Calculated-risk play and teacher mediation in early childhood education

1. Method: PRISMA-lite review 2021–2026

This section fixes the protocol before any reading of the yard. The review is called PRISMA-lite because it takes on the duty to declare outlets, time window, inclusion, exclusion, and final corpus. It is not a registered systematic review, it does not estimate a common effect, and it does not draw a flow diagram. A funnel with counts of records identified, duplicates, or exclusions would be fiction: those counts were not audited. The only count this article reports is that of the cited corpus: twenty-four publications, each with a DOI checked in Crossref and a year between 2021 and 2026. The close on the kit’s verified bank was made on 1 October 2026.

Eligible outlets were Early Childhood Education Journal, Early Years, International Journal of Early Childhood, Journal of Early Childhood Research, International Journal of Play, Journal of Adventure Education and Outdoor Learning, and, with an early-childhood focus and a verified DOI, Children and the International Journal of Environmental Research and Public Health. The corpus also retains Australasian Journal of Early Childhood, European Early Childhood Education Research Journal, Contemporary Issues in Early Childhood, Journal of Outdoor and Environmental Education, and Early Years Educator. Journal of Early Childhood Research, Children, and IJERPH contribute none of the twenty-four citations.

The year filter is closed: 2021 to 2026. An earlier text may be foundational and still stays outside this citation. If the DOI is not in the kit’s bank, it is not cited. Pages, volumes, samples, and coefficients absent from the record are not reconstructed. When the object is known from the title, the gloss stays on that object. What enters is work on risky play, risk in outdoor play, teacher permission, or family permission in early childhood education, work that lets a craft be read —perceptions, permission practices, attitudes, challenges of unstructured play, environmental quality, affordances, or learning in child-initiated play— and not only the slogan of going outside. The outlet has a DOI and the record is not retracted. Every retracted article, every paper without an early-childhood focus, and every unverified DOI is out. The retracted identifier is not reproduced here.

The twenty-four pieces are grouped into four clusters and are not added up as one sample. Teacher permission and perceptions: Bowden et al. (2026), Turgut-Kurt and Sevimli-Celik (2026), Tsujitani (2024), Liu and Birkeland (2022), Višnjić Jevtić et al. (2022), Giles and Johnson (2026), Cheng et al. (2023), and LeMasters and Vandermaas-Peeler (2023). Families: McCrorie et al. (2025), MacQuarrie et al. (2022), Sak et al. (2025), Sak et al. (2026), and Kybartas et al. (2025). Infants and toddlers: Quinones (2023), Little and Stapleton (2023), and Chen and Hamel (2023). Benefits, yard, and nature: Cetken-Aktas and Sevimli-Celik (2021), Zakharova-Goodman (2026), Barrable and Robinson (2026), Battelley et al. (2026), Speldewinde (2025), Aslanian et al. (2024), Senoz and Kol (2025), and Axelsson (2021). Surnames follow the verified record; initials are not invented, except the one LeMasters already carries.

Each claim carries a mark. Empirical finding names the object of the title —perspectives, practices, perceptions, attitudes, challenges, preferences, or relationships— and does not turn it into a coefficient. Framework names an organization this review does not recode: risky-play categories, risk-benefit balance, a socio-ecological lens, a safety procedure that stalls attitude. Design inference names a hypothesis about the craft. The criteria in section 6 and the contrast in section 7 are inferences, not norms or product trials. The protocol sustains a category argument. It does not rank countries, prescribe a model yard, or certify an app.

2. State of the art: risky play and the teaching craft

The object is calculated risky play and the mediation that makes it livable in early childhood education. In recent literature it appears as risky play: the child tries a physical or social challenge whose outcome is not guaranteed, and an adult may allow, come close, or stop. Calculated risk names a judgment, not an index: it is weighed by someone who knows the child, the group, and the place. The craft enables the challenge, prepares the competence it requires, supervises closely without confiscating the attempt, arranges the yard, and talks with families about the criterion, not only about the signature.

This object does not rewrite four neighbors. Outdoor play is broader: a group can go into the yard and still allow no challenge. Open materials organize material exploration; they may host risk and they do not define it. Guided play is an adult intention inside play; permission for a physical challenge is not that guidance. Role play organizes a fiction; a run down a slope or bounded rough-and-tumble is not theater for that reason. They share a day. If they are fused, any hour in the yard already looks like support for calculated risk.

Zakharova-Goodman (2026) titles the risk of risky play to pedagogical possibilities. It is a framework: there is loss if the challenge falls into negligence, and there is loss if pedagogy protects itself by erasing the challenge, the competence, and the child’s judgment. No protocol and no injury datum are extracted from that piece. Axelsson (2021) places outdoor play in the early years curriculum, as exercise of muscles and minds. The calculated risk that occurs inside that play is situated teaching, not a pause between lessons.

The craft is not a single common sense. Liu and Birkeland (2022) compare kindergarten teachers’ perceptions of risky play in Norway and in China. The finding is that object: permission travels with institutional histories, and this review builds no country ranking. Tsujitani (2024) asks, in Japan and by questionnaire, about perceptions anchored in the practitioner’s own childhood and in present practice. Whoever mediates arrives with a biography and with a practice that has already taken a side. Višnjić Jevtić et al. (2022), from Croatia, study attitudes that teacher education develops and that safety procedures impede. The obstacle can be a procedure that makes permission unthinkable.

Giles and Johnson (2026) add preschool teachers’ qualitative view of outdoor play. Their object is educators’ experience of the outdoors, not a yard score. Bowden et al. (2026) lower the age: educators’ perspectives on risky play from birth to three years, in early childhood services. The challenge stops being a privilege of the child who already runs with ease. Infants and toddlers meet slopes, rolling objects, and other children’s bodies. The craft starts there, with another calibration. Cheng et al. (2023) complete the map with perceived challenges in promoting unstructured outdoor play, read through ecological systems. Permission is decided among center rules, family expectations, yard conditions, and wider frames. An artifact that ignores those planes simulates an individual decision where the title shows a system.

3. Findings: risk categories and benefits

The field organizes risky play into recurrent categories. Here they are a framework, not a coding of the twenty-four sources: mentions were not counted and inter-rater agreement was not calculated. The craft’s vocabulary names height, speed, tools, bounded rough-and-tumble, proximity to what is perceived as dangerous —deep water, fire, a real drop— and play at disappearing for a moment. It serves to see the challenge. It does not serve to score a child or to declare safe a yard that lacks a category. Whoever looks only at height leaves speed, tools, rough-and-tumble, the edge, and the brief hiding place unnamed.

Cetken-Aktas and Sevimli-Celik (2021) examine opportunities for risky play in preschool outdoor areas. The empirical finding is environmental: the yard offers or withholds occasions. It is not translated here into meters or an index. If there is no moderable height, no slope, no tool under a rule, and no corner in which to be lost from sight without being truly lost, permission has nowhere to land. LeMasters and Vandermaas-Peeler (2023) study, with mixed methods, the quality of those environments and teachers’ perceptions. They join the place and the belief of the person who inhabits it, without this review melting them into a single number.

Benefits are read with the same discipline. Battelley et al. (2026) state that risky play builds physical development and emotional resilience: a professional framework, not a controlled trial. Senoz and Kol (2025) explore outdoor-play frequency and emotional intelligence at ages five and six. The object is in the title; sign and magnitude are not extracted. Aslanian et al. (2024) describe holistic learning in child-initiated outdoor play in a Norwegian kindergarten. Axelsson (2021) adds muscles and minds. The thinkable benefit is wide, in play the child starts and the adult sustains. There is no improvement percentage here, and no poster that replaces mediation.

Barrable and Robinson (2026) argue the step from risky play to dirty play, and the reason young children need nature play that gets them dirty. It is a framework. Calculated risk is not exhausted by the clean climb in a catalog. It includes mud, contact, and an adult tolerance for the body not leaving play as it entered. Speldewinde (2025) connects young children’s risky play in nature to science education in Australian bush kinders. The title carries a local rule —don’t pick, don’t lick— that is already mediation: there is permission and there is a named limit. The empirical object is that link, not an incident rate. Together, the two pieces block an aseptic version of risk. The challenge is also sensory, and the rule is also teacher language.

The joint reading is a design inference. Categories serve to see the challenge. The benefits the corpus states —physical development and resilience as a professional thesis, emotional intelligence as an object tied to outdoor-play frequency, holistic learning, a curriculum of muscles and minds, dirty play, science inside natural risk— describe why the challenge deserves permission. None of those pieces is an app trial, and none delivers a threshold past which the yard is “already safe” because a score fell. When an indicator rewards the elimination of the challenge, it rewards the disappearance of the object whose benefits this literature studies.

4. Findings: teacher permission, families, and the yard

Permission is the central teaching act. It is not a distracted yes or a signature at the start of the year. It is the repeated decision to let a challenge happen because someone judged the balance between what the child may gain and what may happen, and because there is a way of staying near. Turgut-Kurt and Sevimli-Celik (2026) study the giving of permission: teacher variables and reported practices. The empirical finding is that object. Permission is investigated as a practice stated by those who teach and is related to variables of the person who teaches. This review does not list those variables or invent which one weighs more. The declared design is enough: permission is not a fixed moral trait. An identical checklist for the whole staff simulates a practice the study treats as variable.

With infants and toddlers, permission changes scale and does not disappear. Bowden et al. (2026) gather educators’ perspectives from birth to three. Quinones (2023) describes outdoor risky-play practices with infants and toddlers in an Australian center. Little and Stapleton (2023) explore toddlers’ rituals of belonging through outdoor risky play. Chen and Hamel (2023) study toddlers’ play preferences and the teacher’s role in the outdoor environment. Kybartas et al. (2025) investigate toddler risky play in a single school, informed by parents’ and teachers’ voices. Five objects, none interchangeable: professional perspectives, practices in one center, ritual belonging, child preference with an adult role, and home and school voices in the same place. This review does not melt them into “adults allow” or “adults forbid.”

The craft reading is a design inference. With infants the challenge is usually low and closely watched: a short drop, a rolling object, another body nearby. With those who walk, brief speed, the height of a log, a simple tool, and rough-and-tumble that still resembles a hug appear. Little and Stapleton (2023) make outdoor risk a form of belonging. Chen and Hamel (2023) put the teacher’s role inside the title. Kybartas et al. (2025) make permission a conversation between home and center. Quinones (2023) studies the practices of one service, not a general policy.

Families enter with their own objects. McCrorie et al. (2025) study how parental and practitioner perceptions —danger and benefit— are associated with young children’s outdoor play experiences. The title joins both sides of the balance; a score that only subtracts incidents has no cell for benefit. This gloss does not state the sign or the size of the association. MacQuarrie et al. (2022) read parental perceptions through a socio-ecological lens: tolerance is explained in circles, not only by character. Sak et al. (2025) examine parents’ gender, attitudes, and permission. The object is that relationship; it does not authorize saying that some forbid and others allow. Sak et al. (2026) examine school readiness at ages four to six and parental tolerance of risky play, without any claim here that one predicts the other. Parental permission remains part of school life, not a side errand.

The yard is the third term. Cetken-Aktas and Sevimli-Celik (2021) examine risk opportunities in the outdoor area. LeMasters and Vandermaas-Peeler (2023) tie environmental quality to teacher perception. Giles and Johnson (2026) return educators’ gaze. Cheng et al. (2023) place the challenges in several systems, not only in one person’s character. A poor yard makes the order to “allow” false. A rich yard without permission is used as a corridor. Without dialogue with families, conflict waits at the door. The finding, at the level of objects, is that triad. The inference is that support exists when permission, families, and yard are worked together.

5. Anonymized classroom case (ages 3–5, yard / woodland)

The case is an anonymized reconstruction, composed in order to examine mediation. It does not come from a deposited field notebook, it does not belong to the corpus, and it is not a sample. It takes place in a public early childhood center whose name, city, and neighborhood are omitted. The group brings together children of three, four, and five. The place has a dirt yard and a wooded edge to which the same group goes out, not a club of the most able. The duration is a morning block, from going out to the yard until returning to the room. The exact clock is not fixed: fixing it would simulate a datum. What is narrated is what kind of challenge appears and what the person who mediates does. An injury outcome is not narrated, because that outcome is not claimed.

At height, two children try a low, wide log. The teacher stays one step away, names the support of the feet, and waits for the next attempt or the chosen descent, without lifting the child who hesitates. At speed, a gentle slope is used to come down faster than a walk; the rule is to watch whoever comes the other way, and an adult stands on the open side. With tools, a child digs beside a root with a short spade. The teaching covers how to carry it, where not to lift it near another face, and when it returns to the basket. Rough-and-tumble is bounded: two children push shoulders and laugh, with a stop-word anyone may say. If one says it, the play stops. There is no zero-contact slogan; there is a distinction, with the body near, between the play both seek and the play one no longer wants. A brief hiding place closes the scene: behind a bush, still visible at one meter. The boundary is said aloud —the bush yes, the path that leaves the wood no.

The case claims neither injuries nor their absence, nor a reduction attributable to the design: there is no record. It claims no number of children, of minutes, or of mean age, no comparison with another group, no claim that the teacher applies a published program, and no claim that families approve the scene. At the door she tells a mother about the log and the spade rule, and she hears a worry. That conversation illustrates the craft. It is not a satisfaction datum. The wooded edge made the log, the spade, and the hiding place possible; it is not claimed that a wood is superior to an urban yard.

The reading is a design inference. The teacher calculates. She does not remove the slope, lock away the spade, forbid contact, or ban the bush. She does not absent herself either. She is near at the height, on the open side of the speed, in the tool rule, and at the stop signal. Scaffolding happens before and during: the stop-word was practiced, the spade was taught, the boundary was said. If a safety score marked the slope, the spade, the contact, and the brief loss of sight as defects, the whole episode would be a bad result. That sentence does not try a brand. It shows why the genre described at the start does not constitute support: its administrative success would coincide with the disappearance of the scene.

6. Discussion: criteria for mediating calculated risk

The four criteria are design hypotheses for the craft. They are not a validated norm, they have no cut point, and they should not be printed as an inspection rubric. They serve a team discussing a real yard. Each one rests on objects from the corpus and marks the step toward inference.

First: calculated risk and negligence are different acts. Negligence is a challenge nobody watched, a tool without a rule, or rough-and-tumble that continues after the no. Calculated risk knows the challenge, accepts a bounded uncertainty, and keeps a way to intervene in time. Višnjić Jevtić et al. (2022) show safety procedures that stall the attitude education tried to form. The inference does not repeal the procedure: it asks the procedure to distinguish negligence from challenge. Treating every height and every tool as a fault produces paper safety. Zakharova-Goodman (2026) recalls that erasing the challenge in order not to judge also loses pedagogical possibility.

Second: competence is scaffolded before it is forbidden. Forbidding is right when the challenge exceeds what that child, that day, can manage even with help. The order starts from the competence that is missing and from whether it can be taught there. Chen and Hamel (2023) place the teacher’s role inside play preferences: the adult does not cancel by default. Bowden et al. (2026) require that scaffolding from birth to three: how to bring a foot down, how to let go, how to wait. Little and Stapleton (2023) add belonging to the group. Quinones (2023) recalls that the practice belongs to a center, not to a universal poster.

Third: supervision is calibrated and near. Near means being able to touch, speak, and stop in human time. Calibrated means the distance changes with the challenge and with the person. At the log, one step. On the slope, the open side. In rough-and-tumble, the distance of whoever hears the stop-word. Entirely far is negligence. On top of the child, solving every foothold, is confiscation of the challenge. Tsujitani (2024) ties perception to present practice: supervision is not deduced only from a stated belief. Giles and Johnson (2026) return that practice to educators’ gaze. Turgut-Kurt and Sevimli-Celik (2026) study permission as reported practice. The inference joins the three objects: calibrated nearness is the bodily form of permission. An incident dashboard is not near. An assessment generated at a desk is not near.

Fourth: technology serves the reflective record, not a score that eliminates the challenge. Noting or photographing with an ethical criterion returns the scene to the team’s conversation. The score rewards the yard where nothing classifiable as risk happened and punishes calibrated exploration, which from the outside looks like carelessness. McCrorie et al. (2025) join danger and benefit; an indicator of mere absence of incident erases half. Cetken-Aktas and Sevimli-Celik (2021) and LeMasters and Vandermaas-Peeler (2023) treat the yard as an environment observable by people. The four criteria hold together and are not a psychometric instrument. They are a hypothesis for teams that already have a yard, children, and families at the door.

7. Contrast: score apps versus situated judgment

The contrast is categorical. No commercial app was evaluated, no demo account was opened, and no experiment was fabricated with areas under the curve, precision, or incident reduction. The argument does not depend on a brand. It depends on the genre. Four genres circulate as if they were support for risky play. They can reinforce one another in the same center: the score feeds the dashboard, the generative text writes the report the dashboard seems to request, and the checklist files a checkbox that calms administration.

The first genre assigns a safety score to the yard. Its grammar is subtraction: each height, each slope, each tool, and each corner with partial visibility deducts, and the virtuous yard is the one that does not deduct. That grammar rewards elimination of the challenge and punishes calibrated exploration, which from the dashboard looks the same as carelessness. Cetken-Aktas and Sevimli-Celik (2021) examine opportunities for risky play: a yard with high opportunities would be, for the score, a suspect yard. The contradiction is structural. Situated judgment reads the same height and asks who climbs, what competence they have today, who is near, and what benefit sustains the permission. The score asks how much to deduct.

The second genre produces “play risk assessments” with a generative model. The text can be coherent and still not have seen the child, the group, the log, or the family at the door. Turgut-Kurt and Sevimli-Celik (2026) tie permission to someone who teaches and to what that person reports doing. Tsujitani (2024) ties perception to a childhood and to a present practice. A language model has neither that biography nor that practice. It may, as a tightly bounded inference, order questions the team has already decided to ask. It replaces judgment when the report is filed as if the yard had been seen. This review does not measure the model’s error rates. It marks the category jump: from plausible text to situated assessment.

The third genre counts incidents without an observation protocol. A fall inside a calculated challenge, negligence, an ordinary stumble, or rough-and-tumble past the stop signal fit in the same number. Cheng et al. (2023) place the challenges in ecological systems: a dashboard that ignores rules, families, and yard turns a system problem into a personal figure. Giles and Johnson (2026) ask educators how they see outdoor play. That question is slower than the dashboard, and speed does not replace it. The fourth genre files digital parental permission: a yes and a time. Sak et al. (2025, 2026), MacQuarrie et al. (2022), McCrorie et al. (2025), and Kybartas et al. (2025) study dense relationships —attitudes, tolerance, danger, benefit, voices from home and from the center. The checkbox may record that a conversation happened. If it replaces the conversation, there is a file and there is no shared criterion.

The closing sentence is an inference and means to look like what it is. None of the four genres constitutes support for calculated risky play. Support is recognized in another scene: a person who distinguishes negligence from challenge, who teaches competence before forbidding, who stays near at a variable distance, and who uses technology, if at all, to look again at what happened. Apps may sit beside that scene. They take its place when the score, the generated report, the dashboard, or the checkbox are presented as the scene itself.

8. Limits

This is not a systematic review with a flow diagram: there is no count of records identified or excluded other than the corpus of twenty-four. Literature before 2021 is not cited; the window prevents it, and the gap is not filled with an authority absent from the bank. There is no reanalysis either. Titles that announce relationships —perceptions and experiences, play frequency and emotional intelligence, tolerance and school readiness, parental gender and permission, teacher variables and practices— receive neither direction nor size here. There is no N, d, or r. Battelley et al. (2026), Barrable and Robinson (2026), Axelsson (2021), and Zakharova-Goodman (2026) are read as frameworks.

Geography is that of the titles, not a map of the world: Japan (Tsujitani, 2024), Norway and China (Liu and Birkeland, 2022), Croatia (Višnjić Jevtić et al., 2022), an Australian center (Quinones, 2023), Australian bush kinders (Speldewinde, 2025), and a Norwegian kindergarten (Aslanian et al., 2024). Where the title does not declare a country, none is inferred from the journal or the surname. Journal of Early Childhood Research, Children, and IJERPH were eligible and contribute no citation. The case is an illustration, not evidence of effectiveness or of safety. General outdoor play, open materials, guided play, and role play are not redone here. No paragraph evaluates a concrete digital product: what data it stores, and whether the score pushes toward removing the challenge, is a local evaluation this corpus does not contain.

9. Conclusions

The restrictive thesis holds at the level the method can reach. A safety-score app, a generative package of play risk assessments, an incident dashboard without an observation protocol, and a digital parental-permission checklist do not constitute support for risky play in preschool when they lack the pedagogical criterion of calculated risk. That criterion comprises the balance of risk and benefit, the scaffolding of competence, calibrated supervision, the design of affordances, and dialogue with families. The craft that does constitute support is teacher mediation that enables situated physical and social challenges. Safety and the elimination of risk are not synonyms. The first requires judgment. The second can be a way of not judging.

The corpus verifies objects, not a ranking of interventions or a software certificate: permission as a variable practice; perceptions that travel with context, biography, and procedure; risk studied from birth; families who enter through perception, tolerance, and permission; a yard that offers or withholds opportunities; benefits formulated as body, emotion, holistic learning, curriculum, dirty play, and science in nature; and a risky play that puts pedagogical possibilities at stake. Challenges remain at an ecological scale. Categories serve to see, not to score. The four criteria are design hypotheses. The case illustrates teacher presence and does not prove an absence of injuries. Whoever invests in situated judgment —team time, conversation with families about benefit and danger, teachable challenges— supports calculated risky play. Technology may remember what was seen. A score that wins when the challenge disappears cannot be given the permission.

10. Implications for teacher education

To educate teachers is to teach the distinction between safety and the elimination of risk. Višnjić Jevtić et al. (2022) show attitudes that teacher education develops and that safety procedures impede. A program that preaches permission and assesses the student by a yard without challenges reproduces that clash. If the rubric only subtracts potential incidents, it trains the score. Permission practice deserves hours in a real yard. Turgut-Kurt and Sevimli-Celik (2026) tie permission, teacher variables, and reported practices: education can ask for a narrated decision, with the challenge, the scaffolded competence, the place of the body, and the border with negligence. Tsujitani (2024) justifies looking at what play each person knew and how that memory pushes practice. Liu and Birkeland (2022) block a single national common sense. Giles and Johnson (2026) anchor continuing education in the gaze of someone already in preschool.

Dialogue with families is practiced. McCrorie et al. (2025) join danger and benefit, families and practitioners. MacQuarrie et al. (2022) read tolerance through a socio-ecological lens. Sak et al. (2025) and Sak et al. (2026) make permission, attitudes, and tolerance studiable, the last of these in relation to school readiness from ages four to six. Kybartas et al. (2025) gather parents’ and teachers’ voices in a single center. In practicum, the student shows a real challenge —log, slope, or spade— and agrees a criterion. The checklist, if it exists, is filled afterward, as a record.

Age is educated too. Bowden et al. (2026), Quinones (2023), Little and Stapleton (2023), and Chen and Hamel (2023) prevent reserving risk for age five: infants and toddlers need a small challenge, belonging, preference, and an adult role in real outdoor settings. Cetken-Aktas and Sevimli-Celik (2021) and LeMasters and Vandermaas-Peeler (2023) give a yard exercise: what risk opportunities it offers and what quality the person who inhabits it perceives. Cheng et al. (2023) require naming challenges at more than one level of the system. Technological literacy, also hypothetical, asks the graduate to read a score, a generative report, a dashboard, and a digital permission, and to say which part of the craft each genre does not see. A team note serves to adjust distance the following week. That criterion, practiced in the yard and with families, prepares the craft.

Laboratorio Editorial de NEXTECH.IA / Ingeniero Mitre