thyroid cancer thanks to evacuation and low levels of released ¹³¹I. Moreover, it is assumed that dosimetric measurements will only increase public fear. However, this leads to a crisis with extensive protests and attention in the media. This leads to another critical decision by another agent, the local authorities, who launch the so-called Fukushima Health Management Survey late in 2011 to calm the situation (Yasumura et al., 2012). An ultrasound thyroid screening of about 300 000 people aged between 0 and 18 years at the time of the accident is carried out, showing, against all expectations, that 49 percent carry thyroid abnormalities that may lead to cancer. Actual thyroid cancer is verified in 113 people, which is equivalent to a prevalence of 386 cases per million people. Since screenings for thyroid cancer have never been carried out in Japan before, there are no data to compare this value with. The narrative then reaches its climax when a third agent, a group of researchers led by Prof. Tsuda, mathematically convert the prevalence values from the screening into incidence values, compare these with existing Japanese incidence data, and claim that there is a 30 times higher risk of acquiring thyroid cancer in Fukushima compared to the rest of the country (Tsuda et al., 2016). Again, there is extensive attention in the media as well as public fear and upset. Unlike the typical plot structure of a science story, however, our narrative does not end with a conclusion or resolution to the problematic situation. Instead, it ends with a new problematic situation arising as a final agent, namely other researchers whose results contradict those of Tsuda and colleagues’, enters the plot (Wakeford et al., 2016). The research concerns both conflicting results using other methodologies, and arguments against the methodology employed by Tsuda et al. This ambiguous ending constituted an essential element in the narrative that we hoped would create a kind of final ‘narrative appetite’, to paraphrase Klassen (2009), thereby stimulating subsequent discussion about the narrative among the students. Moreover, interspersed in the narrative are short, expository accounts aiming at clarifying challenging concepts or relationships that were considered essential for understanding the narrative. In that sense too, our narrative does not count as a science story in the strict sense postulated by Froese Klassen (2014). On the other hand, this mixing between narrative and expository elements is reminiscent of the way teachers normally employ narrative in the classroom (Boström, 2006). (see also Supplementary material Appendix A).

NOS in the narrative

We developed the narrative with the primary purpose of providing students with an account of disagreement between scientists in the setting of a risk-related SSI. This means that we did not work from any preconceived ideas of which NOS-themes should be included in the narrative. Therefore, we conducted an analysis of NOS-themes in the narrative from transcripts of Wojcik’s presentation. We first divided the narrative into elements (Froese Klassen, 2014; Klassen, 2009). Each element and intermediate expository parts were then assigned to relevant NOS-themes from the three NOS-frameworks. Table 1 shows a translated extract from the original Excel coding sheet showing the division into narrative and expository sequences and their assignment to narrative elements and NOS. To increase transparency we added a column with reference to corresponding parts in the short version of the narrative provided in Appendix A.

Table 1. Translated extract from original Excel coding sheet, illustrating the sequencing of the narrative and coding of narrative elements and expository incursions into NOS-themes.

Storyline		Narrative element or expository incursion	NOS-framework			Lines in App. A
Sequence	Content		Cons	WS	FRA
4	From the Chernobyl accident scientists learnt that ^131I is released in a nuclear accident, and that there is a correlation between exposure to ^131I and incidence of thyroid cancer.	Event 1	- Empirical - Subjective			3–6
5	Explains dose-response relationships and their significance for determining causality. Explains relative risk and confidence interval.	Expository	- Empirical - Subjective	- Evidential relevance - Statistical analysis of error - Correlation vs. causation- Norms of handling scientific data - Nature of graphs		Not included
6	Conclusion by the research community: Increase of thyroid cancer in Chernobyl area was caused by exposure to high doses of ^131I.	Agency (experts)	- Empirical - Subjective	- Evidential relevance	Social certification and dissemination	8–9
7	Two differences were observed between Chernobyl and Fukushima: (a) Fukushima had only 7% av Chernobyl’s emissions of ^131I and (b) a smaller area was expose. So, lower dose and fewer people exposed!	Event 2	- Empirical - Subjective	- Evidential relevance		10–11
8	A group of experts concluded that no effects of exposure to ^131I would be possible to detect, based on comparisons between ‘baseline liftime risk’ and ‘lifetime risk for Fukushima inhabitants’. But note that it was only estimations, not real data. On that basis, the Japanse government decided not to carry out dosimetric measurements in order to avoid panic. Big mistake.	Agency (government and experts)	- Subjective	- Evidential relevance - Completeness of evidence.- Norms of handling scientific data		12–14
9	But people were upset and frightened, and the decision not to conduct dosimetric measurements decrased the public’s trust in authorities and the government. Protests. ‘The government is trying to mislead us’.	Agency (the public)		- Credibility		14–16

Below we explain the coding of three sequences from Table 1 in more detail. Sequence no. 4 was coded as Empirical NOS(Cons) because it concerned data on emissions of ¹³¹I and as Subjective NOS (Cons) because it dealt with scientists making inferences about correlations based on that data. Sequence no. 5, which is an example of an expository incursion, was also coded as Empirical and Subjective NOS on the same grounds as sequence no. 4. In addition, it was coded as Evidential relevance (WS) because it dealt with issues of the kind of needed to determine causality (viz., dose and response measures). In addition, in dealing with causality, Correlation vs. Causation (WS) and The nature of graphs (WS) were dealt with as part of the explanation. Finally, the notion of confidence intervals touched upon the WS-NOS theme Norms of handling scientific data. Sequence no. 6 was coded as Subjective NOS (Cons), again on the grounds that it concerned conclusions made by scientists (high doses of ¹³¹I cause increased incidence of thyroid cancer), and as Empirical NOS (Cons) and Evidential relevance (WS) because these conclusions were based on certain evidence in the form of real data from the Chernobyl accident. Finally, this sequence pointed out that this conclusion was reached by the scientific community, rather than individual scientists, thereby being coded also as Social certification and dissemination (FRA).

We identified 19 NOS-themes in total, eight of which were judged to be central throughout the narrative (Table 2). All eight themes were identified in narrative sequences and are, thus, directly tied to the narrative. Another three NOS-themes were not central but still emphasised at some point. Of these, however, two (Correlation vs. causation and Nature of graphs) were only coded in one expository sequence (sequence no. 5, Table 1). As such, they are not directly tied to the narrative but are dependent on which expository clarifications the narrator (here, Wojcik) chooses to invoke to supplement the narrative. Although they were interpreted as ‘emphasised’ in our coding, they nevertheless played a very minor role in the intervention, as they did never emerge as part of the students’ discussions (see details in the Results section). Remaining themes were present but not emphasised (Table 2).

Table 2. The 19 NOS-themes identified in the narrative and whether they had a central role in the narrative, were emphasised at some or a couple of instances, or were present but not specifically emphasised. Cons = Consensus NOS-framework (Lederman, 2006), WS = Whole Science NOS-framework (Allchin, 2011), FRA = Family Resemblance Approach to NOS-framework (Erduran et al., 2019).

Nature of science			Role in narrative
Type	Theme		Central	Emphasised	Present
Cons	empirical		x
Cons	subjective		x
Cons	creative				x
Cons	cultural				x
WS	evidential relevance		x
WS	completeness of evidence		x
WS	alternative explanations		x
WS	peer review		x
WS	nature of graphs			x
WS	correlation vs. causation			x
WS	norms of handling scientific data			x
WS	role of probability in inference				x
WS	statistical analysis of error				x
WS	role of imagination and creative synthesis				x
WS	role of cultural beliefs				x
WS	credibility				x
WS	resolving disagreement				x
FRA	methods and methodological rules		x
FRA	social certification and dissemination		x
Total	19		8	3	8

Testing the narrative in class

We tried the assignment in three sessions amounting to two iterations with slight modifications between them. We formulated follow-up questions to support the student discussions about the narrative. The narrative and the questions together constituted an assignment that could be run in a regular upper secondary science class.

Session 1 (iteration 1) was conducted in an upper secondary school in a large Swedish city in November 2020. The narrative was presented on-site by Wojcik to a group of three grade 11 and four grade 12 students who had volunteered to participate in the session after school. The session lasted 80 minutes. The presentation took 37 minutes. After the presentation, the students were divided into two grade mixed groups, and a sheet with five questions was handed out. The two groups were audio recorded. The questions were:

Q1. What was the reason for the researchers disagreeing on whether thyroid cancer increased in those who were children at the time of the accident in Fukushima 2011?

Q2. Can you come up with more examples of areas where researchers disagree?

Q3. How should we as citizens cope with the fact that experts disagree?

Q4. Tsuda et al.’s results that the Fukushima accident led to an increase in thyroid cancer received considerably more attention in media that those researchers claiming that there was no increased risk. (a) Why is that so? (b) To what extent could it become a problem? (c) Who has the responsibility for how research results are communicated in the media?

Q5. Can you think of other ways for investigating if the Fukushima accident has led to an increase in thyroid cancer?

Based on the experiences from the first session, we removed the Q5 since it turned out to be difficult for the students. We also made a video recording of the presentation. The recorded version was 9 minutes shorter but retained all the information from the on-site version (slight reorganisation of content and removing one slide). The shorter time was primarily accomplished by shortening the expository accounts interspersed between the narrative elements.

Sessions 2 and 3 (iteration 2) were conducted within one week in March 2021 in another upper secondary school located in a municipality just outside the city where session 1 took place. This time, the assignment was run in two regular grade 12 science classes. We audio recorded three groups of three students in each session (i.e. six group discussions in total).

The students in the two schools had not received any prior explicit teaching about NOS.

Data analysis

We employed the analytic framework Practical Epistemology Analysis (PEA) (Kelly et al., 2012; Wickman & Östman, 2002) to structure the flow of reasoning in the student groups. PEA conceptualises classroom discourse as a rhythm between gaps that students notice in encounters with the world (e.g. an observation, a written question, a statement from a peer), and relations that they establish to fill these gaps (Wickman, 2006). Excerpt 0 illustrates this.

Excerpt 0.

Table

1	Student 1	‘Can you come up with more examples of areas where researchers disagree?’ [reads Q2 from the sheet]. Corona.
2		[laughter]
3	Student 2	Well, I mean …
4	Student 1	Always …
5	Student 3	… very relevant.
6	Student 2	There are certain things that you usually regard as universal truths, but I mean it’s extremely little that belongs to that zone, pretty much is still ‘up for debate’ [in English] what’s true, what’s not, what can be proven, what cannot, that kind of stuff.

The question that was read aloud also constitutes the gap noticed in this encounter (Turn 1). The students filled the gap with the relations ‘corona – always very relevant’ (Turn 1-5), ‘certain things – universal truths – but extremely little’ (Turn 6), and ‘much – still up for debate – what’s true or not – what can be proven and not’ (Turn 6). Performing PEA on the entire transcripts from all eight groups, we ended up with lists in which the group conversations were converted into the gaps noticed and the relations established to fill them.

Thereafter, the gaps and relations were analysed in terms of whether they related to a NOS-theme, and in that case, which theme (RQ1). For instance, the first relation above (corona – always relevant) did not contain any NOS-theme. The second and third relations included both Tentative NOS (Cons) (‘up for debate’), Empirical NOS (Cons) (‘can be proven – cannot’) and Subjective-NOS (Cons) (the whole relation implies that different researchers can hold different opinions). We scrutinised each gap and relation against all three frameworks in an iterative process where labels for individual gaps/relations were checked and re-checked against each other to insure consistency of coding. Moreover, NOS-themes appeared both explicitly and implicitly in student discussion and their identification did not depend on the level of student understanding of the NOS concerned. For instance, in most groups where we identified the NOS-theme Peer review (WS), students displayed an almost total disregard for this kind of communication as part of establishing scientific knowledge. In that case, we could clearly identify the theme, although instead through students’ lack of understanding of it.

Second, we analysed what understanding that could be discerned about the identified NOS-themes (RQ2). We coded the transcripts concerning how the identified NOS was used for coping with disagreement. For instance, in excerpt 0 the students not only dealt with the three NOS-themes, but also seemed to appreciate that scientific knowledge is tentative, empirical, and subjective. Hypothetically, the students could have established relations equating scientific knowledge with things that are not up for debate or suggested that what eventually counts as scientific knowledge is based on personal preference rather than evidence. In that case, we would still have identified Tentative and Empirical NOS in the discussion, but it would have indicated a weaker understanding. Finally, we conducted thematic analysis (Braun & Clarke, 2006) of initial codes, bringing them together into increasingly comprehensive groups. This way, we ended up with four ‘salient outcomes’ concerning what understanding that could be discerned about NOS in student discussions.

Methodological considerations

In all, we collected data from just over 135 minutes of audio recordings corresponding to 80 pages of transcribed student talk. PEA-analysis of this data generated 195 identified relations that were subsequently interpreted through the three NOS-frameworks. This focus on detailed, in-depth analysis of what happened in the students’ discussions is in line with the exploratory research interest of this study. At the same time, it should be noted that the study was part of a project with an overall research design based on DBR, with a special emphasis on joint collaboration concerning both design and analysis of interventions. In that sense, some of the kind of additional data that might otherwise be collected for purposes of triangulation (such as teacher interviews) was already included in the collaborative analytic work conducted within the research group.

The joint analytic work also meant that regular interrater reliability measures were not conducted in this study, since they would have constituted significant departures from the ways of doing joint analysis cultivated within the research group. The initial analyses of NOS in both the narrative and student discussions were made jointly by four of the six team members (Hamza, Wojcik, Arvanitis, and Haglund) on several occasions. Based on these, the first author subsequently refined the analyses, whereafter they were shared among the team members and discussed in the research group on two occasions, before being finally established. The more interpretive analyses of students’ understanding as well as the subsequent thematic analysis were originally made by the first author alone, but the categories were subsequently discussed and refined in the research group. We feel confident that this way of conducting analyses ensured consistency and credibility of the results. Moreover, the basic reasoning behind our coding is made transparent through the detailed analyses of transcripts provided in the results section below.

Results

NOS-themes appearing in student discussions

The number of NOS-themes identified in the narrative (19 themes, Table 2) exceeded the number of NOS-themes dealt with by any single group (6–14 themes, Table 3). Nevertheless, considering all eight groups, a wide range of NOS-themes were present in students’ discussions. Together, the eight groups dealt with 19 themes, 14 of which were also present as themes in the narrative (Table 3). Empirical and Subjective NOS (Cons) were present in all eight groups and Peer review (WS) and Social certification and dissemination (FRA) were present in all groups except one. These four themes were also among the eight central ones in the narrative. In fact, the top seven themes dealt with by the groups were also identified as central themes in the narrative (Table 3).

Table 3. Distribution of NOS-themes in the group discussions and total number of groups treating each theme. The themes are sorted by whether they appear in the narrative or not and by the number of groups treating them. The eight central NOS-themes in the narrative (cf. Table 1) are indicated with *. Cons = Consensus NOS-framework (Lederman, 2006), WS = Whole Science NOS-framework (Allchin, 2011), FRA = Family Resemblance Approach to NOS-framework (Erduran et al., 2019).

Nature of science		Narrative	Groups								Groups total
Type	Theme		1	2	3	4	5	6	7	8
Cons*	empirical	x	x	x	x	x	x	x	x	x	8
Cons*	subjective	x	x	x	x	x	x	x	x	x	8
WS*	peer review	x	x	x	x		x	x	x	x	7
FRA*	social certification and dissemination	x	x	x	x		x	x	x	x	7
WS*	alternative explanations	x	x	x		x	x	x	x		6
WS*	evidential relevance	x	x	x		x	x		x		5
WS*	completeness of evidence	x	x	x		x			x		4
WS	role of cultural beliefs	x		x	x	x		x			4
Cons	cultural	x		x	x	x		x			4
WS	credibility	x	x	x		x					3
FRA*	methods and methodoligical rules	x	x				x				2
WS	resolving disagreement	x	x								1
WS	role of probability in inference	x							x		1
WS	statistical analysis of error	x							x		1
WS	social resp. of scientists		x	x	x	x	x		x	x	7
Cons	tentative		x	x	x			x		x	5
WS	credibility of sci journals and media		x		x						2
WS	conceptual change				x			x			2
FRA	scientific ethos		x								1
Cons	creative	x									0
WS	correlation vs. causation	x									0
WS	role of imagination and creative synthesis	x									0
WS	norms of handling scientific data	x									0
WS	nature of graphs	x									0
Themes total	24	19	14	12	10	9	8	9	10	6

Students also went beyond the NOS of the narrative to include other NOS-themes as well. Five of the themes that the student groups dealt with were not identified as themes in the narrative (Table 3). The most common themes unique to the student discussions were the Social responsibility of scientists (WS), which was expressed in seven groups, together with Tentative-NOS (Cons) which was present in five groups (Table 3).

Conversely, five of the themes present in the narrative did not appear in the group discussions (Table 3). However, none of these belonged to the group of eight central NOS-themes, although three were emphasised on at least one occasion (cf. Table 2).

We checked to what extent the NOS-themes appearing in the group discussions were tied to the questions handed out after the narrative. Nine of the 19 themes identified in the discussions appeared as part of three or all four questions (Table 4). Seven of these (Empirical NOS, Subjective NOS, Tentative NOS, Completeness of evidence, Evidential relevance, Alternative explanations, and Resolving disagreement) may be considered generic, meaning that we could not find any distinct pattern connecting them to any one question (Table 4). On the other hand, the remaining two (Peer review and Social certification and dissemination; Table 4) were almost entirely tied to Q4, which explicitly addressed the dissemination of research in media. This is also true for one of the themes in the lower part of Table 4 (Social responsibility of scientists), which displayed the same pattern as the theme Peer review. In addition, a few less conspicuous themes were unique to particular questions (Table 4). Finally, it may be noticed that of the nine themes appearing in three or all four questions, seven Belonged to the eight themes identified as central in the narrative (cf. Table 2 and 3).

Table 4. Distribution of NOS-themes between the four questions handed out to the students as support for discussing the narrative. The themes are sorted by the number of questions in which they appear in the student discussions. The eight central NOS-themes in the narrative (cf. Table 1) are indicated with *. Cons = Consensus NOS-framework (Lederman, 2006), WS = Whole Science NOS-framework (Allchin, 2011), FRA = Family Resemblance Approach to NOS-framework (Erduran et al., 2019).

Nature of science			Number of groups discussing the theme under each question				No. of questions including the theme
Type	Theme		Question no.
			1	2	3	4
Cons*	empirical		4	6	7	4	4
Cons*	subjective		4	7	8	7	4
WS*	completeness of evidence		2	2	1	1	4
Cons	tentative		0	3	4	3	3
WS*	evidential relevance		3	0	2	2	3
WS*	alternative explanations		3	4	3	0	3
WS*	peer review		0	1	1	7	3
WS	resolving disagreement		0	1	1	2	3
FRA*	social certification and dissemination		0	1	2	7	3
WS	credibility		0	1	0	3	2
WS	social responsibility of scientists		0	1	0	7	2
Cons	cultural		0	4	0	0	1
WS	role of probability in inference		0	0	1	0	1
WS	statistical analysis of error		0	0	1	0	1
WS	conceptual change		0	2	0	0	1
WS	role of cultural beliefs		0	4	0	0	1
WS	credibility of sci journals and media		0	0	0	2	1
FRA	scientific ethos		0	0	1	0	1
FRA*	methods and methodological rules		2	0	0	0	1

Discerning understanding of NOS in student discussions

The analysis of what understanding of NOS that could be discerned in students’ discussions concerning disagreement between experts produced four salient outcomes, which we labelled (1) Taking basic NOS for granted, (2) Rationalising disagreement, (3) Peer review missing, and (4) Accepting vs. dealing with disagreement. The first two salient outcomes indicate that students were able to appreciate several central NOS-themes from all three frameworks: Empirical, Subjective and Tentative NOS from the consensus NOS-framework; Evidential relevance, Completeness of evidence, and Conceptual change from the Whole Science NOS-framework, and Methods and methodological rules from the FRA NOS-framework. The third outcome, on the other hand, indicates that in general the students did not appreciate the Social certification of knowledge (FRA) and Role of peer review (WS) for the development and establishment of scientific knowledge. The fourth outcome, finally, implies that although students appreciated many basic NOS-tenets, they still struggled with how to act as citizens in the face of uncertainty and disagreement.

1–2. Taking basic NOS for granted and rationalizing disagreement

The students in general seemed to regard disagreement between experts as unproblematic, taking Tentative, Empirical, and Subjective NOS (Cons) for granted. In excerpt 1, group 2 discussed other examples of disagreement between researchers (Q2).

Excerpt 1.

Table

7	Student 1	If it’s in general there are, you know almost every subject in that case, where there’s research, there researchers may, disagree
8	Student 2	Yeah, especially when there’s something pretty new that you don’t know that much about or if it, well, that it’s like now, corona, there are many, you know, there are many experts who disagree about, well, how you should, or well, which measures to take to decrease, well, to decrease risk of infection cause well, and then get more people to survive and not get infected.

These students established relations between disagreement and ‘almost every subject’ (Turn 7), indicating acceptance of Subjective NOS (Cons). At the same time, they established additional relations between disagreement and new topics where knowledge is scarce (Turn 8). Thus, subjectivity was not connected to relativism, and they talked about science as changing as knowledge accumulates, suggesting acceptance of both Tentative and Empirical NOS (Cons). Other groups reasoned in similar ways. Group 1 established relations to disagreement being part of the scientific process suggesting both Subjective and Tentative NOS (Cons) (Supplementary material Appendix B, excerpt B1). Likewise, group 6 (Supplementary material Appendix B, excerpt B2) invoked Empirical and Tentative NOS as they established relations to knowledge changing over time and that research can be disproven.

While disagreement was accepted as a matter of fact by the students with reference to Tentative, Empirical, and Subjective NOS (Cons) our analyses also revealed nuances in their rationalizations for the disagreement. These nuances were revealed primarily through the way in which Whole Science-NOS dimensions entered discussions. For example, a common explanation for disagreement was reference to Incomplete evidence (WS), as shown in excerpt 1 (Turn 8). Another rationalisation was to explain disagreement with reference to Methods and methodological rules (FRA) and Evidential relevance (WS), as in excerpt 2, in which group 5 discussed why the researchers in the narrative disagreed on whether thyroid cancer increased or not after the Fukushima incident (Q1).

Excerpt 2.

Table

9	Student 1	And they also felt that … that, the results won’t show any direct c- … The biggest difference is that they didn’t have a control. They didn’t have any … They didn’t have any way of comparing their results with … cause, how he did it was only an estimate.
10	Student 2	Yeah
11	Student 1	He looked at, how often there is, outside and then it was, thirty times less. And then he said: ‘Ok then there has to be some, effect’.
12	Student 2	Right.
13	Student 1	Whereas others said that … not exactly it was, it was only about, now we’ve tested this many people in the area so, we see … we’ve got a figure, but we have no control.
14	Student 2	Right.

Here, the students explicitly established relations between researchers having different opinions and the worth of the employed methodology (turn 13), linking also to the relevance of evidence (Turns 11–13). At the same time, they displayed appreciation of Empirical NOS (Cons), making a distinction between different types of data (results vs ‘only an estimate’, Turn 9). Also, appreciation of Subjective NOS is evident in that they accepted that different researchers may have different opinions (Turns 11-14).

A third rationalisation for disagreement was to invoke conceptual change (WS) as something inherent when new research challenges old beliefs (also suggesting an appreciation of Tentative NOS). This was most pertinently expressed by group 3 (Supplementary material Appendix B, excerpt B3).

3. Peer review missing

Unlike the first two salient outcomes, our analyses suggested poor understanding of the role of Peer review (WS) and the Social certification of knowledge (FRA) when discussing disagreement between scientists. In all groups except one, these themes appeared in connection to Q4. Moreover, only group 1 displayed some understanding of peer review as central to the scientific process. This understanding was hinted at already as they stated that disagreement is part of the scientific process (Supplementary material Appendix B, excerpt B1), and was further touched upon as these students addressed the issue of who is responsible for the dissemination of research in the media (Q4c), as seen in excerpt 3.

Excerpt 3.

Table

15	Student 1	Right. And it, well, no, but the whole thing. So, then it might be more important that sort of the scientific process is honoured so that what gets out to the public …
16	Student 2	Yes
17	Student 3	… gets the same status as the bad stuff, yeah.
18	Student 1	That, yeah, no, but no, that the bad stuff doesn’t sort of get out before it is, you know …
19	Student 3	Before it’s proven, yeah.
20	Student 1	… proven and sort of it’s, of course everything doesn’t get proven, but you know, but goes through … [interrupted by the teacher]

The students hinted at a process of internal scientific communication that should occur before results are communicated to the media (Turns 15 and 18). It is true that they never established relations to ‘communication’ or ‘peer review’. But considering other instances in which they established relations between the scientific process and disagreement between scientists, it is reasonable to assume that they included communication between scientists as part of the process of establishing scientific knowledge (Turns 15 and 20). At the same time, these students established relations indicating the somewhat unrealistic view that a process of purely scientific communication can take place separate from communication to and in the media (Turns 15–20). It is as if there were no science journalism or informed citizens capable of reading scientific papers.

None of the other seven groups established any similar relations between communication and the scientific process. Some groups allowed for scientists presenting their results freely without consideration of how the results are received. But none made any connections between such publication and the certification of scientific knowledge, nor did they distinguish clearly between presenting results to peers and to media and the public. Excerpt 4 from group 4 exemplifies this.

Excerpt 4.

Table

21	Student 1	Yes. But those who publish need to … think a little extra.
22	Student 2	Yeah. Maybe more responsibility, yeah well …
23	Student 3	But how can they be more responsible? Cause they, they only present their results. And then what’s done, they can’t control that.
24	Student 1	But … they have to check that …
25	Student 3	Cause they can’t be after everyone. I mean that’s simply impossible.
26	Student 1	No, that’s true.
27	Student 3	It’s only because really, what I think, about research results that people for example only take part of the research that agrees with, your own opinion and then it’s a … misinterpretation of the whole, research result.
28	Student 1	Yes.
29	Student 3	Like that. The researchers can’t be blamed for that.

Relations such as ‘those who publish – think a little extra’ (Turn 21) and ‘they – only present – their results’ (Turn 23) indicate no role for scientific communication for establishing scientific knowledge and suggests that they primarily connected communication by researchers with disseminating results to the public. Still other groups even expressed that publication of research results should not occur at all until it has become established knowledge (Supplementary material Appendix B, excerpts B4 and B5).

4. Accepting vs. dealing with disagreement

Although students accepted disagreement as unproblematic, they struggled with how to cope with it as citizens. When making the move from appreciating that scientists disagree to reasoning about how to act on this fact, the students took recourse either to standard attitudes possibly learned in school or to ad hoc solutions.

Almost all groups invoked open-mindedness together with source criticism as ways of coping with disagreement. Excerpt 5, in which group 3 discussed Q3, is an example of this.

Excerpt 5.

Table

30	Student 1	So how … how should you act, if you see two researchers who disagree?
31	Student 2	In that case you read about it.
32	Student 1	Right
33	Student 2	And check what you think. And choose that which …
34	Student 3	Yeah, you have to choose yourself who …
35	Student 2	Yeah
36	Student 1	Yes, you can choose for yourself. Or you wait until they agree.
37	Student 3	Yes.
38	Student 1	But you never know, that could take like a hundred years. Uhm …

This group first filled the gap of how to act in the face of disagreement with the relations ‘read about it’ (Turns 31–32; ‘source criticism’, although this feature was more obvious in some other groups) and ‘choose for yourself’ (Turns 33–36; ‘open-mindedness’) as well as to wait until disagreement ends (Turns 36–37), noting though that this may take too long to be reasonable (Turn 38).

In addition to this general feature, other more ad hoc ideas emerged in individual groups. Group 6 suggested that when experts disagree, we should mainly rely on our basic knowledge of how things work combined with common sense, rather than wait for quarrels on empirical evidence to settle (Supplementary material Appendix B, excerpt B6). Another example is the sequel to group 3’s discussion in excerpt 5, in which they suggested a risk analysis and risk minimisation (Supplementary material Appendix B, excerpt B7). Some groups left out any considerations of empirical evidence, suggesting that you simply decide on personal preference.

Discussion

Our purpose in this study was to examine what possibilities for addressing NOS emerged as students engaged in discussions around a contemporary historic narrative on disagreement in science connected to a recent SSI. As Zeidler (2014) noted, the ways that NOS may become interwoven with SSI contexts are subtle and complex, and require critical pedagogical skills on the part of the teacher as well as research-based knowledge about how NOS and SSI interact in students’ reasoning. We contend, therefore, that although we did not inquire into whether or what students eventually learnt from the narrative, the kind of more open-ended knowledge generated in this study, showing the range of possibilities for potentially addressing NOS that arise as students encounter this kind of narrative, constitutes important basic knowledge for teachers and researchers who want to design NOS-teaching through SSI. The students did address NOS to a considerable degree when discussing the narrative. The consistent appearance of a common core corresponding to the most central NOS in the narrative suggests that the narrative may be reliably used as an authentic point of departure and as a concrete example for teaching about the interplay between empirical and subjective aspects of NOS at minimum, and most likely also concerning the remaining six core themes (cf. Table 3). At the same time, the finding that five of the themes surfacing in the groups did not have any correspondence in the narrative suggests that it has the potential to also constitute a basis for discussions around NOS that are more closely tied to what the students happen to bring to the discussions.

However, it was also possible to glean some information concerning the students’ understanding of NOS from their discussions. We grant that the understanding that we could discern cannot be equated to that derived from specifically designed instruments or interviews. But that kind of reliable assessment is not necessary for the purposes of this study. Every science teacher knows that they cannot use classroom discourse alone to unequivocally assess individual students’ understanding. Yet, they also know that however piecemeal and incomplete, such information is crucial for making decisions on how to proceed in class. Thus, any observed variation in students’ understanding arising from their discussions on the narrative constitutes a potential resource for teaching NOS with increasing complexity and detail. This finding, then, adds an important dimension to the possibilities of using the narrative as a basis for addressing NOS. For instance, despite being remarkably comfortable with the existence of disagreement in science and showing reasonable understanding of basic NOS, students nevertheless struggled with how they should cope with situations where experts disagree. This discrepancy between general acceptance of disagreement and facing it in concrete situations opens for deepened discussions of NOS. For example, it can be used as an opportunity for having the students reflect on the more detailed and complex reasons for why scientists disagree, such as differing judgments on completeness or relevance of evidence (which some groups did in fact touch upon). The discrepancy may also be taken as a starting point for distinguishing between different contexts for dealing with uncertainty, such as settling an issue for purely scientific reasons (which may very well ‘take a hundred years’, cf. excerpt 5) vs. having to decide right away regardless of uncertainty (for instance whether to lock down society in the face of a new virus). Sadler et al. (2004) observed a larger variation than we did, both in terms of understanding of empirical NOS and how students rationalised that scientists may come to different conclusions from similar evidence. However, the studies differ in the framing of the SSI and in how opposing viewpoints were framed (specifically authored science briefs of equal weight in Sadler et al. vs. authentic narrative of arguments of unequal weight in our case, see Supplementary material Appendix A).

Moreover, the poor appreciation of communication between scientists as a crucial part of establishing knowledge, and confusion over scientific communication in the media, may be said to partly offset students’ more informed appreciation of other NOS-themes (Allchin, 2011; Priest, 2013). In any case, that students appreciate that scientific knowledge is empirically based and subjective but lack basic understanding of the processes through which this knowledge is established, constitutes important information for a teacher to consider when planning for how to build teaching off the narrative. To begin with, given students’ unawareness of it, simply introducing peer review as a phenomenon seems to be a clear possibility, as well as broadening their ideas about communication both in and of science through different media. In addition, however, the group discussions suggest ample opportunity to deepen students’ NOS-understanding by making them reflect on possible inconsistencies between simultaneously accepting Empirical and Subjective NOS and disregarding how empirical evidence turns into scientific knowledge.

The suggestions above are meant only as examples of possibilities for addressing NOS through the narrative presented here. More generally, our results could be interpreted as a promising example of the potential for addressing NOS through socio-scientific issues by framing them in the form of narratives from contemporary history of science. Possibly, this approach facilitates the retention of complexity and authenticity of the historic narrative since the science of SSIs is often highly contextual science-in-the-making (Christensen, 2009; Tytler et al., 2001). In that way, some of the critique against using simplified and linear narratives of the history of established science is accommodated (Allchin, 2003; Metz et al., 2007). The other way around, presenting the SSI as a narrative allows teachers to have students discuss its foundations and general aspects of decision-making in the face of disagreement on the science behind the SSI, instead of the common but criticised practice of asking students to take decision on the SSI as such (Allchin, 2011; Oulton et al., 2004).

We ended up with the current SSI and the current historic narrative on disagreement because of the specific competency in our research group (Schenk et al., 2021). But disagreement is intrinsic to science and plays a crucial role in how citizens cope with socio-scientific issues (Ryder, 2001), and other examples prevail. As an obvious case in point, the COVID-19-pandemic will likely soon be officially over and will then turn into the sort of contemporary history of science that was used here, providing material for narratives ranging from the science behind the use of face masks to possible side-effects of the vaccine. Given the results presented here, mining contemporary history of science for similar cases seems like a potentially fruitful way for preparing young people to deal with socio-scientific issues in the face of uncertainty and conflicting evidence.

Ethics statement

The study complies with the research ethics guidelines of the Swedish Research Council (2017). Ethical considerations and measures were described in the grant application (Grant number 2017-00028) and approved, including that the study did not need additional approval from the Swedish Ethical Review Authority.

Acknowledgment

We would like to thank the physics teachers at Tumba gymnasium for valuable input to the assignment.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Skolforskningsinstitutet under Grant number 2017-00028.