“When something like a ladybug lands on you”: Origins and development of the concept of luck

Abstract

In Study 1, 103 children ages 4 through 10 answered questions about their concept of and belief in luck, and completed a story task assessing their use of luck as an explanation for events. The interview captured a curvilinear trajectory of children’s belief in luck from tentative belief at age 4 to full belief at 6, weakening belief at 8, and significant skepticism by 10. The youngest children appeared to think of luck simply as a positive outcome; with age, children increasingly considered the unexpected nature of lucky outcomes and many came to view luck as synonymous with chance. On the story task, younger children attributed a stronger role to luck in explaining events than did older children. Studies 2 and 3 explored two potential sources of children’s concepts. Study 2 explored adult use of the words luck and lucky, and found that most of this input consisted in using lucky to refer to positive outcomes, although the nature of use changed with the ages of the children. In Study 3, we examined children’s storybooks about luck and found them to be rich potential sources of children’s concepts.

A pocketful of luck is as good as a sackful of wisdom.

– Proverb.

References to luck are ubiquitous in Western society: College campuses are alive with “good luck!” wishes before exams; we remark that it must be our “lucky day” when we find a good parking space; a parent tells her child that she is a “lucky girl” when she receives gifts at her birthday party. These examples reflect three different ways the concept of luck is commonly used in everyday life. First, luck is often thought of as a supernatural causal force, one that can be harnessed through the use of lucky objects and behaviors to confer positive outcomes. For example, during the 2016 election campaign, President Obama purportedly attempted to recruit luck by playing basketball before each primary election; Michael Jordan wore his college practice shorts underneath his NBA uniform for good luck. This practice of possessing “lucky charms” or engaging in “lucky” behaviors has been present throughout history, and is even documented in a collection at The British Museum entitled, “Guardians and Lucky Charms.”

Second, we recruit luck as an explanation for unusual or unexpected positive (and sometimes negative) events (e.g., finding a parking space in a crowded city). That these sorts of unexpected outcomes occur is an objective reality that we often refer to as “chance.” However, when these random or chance outcomes also have a particularly strong emotional valence or seem unlikely, they are often referred to as lucky. For example, the press has been known to attribute near misses in the wake of natural disasters to luck, as they did regarding the limited damage in Mexico City from a potentially devastating earthquake: “Luck, not tougher building standards, spared Mexico in quake” (New York Times, 9/23/2017). Finally, luck can be thought of as a personal trait, as when a parent tells her child that she’s a “lucky girl.” Simply put, some people are thought to be lucky. Our research aims to address each of these three senses of luck, as well as the extent to which children view it as a real-world phenomenon.

Many uses of the words “luck” and “lucky” can be considered colloquialisms—we simply want to indicate our support for the exam-taker; we like our children to feel fortunate. Yet research documents beliefs in luck as a supernatural causal force in adults (Andre’, 2006; Day & Maltby, 2003; Wiseman & Watt, 2004). Albas and Albas (1989) show that “good luck” wishes on college campuses reflect more than just colloquialisms: In a 13-year qualitative study of 300 college students’ behaviors around exam times, one fifth to one third of students reported using magical thoughts and behavior to garner good luck. Students reported two types of behaviors: (1) possessing lucky objects or wearing particular clothing, and (2) engaging in particular behaviors, such as taking a certain path on the way to class. One student was so convinced of luck’s efficacy that he would make himself late to exams by searching for a “lucky” coin at a bus stop, sacrificing the benefits of having a full period to work for the supernatural advantage of a lucky coin.

Darke and Freedman (1997) developed the “Belief in Good Luck Scale” to quantify these beliefs. They uncovered reliable individual differences, with some participants exhibiting belief in luck as a stable, internal force, and believing that they possessed it. Others thought of luck as external, random, and unreliable, akin to chance. Participants who reported believing in luck as a stable force acted as if experiencing good luck would help them in the future. Importantly, these participants’ beliefs in good luck as a force were distinct from simply thinking of luck as a synonym for being fortunate.

In contrast to Darke and Freedman’s (1997) view that luck concepts range on a continuum from belief in luck as a supernatural force to believing it to be equivalent to chance, Thompson and Prendergast (2013; Prendergast & Thompson, 2008) suggest that there are two discrete dimensions of luck beliefs, with no relations between them: (1) beliefs in luck as a supernatural force, and (2) beliefs about personal luckiness. They reported that even disbelievers in luck as a supernatural force nonetheless still have beliefs about their own personal luckiness.

These senses of luck are reflected in research on a variety of adult populations. Pritchard and Smith (2004), for example, report that gamblers think of luck as referring to people and chance as referring to events, believing that some people are luckier than others, but that chance is the same for everyone. Supporting Thompson and Predergast (2013), Risen (2016) acknowledges that adults who do not believe in luck often ignore their sensibilities and participate in lucky rituals anyway. This dissociation between beliefs and behavior has been identified in other forms of magical thinking (see e.g., Subbotsky, 1993). This apparent suspension of disbelief along with the great lengths many go to in order to “get” luck makes it a unique concept in need of exploration.

Where do beliefs in luck come from? Research and theory on adult superstitions proposes that they result from various inferential errors and biases used to navigate the uncertainties of daily life (e.g., Risen, 2016; Vyse, 1997). But this explanation begs the question of the origin and development of these concepts. Historically, children are thought to be the strongest believers in concepts like luck and magic (Woolley, 1997). Yet there is little evidence of what children’s luck concepts might be like. Three dimensions of luck concepts that warrant consideration are whether children (1) use lucky objects or behaviors to influence outcomes (luck as supernatural causal force), (2) appeal to luck to explain events (luck as explanation), or (3) think of luckiness as a personal trait (luck as personal attribute). Regarding the first, Opie and Opie (1959) report that children often use good luck charms as a way to improve their performance in school and games. Epstein (1993) reports that 45% of children claim to have at least one good luck charm. To address the development of belief in luck as a causal force, we interviewed children from a wide age range regarding their experiences with both lucky objects and rituals. Previous research shows that children attribute causality to natural invisible entities such as beliefs and desires (Bartsch & Wellman, 1995) and germs (Kalish, 1996). We expect our findings to provide insight into the extent to which children extend causal powers to invisible supernatural forces as well, thus broadening our knowledge of children’s beliefs about causality.

Regarding the second dimension, children endorse a range of supernatural explanations, including magic (Phelps & Woolley, 1994; Subbotsky, 1993), miracles (Woolley & Dunham, 2017), and karma (Banerjee & Bloom, 2017). Woolley and Cornelius (2017) showed that 6- and 7-year-olds agreed with proposals that luck caused a range of improbable and extraordinary events. Skinner (1990) found that children aged 7 through 12 identified luck as an explanation for school events. The present study builds on this work by systematically assessing the roles played by age, emotion, and likelihood of outcome on children’s use of luck as an explanatory tool. Although most work on children’s explanations has addressed their use of natural (physical, mental, and biological) explanations, knowledge of when children also marshal supernatural explanations is essential to telling a complete story. If we can identify the conditions under which children believe luck to be an appropriate explanation, we can look for convergences and divergences with factors that affect their use of natural explanations, and ultimately, explore when children endorse one over the other.

Regarding the third dimension of luck concepts, there is evidence that children think of luckiness as a personal trait. Amir and Williams (1999) reported that 72% of 11- to 12-year-olds agreed that some people are luckier than others. Skinner (1990) found that children aged 9 through 12 who reported feeling unlucky and thought that luck was important in determining their grades tended to be rated by teachers as being disengaged with classroom material; they also hold lower beliefs about their ability (Skinner, Wellborn, & Connell, 1990). Olson et al. (2008) explored whether 4- to 12-year-olds prefer lucky to unlucky people and whether children think that being lucky determines how people act. “Lucky” and “unlucky” people were operationalized as characters to whom random good or bad things happened, respectively. Children preferred lucky to unlucky people and also expected lucky people to be more likely to perform good actions; children also judged lucky people as nicer. In our work, we assess directly whether children think that they themselves and others are lucky. Children’s beliefs about whether luck is a trait possessed by themselves and certain others might have important implications for theories of motivation and achievement (Young, Chen, & Morris, 2009).

The findings that children have good luck charms, use luck as an explanatory tool, and reason about “lucky” people suggest that all three senses of luck are present during childhood. Using a structured interview and a set of vignettes, our first study explored these three facets of understanding, as well as explicitly assessing children’s belief in luck. Our second and third studies explored the nature of the input children might use to form these concepts and beliefs, specifically, adult speech and storybooks. Clearly, luck is not taught in school. To further complicate matters, luck is thought to be an invisible process, and invisible processes and entities, both real and not-real, pose unique challenges to learning (Harris et al., 2006; Woolley & Brown, 2015). Whereas parents report encouragement of other supernatural concepts, like magic (Rosengren, et al., 1994) and Santa Claus (Prentice & Gordon 1986), it is not clear whether they similarly encourage or discuss luck. Parents might endorse luck implicitly, such as by giving children lucky pennies to throw in fountains, but not discuss it further. Thus, children might form their concepts both from the things that people say (e.g., “good luck!”) and do (e.g., throwing pennies in fountains). Their concepts might also be influenced by exposure to various forms of media such as storybooks, television, and movies. Specifically, here we sought information regarding the extent to which storybooks about luck present it as a real-world causal phenomenon.

Study 1

In Study 1, we developed a comprehensive interview probing various facets of children’s concepts of luck and constructed vignettes to explore their beliefs about its explanatory power. The interview addressed children’s ideas about what luck is, who has luck, good versus bad luck, and their use of lucky charms and rituals. The vignettes were aimed at understanding what parameters define an event that can be explained by luck. Rescher suggests that, “Luck is a matter of having something good or bad happen that lies outside the horizon of effective foreseeability” (Rescher, 1995). Research also shows that negative events provide stronger motivation to infer causal explanations (Banerjee & Bloom 2017; Gray & Wegner, 2010). Thus, two components were central: emotional valence and likelihood, that is, whether an outcome is expected or unexpected. We developed vignettes that varied these two factors to explore their role in children’s use of luck as an explanatory tool. We also varied, on a subset of the stories, whether verbal wishes or lucky objects were involved. We expected that, if children interpreted “good luck” wishes as simply generic offers of support, that they would refrain from explaining outcomes in which these wishes were expressed as involving luck.

Finally, we also explored potential relations between luck and family religiosity. Although findings regarding relations between religious and supernatural beliefs among Western adults are mixed (see e.g., Rice, 2003), one might expect that if one holds a determinant world view, there should be little room for concepts of luck, either as a supernatural causal force or as an explanatory tool. The idea of some people being luckier than others might also be incompatible with a religious world view. To the extent that those beliefs are transmitted to children, we might find a negative relation between family religiosity and luck. At the same time, studies with children have shown that their religious beliefs are often intertwined with emerging fantastical and superstitious beliefs. For example, Woolley and Phelps (2001) note various commonalities in children’s concepts of wishing and prayer, and Clark (1995) reports that children often think similarly about Santa Claus and God. In addition, in certain other cultures (e.g., Japan), the concept of luck intersects with religious and spiritual beliefs (Daniels, 2003). From this perspective, studying children’s beliefs about luck has the potential to inform our understanding of their developing religious cognition.

Method

Participants

One-hundred-three children participated: 26 4-year-olds (M = 4.47 years; range = 4.06 to 4.95 years), 25 6-year-olds (M = 6.46 years, range = 5.96 to 6.95 years), 27 8-year-olds (M = 8.46 years; range = 8.03 to 8.87 years), and 25 10-year-olds (M = 10.37 years; range = 10.10 to 10.90 years). Sample size was determined by consulting similar exploratory studies of children’s abstract concepts (e.g., Banerjee & Bloom, 2017; Rosengren & Hickling, 1994; Woolley & Dunham, 2017) and was fixed before starting data analysis. Fifty-eight girls and 44 boys participated (the gender of one child was not volunteered) between August 2016 and May 2017. Seventy-four children were White, 4 were Asian, 2 were Black, 10 reported multiple races, and, for 13, race was unreported. Children were recruited from the child participant database at a large public university or from the university’s preschool, and tested in a private room on campus. The research was approved by the University of Texas Institutional Review Board: Children’s conceptions of luck, #2016-06-0005. Although socioeconomic status was not assessed, participants in our database and at the preschool are largely the children of well-educated parents (40% of our sample had at least one parent with a graduate or professional degree and an additional 30% had at least one parent with a bachelor’s degree).

Materials

There were three components: (1) an interview, (2) an explanation task, and (3) a parent religiosity questionnaire. The interview asked children about their experiences with, concepts of, and beliefs about luck, both good and bad, and their sense of personal luckiness (see Study1 Interview Questions in the Supplemental Material available online). The explanation task comprised ten vignettes designed to vary in their content, emotional valence, and in the likelihood of the focal event. In the Lucky Object stories (n = 2), a positive outcome occurred when a child was carrying a lucky rock, and a negative outcome occurred when a child forgot her lucky bracelet. In the Good Luck Statement stories (n = 2), a positive outcome resulted after an adult said “good luck,” and a negative outcome resulted after an adult neglected to say it. The remaining vignettes (Likelihood stories; n = 6) contrasted likely and unlikely events. For both levels of likelihood (likely, unlikely), we designed a story with neutral, positive, or negative content. For all stories, children were asked to indicate how much they thought luck brought about the outcome and given three possible answers: “not at all,” “a little,” and “a lot.” (See Supplemental Material: Study 1 Explanation Vignettes for the stories.)

To explore the role of religious upbringing, parents completed a religiosity questionnaire (see Supplemental Material: Study 1 Religiosity Questionnaire). The questionnaire was developed by the senior author and has been used in many of our studies of children’s religious concepts (e.g., Woolley & Dunham, 2017; Woolley & Phelps, 2001). Others have also used these questions to predict key measures (see e.g., Lane, 2020; Lane, et al., 2016).

Procedure

Children were tested individually. Verbal consent was obtained from 4- and 6-year-olds; 8- and 10-year-olds additionally gave written consent. Half of the children received the interview first and half received the explanation task first. Interview questions were asked in a standard order (see Appendix A), beginning with asking children if they had heard of luck, which we intended as a conversation starter. To familiarize children with the format of the explanation task, children were first presented with three versions of a short vignette about the causes of a child getting sick. These practice stories were designed to solicit each of the three possible responses used in the main task: (1) not at all (wearing a blue shirt causing the child to get sick), (2) a little (other kids nearby causing the child to get sick), and (3) a lot (being coughed and sneezed on causing the child to get sick). Children were then reminded that they could pick any of these responses and were presented with each of the ten test stories in a random order.

Coding of the interview.

For all yes/no or other forced choice questions, responses to the two options were tallied. Responses to questions 1b (“Where have you heard about luck?”), 1c (“Have you ever read any books about luck?”), and 1d (“Have you ever seen any TV shows or movies about luck?”) were combined and coded into three categories: media, testimony, and both. Question 2 (“What do you think luck is?”) was coded into four categories: positive event (e.g., “when you get 5 dollars”), life circumstance (e.g., “if you have a good family”), causal force (e.g., “brings you good stuff”), and chance (e.g., “it is random, things can happen, luck is if it goes good or bad”). Children’s responses to Question 3a “Tell me about a time when you had good luck”) were coded into 3 categories: positive event, accomplishment (e.g., “won a spelling bee”), and life circumstance. Responses to Questions 4 (“Do you have something that gives you good luck?”) and 5 (“Is there anything you do that brings you good luck?”) were tallied and coded as personal or cultural. Two coders coded 100% of the sample. Cohen’s kappa indicated almost perfect agreement across all questions: κ = .90 – .99.

Scoring of religiosity questionnaire.

Affirmative responses were scored as 1 and negative responses as 0. For the questions about religious involvement, 1 was given for answers indicating high involvement, .5 for moderate involvement, and 0 for little to no involvement. For religious activities, a score of 1 was given when more than 3 activities were indicated, a score of .5 for 1–2 activities, and 0 for no activities. Scores were totaled for a composite score of family religiosity.

Results and Discussion

Interview

Although our first question, whether children had heard about luck was intended as a conversation starter, we first note their responses to it. Children’s reported familiarity with the concept of luck increased significantly with age (Table 1). Whereas approximately half of 4-year-olds claimed to have heard about luck, the majority of 6-year-olds claimed that they had, and all but a few of the oldest children did as well (see statistics in Table 1). Follow-up Fisher’s exact tests indicated that the greatest differences were between 4- and 8-year-olds (4 vs 8: χ²(1) = 16.11, p < .001; 4 vs 10: χ²(1) = 12.4, p < .001) and between 6- and 8-year-olds (6 vs 8: χ²(1) = 7.33, p = .009; the difference between 6- and 10-year-olds was also significant 6 vs 10: χ²(1) = 4.37, p = .044).

Table 1.

Children’s responses on the luck interview by age.

Percentage of “yes” responses
Item	4-year-olds	6-year-olds	8-year-olds	10-year-olds	χ2(3, N = 103)
Heard of luck	54%	76%	100%	96%	23.48, p < .001
Experienced good luck	69%	88%	93%	81%	5.83, p = .12
Experienced bad luck	38%	60%	78%	81%	12.99, p = .005
Lucky object	35%	44%	22%	31%	2.83, p = .42
Lucky behavior	31%	42%	19%	44%	4.77, p = .19
Are you lucky?	64%	68%	56%	48%	10.91, p = .091
Is everyone lucky?	26%	70%	67%	75%	17.93, p < .001
Belief in luck	72%	88%	100%	60%	14.86, p = .002

Despite only about half of the 4-year-olds responding that they had heard of luck, many went on to provide sensible responses to later questions. Thus, we included all 4-year-old data in our analyses. We explored where children heard about luck by asking an open-ended question and then following it with specific queries regarding books and TV. Children reported hearing about luck primarily through books and TV (46% of answers), however there were also frequent references to the testimony of other people (33%). A number of children claimed that they had heard about it from both sources (21%)

To explore whether children conceive of luck as a causal force, we asked them what they thought luck was. Responses fell into 4 categories: (1) Positive event, (2) Life circumstance, (3) Force, and (4) Chance/randomness. Of the 26 4-year-olds, only 7 produced codable responses and, of those, 6 (86%) referred to luck as a positive event (e.g., “when you have fun”). Many 6-year-olds’ responses also referred to luck as a positive event (69%) but responses were more nuanced, with some children recognizing the unexpected nature of lucky events (e.g., “when you get something you want but didn’t know you’d get it”). Other 6-year-olds referred to luck as a force, supernatural or otherwise (e.g., “brings you good stuff”). Positive event responses were also most common among 8-year-olds (67%) but children also referred to chance as well as supernatural or other forces. The modal response among 10-year-olds was chance or randomness (50%; e.g., “it is random, things can happen, luck is if it goes good or bad”). Ten-year-olds also referred to positive events, albeit in a more nuanced fashion than did the younger children, often noting the event’s unexpected nature (e.g., “a ton of events that are unexpected that make you happy”), and to luck as a force. Questioning of luck’s reality appeared in these older children, with one child responding that luck is “something people make up to help others turn to something to make things not their fault” and another that “it isn’t a thing.” In sum, with age, there appeared to be increasing understanding of the role of expectations and chance in attributing an event to luck.

As shown in Table 1, most children in each age group reported that they themselves had experienced good luck at some point in their lives. (More 4-year-olds responded that they had experienced good luck than that they had heard of luck. It is possible that asking about “good luck” cued thoughts that simply asking about “luck” did not.) The examples children gave reinforced the patterns above. Of the 4-year-olds who responded affirmatively, 100% referred to positive events (e.g., “making friends, getting ice cream, playing with mommy”). Six-year-olds also referred to positive events (100%) but responses were more nuanced, sometimes citing a specific accomplishment (25%; e.g., “won a soccer game without practice). Increased reference to accomplishments held for 8-year-olds (29%), and increased among 10-year-olds (40%).

Children’s claims that they had experienced bad luck were less frequent than reports of experiencing good luck, and increased with age (Table 1). Four-year-olds were significantly less likely than both 8- (χ²(1) = 8.43, p = .004) and 10-year-olds (χ²(1) = 9.67, p = .002) to claim that they had experienced bad luck. The source of this difference between reports of good versus bad luck is unknown, but it seems possible that parents talk to their children more about good than about bad luck. This possibility is suggested by Rosengren and Hickling’s (1994) finding that parents promote positive generic fantasy figures (e.g., fairies) more than negative ones (e.g., witches), and is explored in Study 2.

To further probe children’s sense of luck as a causal force, we also queried them about ways in which they attempted to influence luck. First, we asked whether they had something that brought them good luck. Approximately one third of children claimed to have a lucky object; this was constant across age groups (Table 1). These were coded into personal objects (58%), such as “a white necklace,” and cultural objects (42%), such as a lucky penny or tiger’s eye rock. Cultural lucky charms first appeared in 6-year-olds. Second, we asked if they engaged in any luck-specific behaviors. Responses were similar, with approximately one third overall engaging in some sort of behavior, and this percentage remaining constant across age groups. Similarly to the objects question, approximately half (48%) of the responses were personal rituals (e.g., “take socks off for a test”). Unlike with objects, there were very few cultural rituals, but a new category emerged that reflected engaging in prosocial behaviors, or more generally, just being good; one child responded “do good in school and do your chores,” and another offered simply, “being nice.” These examples fit more with the standard definition of karma, in which outcomes result from past actions, rather than with luck, in which outcomes are conceived of either the result of a supernatural force or are simply random. One interpretation of these responses is that young children’s concept of luck might overlap significantly with their concept of karma and differentiate with development. Future research should explore relations between luck and karma beliefs.

When we asked children whether they themselves were lucky, the majority claimed they were, reflecting a strong concept of luck as a personal attribute. Only in the older groups, the 8- (31%) and 10-year-olds (30%), did children respond that they were lucky “sometimes,” perhaps reflecting an increasing understanding of luck as chance (only 8% of the 4- and 4% of the 6-year-olds responded “sometimes”, χ²(2) = 10.40, p = .006). Regarding whether everyone is lucky, the majority of 4-year-olds (74%) claimed that only some people were lucky, whereas the majority of children in the older groups (70%) replied that everyone was lucky, χ²(3) = 14.62, p < .002. Darke and Freedman (1997) suggest that with a more advanced conception of luck, in which it is equivalent to chance, one would allow that one person is just as likely to be lucky as another, whereas the holder of a less advanced view would be likely to think that luck is a personal and stable attribute. Our developmental findings support this proposal.

Finally, we assessed children’s belief in luck. As indicated in Table 1, belief increased between ages 4 and 8, and then decreased significantly between 8 and 10. If children indicated that they believed in luck we also asked how much they believed, with response choices “a tiny bit,” “pretty much,” and “totally.” To obtain an ordered variable, we assigned a “No” response to the initial belief question a 0, “a tiny bit” a 1, “pretty much” a 2, and “totally” a 3. As shown in Figure 1, “totally” and “pretty much” responses both formed inverted U-shaped curve patterns, the former peaking at age 6, and the latter peaking at age 8, χ²(9) = 26.85, p = .001. We appear to have captured a meaningful trajectory of children’s belief in luck from tentative belief at age 4 to full belief at 6, weakening belief at 8, and significant skepticism by age 10.

Figure 1.

Children’s responses to the question, “How much do you believe in luck?”

Explanation task

Our vignettes varied three factors: (1) luck form – physical object vs. verbal wish, (2) likelihood of the outcome, and (3) emotional valence of the outcome (positive, negative, neutral). Regarding luck form, a 2(Luck form: Object, Verbal wish) × 2(Valence) × 4(Age) Analysis of Variance (ANOVA) on children’s ratings of luck’s involvement revealed no effect of luck form or valence, but a main effect of age F(3,99) = 5.60, p = .001, η_p² = .145. Post hoc tests (Tukey HSD) indicated that 4-year-olds (M = 1.47, SD = .47) tended to judge luck as playing a stronger role than did the 8-year-olds (M = 1.11, SD = .65, p = .09) and judged luck as playing a stronger role than did 10-year-olds (M = .92, SD = .55, p = .003). Six-year-olds (M = 1.41, SD = .50) also judged luck as playing a stronger role than did 10-year-olds (p = .011). That children responded similarly to stories in which good luck wishes and lucky objects were involved is suggestive evidence that children really do interpret these statements as conferring luck.

Our second set of analyses explored the roles of likelihood and emotional valence of the outcome. A 4(Age) × 2(Expected, Unexpected) × 3(Valence) ANOVA revealed a main effect of valence with luck being rated as playing a stronger role in both positive (M = 1.08, SD = .88) and negative (M = 1.22. SD = .81) versus neutral (M = .82, SD = .80) outcomes, F(2,192) = 20.28, p < .001, η_p² = .17. There was also a significant effect of likelihood, with children rating luck as more involved in explaining unexpected (M = 1.22, SD = .62) versus expected (M = 1.04, SD = .64) outcomes, F(1,99) = 13.49, p < .001, η_p² = .12. This finding was qualified by an expectation by age interaction, F(3,96) = 3.94, p = .011, η_p² = .11. Follow-up t-tests indicated no effect of outcome in the 4- and 6-year-olds, a trend toward an effect in the 8-year-olds, t(26) = 1.99, p = .058, and a significant effect in the 10-year-olds, t(24) = 4.39, p < .001. Thus, an understanding of the role of likelihood in luck appears to increase with age.

Relations between interview and explanation task

We also explored relations between children’s interview responses and their willingness to grant luck an explanatory role in the vignettes. We first conducted a univariate ANOVA examining the effect of belief in luck as a dichotomous variable (yes, no) on endorsement of luck in the vignettes (average endorsement (0–2) across the 10 vignettes). Children who claimed to believe in luck (n = 81) endorsed luck as an explanation (M = 1.22, SD = .53) significantly more often than did children who claimed not to believe (n = 20; M = .80, SD = .76), F(1,100) = 2.79, p = .005, η_p² = .08. Because there were so few children who denied belief in luck, we also utilized children’s responses to the follow-up question regarding how much they believed. A univariate ANCOVA with age as covariate revealed a significant effect of belief level on children’s endorsement of luck as an explanation, F(75) = 6.77, p < .001, η_p² = .22. Although it seems intuitive that these would be related, few studies of children’s supernatural thinking have included measures of both explicit belief level and use of these concepts to explain events. In fact, implicit tasks often reveal magical thinking while participants deny such beliefs in response to explicit queries. For example, Subbotsky (1993) reports that participants who initially denied the viability of a magic box went on to perform behaviors consistent with a belief in the box’s magical powers. These dissociations raise the question of how to characterize the nature of belief (Risen, 2016). The strong positive relation between explicit and implicit beliefs in our sample suggests a high level of depth and coherence to children’s luck beliefs. Future research should explore the developmental trajectory of this coherence.

There were no significant sex or religiosity differences, nor were there any order effects. We discuss the lack of an effect of religiosity in the General Discussion.

Study 2

Learning about unobservable constructs like luck presents unique challenges, because children cannot experience these constructs first-hand (Harris & Koenig, 2006). As such, adult testimony might be particularly important in these cases (Canfield & Ganea, 2014). Parents often encourage children to believe in supernatural concepts including magic and Santa Claus (Rosengren & Hickling, 1994; Goldstein & Woolley, 2016). Rosengren, Johnson, and Harris (2000) suggest that interactions with parents and others “exert more immediate influence than broader forms of cultural support” (p. 82) on children’s concepts of magic, and this may be the case with luck as well.

We used the CHILDES database (MacWhinney & Snow, 1990), to explore adult references to luck in their conversations with children, utilizing the same three guiding questions as in Study 1. First, we explored the extent to which adults referenced lucky items or behaviors causing outcomes. We expected adults to encourage belief in lucky items and behaviors, mirroring their encouragement of fantastical beings. Second, we explored the roles of age, emotion, and likelihood on adults’ use of luck to explain events. Because, with age, children in Study 1 increasingly used luck to explain unexpected events, we anticipated that adults might refer to luck when talking about unlikely events, but might do this more with older children. Finally, because children in Study 1 primarily reported that they had experienced good (vs. bad) luck, we expected that adults would primarily refer to good luck; however because we saw increased references to bad luck with age, we suspected we might see a similar pattern in adult speech. Rosengren et al.’s (1990) findings that parents report stronger encouragement of positive fantasy figures also predicts a greater frequency of references to good luck. Finally, we explored the extent to which adults referred to children or others as lucky.

Method

Participants

Our sample consisted of transcripts of conversations between children and adults from the CHILDES database (MacWhinney & Snow, 1990), which is made up of recordings of natural conversations in the home, laboratory, and at school. A total of 53 corpora, or CHILDES datasets from 1962–2000 were searched for conversations that met the following criteria: (1) the transcript was in English, (2) the focal child was between 1 and 7 years of age, (3) the child was typically-developing, and (4) the transcript included an utterance of the words luck, lucky, or unlucky. Utterances that were incomplete, inaudible, or ambiguous (e.g., an utterance transcribed as luck that appeared more likely to be look based on context) and utterances used as a proper noun (e.g., a dog named Lucky) were excluded. Utterances of luck or lucky were also excluded if they were read directly from a storybook, however they were included if they were part of a discussion about a book. Based on these criteria, the final sample consisted of 797 utterances using the target words luck or lucky. There was only one instance of the word unlucky so its use will not be considered. Demographic information for children in some of the corpora is not available; however, most of the children were white and middle to upper class.

Coding

The coding scheme consisted of five dimensions assessing our three guiding questions. First, to assess input regarding luck as a supernatural causal force, we noted (1) whether adults discussed luck with reference to supernatural forces like magic or superstition or to lucky objects or behaviors (see Referent below). Second, to assess how luck is used to explain events, we coded both (2) the likelihood of the event and (3) its emotional valence. Third, to assess whether adults discussed luck as a personal attribute, we coded (4) the referent of the utterance, that is, how often adults used the target words to refer to people versus objects, and (5) whether the utterance was used for social comparison. Examination of the context of the conversation was emphasized during coding to ensure a robust understanding of the circumstances in which the target word was being used. For each code, one research assistant coded 100% of the utterances; a second coded 20%. Substantial agreement was obtained for each dimension (κ = .80 – .95). Disagreements were resolved through discussion.

Supernatural causal force.

We coded whether adults used luck and lucky in a supernatural causal manner. If it was impossible to determine whether the response included supernatural content, the utterance was excluded. A total of 739 utterances were given a supernatural code (κ = .80). To further assess use of luck to refer to the supernatural, we also coded both whether adults referred to luck as a causal force at all and how often they referred to specific objects as being lucky (see Referent).

Likelihood.

To explore whether use of luck and lucky varied with the likelihood of the focal event, we rated the likelihood of each event, predicting that luck would most often be used to describe unlikely outcomes. Mundane usage referred to everyday events, which included events that were a part of the child’s daily routine, such as playing with toys and taking a bath. Unlikely usage referred to experiences that were more unusual or infrequent. These included special trips and other unlikely events (e.g., a car fell into the water and no one got hurt). Utterances that did not fall into these categories were excluded. A total of 655 utterances were given a likelihood designation (κ = .87).

Emotion.

We also explored whether luck was used more often to explain positive or negative events. Positive utterances using luck included things like saying “we finally had some luck”. Negative utterances consisted of having “no luck” or something bringing bad luck, such as spilling salt. Sarcastic utterances were also included as Negative.. These included wishing someone luck before a challenging or impossible event (e.g., Mother: “Yeah, well good luck” to father trying to get a child to calm down from a tantrum). Lucky was primarily always used positively, however, if it was used sarcastically, we coded it as negative. A total of 116 utterances were given an emotion code (κ = .92).

Referent.

Each utterance was coded for the referent of the target word—what was lucky or had luck. This allowed us to examine how often luck was referred to as a personal attribute. Utterances were coded into six categories: Target child, Animal, Other person, Event, or Item, or Force. Target child was used when an adult said that a child was lucky (e.g., “Lucky boy, having a toy truck”) or referred to multiple people (e.g., the family) as lucky. Animal was used when the family’s pet was considered lucky for some reason, like getting milk. The Other person was used when luckiness was attributed to an individual other than the child (e.g., “Mary [the researcher] is so lucky you shared with her”). Event was used for lucky events, such as saying that one is having a lucky day. Item was used for lucky objects, such as lucky clothing to be worn during a soccer match. Finally, all uses of “luck” (e.g., something bringing good luck), with the exception of “good luck!” wishes, were coded as Force. (k = .95)

Social Comparison.

Using luck for social comparison, implying that some people are luckier than others, also implies that luck is a personal attribute. We coded whether the target words were used to make a social comparison. For example, one mother remarked that, “..a whole lotta people have been sick through the winter, so we’re lucky.” If it was impossible to designate whether a social comparison was being used (e.g., “You’re a lucky girl having that”, referring to something out of view of the camera with no other context) the utterance was excluded. A total of 739 utterances were given a social comparison code, (κ = .92).

Results and Discussion

Luck as supernatural causal force.

The majority of utterances (86%) contained the word lucky. Reference to luck as supernatural, whether in an object or a person, was rare (9% of total utterances). Adults also rarely referred to luck as a causal force (8% of Referent codes) or encouraged the use of lucky objects (8%). This was somewhat surprising given the number of children in Study 1 who claimed to have lucky objects or engage in lucky behaviors. Thus, it seems unlikely that ideas that children have about luck as a supernatural or otherwise causal force that they can control through actions or objects develop primarily through adults’ use of the words luck and lucky.

Luck as explanation.

Contrary to our expectations, parents used luck and lucky to refer to mundane events (67%) more often than unlikely events (33%). From this, it appears that parent speech may not be at the root of children’s differential use of luck to explain unexpected or unlikely events. They also used the target words to refer to positive (47%) and negative (53%) emotional events equally, which was consistent with children’s endorsement of luck as an explanation for both good and bad outcomes in Study 1. It is, however, somewhat at odds with the finding that children reported that they had more often experienced good versus bad luck.

Luck as personal attribute.

Adults used the target words in reference to the child roughly half of the time (49%). A person other than the child was referred to 16% of the time, and animals 3%. Thus, adults’ utterances most often referred to luck as a personal attribute. Parents also sometimes used lucky to engage in social comparison (25% of utterances), for example, to explain why their child experienced a positive event whereas another child had not. Repeatedly being referred to as lucky across a range of events could potentially result in a sense of luckiness as an enduring personal attribute. Children’s claims in Study 1 that they were lucky could result from this sort of dialogue.

To examine whether parents used the target words differently depending on the age of the child, we separated utterances into toddler (ages 1– 2, n = 420 utterances), preschool (ages 3–4, n = 247 utterances), and school-age (ages 5–7, n = 33 utterances) groups. Supernatural usage differed by age group χ²(2, n = 755) = 13.01, p = .001. The frequency of supernatural use with toddlers was less than that with both preschool, χ²(1, n =715) = 7.00, p = .008, and school age children, χ²(1, n = 481) = 10.85, p = .001. Thus, it seems possible that adults might consider the topic of supernatural forces more appropriate as children get older. Emotion references also differed by age group, χ²(2, n = 127) = 12.90, p = .002. Luck references with negative emotionality were used more often with school age children than with both preschool-age children, χ²(1, n = 56) = 8.09, p = .004, and toddlers χ² (1, n = 65) = 6.48, p < .011. There were no age differences for Likelihood or Social Comparison.

Study 3

Our analysis of adult speech indicated that the majority of their use of luck and lucky did not address the causal properties of lucky objects and behaviors, nor were these words used to explain events. Instead, they were primarily used to refer to personal luckiness and appeared synonymous with fortunate. It is not clear how this sort of input could lead children to have a rich concept of luck or to aver strong belief in it. Although, with age, children in Study 1 also increasingly adopted the perspective that luck is a more appropriate explanation for unexpected than expected events, there was no parallel pattern in adult speech. Children in Study 1 were more likely to report that they had experienced good than bad luck. We proposed that this might be because parents talk to their children more about the former. This was not the case; adults talked about good and bad luck equally. However, children in Study 1 used luck to explain positive and negative outcomes equally; this could potentially be due to the nature of the input.

All in all, although adult speech using luck and lucky contained numerous references to personal luckiness, the patterns that we observed in children’s luck concepts in Study 1 regarding causality and explanation were not reflected in the input. In Study 1, when we asked children where they had heard about luck, references to storybooks and media were common, in addition to references to testimony. Thus we decided to explore the patterns of input to which children are exposed via storybooks about luck.

Children’s storybooks are a rich source of information about a range of concepts, from mental states (e.g., Dyer, Shatz, & Wellman, 2000) to death (e.g., Gutiérrez, Miller, Rosengren, & Schein, 2014). Children can also learn a variety of complex concepts from storybooks, including evolution (Kelemen, et al., 2014) and camouflage (Ganea & DeLoache, 2011). Although Rosengren et al. (2000) suggested that other forms of media would be less formative in children’s developing concepts of magic, the lack of references to luck as a causal force in adults’ utterances inspired us to look to children’s storybooks as a possible source of children’s concepts and beliefs.

Method

We searched Amazon.com for children’s books written in English, using the keywords luck and lucky, yielding approximately 60 books. We either purchased or obtained these books from local libraries and checked out an additional 4 books via the University library. We read each book and omitted books that were not actually about luck (e.g., one was about a dog named Lucky, but didn’t address the concept). Our final sample comprised 43 books. The intended age ranges for the final sample were from 3 to 12, though several books did not specify the upper bound (e.g., “4 and up”).

Coding

Basic Information.

For each book, we recorded the title, author, publication year, and number of words, pages, and pictures. We included the age range for each book either as stated in promotional materials or on the Amazon site, in addition to noting whether each book was a picture or chapter book. One research assistant served as the primary coder; a second research assistant coded the majority of questions from the majority of the books (79%). Cohen’s kappa indicated substantial reliability across the various codes (κ = .78–86). Disagreements were resolved by discussion.

Luck as causal force.

To assess the extent to which luck was presented as a causal force, we recorded whether each book addressed the source or origin of luck, and, with the aim of determining how each book presented the mechanisms underlying luck, tallied the number of times the source of each lucky outcome was an action or behavior (e.g., knocking on wood), an object (e.g., a lucky charm), or a person or special being (e.g., someone considered to be lucky). We also coded whether these were cultural (e.g. , 4-leaf clover) or personal (e.g., a special necklace) (Causal force κ = .82).

Emotional valence.

We coded whether each book’s primary focus was on good luck, bad luck, or both (Emotional valence κ = .86). We also tallied the number of positive and negative references to luck overall.

Reality status.

To assess the book’s stance on the reality of luck, we first coded whether the characters in the book were real, fantastical, or both. Setting was coded the same way. We also noted whether the main character’s beliefs about luck affected his or her behavior, and, if so, how. Here we noted whether the character’s beliefs caused him or her to engage in a particular behavior, to stop engaging in a behavior, or some combination. We also coded whether each book took a stance regarding the reality of luck, whether the main character attempted to challenge the reality status of luck, and, if so, the outcome of this test. Lastly, we coded whether the book presented an alternative explanation for outcomes that appeared to be the result of luck (e.g., psychological processes like confidence and effort) (Reality status κ = .78).

Results and Discussion

Concept of luck.

As shown in Table 2, luck and lucky were used fairly equally. In contrast to the utterances in Study 2, most books addressed the concept and sources of luck and contained multiple references to it, including references to specific objects (e.g., a “bad luck chair”), actions (e.g., wearing a belt a certain way) or special beings (e.g., a leprechaun), or as situated in the characters themselves, (e.g., “..and wherever he went, misfortune seemed to follow.”) The idea of a specific object (e.g., lucky charm)bringing about good luck seemed the most commonly offered mechanism and it was prevalent in this set of books. References to both objects (e.g., lucky charms, both personal and cultural) and actions (e.g., knocking on wood) were common. Consistent with the Study 1 finding that children more often reported having experienced good than bad luck, use of luck and lucky in the books was more often positive than negative. This finding is inconsistent with the equal frequency of children’s explanations of positive and negative outcomes in Study 1 as being due to luck, and the equal distributions of parental references to good and bad luck in the CHILDES study. The role of input in children’s beliefs about good luck versus bad will need to be clarified in future work.

Table 2.

Concept of luck as presented in storybooks.

Term used in book	Luck (53%)	Lucky (47%)	--	--
Luck or lucky refers to concept of luck as a…	Phenomenon (51%)	Colloquialism (31%)	Both (18%)	--
Luck’s valence is presented as…	Positive (68%)	Negative (8%)	Both (24%)	--
Does the book define luck?	Yes (9%)	No (91%)	--	--
Does the book explain how luck works?	Yes (42%)	No (58%)	--	--
Of books that explain how luck works, how is luck said to work?	Object (42%)	Action (21%)	Personal Trait (26%)	Special Being (11%)
Does the book address the source of luck?	Yes (84%)	No (16%)	--	--
Of books that address the source of luck, what is the source said to be?	Object (54%)	Action (31%)	Personal Trait (15%)	--
Of books in which objects bring luck, what type of lucky objects are there?	Personal (50%)	Cultural (25%)	Both (25%)	--

Reality status of luck.

As shown in Table 3, the majority of books portrayed the main character or characters as real and the setting as realistic, and presented luck as a real phenomenon at some point during the book. Very few books explicitly presented luck as being not real, and these books all began with luck being presented as a real phenomenon. Notably, the books consistently provided evidence of the involvement of luck in real-world causal relations in the character’s lives: either through an object or action causing a character’s luck to change a character’s beliefs about luck affecting his or her behavior. Thus, overall it appears that children’s books primarily present luck as a real phenomenon that is a part and parcel of everyday life.

Table 3.

Reality status of luck as presented in storybooks.

Main character is presented as…	Realistic (62%)	Fantastical (24%)	Both (13%)	--
Setting of story book is presented as…	Realistic (82%)	Fantastical (11%)	Both (7%)	--
Luck is presented as…	Real (36%)	Not Real* (6%)	Reality status unresolved (36%)	Reality status not addressed (22%)
Of books that initially present luck as real, is reality status ever explicitly challenged?	Challenge reality status (30%)	Don’t challenge reality status (70%)	--	--
Resolution of books that challenge the reality status of luck	Reaffirm luck’s reality (9%)	Disconfirm luck’s reality (5%)	Reality status is not resolved (86%)	--
Does an object or action cause luck to change in the storybook?	Yes (57%)	No (43%)	--	--
Do beliefs about luck cause luck to change in the storybook?	Yes (62%)	No (38%)	--	--
Does the book present an alternate explanation to luck?	Yes (36%)	No (64%)	--	--
Of books that present an alternate explanation to luck, what is the alternate explanation?	Effort (38%)	Hope (19%)	Other psychological process (19%)	Story-specific real-world process (19%)

The way each book portrayed the reality status of luck was also apparent in whether the book provided alternative explanations for putatively lucky events. Recent work suggests that children increasingly recognize that positive outcomes that appear to result from luck actually reflect real psychological processes. That is, with age, children become more aware that, for example, someone who forgets her lucky charm might feel less confident, or that hard work is the real causes of apparently lucky outcomes (Kelley, Lamoroux, & Woolley, 2019). As shown in Table 3, a number of the books suggested an alternative explanation for what the focal character believed was luck. Thus it seems that these books about luck might potentially play a role in children’s understanding of the real-world mechanisms underlying apparent luck effects.

Intended ages of books.

Although our first study revealed significant developmental differences in children’s beliefs and concepts, the portrayal of luck in the books remained fairly constant across the intended age groups (see Supplemental Materials for results of analyses).

Overall, the storybooks appeared to be a rich potential source of children’s concepts of luck. Most books situated luck in a realistic versus a fantastical context, potentially facilitating children’s concepts of luck as a real-world phenomenon. Explicit references to the sources of luck and the nature of lucky objects were common. Many books presented luck as a real phenomenon and provided evidence of the involvement of luck in real-world causal relations. A subset of the books provided alternative explanations for apparently lucky events, such as the role of effort. Against our expectations, the majority of the luck content was consistent across the intended age groups of the books.

General Discussion

Cognitive developmentalists often paint a picture of children as explanation seekers, constantly striving to find the causes that underlie events (Gopnik, 1998). In addition to a range of sophisticated natural explanations, children’s repertoire includes a set of supernatural ones, including God, karma, magic, and fate (Banerjee & Bloom, 2017; Rosengren, et al., 1997; Woolley & Cornelius, 2017). Yet we know little about these supernatural explanations—what they are like and where they come from. The goal of our research was to explore the origins and development of one of these supernatural explanations that is particularly common in adult discourse and behavior—luck.

Concepts of luck

The youngest children in our sample appeared to think of luck simply as an event with a positive outcome; with age children increasingly articulated more nuanced conceptions, including referring to the unexpected nature of lucky events, noting a force of some sort, and making comparisons to chance. This increasingly nuanced concept of luck was also reflected in children’s reports of their experiences with luck, with older children including references to lucky events as reflecting specific accomplishments (e.g., passing an exam without studying). References to experiencing bad luck also increased with age. Thus it appears that children’s concept of a lucky outcome in its early form might be synonymous with being “fortunate”—a good thing that happens in the natural course of things. As children get older, they come to conceive of lucky events as positive, randomly determined outcomes.

With a mature concept of luck in which it is equivalent to chance, one person is just as likely to be considered lucky as another. Hence, the more immature view would be to conceive of luck as a personal and stable attribute possessed by some people but not others (Darke & Freedman, 1997). When asked who is lucky, the youngest children in our study seemed to think that only certain people were lucky, indicating support for Darke and Freedman’s proposal (although whether they view luck as stable should be probed in future research). In contrast, the majority of older children claimed that everyone was lucky. This could reflect older children’s increased understanding that everyone has an equivalent probability of experiencing “lucky” (i.e., chance) outcomes.

Belief in luck

Our interview revealed a significant increase in children’s levels of belief in luck between the ages of 4 and 8, and a decrease between 8 and 10. Thus, the early elementary school years appear to be peak years for belief. Not only was belief lower in 10-year-olds, but younger children (4- and 6-year-olds) also attributed a stronger causal role to luck in the vignettes than did older children (8- and 10-year-olds). With the vignettes, we investigated two facets of two facets of children’s endorsement of luck as an explanation—emotional valence and likelihood. Children of all ages more often attributed luck a causal role in outcomes when the events were emotionally charged, either positive or negative. An awareness of the role of likelihood increased with age, with older children more often granting luck a causal role in outcomes that were unlikely.

We did not find a relation between children’s beliefs in luck and their family religiosity. We can suggest a number of possible explanations. One is that perhaps these are only related in certain religions or certain denominations, particularly those that more strongly emphasize that outcomes are predetermined by God. It is also possible that our measure failed to capture religiosity differences that might be predictive. Additionally, these relations may also be stronger in other cultures. For example, in Japanese culture, the concept of luck is intertwined with spiritual beliefs (Daniels, 2003). Another possible explanation is that, in other studies that have found relations between magical and religious beliefs in children, there exists a more direct counterpart to the magical belief—prayer for wishing (e.g., Woolley & Phelps, 2001) and God for Santa Claus (e.g., Clark, 1995). Perhaps the clearest counterpart to luck in religion might be the use of designated objects like rosary beads. It might be interesting to explore whether children, when first introduced to these, think of them as lucky charms.

Origins of children’s concepts

Our second and third studies aimed to explore potential origins of luck concepts and beliefs. Unlike many natural causal explanations, supernatural explanations are not taught in school and, we reasoned, were unlikely to be tutored in the home. Yet, because many children referred to testimony as a source of their concepts, and because parents encourage other supernatural concepts like magic (Rosengren et al., 1994) and Santa Claus (Prentice & Gordon 1986), we considered it important to explore this input.

Conversations.

Analysis of adult use of the words luck and lucky indicated that the great majority of these utterances referred to how fortunate children were, as opposed to referencing unexpected outcomes or special objects. The majority of children in Study 1 considered themselves lucky; perhaps this resulted from hearing this sort of reference over the course of several years. It seems important to know more about how children interpret these references, specifically whether they interpret them as simply indications that they are fortunate versus tapping a more supernatural “luckiness” quality. One possibility is that this changes with development and exposure to media. Supporting this, the majority of uses of lucky in conversations were with reference to the child or other people (73%), with fewer references to lucky events or objects. Based on this sort of input alone, children might initially conclude that lucky refers simply to experiencing good outcomes. In the storybooks, we observed the reverse: lucky was used primarily to refer to objects and behaviors (85%) as causing positive outcomes, with references to lucky people less common. Perhaps, with continued exposure to the idea of luck as a supernatural causal force that works through actions and objects, children to come to view themselves as lucky in a more supernatural (vs. fortunate) sense.

The paucity of references to lucky objects and rituals in the conversations was surprising given that approximately a third of children of all ages in Study 1 claimed to possess lucky charms. Although it does seem that children may get ideas about lucky objects and actions through the books they read, it is important to obtain a better sense of how these lucky objects emerge. One possibility is that they have ties to existing transitional objects. It might also be that children who are involved in sports are more likely to have them, and that various subcultures and social forces within sports promote them. Relatedly, children’s conversations with their peers might be a fruitful source of information about luck and lucky objects. It may also be that the recordings we studied neglected to capture those instances in which parents promote having a lucky charm (e.g., perhaps when a child finds a 4-leaf clover). Finally, it would also be interesting to explore whether and how the prevalence and nature of lucky objects changes over time.

Because only older children in Study 1 seemed to reserve luck explanations for unexpected outcomes, we anticipated that adults might discuss this aspect of luck differently with children of different ages, however there were no age differences in this respect. Adults did use more negative content with older children; this parallels the increasing tendency with age for children in Study 1 to report more experiences with bad luck. Adults also used more supernatural references with older children. These together provide evidence that adults modify their talk about luck as a function of children’s age.

Storybooks.

Because the storybooks focused more on the concept of luck than did children’s conversations, they might prove a more fruitful realm for continued exploration of the origins of children’s concept of and belief in luck. Characters and settings were mostly realistic—luck was not portrayed as part of a fantasy or magical world but as something that operated alongside everyday people and events. As Woolley and VanReet (2006) demonstrated, children are more likely to believe in the real existence of novel entities that are associated with scientific or everyday entities versus those that are embedded in fantasy. Hearing about luck in these realistic storybook contexts could potentially facilitate children’s belief in luck as a real supernatural force.

Not only was luck often portrayed as a real phenomenon in the books, there was also often considerable intertwining of luck with both psychological (e.g., the role of effort) and physical (e.g., character’s actions) causality. Discussion of lucky objects was common. From this, it seems plausible that children’s storybooks, in contrast to their conversations with adults, may be a key influence on their beliefs in luck, their possession of lucky objects, and their use of luck as an explanation. Our suggestions are clearly limited by the small number of books coded (43), but the results are provocative enough to suggest that further research with a larger number of books might be fruitful.

Additional suggestions for future research

Although we have begun to document children’s conceptions of luck, there is much work to do. Future work could investigate the role of children’s understanding of improbability in their luck beliefs—a mature understanding of luck may be predicated on this understanding (Shtulman & Carey, 2007). Relatedly, a mature understanding of luck (vs. thinking it synonymous with “fortunate”) involves acknowledging the interplay between actions and outcomes; people are considered lucky when things turn out well despite their inaction, and unlucky when things turn out poorly despite appropriate action. Thus another research direction might be to address children’s understanding of the role of intention in apparently lucky outcomes. Future work should continue to explore the role of testimony in children’s luck concepts. One promising direction would be to explore adult talk about luck in families representing a wider range of ethnicities and cultures than we included here, as luck may play a greater role in some non-Western cultures. It also seems likely that adults understand colloquial use of luck by other adults as distinct from the supernatural force in books and fiction, but we do not know how children interpret these references. Thus it would be fruitful to explore whether children interpret adults’ colloquial use as “evidence” of the force depicted in books.

Conclusions

Our findings revealed that children have a rich concept of luck, as something that causes events to happen in their lives, as an explanatory tool, and as something they can possess, in part by engaging in various behaviors and through the possession of certain objects. This concept appears to change significantly between the ages of 4 and 10. Our analysis of adult use of the words “luck” and “lucky” suggests that children’s first introduction to the concept of luck might be when adults mark positive outcomes by commenting that the child is lucky. This might result in children viewing themselves as lucky, without necessarily having a concept of luck as anything more than something they possess that is related to good things happening. Perhaps then, through encountering luck in storybooks and other media, they come to consider the possibility that they can play a role in obtaining luck, through rituals and lucky objects, which might result in the increase in beliefs in the reality of luck that we observed between 4 and 8. Our findings broaden our understanding of the range of supernatural explanations children marshal alongside their repertoire of natural ones. The question of how children explain events in their world, via both natural and supernatural means, is central to developmental psychology. As best captured by Nicholas Rescher (1995): “There are, in general, three routes to realizing the good things of life.. we can achieve them by effort and hard work, by coming into them through good fortune (by accident of birth or inheritance), or by obtaining them through sheer luck” (p. 36). The task of differentiating these appears to be a life-long endeavor.