Children’s Recognition of Dangerous Household Products: Child Development and Poisoning Risk

Abstract

Objective Preliterate children may be poisoned because they fail to distinguish safe versus hazardous household products. Methods Study 1: A total of 228 children aged 18–54 months completed four tasks assessing ability to recognize product safety. Study 2: A total of 68 children aged 17–31 months chose products to drink from pairs of dangerous versus beverage bottles. Study 3: A total of 119 children aged 18–42 months sorted 12 objects into toys, things you can drink, and things that are bad/dangerous. Results Left alone, children frequently touched dangerous household products. Children frequently misidentified poisonous products as safe. Some developmental trends emerged. The following packaging features apparently helped children recognize danger: black bottle color; opaque packaging; salient symbols like insects; lack of pointy spouts; squared, not round, bottles; and metal, not plastic, containers. Conclusions Developing cognition helps preliterate children distinguish safe from dangerous household products. Multiple aspects of product packaging may reduce child poisoning risk if implemented by industry or policy.

In the United States, poisoning is the fifth leading cause of unintentional child death (National Center for Injury Prevention and Control [NCIPC], 2014) and leads to >50,000 annual hospital visits among children <5 years of age, costing >$26 million in medical and work-loss costs (NCIPC, 2014). The most recent data from the Toxic Exposure Surveillance System tally >1.1 million annual inquiries to poison control centers in the United States about potential poisonings to children <5 years of age, an inquiry rate far higher than that of any other age cohort (Bronstein, Spyker, Cantilena, Rumack, & Dart, 2012; Gutierrez, Negrón, & García-Fragoso, 2011). Of those, about one third of inquiries concerned exposure to common nonmedication household products, such as cosmetics, personal care products, cleaning substances, topical preparations, and arts, crafts, or office supplies (Bronstein et al., 2012).

A wide range of individual, family, and environment factors intersects to cause particular pediatric poisoning incidents. Prominent family factors relevant to young children’s risk of injuries such as poisoning include (a) the quality and quantity of adult supervision of young children (Morrongiello, 2005), (b) the quality and quantity of safeguarding the homes so that young children are unable to access dangerous household products (Morrongiello, Ondejko, & Littlejohn, 2004), and (c) adult preparedness to react in case of possible injury (e.g., quick access to poison control center phone number; Lovejoy, Robertson, & Woolf, 1994). Environmental factors include those in the home—how many poisonous substances are present and how are they stored—as well as how products are manufactured. Broader policy-based environmental factors are also relevant; among the most influential factors to reduce child poisoning rates in the United States over the past several decades was the Poison Packaging Prevention Act of 1970, which mandated child-resistant caps on a number of dangerous household products.

Although changes to parenting practices, including both supervision and storage of poisons, plus policy changes such as the Poison Packaging Prevention Act of 1970, can and have reduced pediatric poisoning rates significantly, children inevitably will encounter potentially dangerous products when unsupervised and therefore will have the opportunity to ingest those products if they deem them safe and desirable. Thus, another contributing factor in child poisoning risk, and the present focus, is the role of the child him- or herself. Studies examining the role of children in poisoning risk are sparse, but the broader child injury literature suggests individual differences in temperament, cognitive development, and risk-taking tendencies likely contribute to poisoning risk (for reviews, see Schwebel & Barton, 2006; Schwebel & Gaines, 2007). Also likely to be relevant is preliterate children’s understanding of what products are safe versus dangerous. This understanding, and the behavioral patterns that emerge based on young children’s perception of risk, develops quickly during the toddler and preschool years and forms the present topic of study.

Several aspects of cognitive development are relevant to development of the ability to distinguish dangerous household products from safe ones. At a theoretical level, these processes might be considered from the perspectives of Gibsonian perception of affordances (e.g., does an object afford drinkability?), Piagetian development of cognition through a child’s active engagement and experimentation in the world, and neo-Piagetian concepts of working memory and information processing development. At the more specific level of particular cognitive skills relevant to distinguishing safe versus dangerous household products as children develop, two skills seem particularly relevant. First, the ability to categorize objects develops rapidly during the infancy, toddler, and preschool years. Influenced by a combination of perceptual, conceptual, and linguistic cues (Madole & Oakes, 1999; Sloutsky, 2003; Sloutsky & Fisher, 2004), children begin to classify unfamiliar objects into schema, forming prototypes that can be identified and consistently labeled into categories by the preschool years (Bjorklund, 2012; Gelman & Coley, 1990; Gelman & Markman, 1986). This learning occurs especially via observation and modeling of others (Phillips, Seston, & Kelemen, 2012).

Young children rely on multiple aspects of objects to accomplish categorization, and what they rely on likely develops with age. Toddlers categorize objects based mostly on sensory and perceptual information, but preschoolers are increasingly able to group perceptually different objects into categories based on deeper underlying commonalities (Gelman & Coley, 1990; Gelman & Markman, 1986). Shape is used prominently, especially among young children (Smith, 2003), but children also consider colors, textures, detailed parts, and other aspects of objects when determining the category unfamiliar objects might fit within (Jones & Smith, 2002; Macario, 1991; Pereira & Smith, 2009). Nonshape characteristics such as color or texture are particularly relevant to children when they recognize the predictive power of that characteristic, such as the use of color to categorize food products (Jones & Smith, 2002; Macario, 1991). When examining how children categorize potentially poisonous packages, therefore, children are likely to consider a product’s shape, size, labeling, and color, among other characteristics, to decide whether it is a product they can eat/drink or they should avoid. Children likely form prototypes of safe versus consumable products based on a combination of personal experience and especially modeling of how others, such as parents and other trusted adults, use household products.

A second cognitive skill that develops rapidly during the toddler and preschool years and may be relevant to children’s ability to categorize and distinguish safe from dangerous household products is the ability to recognize and interpret labels and symbols on bottles (DeLoache, 2002). Early research on poisoning risk discovered that children have a poor understanding of vague nonrepresentative symbols indicating poison such as Mr. Yuk or a skull and crossbones (Braden, 1979; Fergusson, Horwood, Beautrais, & Shannon, 1982; Vernberg, Culver-Dickinson, & Spyker, 1984). It may be, however, that children can use symbol recognition to identify safe products when those symbols are familiar cues (e.g., juice bottles labeled with pictures of apples or grapes) as well as to identify potentially dangerous products (e.g., insecticide bottles labeled with pictures of unappealing insects). This hypothesis is untested in the published literature.

Given the frequency of poisoning incidents among young children and the concomitant development of several cognitive skills that may help preliterate children categorize household products into safe and dangerous objects, this study was designed to study the accuracy with which children at different ages identify safe versus dangerous household products, and whether aspects of those products’ packaging influences children’s accuracy of judgment. We are aware of just one published study with a similar goal; in that study, a between-subjects design was used to examine how children reacted to different packages and labels. Results were inconclusive (Schneider, 1977).

In this three-experiment study, children aged 17–54 months were presented with safe and dangerous household products in the context of various laboratory- and school-based tasks. We posited three hypotheses. First, we expected children with still-developing cognitive skills would generally perform poorly at distinguishing safe versus unsafe household products. Second, we hypothesized improvement across development in a cross-sectional sample, such that older children would correctly classify products as safe or dangerous more often than younger children. Third, because children in the target age range are developing skills in categorization and symbol recognition, we hypothesized that correct classification by children would depend on multiple factors about the product, including transparency/opaqueness, bottle color, bottle shape, bottle labeling, and bottle material. We expected that hazardous products packaged in a manner similar to beverages familiar to children (e.g., juice) would be more often misidentified by children as dangerous than products packaged in opaque, black, square, and/or metal containers because they would incorrectly fit the categorization prototype of consumable products. We also expected familiar symbols such as pictures of fruit or insects, when present on packages, might help children classify them correctly. Table I provides an overview of the tasks and products used in the three studies.

Study 1

Study 1 was designed to gather initial information addressing all three hypotheses. We examined how well children were able to distinguish safe versus dangerous household products, and whether older children performed better than younger ones. We did so using four different tasks that implemented a variety of assessment strategies—touching of objects when left in a room with them, sorting and identification of products, and selection of a product to drink—to gather data on young children’s ability to recognize dangerous products. We also considered children’s performance on the tasks and whether particular qualities of products made it easier for children to recognize them correctly as safe or as dangerous.

Methods

Participants

A sample of 228 children was recruited from community sources in the Birmingham, Alabama area. Two other children were recruited but did not understand the protocol or were uncooperative and were therefore excluded. The participating sample was a mean age of 36.97 months old (SD = 11.04, range = 18–54 months) and was 47% male, 48% White, and 42% Black. Parents had a median education level of “some college,” and an annual household income between $40,000 and $59,999.

General Protocol

Families arrived to the laboratory for a single 45-min session. Parents provided informed consent and children verbal assent. All children were tested individually and without parents present, although in rare cases of child distress (<10% of sample), the parent was permitted in the experimental room to quietly observe from behind the child (preventing nonverbal cues). The protocol was approved by the university institutional review board (IRB), including replacement of dangerous substances with safe ones of comparable color and viscosity. During the session, children completed four tasks and parents responded to a brief demographics questionnaire. The tasks occurred in a consistent order across all participants: “play,” “sorting,” “identification,” and “choices.” Each is described below, along with the outcome measures derived from them. Following completion of the four tasks, families were thanked, compensated for their time, and debriefed about basic child poisoning prevention strategies.

Demographics

Parents completed a brief form concerning parental education, household income, and children’s race/ethnicity, age, and reading ability. Reading ability was assessed on a 6-point scale (no reading; reads a few words such as his/her name, stop, McDonald’s; reads well enough to read very basic books with cues; reads well enough to read “easy reader” books; reads well enough to read most children’s books; reads well enough to read most everything).

“Play” Task

Adopted from a similar task used by Schneider (1977), a research assistant brought children into a large mostly empty room containing three toys (teddy bear, small toy frog, and book), three bottles in their original packaging that previously contained poisons but had been cleaned for safety (hydrogen peroxide, bleach, and torch fuel), three beverage bottles (apple juice, white grape juice, and lemon-lime soda), and one laboratory-designed bottle placed around the perimeter of a rug on the floor in a previously determined random order that varied across participants. The laboratory-designed bottle, called the “alternative bottle,” was a black plastic bottle with a white cap. The bottle was rectangular in shape, with rounded corners, and was labeled with gray labels and black text as a hazardous household product. The hydrogen peroxide bottle was 32 fluid ounces, rounded, and brown in color. The bleach bottle was a gallon-sized, white opaque bottle with a handle. It was rounded with blue and red labels. The torch fuel bottle was a rectangular 64-fluid-ounce transparent plastic bottle, with a narrower “easy grip” on the back side. It had colorful labeling. The apple juice was a 64-fluid-ounce transparent plastic bottle with a narrower “easy grip.” It had colorful labels that included images of red apples. The white grape juice bottle was 64 fluid ounces and made of transparent plastic with an “easy grip” and a colorful label that included images of green grapes. The lemon-lime soda bottle was a 2-L bright green transparent plastic bottle with colorful green and blue labels. The toys were a soft brown teddy bear measuring 30.5 cm high when placed in an upright seated position, a plastic toy frog that was bright green and sat 5.2 cm tall, and a small 12.5 × 12 cm cloth book with six pages of colorful images and three simple words. Toys were included for three primary reasons: (a) to replicate what had been used previously by Schneider (1977), (b) to provide children with safe objects to play with if they chose (for ethical reasons), and (c) to increase ecological validity of the task, creating a situation that was somewhat more realistic than one with only household products present.

Just after the researcher walked into the room with the child, the researcher exclaimed, “Oops, I forgot something. Can you wait here and play for a few minutes while I go and get it?” The researcher exited the room and shut the door. The researcher then observed the child through a one-way mirror from a neighboring room for 3 min, reentering only in cases of child distress.

Videotapes were coded to record the percentage of time each object was touched by the child, defined as any physical contact between a child and an item in the room. Interrater reliability was established between two independent coders on 10% of the sample and was excellent (average interrater correlation across objects = .98 and minimum correlation = .94). Given substantial skew of the 10 variables (mean skew = 4.72, SD = 2.37, range = 1.63–8.38) and the fact that any contact with a dangerous product might be concerning, analyses were conducted with a dichotomous measure of whether the child touched each object.

“Sorting” Task

Following the 3-min “play” task, the researcher reentered the experimental room, stated, “I found what I needed for us to do our next task,” and placed three large transparent plastic bins in the center of the room. Around the bins, the researcher placed 12 items: The 10 items used in the “play” task, plus two additional laboratory-made bottles. One bottle, called the “black torch fuel,” was the same shape as the original torch fuel bottle and included the original labeling, but it was painted opaque black rather than being transparent. The second bottle, called the “alternative black torch fuel,” was also the same shape as the original torch fuel bottle and was painted opaque black, but had alternative gray and black labeling.

With the 12 items spread in front of the child, the researcher asked the child to sort the items into the three bins. Toys were to be placed in the middle bin, things you can drink into a second bin, and things that are “bad or dangerous” into the third bin. Both order of presentation (drink vs. bad/dangerous) and placement in left or right bins for those categories were randomized across participants. The researcher continued to prompt the child, reminding him or her of the bin assignments and encouraging placement of all items into a bin. For occasional cases when the child was confused by the task, toys were used as examples of how to sort and were then excluded from analyses for those children.

To ensure valid data collection during the sorting task, researchers were trained to follow an explicit written protocol that required them to be mindful of any words, expressions, or body placement that could potentially influence children’s responses. During the task, researchers positioned themselves in front of the middle bin in a seated or kneeling position. If a child asked for help lifting a bottle, the researcher would ask into which bin he or she would like the bottle placed and would follow the child’s direction. If a child simply handed a bottle to the researcher, the researcher would prompt the child by saying, “Is this something you can drink, something that’s bad or dangerous, or is this a toy?” If multiple prompts were required, the researcher would alternate the order of the first two options, such that if “drink” was offered as the first prompt then the next prompt would be, “Is this something bad or dangerous, is this something you can drink, or is this a toy?” Praise was used to maintain children’s motivation and interest but was limited to two phrases: “Good progress” and “good work.” To standardize administration across participants, researchers were instructed to use one of the two approved phrases after each choice/bin placement, regardless of the child’s decision.

“Identification” Task

After sorting was completed, the 12 items used in the sorting task were removed from the bins and placed in the center of the large empty room. Using a randomized order list, the researcher asked the child to identify the following nine items: teddy bear, toy frog, book, hydrogen peroxide, bleach, torch fuel, apple juice, white grape juice, and “Sprite.” Children were permitted to point to or bring each individual item to the researcher, and responses were recorded.

“Choices” Task

In the last task, the experimenter escorted the child to a small adjacent room where they sat together at a child-sized table. The researcher presented 12 pairs of bottles, always including one bottle with a dangerous substance and one beverage, to the child. The following four dangerous bottles were included: torch fuel, toilet cleaner, insecticide, and alternative bottle. The following three beverages were included: white grape juice, apple juice, and lemon-lime soda. Each pairing (one dangerous bottle, one beverage) was presented once, in randomized order, and the child was asked to indicate the bottle in the pair that he or she would rather drink.

Results

Table II lists demographic characteristics of the sample, along with the subsamples of children divided into three age-groups (18–29 months, 30–42 months, 43–54 months). Table III presents results from the “play” task. As expected, more than half the children touched each of the toys. Between 18% and 28% of children touched each of the nontoy items, and there were only small differences between the rate of touching consumable products (juices, lemon-lime soda) compared with the rate of touching dangerous products (torch fuel, hydrogen peroxide, bleach). About one fifth of children touched each of the dangerous product bottles, indicating significant risk of poisoning from those products for unsupervised children. An independent samples analysis of variance (ANOVA) evaluated developmental trends in touching behavior, and two statistically significant differences emerged: A slight increase in the rate of touching the teddy bear as children grew older and a stronger increase in the rate of touching the book.

Table IV presents results from the “sorting” task. Children generally sorted the toys correctly (95%–96% across objects, with age-groups combined), an indication that children understood the sorting task and the sorting data results are valid. Children also generally sorted the safe beverages accurately (81%–86% across products, with age-groups combined). Children were not accurate in their ability to recognize and sort the poisonous product bottles as “bad or dangerous,” with overall rates ranging from 35% to 64%. They performed most poorly on the torch fuel bottle, which only 35% of children recognized as “bad or dangerous,” and best among dangerous bottles on the bleach bottle, which 64% of children recognized as bad or dangerous. Children sorted the alternative torch fuel bottles, which were opaque and black, as dangerous more often than the original transparent torch fuel bottle. One-way ANOVAs were conducted to evaluate developmental trends for sorting products correctly. Statistically significant results emerged for all dangerous product bottles except the torch fuel product, with older children performing better than younger children. There also were significant trends for older children to more accurately sort the book and the white grape juice.

Table V presents results from the “identification” task. Children tended to recognize the toys correctly. About half of the children correctly identified the apple juice and lemon-lime soda bottles, and about a quarter correctly identified the white grape juice bottle. Only 20% correctly identified hydrogen peroxide, 16% bleach, and 10% the torch fuel bottle. One-way ANOVA models were computed to evaluate developmental trends across the age-groups, and statistically significant results emerged for all but one product, with older children performing better than younger ones. The only exception was the torch fuel bottle, which children at all ages struggled to identify correctly.

Table VI presents results from the “choices” task; lower numbers represent greater recognition of danger. Overall, children chose the dangerous product between 20% and 36% of the time. They were most likely to choose the dangerous product when torch fuel or toilet cleaner was presented and less likely with the alternative torch fuel product or the insecticide bottle. No strong trends emerged for selections based on the safe bottle offered. One-way ANOVA models revealed developmental trends for almost all choices task pairings, with older children, especially those in the oldest age-group of 43–54 months old, making safer choices than younger children.

Further analyses tested for differences on all Study 1 tasks based on gender and race; no statistically significant differences emerged, and data are not shown. Parents reported the following reading levels for children in the sample: 120 children (53%) had no reading ability, 92 children (40%) could read just a few words, 8 (4%) children could read very basic books with cues, 1 (0.4%) child could read “easy reader” books, and 1 (0.4%) child could read most children’s books. As expected, children with greater reading skill tended to be older and also tended to perform somewhat more accurately on experimental tasks. More accurate performance was seen especially among better readers in correctly sorting dangerous products and in the choices task.

Discussion

Study 1 illuminates several details about how young children identify safe or dangerous household products. First, as expected based on our first hypothesis, children made many errors in their identification of hazardous products, often categorizing them as safe. Children also touched dangerous products frequently when left alone in a room with them and were unable to sort dangerous products accurately into a bin for “bad or dangerous” items. When presented with a choice of which product they wished to drink, children performed only modestly better than chance on several selections. In fact, among the youngest children (aged 18–29 months), they were more likely to choose to drink the dangerous product than the safe one when presented with torch fuel and white grape juice, and when presented with toilet cleaner and both white grape juice and lemon-lime soda.

Second, as expected in our second hypothesis, older children tended to make safer decisions than younger ones. Age-related differences emerged especially in the sorting task where children sorted dangerous items into a bin, in the identification task where children identified dangerous items when given their names, and in the choices task where children chose which product to drink from a pair of one dangerous and one safe product.

Third, as expected from our third hypothesis, children of all ages seemed to recognize some products as dangerous more often than others. In particular, alternative bottles that were opaque and black seemed to be packaged in a manner that helped children deduce they were dangerous, perhaps because opaque and black bottles are categorized by young children as fitting a “dangerous” schema or prototype. The insecticide bottle also was recognized as dangerous somewhat more often, perhaps partly because it was opaque and black and partly because it included a symbolic picture of a large and rather ugly insect on it. The torch fuel bottle, which was transparent, had a bright label, was packaged in a round container shaped similar to juice containers, and contained a liquid similar to the color of juice, was misidentified as safe by a large portion of children, including older ones. It may have been miscategorized into a “juice” schema or prototype by children. The toilet cleaner bottle also was misidentified frequently. It was brightly colored and included a long pointy spout designed to clean toilets. Anecdotal comments from a few children during and after the study caused us to think children may have miscategorized or misinterpreted the spout as a drinking or pouring mechanism.

Given results from Study 1, Study 2 was designed to investigate further the study’s third hypothesis, that aspects of packaging, labeling, and product coloring might influence young children’s categorization of safe and dangerous household products. We sought to accomplish two goals in Study 2 through a replication of the “choices” task among a new sample of children. First, as an assessment of the role of developing symbol recognition skills, we tested further whether the symbolic nature of the insect on the insecticide bottle helped children identify that bottle as dangerous by seeking to replicate the trend of safer decisions with that particular bottle. Second, we sought to test further whether the long pointy spout of the toilet cleaner product may have caused children to categorize that product as a safe beverage to drink. To test this possibility, we replaced the toilet cleaner product with a similarly sized bright and opaque container that did not include a long pointy spout, a dangerous dishwasher detergent product.

We also adjusted a methodological error from Study 1 in Study 2. In collecting data and talking to parents of children in Study 1, we realized that many young children are taught not to drink carbonated sodas, so we replaced the lemon-lime soda bottle with a milk jug. Because the youngest children in Study 1 were most vulnerable and made the most errors in identifying safe products, we conducted Study 2 with children aged 17–31 months old.

Study 2

Methods

We replicated the choices task from Study 1 using the same methodology. Five of the bottles were the same as used in Study 1: apple juice, white grape juice, torch fuel, alternative torch fuel, and insecticide. Two new bottles were included: a 1-gallon milk jug and a 96-ounce dishwasher detergent bottle that was opaque bright green with green and white labeling and a rounded handle.

A sample of 68 children aged 17–31 months was recruited from eight area preschools (mean age = 24.95 months, SD = 4.22). Eleven other children did not understand the protocol or were uncooperative and were excluded. The included sample was 50% female, 67% White, and 24% Black. The preschools were selected to represent the local area and included religious and nonreligiously affiliated institutions; centers located in urban, suburban, and rural areas; and centers serving children from a range of racial, ethnic, and socioeconomic backgrounds. Children were tested individually with a researcher at quiet locations in their preschools. On rare occasions, a teacher or aide accompanied children to reduce children’s anxiety or distress. Teachers or aides who accompanied children were given explicit instructions to stay quietly behind children and avoid influencing children’s responses. School authorities and parents provided signed informed consent, and children provided verbal assent. All procedures were approved by the local IRB.

Results and Discussion

Table VII presents results of Study 2. Results generally replicated those of the youngest age-group in Study 1. As expected from the first hypothesis, children continued to misidentify dangerous products as safe with great frequency. Largely replicating results from the first study and relevant to the third hypothesis, the torch fuel product was identified as children’s preferred drink between 40% and 50% of the time, depending on the beverage it was paired with. This confirms the possibility that children may have miscategorized that product as a juice. Results from Study 2 also replicated those of Study 1 with respect to the insecticide bottle: Children tended to recognize the insecticide bottle as dangerous at a higher rate than any other bottle with dangerous products inside, and results from Study 2 were stronger in this respect than those in Study 1. This finding suggests the development of symbol recognition, combined with the opaque black nature of the bottle, may have helped children identify the insecticide product as dangerous.

Results from Study 2 indicate that children were able to recognize the dishwasher detergent bottle as dangerous more often than the toilet cleaner bottle in Study 1, and at a rate close to the recognition rate of the insecticide bottle. A possible explanation for this finding is that the pointy spout of the toilet cleaner bottle caused children to categorize the product as drinkable.

Taken together, results from Study 2 generally replicate those of Study 1 and support our hypotheses: Young children are rather poor at distinguishing safe from dangerous household products. Opaque containers appear to be identified as dangerous more accurately than transparent ones with juice-colored dangerous liquids inside, perhaps because early categorization skills are used by children to classify safe products as packaged in opaque containers and juice or drinkable products as packaged in transparent containers. Owing to early symbol-recognition skills, bottles with unappealing symbols children might recognize (e.g., insects) may help children identify dangerous products correctly. Bottles containing dangerous products with pointy spouts, such as toilet cleaner containers, may mislead children about product safety because they are miscategorized as drinkable.

Further analyses tested for differences based on gender and race; no statistically significant differences emerged, and data are not shown. None of the children’s parents reported their child could read more than a few words, so analyses were not conducted concerning children’s reading ability.

Study 3 was designed to extend results from Study 2 and further investigate aspects of Hypothesis 3 focused on how the shape, size, and material of containers may be used by young children to distinguish safe versus dangerous containers. We focused particularly on the torch fuel product in Study 3 partly because it was consistently among the most incorrectly categorized dangerous product in the first two studies. In Study 3, we controlled for color and transparency/opaqueness of bottles and tested how children would categorize a set of “alternative” bottles, all of them opaque and painted black, but of different sizes, shapes, and materials. We hypothesized that children might categorize nonround packaging as dangerous more often than round bottles. We hypothesized children might recognize a metal can as dangerous more often than plastic bottles. Because similar and identical beverages come in various sized packages, we did not expect size of bottle to influence children’s decisions significantly.

We chose to replicate the sorting task for Study 3, as this task provides no cues as to whether products are safe (all could be safe or all could be dangerous) and offers some ecological validity in that children are exposed to unfamiliar bottles and must categorize each one without contextual cues concerning safety. To offer replication of the age-related effects found in Study 1, we included samples of young (aged 18–31 months) and older (32–42 months) children.

Study 3

Methods

Study 3 replicated the sorting task from Study 1. Children were presented with the three toys from Study 1 (bear, frog, and book) and nine bottles: 96-ounce apple juice bottle (transparent plastic, rounded with colorful label and pictures of red apples), 96-ounce white grape juice bottle (transparent plastic, rounded with colorful label and pictures of green grapes), 128-ounce milk jug (white plastic with red and white label), 100-ounce torch fuel bottle (transparent plastic, rounded with colorful label), 100-ounce alternative torch fuel bottle (plastic and rounded), 32-ounce alternative sports drink bottle (plastic and rounded), 128-ounce alternative plastic fuel can (black opaque plastic, squared, with large handle), 32-ounce alternative insecticide bottle (plastic and rectangular), and 128-ounce alternative paint thinner can (metal, rectangular with small metal handle). All alternative bottles were commercially available bottles that were painted black (except the fuel can, which was black already), including the lids, and labeled with black, white, and gray labels.

A sample of 119 children aged 18–42 months (mean age = 30.89 months, SD = 7.07) was recruited from seven area preschools. Eighteen other children did not understand the protocol, were older than our maximum inclusion age, or were uncooperative, and were excluded. Included children were 42% female, 77% White, and 14% Black. For analytic purposes, we divided the children into two groups, those 18–31 months (n = 61, mean age = 25.22 months, SD = 3.91) and those 32–42 months (n = 58, mean age = 36.86 months, SD = 4.09). Children were tested individually in quiet locations at their preschool. As in Study 2, the preschools were selected to represent the local area and included religious and nonreligiously affiliated institutions; centers located in urban, suburban, and rural areas; and centers serving children from a range of racial, ethnic, and socioeconomic backgrounds. Also replicating Study 2, on occasion a teacher or aide accompanied children during the experiment to reduce children’s anxiety or distress. Teachers or aides who accompanied children were given explicit instructions to stay quietly behind children and avoid influencing children’s responses. School authorities and children’s parents provided signed informed consent, and children provided verbal assent. All procedures were approved by the local IRB.

Results and Discussion

Results appear in Table VIII. As in Study 1, children sorted the three toys with high accuracy (95%–100%, depending on toy), indicating an understanding of the task. They also sorted the three beverages accurately (77%–84% correctly identified as drinks). Similar to patterns from Study 1, the torch fuel bottle was incorrectly sorted as a drink by 71% of children, including 74% of the older group and 67% of the younger group. As hypothesized based on development of categorization skills, all alternative bottles were recognized as bad or dangerous with greater frequency (between 42% and 59% of the time) than the original transparent torch fuel product (24%). The alternative fuel can and the alternative paint thinner were categorized as dangerous most often; both containers were identified as dangerous by more than two thirds of children in the older age-group. Thus, squared and metal containers were more likely to be categorized as dangerous by children than rounded and plastic ones. All black opaque containers were categorized as dangerous more often than the transparent torch fuel bottle with a liquid colored similar to juice inside. These results support the hypothesis that children might tend to classify transparent bottles into a “drinkable” schema and opaque black bottles into a “dangerous” schema.

Independent samples t tests were computed to compare responses across the age-groups (see Table VIII). Statistically significant differences emerged for the alternative torch fuel bottle, the alternative sports drink bottle, the alternative fuel can, and the alternative paint thinner bottle, in all cases with older children performing better than younger ones. Analyses were also conducted to examine differences based on children’s gender and race/ethnicity. Girls performed slightly better in distinguishing insecticide and the alternative torch fuel bottle as dangerous compared with boys, but no other statistically significant differences emerged. Just two parents (2% of the sample) reported their children could read “basic books with cues”; the remainder of the sample either could not read or could read only a few words. Given this response pattern, analyses concerning reading ability were not conducted.

General Discussion

Taken together, results from the three experiments indicate that young preliterate children are generally poor at distinguishing safe beverage bottles from dangerous household products (Hypothesis 1), but that older children recognize risk somewhat more accurately than younger children (Hypothesis 2) and that there are cues in product packaging that may aid young children substantially in such distinctions and therefore reduce household poisoning risks (Hypothesis 3). Such cues appear to include (a) labeling products with symbols children can recognize, such as unappealing insects that may improve children’s recognition of a product as dangerous or pictures of fruit that may improve children’s recognition of a product as safe; (b) opaque packaging, with transparent bottles showing dangerous juice-colored liquid increasing risk of children misidentifying products as safe; (c) squared rather than rounded packaging, which increases likelihood of children identification of products as dangerous, (d) metal rather than plastic materials, which increases children’s identification of products as dangerous, and (e) omission of pointy spouts, which seemed to mislead children to categorize a dangerous product as safe. Our results suggest there may also be a smaller benefit in using black rather than brightly colored packaging and in using black, white, and gray labels rather than brightly colored labels to help children distinguish safe versus dangerous products.

The toddler and preschool years are a time when cognitive development happens quickly. Such development includes the ability to classify and categorize objects, to recognize symbols, and to make independent decisions (Bjorklund, 2012; Gelman & Coley, 1990; Gelman & Markman, 1986). Our results suggest children may use their developing skills in symbol recognition and categorization to determine whether a product is safe or dangerous. Symbolic labels—in particular a picture of an insect on the insecticide bottle—seemed to help children recognize that bottle as dangerous. Symbols like apples and grapes on the juice bottles may have helped children recognize those products as safe. The role of developing categorization skills also emerged in our findings. Most dramatically, there were significant differences in how well children recognized the risk of the torch fuel bottle, which was transparent and contained a juice-colored liquid, versus alternative bottles that were black and opaque.

Parents typically permit increasing independence for their children during the toddler and preschool years, but supervision and safeguarding remain critical aspects of child safety. Young children use touching and tasting senses frequently to explore and learn about the world around them, and poisoning is among the leading causes of child mortality among young American children. Thus, prevention strategies are needed. Our results have implications for poisoning prevention for parents, as well as for industry and policymakers.

For parents, our results indicate that children have difficulty distinguishing safe versus dangerous household products and that even the older children we studied, those aged 43–54 months, made many errors in categorizing dangerous products as safe. Parent supervision and safeguarding of the home remain critical to reduce child poisoning risk, and our findings reaffirm the need for theory-based empirically supported prevention campaigns to educate parents about the importance of consistently implementing both child injury-prevention strategies. Also relevant are secondary and tertiary prevention strategies, such as easy access to, and knowledge of, poison control center contact information.

From the perspectives of industry and policy, our results confirm the need for careful consideration of package design and labeling for products frequently stored or used near young children. As reported in preliminary work with adults (Meingast, 2001; Serig, 2000), and corresponding to basic child development research in categorization (Bjorklund, 2012; Gelman & Coley, 1990; Gelman & Markman, 1986), young children seem to categorize particular types of bottles for holding safe products (e.g., those that are transparent and contain juice-colored liquids; those with symbolic labels of apples) and other types of bottles for dangerous products (e.g., those that are in opaque, black, squared containers; those with symbolic labels of insects). Industry should package dangerous products in containers that are more prototypical of dangerous categorical schemas and package safe products in containers that are more prototypical of safer schemas. Along with industry initiatives, policymakers should consider legislation that would require particular products to be manufactured for sale in particular types of containers. As one example, requiring hazardous products to be packaged and sold in opaque containers might reduce child poisoning incidents. The success of the Poison Packaging Prevention Act of 1970 offers an example of how policy can help reduce pediatric poisoning rates.

We stress that our research is preliminary. Before this study, the literature in this field was extremely sparse and dated. Our work also had limitations. We used a combination of laboratory- (Study 1) and school-based (Studies 2 and 3) scenarios to test children’s behavior. Although they had some ecological validity, they did not reflect real-world environments precisely. Our samples were reasonably representative of children in the local area, but were predominantly White and Black, and results may not generalize nationally or globally. We selected a few safe and a few dangerous products, but were not comprehensive; other bottles and combinations of packaging, labeling, shapes, sizes, colors, and symbols should be tested in future work. Further, our experimental tasks each had pros and cons. Most outcome measures were categorical, creating a situation where descriptive data and basic group comparisons were most useful to address our hypotheses. Further research using sophisticated and ecologically valid research designs, elegant data analysis techniques, and diverse and large samples of children throughout developmental stages will be needed to continue to study and understand how children make decisions about whether to handle and consume dangerous household products.

Acknowledgments

Thanks to the UAB Youth Safety Lab for their assistance with data collection and entry. Special thanks to Dr Craig McClure, UAB Department of Chemistry, for his assistance in preparing the bottles for the experiments. We thank Pat Carr and his colleagues for their support.

Funding

The research was supported by Carr & Carr, Attorneys at Law.

Conflicts of interest: We acknowledge a potential financial conflict of interest in this research, as there is business interest in the study results to both the sponsor Carr & Carr and the researcher David C. Schwebel, who periodically serves as an expert witness consultant to Carr & Carr.

References