Foxes caught in the same snare: a methodological review of social radon studies

Abstract

Mitigating risk from exposures to indoor radon is a critical public health problem confronting many countries worldwide. In order to ensure effective radon risk management based on social scientific evidence, it is essential to reduce scientific uncertainty about the state of social methodology. This paper presents a review of methodological (best) practices, and sensitivity to bias, in research on public attitudes and behaviours with regards to radon risks. Using content analysis, we examined characteristics of research design, construct measurement, and data analysis. Having identified certain challenges based on established and new typologies used to assess methodological quality, our research suggests that there is a need for attention to (limitations of) cross-sectional design, representative and appropriate sampling, and a pluralist approach to methods and analysis. Furthermore, we advocate for more comparative research, rigorous measurement and construct validation. Lastly, we argue that research should focus on behavioural outcomes to ensure effective radon risk management. We conclude that for any field to thrive it is crucial that there is methodological reflexivity among researchers. Our recommendations serve as a useful guide for researchers and practitioners seeking to understand and enhance the rigor of social methodology in their field.

Keywords:

• radon
• methodological review
• health behaviour
• public attitudes
• methodological recommendations

Introduction

In order to evaluate and solidify (future) contributions of social sciences to ionizing-radiation risk management, it is essential to take stock of the substantive findings and the methodological state of the art. While researchers continuously do the first, they are far less active with regards to the latter. This is a pity given that conclusions drawn from research are only as solid as the underlying methodologies (Bouder et al. 2021; Van Oudheusden, Turcanu, and Molyneux-Hodgson 2018). Of course, it is impossible to study the state of social methodology of the whole domain of risk research. In line with Verhagen et al. (2001), it makes more sense to focus on a specific research question(s), in order to minimize heterogeneity of outcome measures. In this study we focus on studies on public attitudes and behaviours towards radon risks.

Background on the field of Radon

Indoor radon is the most important source of exposure to ionizing radiation and one of the primary causes of lung cancer in the general population (World Health Organization 2009). To protect against the dangers of this radioactive gas, legal frameworks were established in national and supranational regulation such as the European Basic Safety Standards Directive. The most recent amendments to the Directive introduce legally binding requirements related to radon. Article 103 in particular stipulates that Member States should develop national radon action plans. Member States are also required to inform the public about the importance of performing measurements and reducing radon concentrations in their homes (European Union 2013, p. 37).

While protective measures for radon measurement (testing) and radon risk reduction (remediation) are relatively cheap and accessible, and considerable efforts have been undertaken by authorities to promote them, the uptake of protective measures to desired levels is still not achieved (Cholowsky et al. 2021; Perko and Turcanu 2020).

Potential explanations point to a gap in the rate and rigor by which social science insights are incorporated into radon risk management (Bouder et al. 2021; Turcanu et al. 2020). Social scientific insights can clarify the uptake of protective measures and the factors (e.g. attitudes) influencing their implementation or lack thereof, hereby guiding risk management and improving performance (Renn and Benighaus 2013). Hence, the need for more scientifically-based action (Lacchia, Schuitema, and Banerjee 2020).

However, flaws in design, analysis or reporting, can cause effects to be under- or overestimated, and thereby obstruct effective evidence-based radon management (Leviton 2017; Higgins et al. 2011). To allow policymakers and researchers to determine the trustworthiness and applicability of evidence, they need to be aware of methodological quality.

To date no comprehensive overview of methodological criteria has been published specifically with regards to social methodology for radon research as nearly all reviews in the field focus on exposure to radon (e.g. Su et al. 2022), associations between radon and lung cancer (e.g. Rodríguez Martínez, Ruano Ravina, and Torres Duran 2018), mitigation strategies (Khan, Gomes, et al. 2019), or a specific social method (Martell et al. 2021).

However, to define and evaluate the quality of social methodology, we can draw on various research traditions. An extensive criteriology, developed by (statistical) experts, exists in the field of epidemiology research (Deeks et al. 2003; Sanderson, Tatt, and Higgins 2007; Verhagen et al. 1998; Vandenbroucke et al. 2014). Methodologists from this field, tend to focus on e.g. the study design, setting, sample design, and measurement of variables. They are mainly concerned with the validity of the design, and the likelihood that a research design generates unbiased results (Verhagen et al. 2001). Our evaluation criteria were strongly influenced by these guidelines. Although the validity typology is historically intertwined with experimental research (see Campbell, Stanley, and Gage 1963; Cook and Campbell Donald 1979), it has also extensively been used in social sciences (Shadish 2010; Neall and Tuckey 2014). In practice, we reflect on external validity. Since, to be of use for radon management, evidence needs to be valid beyond a research sample. We also discuss the internal validity, because it is essential that inferences regarding causal relationships can be drawn about what influences people to (not) take protective action against radon.

We also relied on social methodological literature to inform criteria on methodological quality. Because of inherent pluralism, and reliance on direct public assessments, this literature recognizes the pragmatist tradition. According to pragmatist principles, research should be contextually situated without being committed to any philosophical tradition, and take a holistic approach (diverse methods, analytic techniques) to fully understand a problem, here radon (Creswell 2009). The underlying assumption is that research studies individually and in correlation with each other contribute to the quality of evidence, it is a shared responsibility (Loosveldt, Carton, and Billiet 2004). We therefore, do not consider the substantive content as solely important, and include additional criteria that are relevant to the evaluation of quality; such as awareness of design-specific sources of bias, and the degree of diversity in methodological approaches. Here, the completeness of evidence comes under scrutiny.

To summarize, in this review we quantify the nature and extent of limitations to methodological designs within the body of evidence on public attitudes and behaviours regarding radon. We also then reflect on best practices in minimizing threats to validity and bias, and give recommendations to improve the depth and reach of the evidence base.

Methods

In this paper we conducted a methodological review, which refers to a study where methodological characteristics of studies focusing on a similar topic are evaluated, in order to highlight limitations of design, conduct and analysis. Although there is no published tool for appraising the risk of bias in methodological studies, it should be regarded as a type of observational study, therefore bias was avoided by conducting a comprehensive reproducible search to generate a systematic and representative sample (Mbuagbaw et al. 2020).

Unfortunately, relevant research for the present methodological study is generally poorly indexed and dispersed across many different academic disciplines. We chose to direct our focus on peer-reviewed literature, since in principle here we should find methodologically rigorous research.

Article search

From the 23rd of November 2020 till the 1st of December 2020 we searched four databases: Web of ScienceTM, Scopus®, Medline and Sociological abstracts (SA). The search strings, designed in cooperation with an experienced university librarian, consisted of combinations between the risk of interest - ‘radon’ (1) and method-keywords (2). A separate search (3) was conducted for combinations with ‘radon’ and the methods ‘survey’ and ‘experiment’, due to the common use of such methods in natural sciences. In this separate search, limitations regarding the subject (4) were introduced. For an overview of the search strategy, see Table 1.

Table 1. Search string in web of science format.

#1 (Radon) #2 (((field OR case OR comparative OR cohort OR archival) NEAR/2 stud*) OR ((network OR content OR sentiment OR meta OR framework OR media OR discourse OR morphological OR text* OR conversation OR narrative) NEAR/2 analysis) OR ((systematic OR meta) NEAR/0 review) OR ((mixed OR mental OR mixed OR delphi OR q OR economic) NEAR/2 method*) OR “delphi techniq*” OR “focus group*” OR “repertory grid” OR “analytic induction” OR “life history*” OR historiography OR “socio mapping” OR “feeling thermometer” OR “cybermethod*” OR “participatory action” OR bibliograph* OR questionnaire* OR “secondary data” OR “e-research” OR “memory work” OR interview* OR observation* OR ethnography OR phenomenolog* OR RCT OR “randomized controlled trial*” OR workshop OR “public opinion” OR panel* OR omnibus OR poll OR triangulation OR hermeneutic) #3 (survey* or experiment*) #4 (public* OR citizen* OR participant* OR respondent* OR resident* OR person* OR stakeholder*)

Article selection

A total of 22.290¹ identified records were uploaded into Endnote and thereafter to the literature review tool Rayyan (Ouzzani et al. 2016). In the first automated stage 5.700 duplicates were removed.

In the second stage, three reviewers performed eligibility screening of 16.590 records based on title, abstract and meta-data. Eligible records included all peer-reviewed social radon articles, published in the English language. A sample of about 20% was screened by at least two reviewers, and conflicting decisions (30), were resolved in discussion with a third reviewer.

In the penultimate stage, the full texts of 240 articles that met the initial eligibility criteria were examined. Articles that could not be retrieved, presented no primary data (e.g. review articles) – to avoid duplicates, disclosed no methodological information and/or only mentioned, but did not investigate radon and psycho-social concepts, were excluded.

In the final stage of study selection, articles were snowballed from the references of the included database articles. The 28 articles resulting from snowballing were screened on the same criteria as the initial database sample (Figure 1).

Figure 1. Flowchart of the review process.

Data coding

A data extraction form, based on criteriology from classic epidemiological- (RCT) and social methodological research, was created in the online tool Qualtrics©. For each article, we extracted: 1) bibliographical information, 2) study aim, 3) study design, 4) time-frame, 5) population description, 6) sample strategy, 7) sample size, 8) geographical setting, 9) dependent and independent variables (concepts and indicators), 10) quality assessments (reliability, validity, and qualitative counterparts), 11) type of analysis, 12) unit of analysis, 13) analysis software, 14) data availability (e.g. database), 15) main findings and 16) ethical considerations (conflicts of interest, funding and privacy).

Following a pre-test, a total of ten trained coders extracted the data, and every individual article was reviewed by at least two independent coders. Conflicts (10%) between coders were resolved by additional two independent master coders. The extracted data was later grouped in categories during the stage of data analysis.

Data analysis

The systematically obtained articles were studied through content analysis, results are presented descriptively in counts and percentages. In line with previous methodological reviews, we grouped results on three dimensions: 1) research design, 2) construct measurement, 3) data analysis (Ketchen, Boyd, and Bergh 2008; Scandura and Williams 2000; Aguinis et al. 2009; Whitten et al. 2007).

Five components of ‘research design’ were considered: the overall design, time-frame, data collection method, design context and sample characteristics. For ‘construct measurement’ we looked at three components of construct validation: validity analysis, reliability tests, and validation techniques for qualitative research. Lastly, for the ‘data analysis’ we looked at the analytic technique and level of analysis. Table 2 provides an in-depth overview of the codes with each component.

Table 2. Taxonomy for content analysis.

Coding Scheme
Dimension	Component	Code
Research Design	Overall Design	Quantitative, Qualitative, Mixed
	Time-frame	Cross-Sectional, Longitudinal
	Method	Experiment, Survey, Interview, Focus Group, Content Analysis, Other
	Sample Characteristics	Type of Sample Sample Strategy Sample Size
	Design Context	Continent Level (local, regional national, international)
Construct measurement	Validity test for quantitative studies	Factorial, Construct
	Reliability test for quantitative studies	Test-Retest, Internal Consistency
	Types for Qualitative Studies	Interrater, Triangulation, Protocol pre-test
Data Analysis	Analysis Techniques	Descriptive, Association, (CO-)Variance, Regression, Network Analysis, Structural Equation Modelling, Content Analysis, Discourse Analysis, Narrative Analysis, Comparative Methods, Thematic Analysis, Citizen Science, Other, Not Reported
	Level of Analysis	Individual, Group, National

In addition to these data, we added reflections from the authors of the reviewed studies on the limitation (bias/validity) of their design and/or analysis choices articles. This is to provide insight into limitations of some strategies, and guidance for advancement in the field.

We also reviewed whether the studies focused on cognitive components (knowledge of radon, and perceptions), or behavioural components (testing and mitigation) over time. The cognitive components can be regarded as antecedents to radon behaviour, but we wanted to see if they were studied as such. The coding scheme for this section can be found in Appendix 1.

In order to observe trends over time we reviewed methodological characteristics over three approximately equal time-periods: 1987–1998 (first period), 1999–2009 (second period), 2010–2020 (third period).

Results

In Table 3, we give a longitudinal overview of our analytical sample. We evaluated 154 articles that were published over a 34-year period (see full list in Appendix 2). The number of publications has remained relatively stable, with an average of five articles per year. The relative occurrence of topical components, however, has somewhat changed.

Table 3. Summary of study topics.

Study Topic
	First Period N = 59 articles	Second PeriodN = 42 articles	Third PeriodN = 53 articles	Total^4N = 154 articles
Cognitive
Knowledge	36 (61%)	27 (64%)	35 (66%)	98 (64%)
Perception	45 (76%)	30 (71%)	32 (60%)	107 (69%)
Behavioural
Testing	46 (78%)	24 (57%)	38 (72%)	108 (70%)
Remediation	27 (46%)	14 (33%)	15 (28%)	56 (36%)
Behaviour as DV*	16 (27%)	4 (10%)	15 (28%)	35 (23%)

* Dependent variable.

The most studied components overall were radon testing and radon perceptions, represented in 70% and 69% of the articles, respectively, followed by knowledge (64%). A clear downward trend in interest to the topical component of perceptions can be observed: while 76% of the articles investigated this aspect of cognition prior to the year 1999, the relative frequency declined to 60% in present times. Radon testing was the primary component of investigation overall, but from 1999-2009 (second period) it was only discussed in 57% of the articles.

Remediation was consistently studied the least, compared to the other radon components such as risk perception or radon testing, never in more than 46% of the articles and in only 28% in the latest research period. Here too a downward trend in interest can be observed. Furthermore, in only 35 of the 154 articles, testing and/or remediation were studied as dependent variables.

While four major topics can be distinguished, there is great proliferation in concepts and dimensions studied within. Taking radon risk perceptions as an example, Coleman (1993) makes the distinction between personal risk judgements and societal risk judgements. Other authors focus on the perception of the severity of radon risk and susceptibility to radon (Hazar et al. 2014) or radon-induced illness (Weinstein and Sandman 1992), or risk-synergies (Rinker, Hahn, and Rayens 2014). This proliferation suggests that many of the social constructs studied within the field are multidimensional and complex.

All mentions of social constructs were given equal weight in Table 3, although not all studies went into equal depth on the topic because radon was not the primary focus in all articles. A total of 55 articles (37%) investigated radon as part of a broader study on environmental and/or radiological health risks. Park, Scherer, and Glynn (2001) for example, measured people’s risk judgement about four health issues; water contamination, AIDS, heart disease and radon. Some studies considered also interaction effects, particularly the synergistic risk between radon and smoking is frequently explored (e.g. Hampson et al. 2003). The distinction between primary and secondary interest matters because studies exclusively investigating radon have a higher level of detail, and can thus capture behaviours and attitudes more accurately.

Research design

Study design

Concerning methodological approaches, we first looked at the study design. Choices in study design reflect how we see, collect and analyse data. Quantitative explanatory research allows for hypothesis testing with regards to behaviour. Qualitative research, if some criteria are met, allows for more in depth study of the context of cognition, affection and behaviour. No choice is per definition superior to another, as both approaches have their limitations. We would thus, from a pragmatist perspective, expect to find mixed-methods research or relatively balanced frequencies of qualitative and quantitative methods, in order to compensate weaknesses in a single design.

Quantitative research approaches dominated in our analytical sample and were used in 82% of all articles, primarily explanatory (44%), but also descriptive (38%) design (Table 4). Qualitative approaches (10%) and mixed methods (7%) were only marginally used, together only in 27 out of 154 articles. The use of quantitative explanatory research, declined over time, as did the use of mixed methods.

Table 4. Summary of study design.

Study Design
	First Period N = 59 articles	Second Period N = 42 articles	Third Period N = 53 articles	Total N = 154 articles
Overall design
Quantitative Explanation	30 (51%)	16 (38%)	22 (42%)	68 (44%)
Quantitative Description	18 (31%)	19 (45%)	21 (40%)	59 (38%)
Qualitative	5 (8%)	5 (12%)	6 (11%)	16 (10%)
Mixed	6 (10%)	2 (5%)	3 (6%)	11 (7%)
Method
Survey	38 (64%)	30 (71%)	38 (72%)	108 (70%)
Experiment	12 (20%)	6 (14%)	7 (13%)	24 (16%)
Interview	8 (14%)	5 (12%)	3 (6%)	16 (10%)
Focus Group	2 (3%)	4 (10%)	1 (2%)	7 (5%)
Content Analysis	1 (2%)	1 (2%)	1 (2%)	3 (2%)
Other	3 (5%)	4 (10%)	4 (8%)	10 (6%)
Time frame
Cross-sectional	41 (69%)	32 (76%)	47 (89%)	122 (79%)
Longitudinal	18 (31%)	10 (24%)	6 (11%)	32 (21%)

Use of data collection methods (n = 168² methods used) follows that of research approaches. In 70% of all articles surveys were used, a consistently popular method over the years, and especially used in the most recent decennia. Moreover, 16% of the articles conducted experiments, amongst them survey-experiments. Other methods such as interviews, focus groups, media content analysis, educational activities and eye-trackers were less used.

With the primacy of survey-methodology in the field comes awareness of its limitations. Almost 30 articles refer to challenges inherent to surveys. In survey interviews, the presence of a researcher can cause respondents to answer what they think the researcher wants to hear, meaning there is a social desirability bias (remarked by i.a. Cronin et al. 2020). However, self-reports are also susceptible to social desirability (Neri et al. 2018; Gleason, Taggert, and Goun 2021 Nov-Dec 01; Nicotera et al. 2006). Respondent answers related to action (testing and/or remediation) sometimes reflect more how they want to be perceived than what they will actually do or have done (Losee, Shepperd, and Webster 2020; Jones, Foster, and Berens 2019).

Describing past actions in particular, poses a risk of recall bias leading to misclassifications (Cronin et al. 2020; Smith and Johnson 1988; Smith et al. 1988). Question wording plays a crucial role too. For instance Denu et al. (2019) notes that when presented with the question ‘Have you tested for radon in this home?’, respondents can focus on the ‘you’ and report ‘no’ even though someone else did or coordinated testing (Denu et al. 2019).

Closely related to the issue of prospective and retrospective measurement is the time frame in which studies are conducted. In the majority of reviewed articles observations were made one-off, 79% are cross-sectional studies, and only 21% studied subjects over an extended period of time through longitudinal design in order to assess for instance if intentions to test or remediate have been translated to action (e.g. (Ford et al. 1996; Smith, Desvousges, and Payne 1995). There is also a steady increase in the relative frequency of cross-sectional articles, from 69 % in the research produced before the year 1999, to 89% after the year 2010 and a complementary decrease in longitudinal design in the respective time-windows.

The evolution towards cross-sectional studies is counter-intuitive considering that approximately 20 articles mention it as a limitation of the study. Authors remarked that such studies do not allow to determine the temporal relationship between variables (e.g. was the public that tested for radon more knowledgeable of radon prior to testing, or did they become more knowledgeable because they tested). Furthermore, non-experimental cross-sectional design does not allow for causal inferences, and there is strong reliance on potentially inaccurate self-reports of future behaviour, whereby respondents give socialy desirable answers to behavioural intention questions, but don’t actually change their behaviour (Hampson et al. 2000; Laflamme and Vanderslice 2004; Hill, Butterfield, and Larsson 2006; Rinker, Hahn, and Rayens 2014; Huntington-Moskos et al. 2016; Cronin et al. 2020; Larsson et al. 2009; Poortinga, Cox, and Pidgeon 2008; Nwako and Cahill 2020; Martin et al. 2020; Poortinga, Bronstering, and Lannon 2011).While we have seen that many studies try to circumvent the limitation of a one-off design through retro- and prospective questions, for instance by asking people if they “have taken actions to reduce radon exposure” or “intended to engage in remedial actions” (e.g. Mazur and Hall 1990), these are susceptible to bias due to unknown practical barriers (e.g. ability and willingness to pay for mitigation), and social desirability.

Population sampling

Three parameters are important for population sampling: who are the respondents, their number and how they are selected. In Table 5, we summarized the characteristics of the studied populations. When more than one population was studied in the same article, all of these were taken into account.

Table 5. Summary of sample characteristics.

Sample Characteristics
	First Period N = 59 articles	Second Period N = 42 articles	Third Period N = 53 articles	Total N = 154 articles
Sample Type
Residential population	57 (96%)	33 (79%)	39 (74%)	129 (84%)
Professional Population	6 (10%)	9 (21%)	9 (17%)	24 (16%)
School Population	3 (5%)	6 (14%)	12 (23%)	21 (14%)
Specific Population	1 (2%)	3 (7%)	6 (11%)	10 (6%)
Sample Strategy
Probability Sampling	27 (46%)	16 (38%)	15 (28%)	59 (38%)
Non-Probability Sampling	34 (58%)	33 (79%)	40 (75%)	105 (68%)
Sample size
Average for Quantitative	2554	2681	790	2008
Average for Qualitative	139	45	175	120

Social radon research has been orientated mostly towards the general residential population (84%). Frequently, however, a detailed description of the study sample was missing. In 88 out of 129 articles the respondents could at most be classified as ‘occupants’. In 37 articles, however, the authors specified that they studied homeowners, in three articles - tenants, and in one - landlords. This distinction matters, because people can have different responsibilities with regards to testing and remediation, and differences in socioeconomic status may affect results if not controlled for (Hahn, Rayens, Kercsmar, Adkins, et al. 2014; Hahn, Rayens, Kercsmar, Robertson, et al. 2014; Rinker, Hahn, and Rayens 2014).

Studies were also interested in the cognition and behaviour of professionals (16%), specifically health professionals (13), scientific experts (6), and workers in public and private industries such as real estate officers and government employees (5).

Over time, interest in schools increased. Interest in specific other populations also increases. For the school population the relative frequency increased from 5% to 23%, meaning more university students (8), university staff (2), school children (4), school staff (3) and day-care personnel (1) were studied. This also includes research on an aggregate school (2) or day-care (1) level. The category of specific populations includes: parents or guardians (5), media publics (3), radon mine visitors (1) and tourists (1).

Regarding the selection of respondents, random selection was reported in 38% of all articles; including random sampling (55), systematic sampling (3) and cluster sampling (1). Yet, the relative frequency of random sampling undergoes a decline over time, from 46% of studies in the first observed time-window to 28% in the most recent one (from 30 to 15 studies in absoute terms). Non-probability sampling strategies were used in 68% of studies and included convenience sampling (94), quota sampling (9) and snowball sampling (2). This relative frequency increased to 75% in the most recent time period.

It is important to note that sampling strategies were often poorly described, and that for at least 44% of the articles classification was made by the reviewers based on the description of the sampling procedure. Furthermore, justifications for using a sampling technique were rarely given in terms of appropriateness for the research aim and design. As a consequence, sampling strategies did not always align with study objectives. For instance, sampling parents to assess the knowledge and perceptions of the national population at large.

In nearly 70 articles, authors recognised the prevalence of non-probability sampling techniques as a clear limitation. They acknowledged that results from non-random samples cannot be generalized to more diverse populations because of, among others, socio-demographical differences (e.g. Adams, Dewey, and Schur 1993; Baldwin, Frank, and Fielding 1998; Clifford, Hevey, and Menezes 2012; Cronin et al. 2020; Evans et al. 2015; Sanborn et al. 2019; Huntington-Moskos et al. 2016; Butler et al. 2017). Only in 23 articles the authors state, legitimately or not, that findings can be generalized to the broader population. Thus, the studies are limited in external- and internal validity. Furthermore, convenience sampling risks a self-selection bias towards those most interested in radon, or who have recently tested their house (Clifford, Hevey, and Menezes 2012). This has consequences for conclusions which can be drawn from such data.

However, random samples too, are sometimes difficult to generalize if sample sizes are low. The average sample size has steadily declined for quantitative studies, from 2554 respondents on average per article in 1987-1998, and 2681 respondents in the second wave of research, to 790 respondents on average in the most recent period. All samples fall in the range between the maximum size of 125.000 respondents (Kendall et al. 2016) and the minimum of 44 respondents (Evdokimoff and Ozonoff 1992).

Even with large sample sizes, studies can still be limited by low (item) response rates, as people may be unwilling (Neri et al. 2018) or unable to respond (Ford and Eheman 1997). For example the study by Ford and Eheman (1997) originally had a nationally representative sample of 40.000 respondents, however only 70% had heard of radon, and only 5% had tested, which resulted in large standard errors and low external validity, as stated by the author of the study. Large nationally representative samples may thus not fit every radon study objective.

In qualitative research generalizability is not the objective and consequently sample sizes are generally smaller. The average sample sizes observed in the reviewed articles were considerable for what is common in qualitative research, with an overall average of 120 respondents per study. The sizes range from 4 interviews (Alsop and Watts 1997) to 1491 telephone calls (Macher and Hayward 1991). There were two studies for which the exact sample size could not be determined (Immé et al. 2014; Mazur 1987).

Geographical context

When describing the studied populations, the geographical context becomes relevant as this too is a defining characteristic of the respondents. In Table 6, we describe the location and administrative level on which the studies were conducted. The study location has changed considerably over time, but North American literature (71%) still dominates the field, namely the United States (104) and Canada (6). The relative a frequency of studies conducted in North America declined from 92% to 58% in the latest time period. The latter is linked to the rise of European studies (23); United Kingdom (13), Italy (7), Sweden (4), Ireland (3), Italy (3), Switzerland (3), Romania (1), Belgium (2), The Netherlands (1), Finland (1), Bulgaria (1) and Slovenia (1). Pioneering research was conducted in the United Kingdom, Ireland and Sweden. From the 2000s on, we see research conducted in Asia, first in Pakistan (3) and later also in Turkey (2), Saudi-Arabia (1), Iran (1) and Korea (1).

Table 6. Summary of context characteristics.

Context Characteristics
	First Period N = 59 articles	Second Period N = 42 articles	Third Period N = 53 articles	Total N = 154 articles
Continent
North America	54 (92%)	25 (60%)	31 (58%)	110 (71%)
Europe	5 (8%)	14 (33%)	17 (32%)	36 (23%)
Asia	0 (0%)	3 (7%)	5 (9%)	8 (5%)
Context level
Local	24 (41%)	19 (45%)	24 (45%)	67 (44%)
Regional	25 (42%)	14 (33%)	17 (32%)	56 (36%)
National	10 (17%)	9 (21%)	11 (21%)	30 (19%)
International	0 (0%)	0 (0%)	1 (2%)	1 (1%)

The leading role of literature from the United States has its implications for radon research as many findings and measurement instruments originate from this one country. Language and cultural differences can influence the accuracy and applicability of these instruments elsewhere. Consequently, there is uncertainty with regards to the generalizability of the findings and the policy advice to settings with different demographic compositions, cultures, media-, economic- and or political systems and histories. From the review we learned that the degree to which people have tested for radon, their risk perceptions and knowledge varied in function of communication strategies and other radon policies e.g. related to financial support (Hahn, Rayens, Kercsmar, Robertson, et al. 2014; Neri et al. 2018; Petrescu and Petrescu-Mag 2017; Hazar et al. 2014; Jones, Foster, and Berens 2019). It is thus important to take context characteristics into account when studying peoples’ cognition and behaviour (Poortinga, Cox, and Pidgeon 2008).

Comparative research can shed light onto the transferability of results across settings, which is why we looked at the level on which a study was conducted. Most frequently, the setting was defined on a local (67) or regional (56) level. Regional studies are those in which a geographical region or (American) state is mentioned, while local studies described one or multiple village(s), city/cities or province(s). To a lesser extent settings were described on an aggregate national (30) - meaning country - level. There was also only one international study, which studied the Belgian and Slovenian population (Perko et al. 2012). However, despite little cross-country comparison, tens of studies included participants from different regions within the same country, so there was some comparative research on a lower geographical level.

Construct measurement

Reliability tells us something about how reproducible findings are under the same circumstances, controlling for random error, while validity determines the extent to which methods measure what they claim to measure (Mohajan 2017). From Table 7, we deducted that most studies did not report on construct validation.

Table 7. Summary of construct measurement reports.

Construct Measurement
	First Period N = 59 articles	Second Period N = 42 articles	Third Period N = 53 articles	Total N = 154 articles
Validity
Factorial validity (Explanatory)	1	1	2	4
Construct validity (Confirmatory)	1	2	1	4
Reliability
Test-Retest	0	0	0	0
Internal Consistency	5	3	8	16
Measurement for Qualitative Studies
Interrater Agreement	1	4	1	6
Triangulation	0	1	0	1
Protocol pre-test	1	0	0	1

Table 8. Summary of data analysis techniques.

Data Analysis
	First Period N = 59 articles	Second Period N = 42 articles	Third Period N = 53 articles	Total N = 154 articles
Analysis Techniques
Descriptive	15 (25%)	10 (24%)	12 (23%)	37 (24%)
Association Chi-square (37) Fisher’s exact (3) Pearson correlation (6) spearman’s p (3) Gamma correlation (2) Intra-class correlation (2) Kendall correlation (1) non-specified correlation (15)	25 (42%)	17 (40%)	27 (51%)	69 (45%)
(CO-)Variance T-test (24) ANOVA (12) MANOVA (2) ANCOVA (1) Mann-Whitney U (5) Kruskall-Wallis test (9) Wilcoxon (2) Tukey’s test (2)	19 (32%)	13 (31%)	25 (47%)	57 (37%)
Regression Logistic regression (38) Linear regression (25) Binominal regression (4) Poisson regression (1) Multilevel regression (2) Probit regression (2) Tobit modem (2) Multiple discriminant analysis (1)	34 (58%)	16 (38%)	25 (47%)	75 (49%)
Network Analysis	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Structural Equation Modelling	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Content Analysis	0 (0%)	0 (0%)	2 (5%)	2 (2%)
Discourse Analysis	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Thematic Analysis	0 (0%)	0 (0%)	3 (7%)	3 (2%)
Comparative Method (Grounded Theory)	1 (2%)	1 (2%)	0 (0%)	2 (2%)
Citizen Science	0 (0%)	1 (2%)	3 (7%)	4 (3%)
Other	1 (2%)	4 (10%)	0 (0%)	5 (3%)
Not reported	8 (14%)	4 (10%)	2 (5%)	13 (10%)
Level of Analysis
Individual	34 (58%)	25 (60%)	27 (51%)	86 (56%)
Group	23 (39%)	15 (36%)	25 (47%)	64 (42%)
National	2 (3%)	2 (5%)	2 (4%)	6 (4%)

The highest frequency of reporting was found for reliability (16), with the preferred reliability criterion being Cronbach’s alpha. This psychometric test gives an indication of internal consistency, i.e. how well multiple items measure the same concept. A potential explanation for the low rate of reliability testing is the use of single indicators, in 60% of the articles. This suggests that much of the research assumes constructs are unidimensional, without verification. Furthermore, lack of test-retest reliability or reliability across time assessments, could likely be explained by the prevalence of cross-sectional studies.

Validity assessments were even less frequently performed than reliability assessments; only in 8 articles construct and/or factorial validity were evaluated. These measures give indications of how well the measures fit the theories for which they were designed.

Notably, 20 articles reported using previously validated instruments, without reporting a reliability or validity measurements themselves. However, using existing instruments does not guarantee that the produced results will correspond to real-world values. Factors such as characteristics of participants, the setting of the study and the stimuli, could differ from the original research from which the instruments originate and affect psychometric properties.

Reliability and validity can only tell us something about the quality of quantitative research, for qualitative studies other procedures exist for assessing rigor: interrater agreement (6), triangulation (1) and protocol pre-testing (1). However, in the most-recent time period such an assessment was nearly absent.

Data analysis

Lastly, we looked at how data was analysed for radon research and whether analyses allowed for theory testing, and uncovering of complexities and mechanisms of radon behaviour (see Table 8).

We found that 24% of the articles exclusively presented descriptive statistics, a relative frequency that remained consistent over time. All other articles used multiple analytic approaches. However, among them are associations/correlations (45%), and (co)variances (37%), which also don’t allow hypothesis and causation testing.

Regressions were the most prevalent approaches used (49%). However, multilevel modelling, which is a very useful technique to simultaneously analyze individual and contextual data, was only used in two studies (Poortinga, Bronstering, and Lannon 2011; Poortinga, Cox, and Pidgeon 2008). Other lesser used techniques included citizen-science like educational activities (3%), content analysis (2%), thematic analysis (2%), and grounded theory (2%).

It is interesting to further diversify the techniques through which data are processed. For example, by using network analysis and structural equation modelling, multiple, indirect and/or interrelated influences between actors or variables could be modelled.

We also looked at the level of analysis, in order to know where the agency was usually situated in the social radon studies. Agency is primarily located at the individual level (56%), closely followed by the group-level (42%). The relative high frequency of the latter is explained by the prevalence of household studies, in which family members responded on behalf of their household. Furthermore, experimental studies compared groups over individuals. Since radon testing and remediation is in essence an individual responsibility, the lower level of analysis is understandable. However, systemic characteristics can influence behaviour too.

In 4% of the articles, the analysis was performed on an aggregate national level.

The methodological road ahead

Among the main findings, we identified six key methodological challenges related to data collection, amalgamation, and analysis which affect the advancement of radon risk research, by threatening the validity of findings. Studies in the field generally lack contextual comparison, methodological diversity, adequate samples, longitudinal inquiry, behaviour-explanatory focus and construct validation. In the following discussion, we will consider needs and directions for research and practice. Awareness of-, and attention to biases and threats to validity from methodological practices, could stimulate better, meaning validated and holistic, social research in the field of radon.

Attend to socio-cultural diversity

The ultimate purpose of social radon research is to learn how to improve the health of the public, through their knowledge and behavior, and to use this knowledge to inform public health practice via policy and other interventions. However, translation from research to practice often lags in the larger area of cancer prevention and control, because policy-makers are unable to determine the generalizability and breadth of applicability of research findings (Steckler and McLeroy 2008). Furthermore, radon is a global issue, but the socio-economic, political, cultural and historical specificities that shape behaviour differ over contexts (Poortinga, Cox, and Pidgeon 2008). Yet, it is impossible to test theories and interventions under all possible conditions, due to conceptual, methodological and logistic complexity, and generally costs (Salway et al. 2011). In applied disciplines such as radon, it is thus necessary to empathize and strengthen external validity.

The concept of external validity is dual. First, it refers to generalizability; or the degree to which findings are transferable to similar contexts in terms of inter alia, social- political, historical, and cultural characteristics. Second, it refers to applicability; or the degree to which findings are transferable to dissimilar contexts.

From the review we learn that there is great uncertainty in the degree to which results are generalizable to the same-, or applicable to contexts with different demographical compositions, cultures, media-, economic- and or political systems and histories (Hahn, Rayens, Kercsmar, Robertson, et al. 2014; Neri et al. 2018; Petrescu and Petrescu-Mag 2017).

Strengthening external validity requires that researchers better describe context features, as well as capture-, and explain those features that affect outcomes related to protective behaviour in the context of radon (Leviton 2017; Poortinga, Cox, and Pidgeon 2008).The latter can be achieved through replication of research in similar and dissimilar contexts, under conditions as close as possible to the original research (Steckler and McLeroy 2008). It can also be achieved through more (international) comparative research. Besides providing insights on how contextual characteristics impact health behaviour, it gives the opportunity to explore local, national and global influences on health, identify common barriers to protective behaviour, and finally, critically assess current policies, and interventions in order to identify effective practices and best-practices (Salway et al. 2011). Yet, from the review we know that there is only one cross-national comparative study, and only a few within-country studies encompassing different regions, alluding that context-dependency is seldom studied.

It is important to note that while cross-national research is an interesting avenue for further research, comparison of results warrants caution. Comparability issues arise whenever studies addressing the same question, here: how to motivate people to test and mitigate for radon, are compared over different cultures and time-periods. Meaningful comparison requires equivalence; using the same mode, sampling procedure, questions in terms of wording and meaning, in the same context (question order). The same approach is also needed for comparison within culturally diverse nations, with for instance multiple national languages (Weisberg 2005, p. 303–305). A challenging but innovative approach for the field of radon would thus be to develop new studies to be conducted in parallel in several nations.

Currently most instruments for measuring the attitudes and behaviours of the public originate from a single country. These instruments however, are not necessarily appropriate for use in comparative studies. When instruments are used in contexts other than those in which they were developed, adaptations to new populations (differing in language and/or country), settings, or periods, may be necessary. One well known form of adaptation is culture-sensitive translation. How such adaptations should look like, in order to remain consistent with the original theoretical foundation of concepts and theories, is an empirical question which warrants further investigation (Leviton 2017). There is thus a need for more cross-country construct validation.

All these considerations will likely become increasingly important, since we see that funding streams such as the European Commission’s Framework programmes increasingly encourage international collaborations and cross-national comparative research (Salway et al. 2011).

Use a multimethod approach

Social studies in the field of radon are dominated by survey research, represented in seven out of ten studies. As all methods do, surveys have limitations, and some were explicitly mentioned by the authors of the reviewed articles (Neri et al. 2018; Losee, Shepperd, and Webster 2020; Denu et al. 2019; Cronin et al. 2020). Although these accounts of potential sources of error are useful, we know that the field of survey research has made many advancements in documenting and addressing response-, and other survey errors (Battaglia et al. 2016; Tourangeau, Rips, and Rasinski 2000; Tourangeau 2020). For instance, to overcome recall bias, one could ask a second person in a household to fill in the survey, or use event timelines based on well-known national events or individual/local history, stimulating accurate situating of behavior in time (Moreno-Serra et al. 2022). However, we observe from the review, that knowledge about biases and remedial techniques from the larger survey research field, is outdated or incomplete in the field of radon. Furthermore, there has been little research measuring the impact of biases on findings. So, the field would benefit from introducing robustness tests or instruments (e.g. social desirability scale), to assess existing biases. We acknowledge that the use of additional instruments can introduce new risks, like survey overload and non-response. These may already cause problems in the field, so trade-offs and optimal approaches specific to the field of radon, need to be empirically evaluated.

Despite the predominance of the survey method in the reviewed studies, and the relevance of strengthening this method given its popularity, we do not want to take an isolationist perspective, and encourage the use of a narrow range or single method only. Although methods must serve the study aim, there are always multiple approaches to a research problem. Pluralism in views and skills is often recognized as an asset, so it is important that pluralism in methods for obtaining and analyzing data is too (Dow 2012).

Diversity in methods can stimulate collaboration and knowledge transfer across disciplines, and promote interdisciplinarity in the field of radon. Pluralism and mixing of data or methods also stimulates data triangulation, which can help validate claims and enhance the power of prediction. Mixing of methodologies (e.g. survey and interview), a profound form of triangulation, aids at transcending research paradigms and explaining contradictions and connections between quantitative and qualitative data (Olsen 2004; Petrescu and Krishen 2019; Shorten and Smith 2017). However, the main reason for advocating for more diversity is that it could increase methodological knowledge in the field of radon, and insights on how to overcome the radon risk.

Study longitudinal change

The radon research in this review was conducted in a time range of 34 years, with varying time frames, measuring past, current, or future behaviours. Some of these measurements are more reliable than others. Past behaviour may be less well remembered than current behaviour, and actual future behaviour may differ from predicted and reported future behaviour due to unforeseen or undisclosed hindrances (Cronin et al. 2020; Larsson 2015). This bias can partly be overcome through longitudinal research, more specifically panel analysis that studies the same individuals on the same characteristics over time, or by questioning respondents directly after they perform critical actions such as testing or mitigating. This would require cooperation between, social- and technical researchers and practitioners. Unfortunately, this type of studies are exceptional and paradoxically its prevalence is in decline.

Given that almost all psychological theories are theories of change (Wang et al. 2017), and the strive towards behaviour change is the essence of radon research, it is only appropriate to observe actual change and strengthen confidence in inferences about causality of predictors to this behaviour change, through use of longitudinal research. Other arguments for more longitudinal research are that the field would better understand temporal relationships between variables, evolutions in attitudes and behaviours over time, as well as time-duration of intervention effects (Peterson and Howland 1996; Hampson et al. 1998; Laflamme and Vanderslice 2004).

Optimize the sampling design

Most research studies a general residential population, but we now see a tendency towards the study of more specific audiences. There is still room to improve the detail by which the study sample is described, since individual differences e.g. with regards to homeownership, if not controlled for, can affect results, as discussed by (Hahn, Rayens, Kercsmar, Robertson, et al. 2014; Larsson et al. 2009).

Moreover, we see that only in about one in three studies, participants were randomly selected, and that use of this technique is declining over time. This means that in the majority of cases, results cannot be generalized due to self-selection bias. To allow for extrapolation of findings, researcher must randomly sample.

However, even studies with randomized sampling designs, suffer from a generalization issue and unviability of complex statistical modelling, due to small sample sizes resulting mainly from low response rates. Findings from the review indicate that if the objective is to understand predictors for testing, and especially remediation, samples should be drawn from areas with (likely) more testers and/or remediators, or at least by researchers familiar with the nuances of conducting research with- and defining of- radon risk populations. Large national studies (Ford and Eheman 1997) or omnibus studies focusing on several topics or risks, do not necessarily operate with optimal sample compositions. For instance, a considerable number of individuals with no or little familiarity with radon could leave (sophisticated) questions about radon unanswered, lowering the sample size. The central argument here is that appropriate sampling designs must be adopted in function of the research aim.

Lastly, more than half of the radon studies focus on differences between individuals, and when groups are studied this is usually in an experimental design. While this is not problematic per se, it neglects that civil society exists. Radon research would benefit from a social psychological approach, which evaluates the possible discrepancy between what someone does and (normatively) ought to do, as well as how individual actions are shaped by others through processes of social influence. This could be achieved through comparison between study populations, or analysis on aggregate and/or multiple levels.

Measure in the plural way

Scientific research aims at producing reliable and reproducible results. However our review points out that most studies used items of unknown construct validity. This methodological challenge seems to afflict the wider field of radiological protection (Perko et al. 2019). Obviously, this lack should not necessarily be attributed to researchers’ unwillingness to assess the quality of instruments, the underlying cause is likely that use of single items impedes validation to be performed.

Furthermore, there is a risk of respondent-related measurement errors, caused by respondents giving inaccurate or false answers. Researchers that seek to minimize such error must optimize their question-wording, and study the cognitive response process. According to Tourangeau, Rips, and Rasinski (2000) in all stages of the response process bias can occur. Many errors arise from misinterpretation of questions and/or instructions. Being uninformed about a topic also does not deter people from answering, and the answers of uninformed, or unmotivated, respondents differ from informed respondents, thus creating response bias (Goldsmith and Walters 2015). From the review we learn that those uninformed about radon are questioned, yet we do not know how this impacts the validity of the aggregated measures of public opinion and behaviour. Thus there is a need to identify and control for response bias. An explanatory text or video about radon, and a simplified and shorter questionnaire could be given to those with no-, or low radon knowledge. What the impact of this experimental approach would be on people’s knowledge, but also on their response style, needs further investigation. Ambiguity in the question-wording can also intrude comprehension. For instance ambiguity over the referent of radon risk (general/personal risk): “Are you aware of the health risks associated with exposure to radon?”, or location (current/past residence): “have you ever tested your home for radon”. The optimal conceptualization depends on the purpose of the study, but it is essential that researchers reflect on the wording they choose (Tourangeau, Rips, and Rasinski 2000).

By using multiple items, one can neutralize measurement error unique to some indicators, and measure complex constructs more rigorously. However, it is essential that research assesses whether the indicators measure the same underlying construct, by means of validity testing.

Take a multidimensional approach to study behaviour

From this review we conclude that a great proportion of radon studies focus on people’s knowledge and perceptions, and that use of single items can result in less nuanced measurement of these concepts. We also know that cognitive awareness of the radon risk; being knowledgeable and concerned is not the most important explanatory variable for taking action against radon (Johnson and Luken 1987; Cronin et al. 2020). One action, testing, has been studied the most, however only in few cases were the causal mechanisms explored. Furthermore, research on remediation is declining over time, even though detecting radon levels itself does not reduce the risk, indicating the need for insight on predictors of remediation (Weinstein, Man, and Roberts 1990). Good practice, of making the explanation of health behavior the central research aim, can be seen in studies of Mazur and Hall (1990) and Khan, Krewski, et al. It is also essential that complex constructs that explain behaviour, e.g. risk perceptions, are measured rigorously by capturing conceptually relevant dimensions from literature. For instance, Coleman’s (1993) study is dimension rich, in that she considered the affective and cognitive dimensions of risk, and distinguished between social and personal risk judgements. We conclude that there is a need for dimension-rich and behaviour focused radon research.

Conclusion

There is a growing consensus on the desirability to incorporate social insights into scientific risk analysis as well as the publics’ understanding of risk (Klinke, Renn & Goble, 2021). This trend is also apparent in the field of ionizing radiation where the large EURATOM Horizon 2020 Radonorm project that aims at managing risk from radon explicitly acknowledges the importance of social considerations. Yet, less attention is given to the specificities of social methodologies in the field of radon and the implications of methodological design choices on knowledge creation, bias and validity. The findings from our review are therefore new, and offer relevant insights to the field.

Our review of the social methodology in radon risk research shows that there is over-reliance on some, and disregard for other aspects of methodology, which limits internal, external, and construct validity of research. The most important threats were: limited consideration of context, predominance of cross-sectional evaluations and survey research, inadequate sampling, little construct validation and finally, insufficient causal explanations of testing and remediating behaviour. This knowledge is important for evaluating conclusions researchers have drawn, and can currently draw, about attitudes, and behaviours of people concerning radon.

Our recommendations to limit, or at least reflect on, methodological threats, can strengthen and expand the social scientific evidence on how to manage risks from radon by stimulating more diverse techniques for data collection and analysis, validation, and better methodological design specifications. This could generate evidence which is rich in perspectives, and robust, for policy-makers and researchers to use and build on. We also hope our review will contribute to improving the quality of reporting about methodology in social radon studies.

Nevertheless, more methodological reviews are needed to uncover specific challenges in other fields of risk research, and further develop critereology to evaluate social contributions. We also recommend more systematic sharing of research tools, to promote standardized and validated instruments.,

Finally, we want to acknowledge some limitations to our study. Due to search strategy being based on the presence of keywords in the abstract, we likely missed articles that did not describe their method in the abstract, which might particularly be true for qualitative research. Furthermore, some very specific method-keywords might be overlooked. However, we believe that the search strategy was appropriate for our study objective and that the most common methods in social science were included in the search. Second, our paper did not include grey literature, and we only focused on English-language publication. Therefore, we likely missed literature, particularly from Asia. However, our sample certainly includes accessible and peer-reviewed studies, these practices could inform policy or research. Lastly, we did not directly link the design-, and analysis characteristics of studies to research questions and hypotheses. Allthough this could be seen as a limitation, we indirectly captured the research objectives by only focusing on studies that measure public attitudes and/or behaviours regarding radon. There is also consensus that some practices, e.g. random sampling, reduce systematic bias and produce (quantitative) data of better quality. While other practices, e.g. longitudinal study, are not superior to cross-sectional research, they offer important insights to the literature through their unique perspective, which now may be lacking. So, our methodological recommendations remain valid regardless of the research objectives of the individual studies. Furthermore, validity threats from methodological choices are not unique to radon, the suggestions made can bring awareness, trigger reflection, and serve as guidance for the broader field of risk management.

Acknowledgments

We would like to thank Johanne Longva at the NMBU library for her help with the development of the search protocol. This project has received funding from the Euratom research and training programme 2019–2020 under grant agreement No 900009, as well as the Research Council of Norway, grant nr313070 and 223268.

Disclosure statement

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

Notes

1 An important pathway of radon to the atmosphere are the NORM (naturally occurring radioactive materials) residues. We thus initially conducted a wider search, the findings of this search fall beyond the scope of the present article. The NORM articles unrelated to indoor radon were excluded in the full text screening stage. The full search strategy can be found in appendix 3.

2 There are several articles in which more than one quantitative, qualitative or a combination of both are used, as such the number of methods in higher than the number of articles.

3 https://www.radonorm.eu/

4 First Decade: 1987-1999 1998, Second Decade: 2000-2010 1999-2009, Third Decade: 2011 2010-2020

Appendix 1.

Coding of study topics

Table

Construct	Definition	Variable
Knowledge (n = 98)	Measure of people’s consciousness (awareness) or the nature and risk of radon, health consequences and/or their diffuse or specific level and understanding of radon (knowledge) at the present time based on (factual) information or experience (Trevethan 2017), as well as the skill to cognitively process.	Knowledge personal Knowledge societal Awareness Self-efficacy Confidence in own knowledge
Perception (n = 107)	Measure of people’s beliefs, attitudes, judgments, and feelings, as well as the wider social or cultural values and dispositions that people adopt, toward hazards and their benefits (Hevey 2017).	Acceptability of risk Treat perception personal Threat perception societal Severity Susceptibility Emotional response Salience of risk Trust in authorities Synergistic risk Relative risk
Testing (n = 108)	Measure of people’s present or future action(s) regarding detection of indoor radon concentrations, as well as of direct barriers and/or facilitators to those actions.	Actions in past Intended actions Perceived ease/difficulty Reasons for (not) testing
Remediation (n = 56)	Measure of people’s present or future action(s) regarding reduction of indoor radon concentrations, as well as of direct barriers and/or facilitators to those actions.	Actions in past Intended actions Perceived ease/difficulty Reasons for (not) remediating

Appendix 2.

List of included articles

*Adams, Mary, James Dewey, and Paul Schur. 1993. “A Computerized Program to Educate Adults About Environmental Health Risks.” Journal of Environmental Health 56 (2): 13–17.

Åkerman, Jeanette, F. Reed Johnson, and Lars Bergman. 1991. “Paying for Safety: Voluntary Reduction of Residential Radon Risks.” Land Economics 67 (4): 435–446. doi: 10.2307/3146550.

Alaamer, Abdulaziz. 2012. “Radon Awareness among Saudi People in Riyadh, Saudi Arabia.” World Journal of Nuclear Science and Technology 2: 165–168. doi: 10.4236/wjnst.2012.24025.

Alsop, Steve, and D. Watts. 2000. “Interviews-about-Scenarios: Exploring the Affective Dimensions of Physics Education.” Research in Education 63: 21–32. doi: 10.7227/RIE.63.3.

Alsop, Steve, and Mike Watts. 1997. “Sources from a Somerset Village: A Model for Informal Learning about Radiation and Radioactivity.” Science Education 81 (6):633–650. doi: https://doi.org/10.1002/(SICI)1098-237X(199711)81:6<633::AID-SCE2>3.0.CO;2-J.

Baldwin, G., E. Frank, and B. Fielding. 1998. “U.S. Women Physicians’ Residential Radon Testing Practices.” American Journal of Preventive Medicine 15 (1): 49–53. doi: 10.1016/s0749-3797(98)00030-0.

Bianco, A., C. G. Nobile, F. Gnisci, and M. Pavia. 2008. “Knowledge and Perceptions of the Health Effects of Environmental Hazards in the General Population in Italy.” International Journal of Hygiene and Environmental Health 211 (3–4): 412–419. doi: 10.1016/j.ijheh.2007.07.025.

Bostrom, A., C. J. Atman, B. Fischhoff, and M. G. Morgan. 1994. “Evaluating Risk Communications: Completing and Correcting Mental Models of Hazardous Processes, Part II.” Risk Analysis 14 (5): 789–798. doi: 10.1111/j.1539-6924.1994.tb00290.x.

Bostrom, Ann, Baruch Fischhoff, and M. Granger Morgan. 1992. “Characterizing Mental Models of Hazardous Processes: A Methodology and an Application to Radon.” Journal of Social Issues 48 (4): 85–100. doi: https://doi.org/10.1111/j.1540-4560.1992.tb01946.x.

Burger, J. 1998. “Environmental Attitudes and Perceptions of Future Land Use at the Savannah River Site: Are There Racial Differences?” Journal of Toxicology and Environmental Health A 53 (4): 255–262. doi: 10.1080/009841098159268.

Burger, Joanna, Malcolm Martin, Keith Cooper, and Michael Gochfeld. 1997. “Attitudes Toward Environmental Hazards: Where Do Toxic Wastes Fit?” Journal of Toxicology and Environmental Health 51 (2): 109–121. doi: 10.1080/00984109708984015.

Burger, Joanna, Donald E. Roush, Jessica Sanchez, Jeanine Ondrof, Robert Ramos, Michael J. McMahon, and Michael Gochfeld. 2000. “Attitudes and Perceptions about Ecological Resources, Hazards, and Future Land Use of People Living near the Idaho National Engineering and Environmental Laboratory.” Environmental Monitoring and Assessment 60 (2):145–161. doi: 10.1023/A:1006104329182.

Butler, K. M., L. Huntington-Moskos, M. K. Rayens, A. T. Wiggins, and E. J. Hahn. 2019. “Perceived Synergistic Risk for Lung Cancer After Environmental Report-Back Study on Home Exposure to Tobacco Smoke and Radon.” American Journal of Health Promotion 33 (4):597–600. doi: https://doi.org/10.1177/0890117118793886.

Butler, K. M., M. K. Rayens, A. T. Wiggins, K. B. Rademacher, and E. J. Hahn. 2017. “Association of Smoking in the Home with Lung Cancer Worry, Perceived Risk, and Synergistic Risk.” Oncology Nursing Forum 44 (2): 55–63. doi: 10.1188/17.onf.e55-e63.

Carlo, G. L., N. L. Lee, K. G. Sund, and S. D. Pettygrove. 1992. “The Interplay of Science, Values, and Experiences among Scientists Asked to Evaluate the Hazards of Dioxin, Radon, and Environmental Tobacco Smoke.” Risk Analysis 12 (1): 37–43. doi: 10.1111/j.1539-6924.1992.tb01305.x.

CDC. 1999. “Radon Testing in Households with a Residential Smoker -United States, 1993-1994.” The Morbidity and Mortality Weekly Report (MMWR) 48 (31): 683–686.

Chess, C., K. L. Salomone, and P. M. Sandman. 1991. “Risk Communication Activities of State Health Agencies.” American Journal of Public Health 81 (4): 489–491. doi: 10.2105/ajph.81.4.489.

Cinar, Nursan, T. C. Alvur, Cemile Dede, Pinar Topsever, and Sevin Altinkaynak. 2014. “Parent’s Knowledge and Perceptions of the Health Effects of Environmental Hazards in Sakarya, Turkey.” Journal of Pakistan Medical Association 64 (1): 38–41.

Cinar, Nursan, and Cemile Dede. 2011. “Knowledge and Attitudes of University Students on Health Effects of Environmental Risk.” HealthMed 5: 217–222.

Clifford, Susan, David Hevey, and Gerard Menezes. 2012. “An Investigation into the Knowledge and Attitudes Towards Radon Testing Among Residents in a High Radon Area.” Journal of Radiological Protection 32 (4): N141–N147. doi: 10.1088/0952-4746/32/4/n141.

Coleman, Cynthia-Lou. 1993. “The Influence of Mass Media and Interpersonal Communication on Societal and Personal Risk Judgments.” Communication Research 20 (4): 611–628. doi: https://doi.org/10.1177/009365093020004006.

Coppola, F., G. La Verde, F. Loffredo, M. Quarto, V. Roca, and M. Pugliese. 2019. “Preliminary Results of the Risk Perception of Radon Exposure.” Nuovo Cimento C Geophysics Space Physics C 41: 221. doi: 10.1393/ncc/i2018-18221-6.

Cronin, Chrysan, Michael Trush, William Bellamy, James Russell, and Paul Locke. 2020. “An Examination of Radon Awareness, Risk Communication, and Radon Risk Reduction in a Hispanic Community.” International Journal of Radiation Biology 96 (6): 803–813. doi: 10.1080/09553002.2020.1730013.

Denman, A. R., P. S. Phillips, R. Tornberg, and C. J. Groves-Kirkby. 2005. “Analysis of the Individual Health Benefits Accruing from a Domestic Radon Remediation Programme.” Journal of Environmental Radioactivity 79 (1): 7–23. doi: 10.1016/j.jenvrad.2004.05.011.

Denman, Antony, Karen Timson, George Shield, Christopher Groves-Kirkby, Stephen Rogers, Jackie Campbell, and Paul Phillips. 2009. “Local Health Campaigns to Reduce Lung Cancers Induced by Radon and Smoking – Who Responds?” Health Policy. doi: 10.1016/j.healthpol.2009.07.011.

Denu, Ryan A., Jessica Maloney, Carrie D. Tomasallo, Noah M. Jacobs, John K. Krebsbach, Amber L. Schmaling, Enio Perez, et al. 2019. “Survey of Radon Testing and Mitigation by Wisconsin Residents, Landlords, and School Districts.” WMJ: Official Publication of the State Medical Society of Wisconsin 118 (4): 169–176.

Desvousges, William H., V. K. Smith, and Hillery H. Rink Iii. 1992. “Communicating Radon Risks Effectively: The Maryland Experience.” Journal of Public Policy & Marketing 11 (2): 68–78.

Dipofi, Jackie A., Michael S. Latour, and Tony L. Henthorne. 2001. “The New Social Marketing Challenge to Promote Radon Testing.” Health Marketing Quarterly 19 (1): 79–90. doi: 10.1300/J026v19n01_06.

Dixon, J. K., K. C. Hendrickson, E. Ercolano, R. Quackenbush, and J. P. Dixon. 2009. “The Environmental Health Engagement Profile: What People Think and Do About Environmental Health.” Public Health Nursing 26 (5): 460–473. doi: 10.1111/j.1525-1446.2009.00804.x.

Dowdall, A., D. Fenton, and B. Rafferty. 2016. “The Rate of Radon Remediation in Ireland 2011–2015: Establishing a Base Line Rate for Ireland’s National Radon Control Strategy.” Journal of Environmental Radioactivity 162–163: 107–112. doi: https://doi.org/10.1016/j.jenvrad.2016.05.001.

Doyle, James K., McClell, Gary H., William D. Schulze, Steven R. Elliott, and Glenn W. Russell. 1991. “Protective Responses to Household Risk: A Case Study of Radon Mitigation.” Risk Analysis 11 (1): 121–134.

Dragojevic, Marko, Robert A. Bell, and Matthew S. McGlone. 2014. “Giving Radon Gas Life Through Language: Effects of Linguistic Agency Assignment in Health Messages About Inanimate Threats.” Journal of Language and Social Psychology 33 (1): 89–98. doi: 10.1177/0261927 × 13495738.

Duckworth, Ladda Tammy, Marilyn Frank-Stromborg, William Oleckno, Pam Duffy, and Kenneth Burns. 2002. “Relationship of Perception of Radon as a Health Risk and Willingness to Engage in Radon Testing and Mitigation.” Oncology Nursing Forum 29: 1099–1107. doi: 10.1188/02.ONF.1099-1107.

Eheman, C. R., E. Ford, P. Garbe, and N. Staehling. 1996. “Knowledge about Indoor Radon in the United States: 1990 National Health Interview Survey.” Archives of Environmental Health 51 (3): 245–247. doi: 10.1080/00039896.1996.9936023.

Erickson, B. E. 2007. “Toxin or Medicine? Explanatory Models of Radon in Montana Health Mines.” Medical Anthropology Quarterly 21 (1): 1–21. doi: 10.1525/maq.2007.21.1.1.

Evans, K. M., J. Bodmer, B. Edwards, J. Levins, A. O’Meara, M. Ruhotina, R. Smith, et al. 2015. “An Exploratory Analysis of Public Awareness and Perception of Ionizing Radiation and Guide to Public Health Practice in Vermont.” Journal of Environmental Public Health 2015. doi: 10.1155/2015/476495.

Evdokimoff, V., and D. Ozonoff. 1992. “Compliance with EPA Guidelines for Follow-Up Testing and Mitigation After Radon Screening Measurements.” Health Physics 63 (2): 215–217. doi: 10.1097/00004032-199208000-00012.

Ferng, S. F., and J. K. Lawson. 1996. “Residents in a High Radon Potential Geographic Area: Their Risk Perception and Attitude Toward Testing and Mitigation.” Journal of Environmental Health 58:Medium: X; Size: pp. 13-7 2009-12-16.

Field, R. W., B. C. Kross, and L. J. Vust. 1993. “Radon Testing Behavior in a Sample of Individuals with High Home Radon Screening Measurements.” Risk Analysis 13 (4): 441–447. doi: 10.1111/j.1539-6924.1993.tb00744.x.

Flynn, J., P. Slovic, and C. K. Mertz. 1994. “Gender, Race, and Perception of Environmental Health Risks.” Risk Analysis 14 (6): 1101–1118. doi: 10.1111/j.1539-6924.1994.tb00082.x.

Ford, E. S., and C. R. Eheman. 1997. “Radon Retesting and Mitigation Behavior Among the U.S. Population.” Health Physics 72 (4): 611–614. doi: 10.1097/00004032-199704000-00013.

Ford, E. S., C. R. Eheman, P. Z. Siegel, and P. L. Garbe. 1996. “Radon Awareness and Testing Behavior: Findings from the Behavioral Risk Factor Surveillance System, 1989-1992.” Health Physics 70 (3): 363–366. doi: 10.1097/00004032-199603000-00006.

Gagnon, F., P. Poulin, J. M. Leclerc, J. C. Dessau, A. Abab, P. Arsenault, F. El-Turaby, G. Lachance-Paquette, and F. A. Vezina. 2016. “Implementation of a Radon Measurement Protocol and its Communication Plan by Child Care Centre Managers in Quebec.” Canadian Journal of Public Health 107 (3):e319–e325. doi: 10.17269/cjph.107.5339.

Glasgow, Russell E., Shawn M. Boles, Edward Lichtenstein, Michael E. Lee, and Lyn Foster. 2004. “Adoption, Reach, and Implementation of a Novel Smoking Control Program: Analysis of a Public Utility–Research Organization Partnership.” Nicotine & Tobacco Research 6 (2): 269–274.

Gleason, Jessie A., Elizabeth Taggert, and Barbara Goun. 2020. “Characteristics and Behaviors Among a Representative Sample of New Jersey Adults Practicing Environmental Risk-Reduction Behaviors.” Journal of Public Health Management and Practice 27 (6): 588–597. doi: 10.1097/PHH.0000000000001106

Golding, D., S. Krimsky, and A. Plough. 1992. “Evaluating Risk Communication - Narrative vs Technical Presentations of Information about Radon.” Risk Analysis 12 (1): 27–35. doi: 10.1111/j.1539-6924.1992.tb01304.x

Groppi, F. 2018. “Radon Laboratory for Secondary Schools.” Nuovo Cimento C-Colloquia and Communications in Physics 41 (3). doi: 10.1393/ncc/i2018-18127-3.

Groves-Kirkby, C. J., K. Timson, G. Shield, A. R. Denman, S. Rogers, J. Campbell, P. S. Phillips, and M. Ekberg. 2014. “Influences Motivating Smokers in a Radon-Affected Area to Quit Smoking.” Perspect Public Health 134 (1): 44–456. doi: 10.1177/1757913912453406.

Hahn, E. J., Luz Huntington-Moskos, Monica Mundy, Kathy Rademacher, Am Wiggins, a T., Mary Kay Rayens, Jamie L. Studts, and Karen M. Butler. 2019. “A Randomized Trial to Test Personalized Environmental Report Back to Reduce Lung Cancer Risk.” Health Education & Behavior 46 (1): 165–175. doi: 10.1177/1090198118788602

Hahn, E. J., M. K. Rayens, S. E. Kercsmar, S. M. Adkins, A. P. Wright, H. E. Robertson, and G. Rinker. 2014. “Dual Home Screening and Tailored Environmental Feedback to Reduce Radon and Secondhand Smoke: An Exploratory Study.” Journal of Environmental Health 76 (6): 156–161.

Hahn, E. J., Mary Kay Rayens, Sarah E. Kercsmar, Heather Robertson, and Sarah M. Adkins. 2014. “Results of a Test and Win Contest to Raise Radon Awareness in Urban and Rural Settings.” American Journal of Health Education 45 (2): 112–118.

Hahn, E. J., A. T. Wiggins, K. Rademacher, K. M. Butler, L. Huntington-Moskos, and M. K. Rayens. 2019. “FRESH: Long-Term Outcomes of a Randomized Trial to Reduce Radon and Tobacco Smoke in the Home.” Preventing Chronic Disease 16. doi: https://doi.org/10.5888/pcd16.180634.

Hahn, Ellen J., Marissa Hooper, Carol Riker, Karen M. Butler, Kathy Rademacher, Amanda Wiggins, and Mary Kay Rayens. 2017. “Lung Cancer Worry and Home Screening for Radon and Secondhand Smoke in Renters.” Journal of Environmental Health 79 (6): 8–13.

Halpern, Michael T., and Kenneth E. Warner. 1994. “Radon Risk Perception and Testing: Sociodemographic Correlates.” Journal of Environmental Health 56 (7): 31–35.

Hampson, S. E., J. A. Andrews, M. Barckley, M. E. Lee, and E. Lichtenstein. 2003. “Assessing Perceptions of Synergistic Health Risk: A Comparison of Two Scales.” Risk Analysis 23 (5): 1021–1029. doi: 10.1111/1539-6924.00378.

Hampson, S. E., J. A. Andrews, M. Barckley, E. Lichtenstein, and M. E. Lee. 2000. “Conscientiousness, Perceived Risk, and Risk-Reduction Behaviors: A Preliminary Study.” Health Psychology 19 (5): 496–500.

Hampson, S. E., J. A. Andrews, M. Barckley, E. Lichtenstein, and M. E. Lee. 2006. “Personality Traits, Perceived Risk, and Risk-Reduction Behaviors: A Further Study of Smoking and Radon.” Health Psychology 25 (4): 530–536.

Hampson, S. E., J. A. Andrews, M. E. Lee, L. S. Foster, R. E. Glasgow, and E. Lichtenstein. 1998. “Lay Understanding of Synergistic Risk: The Case of Radon and Cigarette Smoking.” Risk Analysis 18 (3): 343–350. doi: 10.1111/j.1539-6924.1998.tb01300.x.

Hampson, S. E., J. A. Andrews, M. E. Lee, E. Lichtenstein, and M. Barckley. 2000. “Radon and Cigarette Smoking: Perceptions of this Synergistic Health Risk.” Health Psychology 19 (3): 247–252. doi: 10.1037//0278-6133.19.3.247.

Hazar, N., M. Karbakhsh, M. Yunesian, S. Nedjat, and K. Naddafi. 2014. “Perceived Risk of Exposure to Indoor Residential Radon and its Relationship to Willingness to Test Among Health Care Providers in Tehran.” Journal of Environmental Health Science and Engineering 12 (1): 118. doi: 10.1186/s40201-014-0118-2.

Hill, W. G., P. Butterfield, and L. S. Larsson. 2006. “Rural Parents’ Perceptions of Risks Associated With Their Children’s Exposure to Radon.” Public Health Nursing 23 (5): 392–399. doi: 10.1111/j.1525-1446.2006.00578.x.

Himes, L., K. Parrott, and R. Lovingood. 1996. “The Radon Project: A Study in Environmental Hazard Education.” Journal of Extension 34 (3): 48–53.

Huntington-Moskos, Luz, Mary Kay Rayens, Amanda Wiggins, and E. J. Hahn. 2016. “Radon, Secondhand Smoke, and Children in the Home: Creating a Teachable Moment for Lung Cancer Prevention.” Public Health Nursing 33 (6): 529–538. doi: 10.1111/phn.12283.

Imme, G., Roberto Catalano, G. Mangano, and Daniela Morelli. 2014. “Radioactivity Measurements as Tool for Physics Dissemination.” Journal of Radioanalytical and Nuclear Chemistry 299 (1). doi: 10.1007/s10967-013-2712-7.

Jansson, Bjarne, Magnus Tholander, and Olav Axelson. 1989. “Exposure to radon in Swedish Dwellings - Attitudes and Elimination.” Environment International 15: 293–297. doi: 10.1016/0160-4120(89)90041-X.

Johansson, K. E., Ch Nilsson, and S. Wachtmeister. 2007. “Measuring Radon in Air, Soil and Water—An Introduction to Nuclear Physics for Schools.” Physics Education 42 (3): 281–288. doi: 10.1088/0031-9120/42/3/008.

Johnson, Branden B. 2014. “Communication Challenges for Complex Policy Issues: An Illustration with Multimedia Radon Mitigation.” Environmental Practice 16 (2): 113–126. doi: 10.1017/S1466046614000039.

Johnson, F. Reed, and Ann Fisher. 1989. “Conventional Wisdom on Risk Communication and Evidence from a Field Experiment.” Risk Analysis 9 (2): 209–213.

Johnson, F. Reed, Ann Fisher, V. K. Smith, and William H. Desvousges. 1988. “Informed Choice or Regulated Risk?” Environment: Science and Policy for Sustainable Development 30 (4): 12–35.

Johnson, F. Reed, and Ralph A. Luken. 1987. “Radon Risk Information and Voluntary Protection: Evidence from a Natural Experiment.” Risk Analysis 7 (1): 97–107.

Jones, S. E., S. Foster, and A. S. Berens. 2019. “Radon Testing Status in Schools by Radon Zone and School Location and Demographic Characteristics: United States, 2014.” Journal of School Nursing 35 (6): 442–448. doi: 10.1177/1059840518785441.

Keller, C. 2011. “Using a Familiar Risk Comparison Within a Risk Ladder to Improve Risk Understanding by Low Numerates: A Study of Visual Attention.” Risk Analysis 31 (7): 1043–1054. doi: 10.1111/j.1539-6924.2010.01577.x.

Keller, C., M. Siegrist, and V. Visschers. 2009. “Effect of Risk Ladder Format on Risk Perception in High- and Low-Numerate Individuals.” Risk Analysis 29 (9): 1255–1264. doi: 10.1111/j.1539-6924.2009.01261.x.

Kendall, Gerald, Jonathan Miles, David Rees, Richard Wakeford, Kathryn Bunch, Tim Vincent, and Mark Little. 2016. “Variation with Socioeconomic Status of Indoor Radon Levels in Great Britain: The Less Affluent Have Less Radon.” Journal of Environmental Radioactivity 164: 84–90. doi: 10.1016/j.jenvrad.2016.07.001.

Kennedy, C. A. 2001. “Domestic Radon Remediation Rates and Reasons for Testing.” Environmental Management and Health 12 (5): 459–464. doi: 10.1108/09566160110404863.

Kennedy, Cassondra Jeanne, Claudia K. Probart, and Steve M. Dorman. 1991. “The Relationship between Radon Knowledge, Concern and Behavior, and Health Values, Health Locus of Control and Preventive Health Behaviors.” Health Education Quarterly 18 (3): 319–329.

Khan, Selim, and Samia Chreim. 2019. “Residents’ Perceptions of Radon Health Risks: A Qualitative Study.” BMC Public Health 19. doi: 10.1186/s12889-019-7449-y.

Khan, Selim, Daniel Krewski, James Gomes, and Raywat Deonandan. 2018. “Radon, an Invisible Killer in Canadian Homes: Perceptions of Ottawa-Gatineau Residents.” Canadian Journal of Public Health 110. doi: 10.17269/s41997-018-0151-5.

Kilpatrick, Nikki, Howard Frumkin, Jane Trowbridge, Cam Escoffery, Robert Geller, Leslie Rubin, Gerald Teague, and Janice Nodvin. 2002. “The Environmental History in Pediatric Practice: A Study of Pediatricians’ Attitudes, Beliefs, and Practices.” Environmental Health Perspectives 110 (8): 823–827. doi: 10.1289/ehp.02110823.

Kojo, Katja, and Päivi Kurttio. 2020. “Indoor Radon Measurements in Finnish Daycare Centers and Schools-Enforcement of the Radiation Act.” International Journal of Environmental Research and Public Health 17 (8): 2877. doi: 10.3390/ijerph17082877.

Laflamme, D. M., and J. A. Vanderslice. 2004. “Using the Behavioral Risk Factor Surveillance System (BRFSS) for Exposure Tracking: Experiences from Washington State.” Environmental Health Perspectives 112 (14): 1428–1433. doi: 10.1289/ehp.7148.

Larsson, Laura S. 2015. “The Montana Radon Study: Social Marketing Via Digital Signage Technology for Reaching Families in the Waiting Room.” American Journal of Public Health 105 (4):779–785. doi: 10.2105/AJPH.2014.302060.

Larsson, Laura S., Wade G. Hill, Tamara Odom-Maryon, and Paul Yu. 2009. “Householder Status and Residence Type as Correlates of Radon Awareness and Testing Behaviors.” Public Health Nursing 26 (5): 387–395. doi: 10.1111/j.1525-1446.2009.00796.x.

Lawson, J. K., and S. F. Ferng. 1997. “Taxpayers’ Attitudes Toward Local Environmental Health Specialists: Salary Levels, Education Levels, and Services Needed.” Journal of Environmental Health 59 (10): 13–17.

Lee, Geon-Woo, Ji-Yeon Yang, Hyo-Jung Kim, Myung-Hee Kwon, Woo-Suk Lee, Gi-Hae Kim, Dong-Chun Shin, and Young-Wook Lim. 2016. “Estimation of Health Risk and Effective Dose Based on Measured Radon Levels in Korean Homes and a Qualitative Assessment for Residents’ Radon Awareness.” Indoor and Built Environment 26 (8): 1123–1134. doi: 10.1177/1420326 × 16664387.

Lee, M. E., E. Lichtenstein, J. A. Andrews, R. E. Glasgow, and S. E. Hampson. 1999. “Radon-Smoking Synergy: A Population-Based Behavioral Risk Reduction Approach.” Preventive Medicine 29 (3): 222–227. doi: 10.1006/pmed.1999.0531.

Levy, Barcey T., Cynthia K. Wolff, Paul Niles, Heather Morehead, Yinghui Xu, and Jeanette M. Daly. 2015. “Radon Testing: Community Engagement by a Rural Family Medicine Office.” The Journal of the American Board of Family Medicine 28 (5): 617–623. doi: https://doi.org/10.3122/jabfm.2015.05.140346.

Lichtenstein, E., J. A. Andrews, M. E. Lee, R. E. Glasgow, and S. E. Hampson. 2000. “Using Radon Risk to Motivate Smoking Reduction: Evaluation of Written Materials and Brief Telephone Counselling.” Tobacco Control 9 (3): 320–326. doi: 10.1136/tc.9.3.320.

Lichtenstein, E., S. M. Boles, M. E. Lee, S. E. Hampson, R. E. Glasgow, and J. Fellows. 2008. “Using Radon Risk to Motivate Smoking Reduction II: Randomized Evaluation of Brief Telephone Counseling and a Targeted Video.” Health Education Research 23 (2): 191-201. doi: 10.1093/her/cym016.

Lion, René, Ree M. Meertens, and Ilja Bot. 2002. “Priorities in Information Desire about Unknown Risks.” Risk Analysis 22 (4): 765–776. doi: https://doi.org/10.1111/0272-4332.00067.

Loffredo, Filomena, Federica Savino, Marcello Serra, Domenico Tafuri, and Maria Quarto. 2020. “Cognitive Investigation on the Knowledge of the Risk Deriving from Radon Exposure: Preliminary Results.” Acta Medica Mediterranea 36 (1265). doi: 10.19193/0393-6384_2020_2_198.

Losee, Joy E., James A. Shepperd, and Gregory D. Webster. 2020. “Financial Resources and Decisions to Avoid Information About Environmental Perils.” Journal of Applied Social Psychology 50 (3): 174–188. doi: https://doi.org/10.1111/jasp.12648.

Macher, J. M., and S. B. Hayward. 1991. “Public Inquiries about Indoor Air Quality in California.” Environmental Health Perspectives 92: 175–180. doi: 10.1289/ehp.9192175.

Mainous, A. G., 3rd, and M. D. Hagen. 1993. “Public Perceptions of Radon Risk.” Family Practice Research Journal 13 (1): 63–69.

Makedonska, Galina, Jana Djounova, and Kremena Ivanova. 2018. “Radon Risk Communication in Bulgaria.” Radiation Protection Dosimetry 181. doi: 10.1093/rpd/ncy096.

Mancl, K., J. Heimlich, A. Fentiman, and R. Christensen. 1994. “General Public Awareness of Sources of Radiation in Their Environment.” Ohio Journal of Science 94 (5): 134–137.

Mardis, H. M., R. J. Guimond, and E. Fisher. 1988. “Radon Reduction in Homes: Research Phase to Operational Phase.” Radiation Protection Dosimetry 24 (1): 537–539.

Martin, Kirsten, Rebecca Ryan, Thomas Delaney, David A. Kaminsky, Scott J. Neary, Ethan E. Witt, Florence Lambert-Fliszar, et al. 2020. “Radon From the Ground into Our Schools: Parent and Guardian Awareness of Radon.” SAGE Open 10 (1): 2158244020914545. doi: 10.1177/2158244020914545.

Mazur, Allan. 1987. “Putting Radon on the Public’s Risk Agenda.” Science, Technology, & Human Values 12 (3/4): 86–93.

Mazur, Allan, and Glenn S. Hall. 1990. “Effects of Social Influence and Measured Exposure Level on Response to Radon*.” Sociological Inquiry 60 (3): 274–284.

Momin, B., C. McNaughton, J. D. Galanek, A. Neri, M. S. Gallaway, and M. Puckett. 2018. “A Qualitative Study of Realtor Knowledge, Attitudes, and Practices Related to Radon Health Effects: Implications for Comprehensive Cancer Control.” Cancer Causes Control 29 (12): 1249–1255. doi: 10.1007/s10552-018-1086-0.

Murphy, Jacquelyn A., Jennifer L. Peel, Tom Butts, Lisa M. McKenzie, and Jill S. Litt. 2019. “Understanding Emerging Environmental Health Concerns and Environmental Public Health–Tracking Priorities Among State and Local Professionals in Colorado.” Journal of Public Health Management and Practice. doi: 10.1097/phh.0000000000001010.

Neri, A., C. McNaughton, B. Momin, M. Puckett, and M. S. Gallaway. 2018. “Measuring Public Knowledge, Attitudes, and Behaviors Related to Radon to Inform Cancer Control Activities and Practices.” Indoor Air 28 (4): 604–610. doi: 10.1111/ina.12468.

Nicotera, G., C. G. Nobile, A. Bianco, and M. Pavia. 2006. “Environmental History-Taking in Clinical Practice: Knowledge, Attitudes, and Practice of Primary Care Physicians in Italy.” Journal of Occupational Environmental Medicine 48 (3): 294–302. doi: 10.1097/01.jom.0000184868.77815.2a.

Nissen, M. J., J. W. Leach, J. A. Nissen, K. K. Swenson, and H. Kehn. 2012. “Radon Testing and Mitigation: An Intervention in a Primary Care Setting.” Journal of Cancer Education 27 (3): 566–72. doi: 10.1007/s13187-012-0346-z.

Nwako, Paschal, and Terrence Cahill. 2020. “Radon Gas Exposure Knowledge Among Public Health Educators, Health Officers, Nurses, and Registered Environmental Health Specialists: A Cross-Sectional Study.” Journal of Environmental Health 82 (6).

Park, Eunkyung, Clifford W. Scherer, and Carroll J. Glynn. 2001. “Community Involvement and Risk Perception at Personal and Societal Levels.” Health, Risk & Society 3 (3): 281–292. doi: 10.1080/13698570120079886.

Perko, Tanja. 2014. “Radiation Risk Perception: A Discrepancy Between the Experts and the General Population.” Journal of Environmental Radioactivity 133: 86–91. doi: 10.1016/j.jenvrad.2013.04.005.

Perko, Tanja, N. Zeleznik, C. Turcanu, and P. Thijssen. 2012. “Is Knowledge Important? Empirical Research on Nuclear Risk Communication in Two Countries.” Health Physics 102 (6): 614–625. doi: 10.1097/HP.0b013e31823fb5a5.

Peterson, Elizabeth, and Jonathan Howland. 1996. “Predicting Radon Testing Among University Employees.” Journal of the Air & Waste Management Association (1995) 46: 2–11. doi: 10.1080/10473289.1996.10467435.

Petrescu, Dacinia Crina, and Ruxandra Mălina Petrescu-Mag. 2017. “Setting the Scene for a Healthier Indoor Living Environment: Citizens’ Knowledge, Awareness, and Habits Related to Residential Radon Exposure in Romania.” Sustainability 9 (11): 2081. doi: https://doi.org/10.3390/su9112081.

Phillips, Paul S., Julia D. Fraser, and Antony R. Denman. 2000. “Effectiveness of Secondary Radon Protection in Northamptonshire Houses.” Environmental Management and Health 11 (4): 337–351. doi: 10.1108/09566160010372752.

Poortinga, W., Karin Bronstering, and Simon Lannon. 2011. “Awareness and Perceptions of the Risks of Exposure to Indoor Radon: A Population‐Based Approach to Evaluate a Radon Awareness and Testing Campaign in England and Wales.” Risk Analysis: An Official Publication of the Society for Risk Analysis 31: 1800–1812. doi: 10.1111/j.1539-6924.2011.01613.x.

Poortinga, W., P. Cox, and N. F. Pidgeon. 2008. “The Perceived Health Risks of Indoor Radon Gas and Overhead Powerlines: A Comparative Multilevel Approach.” Risk Analysis 28 (1): 235–248. doi: https://doi.org/10.1111/j.1539-6924.2008.01015.x.

Prochaska, J. O., W. F. Velicer, J. S. Rossi, M. G. Goldstein, B. H. Marcus, W. Rakowski, C. Fiore, et al. 1994. “Stages of Change and Decisional Balance for 12 Problem Behaviors.” Health Psychology 13 (1): 39–46. doi: 10.1037//0278-6133.13.1.39.

Pugliese, M., G. La Verde, and V. Roca. 2019. “Dissemination about Natural Radioactivity Through Work-Based Learning Experiences.” Nuclear and Particle Physics Proceedings 306: 183–8. doi: 10.1016/j.nuclphysbps.2019.07.026.

Rafique, M., S. Jabeen, and M. I. Shahzad. 2008. “General Public’s and Physicians’ Perception of Health Risk Associated with Radon Exposure in the State of Azad Jammu and Kashmir.” Public Health Nursing 25 (4): 327–335. doi: 10.1111/j.1525-1446.2008.00713.x.

Rahman, S., M. Faheem, S. Rehman, and Matiullah. 2006. “Radon Awareness Survey in Pakistan.” Radiation Protection Dosimetry 121 (3): 333–336. doi: 10.1093/rpd/ncl021.

Rickenbacker, Harold J., Jessica M. Vaden, and Melissa M. Bilec. 2020. “Engaging Citizens in Air Pollution Research: Investigating the Built Environment and Indoor Air Quality and Its Impact on Quality of Life.” Journal of Architectural Engineering 26 (4): 04020041. doi: 10.1061/(ASCE)AE.1943-5568.0000439.

Riesenfeld, E. P., T. W. Marcy, K. Reinier, J. A. Mongeon, C. W. Trumbo, B. E. Wemple, and D. A. Kaminsky. 2007. “Radon Awareness and Mitigation in Vermont: A Public Health Survey.” Health Physics 92 (5): 425–431. doi: 10.1097/01.HP.0000254862.50407.4a.

Rinker, G. H., E. J. Hahn, and M. K. Rayens. 2014. “Residential Radon Testing Intentions, Perceived Radon Severity, and Tobacco Use.” Journal of Environmental Health 76 (6): 42–47.

Rothman, Stanley, and S. Robert Lichter. 2001. “Environmental Cancer.” Society 38 (4): 20–6. doi: https://doi.org/10.1007/s12115-001-1019-4.

Ryan, Daniel, and C. Cecily Kelleher. 1999. “A Survey of Householders’ Mitigation Strategy: Response to Raised Radon Levels.” European Journal of Public Health 9 (1): 62–64. doi: 10.1093/eurpub/9.1.62.

Sanborn, M. D., and E. A. Scott. 1998. “Environmental Health Concerns in Urban and Rural Family Practice.” Canadian family physician Medecin de famille canadien 44: 1466–1472.

Sanborn, M., L. Grierson, R. Upshur, L. Marshall, C. Vakil, L. Griffith, F. Scott, M. Benusic, and D. Cole. 2019. “Family Medicine Residents’ Knowledge of, Attitudes Toward, and Clinical Practices Related to Environmental Health: Multi-Program Survey.” Canadian family physician Medecin de famille canadien 65 (6): e269–e277.

Sandman, P. M., and N. D. Weinstein. 1993. “Predictors of Home Radon Testing and Implications for Testing Promotion Programs.” Health Education Quarterly 20 (4): 471–487. doi: 10.1177/109019819302000408.

Sandman, P. M., N. D. Weinstein, and P. Miller. 1994. “High Risk or Low: How Location on a “Risk Ladder” Affects Perceived Risk.” Risk Analysis 14 (1): 35–45. doi: 10.1111/j.1539-6924.1994.tb00026.x.

Sandman, Peter M., Neil D. Weinstein, and M. L. Klotz. 1987. “Public Response to the Risk from Geological Radon.” Journal of Communication 37 (3): 93–108. doi: https://doi.org/10.1111/j.1460-2466.1987.tb00997.x.

Scivyer, S. 2005. “A New Approach to Increasing the Uptake of Radon Remediation in England.” In Radioactivity in the Environment, 260–264. Elsevier.

Siegrist, M., G. T. Cvetkovich, and H. Gutscher. 2001. “Shared Values, Social Trust, and the Perception of Geographic Cancer Clusters.” Risk Analysis 21 (6): 1047–1053. doi: 10.1111/0272-4332.216173.

Siza, C., M. Morrison, S. Harris, T. Hatch, and M. Tyler. 2018. “Assessment of Community Awareness and Practices Concerning Indoor Air Pollutants - Madison County, Alabama, June 2017.” MMWR Morbidity and Mortality Weekly Report 67 (15): 447–450. doi: 10.15585/mmwr.mm6715a3.

Sjöberg, Lennart, Martin Peterson, Jana Fromm, Åsa Boholm, and Sven‐Ove Hanson. 2005. “Neglected and Overemphasized Risks: The Opinions of Risk Professionals.” Journal of Risk Research 8 (7–8): 599–616. doi: 10.1080/13669870500062576.

Smith, V. K., W. H. Desvousges, and J. W. Payne. 1995. “Do Risk Information Programs Promote Mitigating Behavior?” Journal of Risk and Uncertainty 10 (3): 203–221.

Smith, V. K., and William H. Desvousges. 1990. “Risk Communication and the Value of Information: Radon as a Case Study.” The Review of Economics and Statistics 72 (1): 137–142.

Smith, V. K., William H. Desvousges, Ann Fisher, and F. Reed Johnson. 1988. “Learning about Radon’s Risk.” Journal of Risk and Uncertainty 1 (2): 233–258. doi: 10.1007/BF00056169.

Smith, V. K., William H. Desvousges, F. Reed Johnson, and Ann Fisher. 1990. “Can Public Information Programs Affect Risk Perceptions?” Journal of Policy Analysis and Management 9 (1): 41–59.

Smith, V. K., and F. R. Johnson. 1988. “How do Risk Perceptions Respond to Information? The Case of Radon.” Review of Economics & Statistics 70 (1): 1–8.

Spiegel, Jerry, and Daniel Krewski. 2002. “Using Willingness to Pay to Evaluate the Implementation of Canada’s Residential Radon Exposure Guideline.” Canadian Journal of Public Health. Revue canadienne de santé publique 93: 223–228. doi: 10.1007/BF03405005.

Tahir, S. N., and A. S. Alaamer. 2008. “Radon Awareness in the Provincial Capital Lahore of Punjab Province of Pakistan.” Radiation Protection Dosimetry 129 (4): 491–494. doi: 10.1093/rpd/ncm461.

Trasande, L., J. Boscarino, N. Graber, R. Falk, C. Schechter, M. Galvez, G. Dunkel, et al. 2006. “The Environment in Pediatric Practice: A Study of New York Pediatricians’ Attitudes, Beliefs, and Practices Towards Children’s Environmental Health.” Journal of Urban Health 83 (4): 760–772. doi: 10.1007/s11524-006-9071-4.

Wang, Y., C. Ju, A. D. Stark, and N. Teresi. 1999. “Radon Mitigation Survey among New York State Residents Living in High Radon Homes.” Health Physics 77 (4): 403–409. doi: 10.1097/00004032-199910000-00007.

Wang, Y., C. Ju, A. D. Stark, and N. Teresi. 2000. “Radon Awareness, Testing, and Remediation Survey among New York State Residents.” Health Physics 78 (6): 641–647. doi: 10.1097/00004032-200006000-00006.

Weinstein, N. D., and W. M. Klein. 1995. “Resistance of Personal Risk Perceptions to Debiasing Interventions.” Health Psychology 14 (2): 132–140. doi: 10.1037//0278-6133.14.2.132.

Weinstein, N. D., M. L. Klotz, and P. M. man. 1989. “Promoting Remedial Response to the Risk of Radon: Are Information Campaigns Enough?” Science, Technology & Human Values 14 (4): 360–379.

Weinstein, N. D., M. L. Klotz, and P. M. Sandman. 1988. “Optimistic Biases in Public Perceptions of the Risk from Radon.” American Journal of Public Health 78 (7): 796–800. doi: 10.2105/ajph.78.7.796.

Weinstein, N. D., and J. E. Lyon. 1999. “Mindset, Optimistic Bias about Personal Risk and Health-Protective Behaviour.” British Journal of Health Psychology 4 (4): 289–300. doi: https://doi.org/10.1348/135910799168641.

Weinstein, N. D., J. E. Lyon, P. M. man, and C. L. Cuite. 1998. “Experimental Evidence for Stages of Health Behavior Change: The Precaution Adoption Process Model Applied to Home Radon Testing.” Health Psychology 17 (5): 445–453. doi: 10.1037/0278-6133.17.5.445.

Weinstein, N. D., Peter M. man, and Nancy E. Roberts. 1990. “Determinants of Self-Protective Behavior: Home Radon Testing.” Journal of Applied Social Psychology 20 (10): 783–801. doi: https://doi.org/10.1111/j.1559-1816.1990.tb00379.x.

Weinstein, N. D., Nancy E. Roberts, and Kerry Kirk Pflugh. 1992. “Evaluating Personalized Risk Messages.” Evaluation Review 16 (3): 235–246. doi: https://doi.org/10.1177/0193841×9201600302.

Weinstein, N. D., and P. M. Sandman. 1992a. “A Model of the Precaution Adoption Process: Evidence from Home Radon Testing.” Health Psychology 11 (3): 170–180. doi: 10.1037//0278-6133.11.3.170.

Weinstein, N. D., P. M. Sandman, and N. E. Roberts. 1991. “Perceived Susceptibility and Self-Protective Behavior: A Field Experiment to Encourage Home Radon Testing.” Health Psychology 10 (1): 25–33. doi: 10.1037//0278-6133.10.1.25.

Weinstein, N. D., and Peter M. Sandman. 1992b. “Predicting Homeowners’ Mitigation Responses to Radon Test Data.” Journal of Social Issues 48 (4): 63–83. doi: https://doi.org/10.1111/j.1540-4560.1992.tb01945.x.

Witte, K., J. M. Berkowitz, J. M. Lillie, K. A. Cameron, M. K. Lapinski, and W. Y. Liu. 1998. “Radon Awareness and Reduction Campaigns for African Americans: A Theoretically Based Evaluation.” Health Education & Behavior 25 (3): 284–303.

Zhang, W., Y. Chow, J. Meara, and M. Green. 2011. “Evaluation and Equity Audit of the Domestic Radon Programme in England.” Health Policy 102 (1): 81–88. doi: 10.1016/j.healthpol.2010.09.016.

Appendix 3.

NORM search

Search String in Web of Science Format

Table

#1 (radioactiv*) AND ((natural NEAR/1 (radiation OR “radioactive material”)) OR tenorm OR residue OR remainder OR leftover OR waste OR oil OR gas OR water OR phosphate* OR fuel* OR geothermal OR building* OR “flying ash” OR mining OR mine OR “NORM industries” OR “building material*” OR “alum shale” OR (environment* NEAR/0 remediation)) #2 (((field OR case OR comparative OR cohort OR archival) NEAR/2 stud*) OR ((network OR content OR sentiment OR meta OR framework OR media OR discourse OR morphological OR text* OR conversation OR narrative) NEAR/2 analysis) OR ((systematic OR meta) NEAR/0 review) OR ((mixed OR mental OR mixed OR delphi OR q OR economic) NEAR/2 method*) OR “delphi techniq*” OR “focus group*” OR “repertory grid” OR “analytic induction” OR “life history*” OR historiography OR “socio mapping” OR “feeling thermometer” OR “cybermethod*” OR “participatory action” OR bibliograph* OR questionnaire* OR “secondary data” OR “e-research” OR “memory work” OR interview* OR observation* OR ethnography OR phenomenolog* OR RCT OR “randomized controlled trial*” OR workshop OR “public opinion” OR panel* OR omnibus OR poll OR triangulation OR hermeneutic) #3 (survey* or experiment*) #4 (public* OR citizen* OR participant* OR respondent* OR resident* OR person* OR stakeholder*)

The acronym “norm” has multiple definitions; custom, function, … Thus, in order to find relevant studies we selected keywords related to industries that cause contamination with naturally occurring radionuclides, and combined them with the word “radioactiv*” to find only those where radioactivity was mentioned.