Characterizing Children’s Intuitive Theories of Disease: The Case of Flu

Abstract

To advance the study of children’s knowledge and understanding of disease, we devised a methodology for assessing key features of intuitive theories laid out by Wellman and Gelman (1998). We then assessed a disease-relevant biological ontology, causal propositions involving unobservables, and coherence in explanations of influenza offered by children aged 8 to 13. Use of disease-relevant terms and mention of propositions in a biological theory of flu causality, although not coherence or connectedness of ideas, increased with age. Measures were moderately correlated with one another and with a traditional Piagetian measure of level of disease understanding, each contributing uniquely to the characterization of children’s thinking. In multiple regression analyses, scores were highest for older children, Latino/minority children, and children of more educated parents with other factors controlled. Specific gaps in children’s intuitive theories are identified to guide theory-based interventions aimed at helping children understand and protect themselves from infectious diseases.

Introduction

How well-prepared are children of different ages to understand and protect themselves from the common infectious diseases that threaten their health each year? How can their thinking and the development of their thinking be characterized and changed? During the 2017-2018 flu season, influenza, defined by the Centers for Disease Control and Prevention (2018) as a contagious respiratory illness caused by influenza viruses, was estimated by CDC to affect 48.8 million Americans, including 11.5 million children. Flu and its complications killed an estimated 79,400 people in all in the United States, including a record 183 children, that year. Yet the literature on the development of children’s knowledge and understanding of infectious disease is surprisingly dated and limited. In this paper, we draw on the intuitive theories perspective on cognitive development to introduce a new approach to characterizing children’s thinking about disease, testing it with data from children aged 8 to 13 from varied ethnic backgrounds who were interviewed about influenza.

Early research on children’s understandings of illness and various diseases was either atheoretical or guided by Piagetian cognitive-developmental theory. Piaget’s theory describes invariant stages that represent a universal developmental progression in the structural complexity or sophistication of logical thought. Cognitive-developmental studies of thinking about disease and illness have therefore focused on the complexity of children’s understanding rather than on their knowledge, or the factual correctness of their thoughts. Studies using the similar coding systems developed by Bibace and Walsh (1980) and Perrin and Gerrity (1981) have documented that level understanding of illness causality progresses from vague and magical thinking about factors associated with getting sick in the preschool years, to awareness in elementary school that a disease agent (whether germs, pollution, or cold air) is internalized and causes sickness, and to a more complex physiological explanation in adolescence featuring interacting causes and effects within the body. Level of understanding, whether of illness in general or of particular diseases, is indeed associated with both age and performance on Piagetian tasks assessing preoperational, concrete operational, and formal operational thinking (e.g., Hansdottir & Malcarne, 1998; Perrin & Gerrity, 1981; Schonfeld, Johnson, Perrin, O’Hare, & Cicchetti, 1993; Walsh & Bibace, 1991; and see Burbach & Peterson, 1996, for a review). From a cognitive-developmental perspective, teaching children facts about how people get a disease is not sufficient; children must also understand the causal processes involved in becoming sick (Schonfeld et al., 1993; Walsh & Bibace, 1991).

The study of children’s thinking about disease has been reinvigorated by the intuitive, folk, or naive theories perspective. This perspective holds that, from an early age, children have or construct foundational theories that organize their knowledge of the physical, biological, and psychological worlds (Carey, 1985; Keil, 1992; Wellman & Gelman, 1992, 1998). Intuitive theories define a domain of knowledge, specify the causal explanatory mechanisms that operate in that domain, and are characterized by coherence (Wellman & Gelman, 1998). Researchers who have applied the intuitive theories perspective to the study of children’s conceptions of disease have been concerned with the content and correctness of children’s thinking rather than its structural complexity, paying special attention to when children have a truly biological theory of illness and a concept of germs as invisible biological entities that cause disease. From this perspective, important intuitions about disease are in place early in childhood; moreover, if children are given accurate biological information, they may not need to have attained an advanced level of cognitive development in order to grasp key facts about disease transmission and construct a scientifically correct theory of what causes a disease (Au & Romo, 1996).

In laying out this approach to cognitive development, Wellman and Gelman (1998) identified four key features of an intuitive theory in a domain such as biology: (a) an ontology, or vocabulary of entities and concepts that fall within the domain (e.g., in the case of an intuitive biology, a distinction between living and nonliving things), (b) domain-specific causal mechanisms (e.g., biological processes involved in growth, inheritance, or disease), (c) unobservables, or invisible underlying constructs that help explain phenomena (e.g., the concept of a gene), and (d) coherence, or an organization of and interrelatedness among concepts.

Applied to children’s thinking about infectious diseases and reduced to three main elements, comprehensively assessing intuitive theories of disease would entail assessing whether children: (a) have command of a disease-relevant biological ontology (e.g., know relevant biological concepts such as germ and immune system); (b) can explain illness causality and, in the process, invoke an invisible disease agent such as a germ or virus as the causal mechanism; and (c) demonstrate a coherent linking together of ideas in their explanations. A mature intuitive theory of disease would be both causally sophisticated and biologically correct.

Research guided by the intuitive theories perspective has focused primarily on the early development of a germ theory of infectious disease in which contacts with ill people transmit an invisible causal agent that is biological in nature—a germ or virus (e.g., see Au & Romo, 1996; Kalish, 1996). This work and other research, to be examined below, has revealed a good deal about children’s thinking but it has addressed only parts of what it means to have an intuitive theory of disease. Researchers have yet to devise a methodology for assessing more comprehensively the key features of intuitive theories and describing children’s thinking about disease at different ages with respect to ontology, causal mechanisms involving unobservable disease agents, and coherence.

Previous Research on Key Features of Intuitive Theories of Disease

Prior research—some of it guided by the intuitive theories perspective, some of it not-- provides a preliminary picture of the development of the key elements of an intuitive theory of disease.

Biological Ontology

Even preschool children have the beginnings of a disease-relevant biological ontology or set of concepts. They are often familiar with the concept of germs, although studies disagree on the extent to which they understand that germs are biological entities and what they do and do not cause (e.g., Kalish, 1996; Keil, Levin, Richman, & Gutheil, 1999; Solomon & Cassimatis, 1999). Yet, in a study of kindergarten through sixth grade children’s understanding of colds, children were much more likely to talk of cold weather or the idea of contagion than to mention germs specifically; only 33% did so, although the percentage increased with age (Badani & Schonfeld, 2002). Grasp of the concept of “virus” is rare until adolescence (Wilkinson, 1988). Understanding of the AIDS “germ,” the virus HIV, progresses with age from viewing it as a nondescript germ to appreciating that it is a disease-specific germ with specific characteristics (Au & Romo, 1996; Sigelman, Alfeld-Liro, Lewin, Derenowski, & Woods, 1997). More generally, children often do not appreciate until late childhood or early adolescence that not all diseases are contagious and that germs and viruses are disease-specific (Nagy, 1953; Wilkinson, 1988). Finally, only in later childhood or adolescence are some children likely to show an awareness, limited in comparison to that of adults, of the immune system or a capacity of the body to fight germs and viruses (Jones & Rua, 2008; Landry-Boozer, 2004; Wilkinson, 1988).

Causal Mechanisms Involving Unobservables

The Piagetian research on levels of understanding has carefully described changes in the complexity of children’s explanations of illness and disease but does not differentiate between a sophisticated and biologically accurate account of causal events and a sophisticated but nonbiological and largely wrong account. Nonetheless, it appears that causal complexity and correctness, understanding and knowledge, both increase with age and are interrelated (Sigelman, 2014). By ages 7-10, many children understand that internalizing germs causes illness, although they typically cannot say how (Bibace & Walsh, 1980; Perrin & Gerrity, 1981). There is then a shift at around age 11, with the emergence of formal operational thought, toward identifying multiple causal factors and beginning to understand that the body fights germs and viruses, but it is articulated by only a minority of early adolescents (Perrin & Gerrity, 1981; Walsh & Bibace, 1991). Teaching children and adolescents about the conditions under which the HIV virus can and cannot survive, reproduce, and infect people (Au & Romo, 1996; Zamora, Romo, & Au, 2006), or explaining step-by-step the mechanisms through which risky behavior results in AIDS symptoms (Sigelman et al., 1996), can improve both knowledge and understanding.

Coherence of Ideas

Coherence in thinking about illness and disease has been examined in a variety of ways. For example, Paterson, Moss-Morris, and Butler (1999) used a measure assessing both the sophistication and correctness of thinking about several dimensions of illness in a study of how children aged 7-14 think about asthma and colds. Cronbach’s alpha coefficients reflecting consistency of thinking across different dimensions of illness were large, suggesting coherence (alpha = .69 for colds, .79 for asthma). Similarly, moderate correlations have been reported between measures of elementary school children’s levels of conceptual understanding of the causality, prevention, and treatment of AIDS, colds, and cancer (Schonfeld et al., 1993). These studies have not examined whether coherence changes with age, however. Nor have they assessed the logical connectedness among the concepts and propositions in children’s causal arguments--the approach to examining coherence taken here. Although some degree of coherence can be expected even in young children’s intuitive thinking about causes of illness judging from prior research using structured questioning approaches (e.g., Kalish, 1996), coherence should be expected to increase with age as children’s cognitive and linguistic skills increase and as they acquire and integrate more information (Wellman & Gelman, 1998)

Previous Research on Flu

Children’s understandings of flu have not been studied as much as their understandings of the common cold or AIDS. In a qualitative study of children aged 6 to 12, Flood et al. (2011) documented that most children had heard of flu and could name some of its symptoms. McCann-Sanford, Spencer, Hendrick, and Meyer (1982) reported that scores on a true-false test about upper respiratory illnesses increased with age during the elementary school years, with growth particularly evident between fourth and sixth grade. Analyzing features of concept maps of flu created by students in grades 5, 8, and 11 and their teachers, Jones and Rua (2008) concluded that there was growth toward more accurate and complete knowledge with age and that children know a lot about transmission routes and symptoms but less compared to teachers about the immune system, vaccines, and medications.

In a study of South African children ranging in age from 5 to 15 and adults about AIDS and flu that used methods associated with both the Piagetian and intuitive theories perspectives, Legare and Gelman (2009) reported levels of understanding of causality, prevention, and treatment of both diseases similar to those found for American children and an age-related increase in scores for causality but not for prevention and treatment. Most participants grasped the concept of internalization of a disease agent but showed little ability to explain the internal biology of disease. Content analysis of themes suggested that children of all ages associated flu with casual contagion but, with age, increasingly associated AIDS with blood and sexual contact rather than casual contagion.

Guided by the intuitive theories perspective and focusing on competing theories of colds and flu, Sigelman (2012) found that, although cold weather thinking was more common for colds than for flu, cold weather thinking about both diseases declined as germ theories became more prominent during childhood and adolescence. Au et al. (2008) asked third grade children, college students, and elderly adults in China about colds and flu (jointly rather than as separate diseases). Owing to the prevalence of hot/cold folk beliefs about illness in China, many children cited cold weather-related causes but elderly adults cited them even more. The transfer of germs from one person to another was almost universally recognized as a cause as well but without a biological understanding of what determines whether germs or viruses live or die. Au et al. (2008) implemented a successful program to teach third and fourth graders that germs live a long time in cold weather but are killed by heat and can cause illness only when they are alive. Providing such information reduced cold weather explanations and enabled children to think more coherently about the biology of cold/flu causality.

Predictors of Knowledge and Understanding of Flu

There is strong reason to expect both level of understanding of flu (Burbach & Peterson, 1996; Legare & Gelman, 2009) and knowledge of flu (e.g., McCann-Sanford et al., 1982; Sigelman, 2012) to increase steadily with age in childhood and adolescence. Specifically, older children and adolescents, through a combination of informal and formal education and cognitive growth, should have a more complete biological vocabulary, should be more likely to understand infectious illnesses in terms of the internalization and action of germs or viruses, and should be more able to conceptualize internal, physiological processes. What besides age might contribute to knowledge and understanding of flu?

It appears that boys and girls do not differ in either conceptual understanding or knowledge of illness and disease (e.g., Johnson et al., 1994; McCann-Sanford et al., 1982; Schonfeld et al., 1993). The possibility of ethnic/racial differences has also been examined, and since our sample is heavily Mexican-American we had an opportunity to compare Latino and European American children. Latino and other minority adults often subscribe to hot-cold folk theories of illness emphasizing the ill effects of extreme temperatures and sudden changes in temperature (e.g., Castro, Furth, & Karlow, 1984; Mikhail, 1994; Taveras, Durousseau, & Flores, 2004), although many non-Hispanic whites share at least some of these beliefs (Baer et al., 1999). Being socialized into cultural folk beliefs about disease does not necessarily mean rejecting or being slow to adopt a germ theory of disease as one gets older. Unlike the Piagetian perspective, in which a new mode of thinking replaces the previous one, the intuitive theories perspective allows for the possibility that an individual can hold multiple theories simultaneously, with the theories typically coexisting rather than being integrated (Legare & Shtulman, 2018). Indeed, research has established the coexistence of two or more theories of illness in many children and adults—for example, the germ theory and a cold weather theory (Raman & Winer, 2002; Sigelman, 2012), or the germ theory and a supernatural theory involving witchcraft, evil spirits, or the like (Legare & Gelman, 2008; Legare & Shtulman, 2018). Despite support for the coexistence concept, evidence to date suggests that cold weather beliefs appear to decline with age as belief in a germ theory increases during childhood and adolescence (Sigelman, 2012), suggesting that Mexican American and other minority children might be slower than European American children to master germ-centered biological theories.

In support of this, Au and Romo (1996) reported that low-income Latino children showed less mastery than middle-class non-Hispanic white children of the biological nature of germs; Badani & Schonfeld (2002) observed both Latino and African American children to be more likely than European American children to cite cold weather-related causes of colds and less likely to cite contagion or germs; and McCann-Sanford et al. (1982) found that minority children scored lower than European American children, and lower middle class children scored lower than upper middle class children, on a test about upper respiratory illness.

However, these studies did not control for socioeconomic status (SES) in examining ethnic group differences, leaving us unsure about the relative contributions of ethnic cultural beliefs and SES to such differences. Parent education and family income are indeed linked to more sophisticated illness conceptualizations and beliefs among children (Paterson et al., 1999; Zhu, Liu, & Tardif, 2009) as well as parents (Taveras et al., 2004). Moreover, Sigelman (2012) determined that tendencies of Mexican American children to believe more than Anglo American children in cold weather theories of colds and flu and less in germ theories were reduced in size or even became nonsignificant once ethnic differences in parent education were controlled. The question of sociocultural influences on children’s knowledge and understanding of illness therefore demands further study, especially when still other studies detect little evidence that either race/ethnicity or SES matters (Johnson et al., 1994; Schonfeld et al., 1993).

Personal experience with an illness is another factor that may have implications for knowledge and understanding of a disease if it means being given more information by parents, teachers, or medical professionals. For example, children with asthma have been found to have more sophisticated conceptual understandings than other children of asthma, although not of colds (Paterson et al., 1999; and see Koopman, Baars, Chaplin, & Zwinderman, 2004); experiences such as having a medical professional in the family are associated with greater knowledge and understanding of the immune system (Landry-Boozer, 2004); and greater personal experience with colds and flu is linked to knowledge of flu among fifth to eleventh graders (Jones & Rua, 2008; but see McCann-Sanford et al., 1982 for null findings and Burbach & Peterson, 1986, on mixed findings regarding the implications of experience with illness for illness understanding).

Hypotheses

The aim of this study was to devise and apply a more comprehensive approach to characterizing children’s intuitive theories of disease and their development, using the relatively understudied but important disease influenza as a test case. We operationalized the elements identified by Wellman and Gelman (1998) as key features of intuitive theories—ontology, causal explanation involving unobservables, and coherence. Because the intuitive theories perspective focuses primarily on knowledge, especially with respect to ontology and causal explanatory ideas, rather than on complexity of thinking in the Piagetian sense, we also included a Piagetian measure of level of understanding of flu causality in the study. This allowed us to tie the present study to previous research using Piagetian measures and to correlate and compare age-related differences in knowledge measures informed by the intuitive theories perspective and a Piagetian measure of conceptual understanding. We hypothesized the following:

1. Scores on all measures will increase with age as children develop cognitively and acquire more knowledge about health and illness.

2. The four measures of interest will be moderately correlated, even with age controlled, based on previous research assessing both knowledge and understanding (e.g., Sigelman, 2014) and an expectation that cognitive development and knowledge acquisition interact to drive the development of intuitive theories (Wellman & Gelman, 1998) and that advances in one aspect of theorizing (e.g., building a disease-relevant biological ontology) will spur advances in others (e.g., grasp of causal mechanisms).

3. Although the role of sociodemographic factors in children’s thinking about illness is less clear, we expect advanced parent education and family income (as indexed by nonparticipation in the free lunch program) to be associated with greater knowledge and understanding, owing in part to interference with construction of an accurate and coherent germ theory by cold weather folk beliefs, which are expected to be more common among children with less educated and lower income parents,.

4. With these indicators of SES controlled, any disadvantage that Latino and other minority children may show in comparison to European American children on measures of disease knowledge and understanding will be reduced in size.

Although the literature provides little basis for hypothesizing, we also examine whether experience having the flu is associated with increased knowledge and understanding and whether any gender differences are evident.

Method

Participants and Interviews

In a new analysis of data collected in 1992 for a study primarily focused on AIDS, we characterized interviews about flu conducted with 156 children aged 8 to 13 in third, fifth, and seventh grade. They came from 12 classrooms in three Catholic elementary schools in a southwestern city that were chosen because of their diverse student bodies. The students on which this report is based were randomly selected from the students in the larger AIDS education intervention study and were interviewed individually about the causality of both AIDS and flu in the pretest phase of the larger project. The larger sample represented 87% of the students who were given packets with parent consent forms and parent surveys to take home.

Mean ages of the three grade groups were 8.8, 10.7, and 12.6, respectively. Of the 61 males and 95 females in the sample, 35.4% were European American; 58.9% were Latino, primarily Mexican-American; and the remaining 5.7% were other minorities (African-American, Asian-American, or Native-American). The nine “other minority” students were grouped with the Latino students for purposes of analysis based on previous findings that children from a number of minority groups are, like Latino children, less knowledgeable than European American children about common respiratory illnesses (Badani & Schonfeld, 2002; McCann-Sanford et al., 1982).

Socioeconomic backgrounds were varied: Parent education ranged from 5 years to postgraduate degrees and averaged 14.3 years (the equivalent of some college or completion of an associate degree or technical training program after high school), and 29% of the children participated in the school free lunch program. Although we did not collect information about parents’ countries of origin, time in the United States, or acculturation, we asked a question about their use of English and Spanish. Among the Latino parents, only 5.7% reported using only or mostly Spanish; 31.3% used Spanish and English equally; 48.9% used mostly English, and 11.4% used only English In combination with average years of education of 13.9 years for Latino parents, this suggests that most were quite well acculturated to the United States.

Measures

In individual, open-ended interviews conducted at school, students were asked five basic questions, supplemented by a planned strategy of follow-up probing. Questions concerned: (1) what flu is, (2) a way someone could get it, (3) how the risk behavior or exposure to risk mentioned by the child would give someone flu exactly (and, in a planned follow-up question, whether something gets in their body that makes them get sick and, if so, what it is; or if not, what does happen to make them sick), (4) if people who get chicken pox get little red spots that itch, what happens to people who get the flu (i.e., what symptoms result), and (5) how the disease agent identified by the child works inside the body to make the person experience the main symptoms mentioned by the child. In the Piagetian tradition, interviewers probed for more specific answers where appropriate (e.g., Can you say more about what you mean by “from other people”? How does that work?), encouraging children to flesh out each part of the causal account they were constructing. The order of the flu and AIDS interviews was counterbalanced.

We should note that we did not define flu for children and that, like other developmental researchers (e.g., Jones & Rua, 2008; Legare & Gelman, 2009), we cannot be sure that they defined it as the CDC defines it. Many children and adults in our society have a broad concept of flu or concepts of both a cold-like flu with primarily respiratory symptoms and a “stomach flu” that aligns more closely with the medical concept of viral gastroenteritis than with influenza and can be caused not only by contact with infected people or their belongings but by food or water contaminated with, for example, noroviruses or bacteria. In the southwestern United States, where the present study was conducted, the folk meaning of flu has been found to be broad and to focus more on gastrointestinal symptoms than the CDC definition of influenza (see Baer et al., 1999; McCombie, 1987), although the Centers for Disease Control and Prevention (2018) states that some influenza victims, especially children, experience vomiting and diarrhea. It should be borne in mind that different children may have different illnesses in mind when they describe flu.

Original codings of interview responses were recoded and combined in various ways to create three new measures—ontology, causal propositions, and coherence—designed to capture the key features of intuitive theories outlined by Wellman and Gelman (1998). A fourth score, level of understanding, was a Piagetian measure of the structural complexity of children’s thinking.

Ontology.

To assess whether children had command of a biological vocabulary relevant to disease, we scored as 0 (not mentioned) or 1 (mentioned) each child’s use of five terms in their flu interview: germ, virus, immune or immune system, white blood cell, and cell of any sort. The ontology score was the average of the five items—i.e., the percentage of the five concepts mentioned. This measure assesses extent of relevant biological knowledge rather than conceptual understanding.

Causal propositions.

The second measure, also a percentage correct knowledge index, assessed a child’s causal explanatory framework by scoring the presence or absence of eight propositions that, based on reference to medical sources on infectious disease causality, we judged to be appropriate elements in a scientifically correct and complete account of how engaging in a risk behavior or being exposed to risk results in at least one recognized symptom of flu. As such, this measure was constructed to tap scientifically correct propositions regarding how one gets flu, what happens in the body to produce flu symptoms, and what those symptoms are likely to be. As with any knowledge test, the items varied in difficulty and the specificity of knowledge required; each proposition was coded 0 (not mentioned) or 1 (mentioned) and the codings were averaged to form a total score. Items were positively interrelated, as indicated by a Cronbach’s alpha coefficient of .65. Although mastery of the eight propositions primarily represents accurate knowledge, it likely reflects conceptual understanding as well in the sense that the child with a high score has assembled—although not necessarily woven together—the elements needed to construct a biologically sophisticated causal explanation of flu:

1. You get flu from other people: mention of saliva/mouth contact, physical contact, or airborne contact (e.g., being sneezed on) as the route of transmission

2. The other person must have flu (or the flu disease agent)

3. The contact results in internalizing a disease agent: the child can explain a process of internalization, even if the disease agent is not correctly identified

4. The disease agent is a germ or virus or the like; a correct answer does not require labeling it as a flu germ or virus)

5. The disease agent spreads or multiplies inside the body

6. The disease agent does damages in the body; the child may describe the agent hurting or upsetting the body only vaguely or discuss it attacking specific body parts or cells associated with symptoms

7. Flu symptoms are experienced: child mentions at least one likely head or stomach flu symptom (e.g., tiredness, headache, coughing, sneezing, runny nose, diarrhea, vomiting, stomach ache, general achiness, fever). This list corresponds well with the CDC’s (2018) list of symptoms, which cites “fever or feeling feverish/chills, cough, sore throat, runny or stuffy nose, muscle or body aches, headaches, fatigue (tiredness),” and which ends with the statement, “some people may have vomiting and diarrhea, though this is more common in children than adults.”

8. The body fights the disease agent: child mentions, whether the immune system is named or not, the body fighting the disease agent or disease or, even better, the idea that symptoms result in part from the immune system’s defense.

Coherence.

Guided by the concept of coherence as an “interrelatedness of ideas” (Wellman & Gelman, 1998), we created a measure of the logical coherence of the child’s account of how the disease agent causes flu symptoms. If the causal propositions score is about the number of ideas or elements that belong in a scientifically correct explanation that are mentioned, the coherence measure assesses whether the child’s ideas, right or wrong, are connected to one another in a logical causal argument. It resembles a Piagetian measure of understanding in that it assesses logical thinking independent of factual correctness; however, it assesses the connectedness of one idea to another to form a causal argument rather than the level of cognitive complexity of reflected in an explanation.

Two codings were averaged to form this score; the two codings were intended to be similar and were in fact highly correlated with each other, r (156) = .81, p < .0001. The first coding assessed the linking of a disease agent to damage to the body and damage to the body, in turn, to symptoms, creating at least in vague form a causal chain leading from internalized agent to damage to the body (the mediator) to observable symptom; it was coded 0 (no explanation), 1 (partial explanation, connecting disease agent to damage to the body or damage to an identified symptom but not both), or 2 (full causal argument connecting disease agent to damage and damage to at least one symptom, whether right or wrong).

The second coding assessed the overall logical tightness of the child’s explanation of how at least one disease symptom proposed by the child comes about. It did not require identification of damage to the body as a mediator of the effect of the disease agent and was intended to give credit to other lines of explanation (e.g., the child’s explanation could be entirely behavioral or psychological in nature or could feature a causal mechanism other than damage to the body). Whatever the ideas, it required connecting them to one another to create a coherent, logical explanation (e.g., through temporal ordering, causal phrases such as “and that makes…,” and a lack of tangential material or gaps in the causal chain). It was also coded 0 (no explanation, confusion), 1 (partial explanation that makes some logical sense, or 2 (a full explanation leading from disease agent to symptom with logical connections between elements in the causal chain). These two measures of coherence are described further in Table 1 and illustrated with examples. The summary coherence score averaged the two codes and could therefore range from 0 to 2.

Table 1.

Coding of the Two Coherence Items with Examples

Damage to Body as Mediator between Disease Agent and Symptom	Examples	Logical Tightness of Ideas	Examples
0 = No explanation provided or can’t explain how agent causes damage and symptom	Can you tell me a way that someone can get the flu? BY SHARING A DRINK AND THEY HAVE IT AND DON'T KNOW… ONE PERSON IS SICK AND THE OTHER PERSON DOESN'T KNOW AND THAT PERSON DOESN'T KNOW IT AND THAT PERSON HASN'T GONE TO THE DOCTOR YET AND THE OTHER PERSON CAN GET IT. If someone gets the flu from sharing a drink is there something that gets in their body? I don’t know.	0 = No explanation offered OR can’t think casually (confuses causes and effects, gives ideas without connections between them, offers implausible or unclear ideas)	How does the disease from the flu make them look red in the face? What happen inside their body? THEY START FEELING HOT AND THAT'S HOW COME THEY'RE TURNING RED. RPT. What happens first after the disease gets in their body to make them feel hot? THEY MIGHT TAKE THEIR TEMPERATURE AND THEIR FOREHEAD IF YOU PUT YOUR HAND THERE IT FEELS HOT IF THEIR HEAD IS HOT THEN YOU GO TO THE DOCTOR AND THEY TELL YOU THAT YOU HAVE THE FLU AND YOU FEEL REALLY BAD. So the disease gets in the body then what happens inside to make them feel hot? I DON'T KNOW.
1 = Partial explanation; describes agent doing some damage to the body, but can’t say how that damage creates the specific symptom OR describes how damage to body results in the symptom, but not how the agent initially damages the body	Can you tell me a way someone could get the flu? BY BEING BAREFOOT WHEN IT’S REALLY COLD.… BECAUSE THE FLOOR IS COLD AND WHEN YOU’RE BAREFOOT AND IT'S COLD YOUR FEET GET REALLY COLD AND WHEN YOU’RE SOMEWHERE HOT IT CAN GET TO YOUR FEET AND YOUR FEET WILL START GETTING REALLY COLD .…. IF IT’S GETTING REAL HOT AND YOUR FEET ARE STILL COLD THEN YOU GET IT THAT WAY. If someone gets the flu from their feet getting cold does something get in their body to make them get sick? PROBABLY GERMS FROM THE FLOOR.	1 = Can explain part but not all of a complete causal chain connecting agent to symptom (with or without damage to body) and makes some logical sense, whether mostly right or wrong	Can you tell me what a virus is exactly? IT’S LIKE BACTERIA THAT GROWS IN YOUR BODY UNTIL SOMETHING CONTROLS IT. And then what did you say after that? UNTIL IT MAKES THEM SICK. When people get chicken pox they get little red spots that itch. What happens to people when they get the flu? THEY GET HIGH FEVERS AND THEY THROW UP A LOT. 5. How does this virus make them get a high fever? How does this virus do that? ITS LIKE A SIDE EFFECT, I'M NOT REALLY SURE.
2 = Describes agent damaging body and shows how that damage results in a specific symptom (both steps in causal sequence clear)	How does the germ make them feel hot? What happens inside their body? THE GERM LIKE, I THINK CAN LIKE WALK AROUND IN YOUR BODY, AROUND YOUR BODY AND LIKE LEAVE LITTLE THINGS THAT MAKE YOU GET FEVERISH AND STUFF. Leave what kind of little things? MAYBE CLEAR LITTLE THINGS THAT YOU CAN'T SEE LIKE EGGS… LIKE THEY [THE LITTLE EGGS] THEY CAN, LIKE BREATHE HOT AIR EVERYWHERE AND THEN THINGS GET REAL HOT.	2 = Can fill in whole causal chain connecting agent to symptom, right or wrong, with or without damage to body as mediator; connections are made clear through ordering of ideas, use of terms such as “because,” “and then,” “and that makes.”	THEIR BODY STARTS TO FIGHT BACK AGAINST [THE GERMS] BY SNEEZING, BY SOMETIMES THROWING UP. INSIDE THEIR BODY, THEIR BODY HAS AN IMMUNE SYSTEM; THE WHITE BLOOD CELLS. AND THE WHITE BLOOD CELLS FIGHT BACK AGAINST THE DISEASE. THEY CREATE MUCOUS AND, I CAN'T REMEMBER THE OTHER WORD. AND THEN THE MORE GERMS YOU GET, THE MORE IT BUILDS UP. AND WHEN YOU SNEEZE IT TAKES THE GERMS OUT WITH THE MUCOUS. [No clear damage caused by the germs, but logical explanation of immune system action against germs.]

Piagetian level of understanding.

The systems developed by Bibace and Walsh (1980) and Walsh and Bibace (1991) were adapted for this study to code levels of conceptual understanding of illness: (0) no response or don’t know, (1) a largely magical association between some phenomenon and illness, (2) the association of a specific cause and a specific effect without causal explanation, (3) an account of how an external cause leads to internalization of a disease agent and symptoms involving the whole body, (4) a mechanistic sequence of specific causes leading to specific changes in internal organs/processes and symptoms, (5) an entire causal sequence involving interactions of multiple causes and effects, with the body playing an active role, and (6) a full physiological explanation involving interacting causes and effects. This coding yielded a level of understanding score that could range from 0 to 6, with 6 representing maximum causal sophistication (although not necessarily correct factual knowledge). As in previous studies (e.g., Legare & Gelman, 2009; Schonfeld et al., 1993), level of understanding was treated as an interval scale for purposes of analysis.

Coding reliability.

The original codings of each interview were done soon after the interviews were completed and transcribed. Two coders from a pool of seven trained coders who were blind to the age of the interviewee coded each interview and reached a consensus on disagreements before assigning final codes. For this analysis, however, we combined original variables and collapsed coding categories in various ways to create new intuitive theory measures (e.g., scoring mention of any one of several flu symptoms as knowing flu symptomatology). In order to establish the reliability of these new measures, a new coder was trained and then coded the new variables for a random sample of 25 interviews. Her codings of the new variables were then cross-tabulated with scores on these variables calculated from the original data.

Both interrater agreement percentages and Cohen’s kappa coefficients (which correct for chance agreement) were calculated. For the eight propositions in the causal propositions measure, agreement averaged 96% and kappa coefficients ranged from .63 to 1.00 and averaging .87. For the ontology score, agreement was 92% (κ = .88); for the coherence score, agreement was 84% (κ = .77); and for level of understanding (exact agreement on the 0-to-6 score), agreement was 92% (κ = .86). All kappas were significant at p < .001 or beyond. This provided the justification needed to use in the analyses reported below the new scores as calculated from the original data.

Results

We first conducted analyses of variance to check for possible order of AIDS and flu interview administration effects, comparing the four summary scores of students who received the flu interview first to those of students who received it second. In no case was the order of administration effect significant so it was ignored in further analyses.

At the start of the flu interview, children were asked what flu is. Percentages of children in third, fifth, and seventh grades who at least implied that flu is a sickness, disease, illness, or virus in response to the open-ended question were 83.6%, 94.2%, and 98.0% respectively, a significant increase, χ² (2, N = 156) = 7.63, p < .05. Children who gave unclear responses or did not know, mostly third graders, were asked a multiple-choice question (Is it a kind of bird, a sickness, a medicine, or a kind of car?) and all 13 children asked passed this easier test, resulting in all 156 children interviewed demonstrating at least minimal knowledge of what flu is. Children were also asked if they had had the flu; 54% of third graders, 69% of fifth graders, and 82% of seventh graders said they had, χ² (2, N = 151) = 9.19, p < .01.

Grade Differences in Summary Measures

Table 2 shows means and standard deviations for the four summary measures at each grade level. Ontology scores increased significantly from third to fifth grade and again from fifth to seventh grade, but even seventh graders used only about one or two of the five disease-related terms assessed. In the sample as a whole, 59.0% referred to germs somewhere in their interview, 24.4% to a virus, 16.7% to cells, 6.4% to immune or immune system, and only 1.9% to white blood cells.

Table 2.

Means and F Statistics for the Four Summary Measures

	Grade level
Summary scores	3rd	5th	7th	F
Ontology	.10a (.10)	.24b (.15)	.33c (.19)	33.74***
Causal propositions	.43a (.24)	.62b (.18)	.70b (.19)	6.46***
Coherence	.73 (.64)	.95 (.49)	.96 (.54)	n.s.
Piagetian level of understanding	2.95a (.83)	3.37b (.60)	3.47b (.65)	8.30***
n	55	52	49

Note: Ontology score is the proportion of five disease-relevant biological terms used; causal proposition score is the average of eight 0 (not mentioned) and 1 (mentioned) proposition codes; coherence is the average of two codings of the completeness of causal linkages between the parts of an explanatory argument as no explanation (0), partial explanation (1), or full explanation (2); and Piagetian level of understanding is a 0 to 6 score. Means in a row that do not share a superscript in common are significantly different (p < .05) based on post hoc Tukey’s HSD tests.

^*p < .05

^**p < .01

^***p < .001

Causal proposition scores increased significantly from third grade to the two higher grades; seventh graders communicated 70% of the eight propositions. Coherence scores did not increase significantly with age and were rather low (below the midpoint on average on a scale of 0 to 2). Most children could construct part but not all of a logical, connected causal argument. Children were typically more able to describe how the disease agent gets in and does damage to the body than to say how this results in symptoms.

Finally, scores on the Piagetian level of understanding measure increased from third to fifth and seventh grades; most children (53.2% across grades) expressed at least a vague understanding that something gets inside the body (level 3 understanding), and another 35.3% were able to describe an external cause, an internalized disease agent, and something going on in the body underlying the emergence of symptoms (level 4); most of the rest (8.3%) scored at level 2 or lower. Only 1.3% (two students, both in seventh grade) scored at level 5 and gave a sophisticated explanation involving multiple interacting influences.

Table 3, supplemented by information provided here, elaborates on this account by detailing which of the eight propositions in a biologically accurate theory of flu were mastered at each grade level. The picture conveyed is of steady improvement from grade to grade except on the proposition that the disease agent spreads and/or multiplies within the body, which was mentioned by only 33% to 41% of the children. Fewer than half of the third graders communicated that the person from whom someone gets flu must have flu, that the causal agent is a germ or virus, that the agent spreads or multiplies in the body, that the agent damages something in the body, or that the body tries to fight the agent off. Half or more of these 8-9-year-olds communicated that you get flu from contact with other people (most often from airborne transmission—e.g., from others’ coughing or sneezing), that a disease agent is internalized, and that symptoms associated with flu are experienced. Whereas 45% of the third graders identified the disease agent as a germ or virus, this concept was mastered by almost all fifth and seventh graders. Yet only a few children explicitly said that the germ or virus involved was specifically a flu virus. Percentages doing so were 0%, 13%, and 14% across the three grade levels.

Table 3.

Proportion of Children at Each Grade Level Demonstrating Mastery of Each of Eight Causal Propositions

	Grade Level
Causal propositions	3rd	5th	7th	F
You get flu from contact with other people	.53a	.79b	.88b	9.68***
The other person must have flu	.35a	.42a	.65b	5.51**
Contact results in internalizing a disease agent	.58a	.77ab	.90b	7.46***
The disease agent is a germ/virus	.45a	.83b	.92b	19.77***
The agent spreads/multiplies in the body	.33	.42	.41	n.s.
The agent damages/hurts the body	.29a	.67b	.61b	10.05***
Flu symptoms appear	.85a	.96ab	.98b	3.81*
The body fights the disease agent	.05a	.12a	.29b	6.18**
n	55	52	49

Note: Each causal proposition score was coded 0 (not mentioned) or 1 (mentioned) so each entry is the proportion of children mentioning a proposition. Means in a row that do not share a superscript in common are significantly different (p < .05).

^*p < .05

^**p < .01

^***p < .001

As for symptomatology, the most commonly mentioned symptoms across grade levels were respiratory symptoms typically associated with colds (55.1%), followed by stomach-related symptoms (46.2%), fever (26.9%), and tiredness or weakness (23.7%). This suggests that both respiratory and gastrointestinal symptoms were prominent in children’s models of flu. Finally, the least mastered proposition was that the body fights the disease agent; even at ages 12-13, only 29% referred to the body defending itself.

Relevant to our special interest in cold weather folk beliefs among Latino children, 23.8% of the sample as a whole referred to cold weather or related concepts in explaining how someone could get flu, compared to 54.5% mentioning airborne transmission (e.g., being sneezed on), 21.2% saliva contact, and 7.7% touch or physical contact. Cold weather thinking tended to be more common among Latino children overall, 26.7% vs. 14.5%, but the difference fell short of significance, χ² (1, N = 156) = 3.04, p < .10. However, cold weather beliefs were especially common among third and fifth grade Latino/minority children and declined significantly with age among Latino/minority children, from 32.4% in third grade and 37.1% in fifth grade to only 9.4% in seventh grade, χ² (2, N = 101) = 7.41, p < .05. The corresponding percentages of European American children expressing cold weather beliefs were 14.3%, 17.6%, and 11.8%, respectively (n.s.). Despite this ethnic difference, the two groups were equally able to identify a germ or virus as a disease agent for flu; overall, 73.3% of Latino/minority students and 70.0% of European American students did so, and the two ethnic groups were similar at each grade level.

Correlations between Measures

As a prelude to multiple regression analyses predicting the four summary scores, the correlations among these scores were examined. The zero-order correlations were all statistically significant, ranging from .26 to .58 and averaging .47. Partial correlations controlling for age, which give a more meaningful estimate of interrelatedness, were all statistically significant as well, averaging .40 and ranging from .20 between ontology and coherence (p < .05) and .34 between ontology and level of understanding to .54 between coherence and level of understanding and .53 between causal propositions and level of understanding. Correlations involving the ontological knowledge score tended to be lowest, possibly because this was the only measure not tapping in some fashion the sophistication of the child’s causal account. Overall, the four measures were moderately interrelated, but they were distinct enough to warrant separate analysis.

Multiple Regressions Predicting Scores

Table 4 reports the results of forced-entry multiple regression analyses conducted with SPSS version 24 to assess relationships between the demographic and experiential variables and the four summary scores. The predictors collectively accounted for a high of 35% of the variance in ontology scores, lower but still significant percentages of the variance in causal proposition mastery and Piagetian level of understanding, and a nonsignificant 9% of the variance in coherence scores.

Table 4.

Standardized Beta Coefficients and R² Statistics for Multiple Regressions Predicting Flu Summary Scores Based on Demographic and Experiential Variables

Predictor	Biological ontology	Causal propositions	Coherence	Level of understanding
Age	.54***	.43***	.19*	.30***
Gender	−.00	.09	.11	.12
Ethnicity	−.25**	−.21*	−.17	−.30**
Years of parent education	.17*	.17*	.11	.06
Free lunch program	−.15	−.16	.02	−.17
Had the flu	−.07	.04	.02	.04
R2	.35***	.26***	.09	.19***
F (6, 120)	10.79***	7.15***	1.93	4.77***

Note: Codings are as follows: gender (0 = male, 1 = female), ethnicity (0 = Latino/minority, 1 = European American), free lunch program (0 = not qualified for free lunch, 1 = free lunch participant), had the flu (0 = no, 1 = yes).

^*p < .05

^**p < .01

^***p < .001a

The consistent relationship of age to all four scores is to be expected in light of the age group comparisons presented thus far; age was consistently the best predictor. However, with parent education and free lunch program participation controlled, ethnicity also proved to be a consistent predictor. Latino/minority children scored higher than European American children on the measures of ontology, causal propositions, and Piagetian level of understanding. Meanwhile, years of parent education made its own independent contributions to higher ontology and causal proposition scores. Participation in the free lunch program was not a significant predictor in its own right. Finally, neither gender nor having personally experienced the flu made an independent contribution to intuitive theory scores or level of understanding when age and other factors were controlled.

Discussion

Our operationalization of intuitive theories of influenza in terms of the features highlighted by Wellman and Gelman (1998) resulted in characterizing the flu theories articulated by children aged 8-9, 10-11, and 12-13 in terms of their use of a disease-relevant biological ontology, communication of propositions that belong in a biological causal explanatory theory of flu, and coherence of thought in the sense of connectedness of ideas to one another in a causal argument. The intuitive theory measures, except for the coherence score, as well as the measure of level of understanding of disease in the Piagetian sense, increased with age, consistent with prior intuitive theories and Piagetian research. Moreover, as expected, the four scores were moderately correlated with each other with age controlled. This suggests that each of the intuitive theory measures adds unique information to what the Piagetian approach tells us about children’s thinking about disease.

Multiple regression analyses revealed that, along with the expected age effects on scores, Latino/minority children not only equaled but outscored White children on three of the four measures when two socioeconomic status indicators, parent education and participation in the free lunch program, were controlled—an unexpected finding to be discussed below. Years of parent education made its own independent contribution to higher biological ontology and causal proposition scores (c.f., Sigelman, 2012; Zhu et al., 2009), but free lunch program participation did not. Contrary to Jones and Rua (2008), experience with flu was not a significant predictor, possibly because flu (especially if believed to encompass both respiratory and stomach flu) is so common that even children who had not had it themselves learned about it from the experiences of others who did. Finally, like other studies, this one failed to find differences between girls’ and boys’ illness understandings or knowledge (e.g., Paterson et al., 1999; Schonfeld et al., 1993).

Characterizing Intuitive Theories

We set out to devise a comprehensive approach, informed by the intuitive theories perspective, to characterizing children’s conceptualizations of the causal factors and mechanisms involved in disease. Partial correlations among the three intuitive theory scores and the Piagetian level of understanding score controlling for age were all significant and averaged .40. This suggests that knowledge, assessed primarily by the ontology and causal proposition scores, and logical understanding, whether the coherence or connectedness of ideas in the child’s theory or its level of cognitive complexity in the Piagetian sense, were interrelated. Interestingly, the correlations were not noticeably lower when they involved a measure of accurate knowledge (ontology or causal propositions) and a measure of logical sophistication independent of correctness (coherence and Piagetian level of understanding) than when they involved two knowledge scores or two logical sophistication scores.

This interdependence between knowledge and logical sophistication is consistent with theorizing that knowledge and the capacity for causal thought complement and advance one another in development (Wellman & Gelman, 1998). Although the matter requires longitudinal study, the relationship between knowledge and understanding is most likely bidirectional: General advances in cognitive development of the sort emphasized by Piagetians enable children to better assimilate and organize new information, and exposure to new information helps them think in more sophisticated ways. At the same time, the moderate size of the correlations among our four measures suggests that each contributes something unique to the characterization of children’s thinking about disease, that the intuitive theories approach adds something that is missing from the Piagetian approach, and that a child’s theory of a disease can be advanced in one sense but not in another.

Sociocultural Differences

Based on previous research (e.g., Sigelman, 2012; Taveras et al., 2004), we predicted that controlling for parent education and free lunch program participation would reduce or even eliminate disadvantages in knowledge and understanding of disease often found when Latino and other minority children have been compared to European American children (Au & Romo, 1996; Badani & Schonfeld, 2002; McCann-Sanford et al., 1992). Latino/minority children in this study, as in previous ones, had less educated parents and were more likely to be in the free lunch program at school. However, in multiple regression analyses controlling for SES and other factors, Latino and other minority children in this study actually displayed more sophisticated knowledge and understanding than European American children judging from their ontology, causal propositions, and level of understanding scores.

Latino children were not more likely to report having had the flu. Moreover, although younger Latino/minority children quite often offered the cold weather explanations of illness associated with Latino cultures, this did not diminish their ability to identify germs or viruses as the disease agent for flu. This is further evidence that cold weather and germ theories of disease can coexist even though germ theories become more dominant with age (Sigelman, 2012) and, more generally, adds to evidence that the coexistence of different theories of the same phenomenon is common across cultures and age groups (Legare & Schtulman, 2018).

How do we explain the strong performance of the Latino children, then? Because our Latino parents were generally proficient in English and quite well educated, although not as well educated as the European American parents, it is possible that the Latino children we studied were more advantaged and therefore more knowledgeable than the Latino children in previous studies. Another possible explanation is that Latino parents are more worried about and attentive to signs of physical illness than European American parents and may socialize their children to be more attentive to information about physical health threats and symptoms. Rates of flu are especially high among Latino adults (Riffkin, 2015). Moreover, they obtain flu vaccinations at lower rates while worrying more about both getting sick from flu vaccines and having a high risk of getting the flu if they do not get vaccinated (Santibanez, Singleton, Santibanez, Wortley, & Bell, 2013).

Consider too a study of anxiety in childhood (Varela et al., 2004) in which Mexican and Mexican-American parents, reacting in family discussions to ambiguous situations that might involve anxiety (e.g., explaining why a child felt funny in her stomach on the way to school), offered more somatic interpretations of what European American parents more often saw as anxiety. This finding was attributed to Latino cultural beliefs that make bodily symptoms more socially acceptable than symptoms of emotional distress. Moreover, parents’ use of somatic interpretations of the scenarios was positively correlated with their children’s scores on a measure of worrying. In further support of this line of argument, Latino children are more likely than European American children to report worries about health (Silverman, La Greca, & Wasserstein, 1995). In addition, parents’ fears about flu are associated with their children’s fears, with parents’ communication of flu threat information to children mediating the relationship (Remmerswaal, 2011). Arguably, then, greater parental attention to somatic symptoms and greater worry about illness in Latino families could contribute to children’s enhanced attentiveness to information about common infectious diseases. However, we are aware of no direct evidence of this; as a result, our surprising finding that, with socioeconomic factors controlled, the Mexican American and other minority children in our sample had more advanced knowledge and understanding of flu needs to be replicated and further explicated before too much is made of it. The main message of this analysis is that investigations of racial/ethnic differences in children’s knowledge and understanding of disease must control socioeconomic differences between groups and should also incorporate measures of ethnic folk theories and beliefs that may coexist with a germ theory of infectious disease.

Limitations and Conclusions

This study involved a convenience sample in which European American children’s families had higher education and income levels than Latino children’s families. In addition, the interview protocol may have encouraged children to think in more advanced ways than they might otherwise have thought by asking them, if they did not mention the concept of internalization spontaneously, whether something gets in the body that makes people get sick, by giving them the example of chicken pox to get them thinking about symptoms associated with flu, and by probing them to explain more specifically key events in their causal narratives. At the same time, open-ended questions generally demand more cognitive and linguistic skill than structured or closed-ended questions and are therefore prone to underestimate what children, especially younger children, know about disease (Sigelman, 2014).

In addition, as noted above, the term “flu” is commonly used to describe both respiratory and gastrointestinal symptoms. It seems likely that those children who described primarily “stomach flu” symptoms—and there were almost as many as described primarily respiratory symptoms—were thinking of illness spread through contact with people who are ill rather than through food or water contamination independent of contact with other people. However, we did not attempt to distinguish between alternative conceptions of flu and their causality—an interesting topic for future research—and must assume that the present results pertain to whatever children themselves meant by “flu.” Finally, because the data were collected in the early 1990s, we cannot be sure the findings generalize to today’s children. However, we have no compelling reason to think that they do not generalize, especially since age has consistently emerged as a stronger predictor of children’s knowledge and understanding of diseases than a variety of socioeconomic and sociocultural factors.

The next step in this line of research is to further develop the basic approach to assessing intuitive theories of disease laid out here, extending it to other diseases, to diverse samples of children and adolescents, including both younger children and older adolescents than we studied, and to other aspects of disease such as prevention and treatment. We would encourage devising more comprehensive and multifaceted approaches to assessing both ontologies and causal explanatory frameworks—for example, by assessing more elements of each or defining meaningful subsets of elements, gauging the meanings children attach to concepts like germ and virus, and using structured as well as open-ended questions to optimize children’s opportunities to express intuitions they may have difficulty verbalizing. Because scores on a knowledge index such as the ontology or causal propositions measure could reflect different degrees of knowledge sophistication depending on which items were scored right or wrong, it would also be useful to determine through more elaborate correlational analyses or instructional experiments which specific pieces of knowledge contribute most to the construction of accurate and conceptually sophisticated theories of disease. Finally, because our coherence measure appeared to be relatively insensitive to developmental differences, we would especially encourage exploration of other methods of assessing the coherence of children’s thinking (e.g., by assessing the proportion of propositions or ideas in an explanation that are linked together or by asking children to predict whether or not novel risk behaviors or disease agents can cause flu in order to assess their ability to generate theory-consistent hypotheses). Such research would help determine whether children can go beyond reciting facts they may have learned in or out of school to showing deeper understanding of disease causality.

In the meantime, the findings on children’s concepts of flu reported here have several implications for parents, educators, and health professionals interested in helping children understand and protect themselves from health threats like influenza. The findings point to the value of assisting children in constructing more complete theories by building their ontological knowledge not only about germs but also about disease-specific germs, viruses, and other microbes; by challenging their misconceptions about the role of cold air in colds and flu; by teaching them more about how germs multiply inside and damage the body and about how the body’s immune system mounts a defense against the invaders; and by helping them “connect the dots” in causal chains linking risky contact with people who have a disease to internalization of a disease agent, to invisible events inside the body, and ultimately to observable symptoms. Children can acquire more sophisticated understandings of illnesses—and will even adopt more appropriate disease prevention behaviors as a result—if they are given key facts along with the critical information about underlying causal mechanisms that helps them tie facts together in a coherent theory (Au et al., 2008; Au & Romo, 1996; Myant & Williams, 2008; Sigelman et al., 1996; Weisman & Markman, 2017). By rigorously analyzing children’s intuitive theories of illness, we will be in better position to help children elaborate them so that they can better protect their health.

Flu Highlights.

• A new method was devised to assess key features of intuitive theories of disease.

• Open-ended interviews about flu with children aged 8-13 were analyzed.

• Scores increased with age, were moderately correlated, and showed knowledge gaps.

• Age, Latino/minority ethnicity, and parent education predicted high scores.

• Findings advance use of the intuitive theories perspective in health promotion.