Nursery vs childminder vs nanny: what the quality evidence says

The question

Should you choose a nursery, a childminder, or a nanny — and does the choice change how your child turns out? This topic compares the childcare types. The companion topic [daycare-before-age-two] covers whether starting group care before age two affects development at all.

Short answer

Across the biggest studies, the type of childcare predicts far less about a child's development than the quality of the specific setting or caregiver [1][2][4]. Children in higher-quality care — warm, responsive adults, plenty of talk and play, small groups — score a little better on language, thinking and school skills, and the advantage is still detectable years later, though it stays small [1][4][5][20]. Nurseries vary the most in quality. Early US analyses linked very long hours in group care to slightly more behaviour problems — but the link weakened under stricter methods, and a 2023 cross-national study found no association at all, with no validated "dangerous hours" threshold [4][6]. And price is a poor guide: in one UK study of infants, higher fees bought higher observed quality only with childminders, not with nurseries [3].

What the strongest evidence says

Quality beats type — the most repeated finding in this literature

The UK's biggest study, the Effective Provision of Pre-School Education (EPPE) project, followed around 3,000 children from age three. Its headline result: the measured quality of the pre-school predicted children's intellectual and social-behavioural development at school entry and through the early school years [1]. Differences between types of setting had washed out by age ten, while the effect of quality remained very strong — and higher pre-school quality still mattered at age fourteen for maths and science attainment and social-behavioural outcomes [2]. Integrated centres and nursery schools did best overall, but they also served the most disadvantaged children; private day nurseries showed the widest variation in quality and effects — which is exactly why "nursery" as a category tells you so little [2].

The US NICHD study (about 1,300 children followed from birth) tells the same story from a different angle: higher-quality care — and care that improved in quality over time — predicted better language and pre-academic skills at age four and a half, with small advantages still visible in the teenage years [4][5]. Centre-based experience added a small extra language benefit on top [4].

The unavoidable confounder, stated plainly: none of this is randomized. Families who choose nannies, childminders, or nurseries differ in income, education, and parenting — and those family differences explain far more of children's outcomes than any childcare feature [4]. The studies above all adjust for family background, but no statistical adjustment is perfect.

What "high quality" actually looks like

Researchers split quality into structural features (things you can regulate: ratios, group size, staff qualifications) and process quality (what actually happens: warmth, responsiveness, rich talk). Process quality is what predicts child outcomes; structural features matter mostly because they make good process quality more likely [7][8].

The markers that show up again and again:

• Warm, sensitive, responsive caregiving — the single most consistent predictor across US and UK studies [4][10][11].
• Rich language and stimulating activities, including the back-and-forth "sustained shared thinking" that EPPE found in its most effective settings — which were also the settings with the most highly qualified staff [1].
• Small groups and low child-to-staff ratios. A DfES study of 301 English settings found better staff-child ratios and a higher proportion of trained staff — especially staff with teaching qualifications — predicted higher observed quality, with the presence of a qualified teacher mattering most for the educational side of quality [7]. In the US, centre classrooms meeting recommended standards for ratios, group size, and caregiver education and training produced modestly better child outcomes [8].

The honest limit: structural features' direct link to child outcomes is modest — they work through day-to-day interactions [8]. A setting can meet every ratio on paper and still be cold.

Where the types do differ

Centre vs home-based, for thinking skills: A Danish study used a natural experiment — municipalities varied in guaranteed access to centre-based care — and found children in centre-based care at age two scored about 0.2 standard deviations higher in Danish at age fifteen and were more likely to enter the academic high-school track, with the biggest gains for boys and children of less-educated mothers [9]. In poor US communities, centre care was linked to better cognitive growth, while children in family child-care homes (the US equivalent of childminders) showed more behaviour problems but no cognitive differences [10]. The pattern: group settings seem to add a little cognitive stretch, especially where the home learning environment is thinner — consistent with the companion topic's finding that disadvantaged children gain most from formal early education [daycare-before-age-two].

For babies, home-based care looked warmer: The UK Families, Children and Child Care study directly observed 307 infants at ten months and 331 at eighteen months across nurseries, childminders, nannies, and grandparents. Observed quality was lowest in nurseries at both ages — except that at eighteen months nurseries offered more learning activities — while childminders, nannies, and grandparents differed little from each other [3]. And the clearest driver of nursery quality was the observed child-to-adult ratio: the more babies each adult was juggling, the lower the quality of care [3].

Nannies are the evidence gap. Almost no outcomes research compares nannies with other types — most studies contrast centre-based with family day care. What we know about nannies is descriptive, not comparative (see What remains uncertain).

The long-hours question, stated carefully

Early NICHD analyses linked more hours in care to slightly more behaviour problems as rated by caregivers [4]. But when McCartney and colleagues stress-tested that link, it turned out to be fragile: results varied across model specifications, it wasn't explained by children's own characteristics, and the hours effect was concentrated where children spent more time in large peer groups — with modest overall magnitude [6]. The UK SEED study similarly found that the total amount of early education between age two and school start was not associated with academic outcomes at age seven [12]. The fair reading: the small hours–behaviour association shrinks under stricter causal methods, and a 2023 cross-national study tracking within-child changes found no association at all — there is no evidence for a magic safe (or dangerous) number of hours [daycare-before-age-two].

Ofsted ratings are a broad, imperfect proxy

In 1,423 English settings, "Outstanding" settings averaged higher observed process quality than "Good" ones — but the associations were weak and incomplete, and in research using a quality measure designed for babies and toddlers, Ofsted grades showed few significant associations at all [13]. Treat Ofsted as a useful but blunt screening tool: it captures whether a setting meets regulatory and structural standards, but it is a weak measure of the day-to-day interactions — warmth, language, responsiveness — that predict children's outcomes, especially under 30 months [13]. And in the FCCC infant study, mothers' satisfaction with their child's care tracked observed quality for home-based care — but not for nurseries [3]. For a childminder you are buying one relationship you can observe directly; for a nursery, quality varies room to room and month to month — so visit in person, watch how staff interact with the babies, and ask about ratios and staff turnover rather than trusting the grade on the door.

What it means for the parents

Cost is the sharpest real difference between the types. Coram's 2025 survey (England, after working-parent entitlements): fifty hours a week for an under-two costs £238.95 at a nursery versus £202.09 with a childminder — nurseries about eighteen per cent more; for twenty-five hours it's £70.51 versus £59.76 [14]. In inner London, fifty hours runs £319.24 (nursery) versus £288.84 (childminder) [14]. A nanny is in another league: at 2025 average live-out rates, roughly £490 a week in Greater London for twenty-five hours (about £395 across the rest of the UK), and £975 a week for a full-time daily nanny in London (£790 elsewhere) [15] — and that is the gross wage, before employer National Insurance, pension contributions, payroll admin, and any agency fees. Whether any of this is worth it depends on your earnings and the alternative — the companion topic [daycare-before-age-two] covers what affordable childcare does for parents' employment.

Logistics and flexibility differ by type in predictable ways. In qualitative UK research, parents valued childminders' flexibility — swapping days, covering shift work and school holidays — while the single-person nature of childminding was the main downside: when the childminder is ill, you have no backup [16]. Nannies keep working when your child is mildly ill (nurseries and childminders generally won't take a sick child); nurseries offer the most institutional reliability but the least flexibility. These are features of the services, not research findings about outcomes.

If your child has additional needs, ask directly about experience. A 2018 DfE survey found seventy-seven per cent of school-based nurseries and seventy-two per cent of group providers had children with SEND registered, compared with nineteen per cent of childminders [19] — not a quality judgement, but relevant to how much practice a setting has.

Guilt: the research is about work, not about type. No study compares parental guilt across nursery, childminder, and nanny. What exists is research on employment-related guilt in general: in a longitudinal study of 210 full-time employed mothers of infants, stronger "intensive mothering" beliefs at four months predicted more employment-specific guilt at sixteen months — but perceived control over work schedules softened the link [17]. If guilt is biting, the evidence points at work flexibility and support, not at the type of care.

What we don't know about parents here: no study in this topic measured parental sleep, mental health, or relationship strain by type of childcare. That's a gap, not a finding.

The bottom line for you: the child-outcome evidence doesn't crown a winner by type — so choose on cost, logistics, and fit with your work, and then judge the specific setting or person hard on quality.

What remains uncertain

• Selection effects. No randomized trial of childcare type exists or is ethically possible; families who choose different types differ in ways statistics can't fully adjust for.
• Nannies. Child outcomes with nannies versus other types are barely studied.
• Ratios and child outcomes. UK evidence links ratios to observed quality, not directly to child outcomes — and almost nothing quantitative exists on whether England's 2023 ratio loosening for two-year-olds (one-to-four to one-to-five) changed anything.
• Long-run adult outcomes by type (earnings, health) — no UK study tracks this the way Norway's registers do for childcare overall.
• Parental outcomes by type — sleep, mental health, relationships: essentially unstudied.
• Whether post-pandemic staffing pressures have shifted any of the older findings.

Benefits and risks in absolute terms

The literature doesn't report absolute "X in 1,000" rates by childcare type, so this section is necessarily thinner than our standards would like:

• Higher- versus lower-quality care (any type): cognitive and language scores shift by about 0.1 standard deviations — real, lasting, but small next to family-background effects [4][5]. [B]
• Centre-based versus family day care at age two (Denmark): about 0.2 standard deviations higher Danish grades at fifteen; more academic-track enrolment [9]. [B]
• More hours in group care: slightly more caregiver-rated behaviour problems; fragile across methods, concentrated in low-quality care [4][6]. [B]
• Cost (England, 50 h/week, after entitlements): nursery £238.95 versus childminder £202.09 per week for an under-two; a full-time live-out nanny roughly £790 a week outside London, £975 in Greater London (gross wage at 2025 rates, before employer costs) [14][15]. [survey data]
• Low-quality care is the one condition consistently linked to worse outcomes — in EPPE it conferred little benefit by age eleven [20]. [B]

Practical considerations

1. Judge the setting, not the category. Visit at a busy time; watch whether adults get down to the children's level, talk a lot, and respond warmly; ask about staff turnover and qualifications [1][3][4].
2. Don't use price as a quality signal for nurseries — but for childminders, higher fees did track higher observed quality [3].
3. Check the actual baby-room ratios, not just the legal minimum (one-to-three under two; one-to-five for two-year-olds since September 2023; childminders a maximum of three young children) — observed ratios drove quality differences in the UK infant study [3][18].
4. Treat Ofsted as a screen, not a verdict — "good" or "outstanding" doesn't guarantee the day-to-day quality that matters [13].
5. Match the type to your logistics: irregular or shift hours → a childminder's flexibility [16]; a job where you can't take sick days → a nanny; school-hours reliability → a nursery.
6. Run the money honestly: use the Coram figures above as a starting point, then apply your funded-hours entitlement and any childcare support you're eligible for — entitlements change the maths substantially [14].

When to talk to your doctor, midwife, or pediatrician

• If you have concerns about a setting's safety or your child's wellbeing there: raise it with the setting first, then with Ofsted or your local authority if needed [13].
• If your child shows persistent behavioural changes — sustained aggression, withdrawal, or loss of skills — talk to your GP or health visitor; childcare type is rarely the whole explanation.
• If your child has a health condition or disability that affects their care needs, discuss with your health visitor or GP what to look for in a setting, and ask prospective providers directly about their experience [19].

References

1. Sylva, K., Melhuish, E., Sammons, P., Siraj-Blatchford, I., & Taggart, B. (2004). The Effective Provision of Pre-School Education (EPPE) Project: Technical Paper 12 — The Final Report: Effective Pre-School Education. Research briefing: https://discovery.ucl.ac.uk/id/eprint/10018684/1/RB69_EPPSE_endKeystage1.pdf — [B]
2. EPPE 3–11 / EPPSE 3–14. Type effects washed out by age 10 while quality effects persisted (promoting-equality report: http://dera.ioe.ac.uk/id/eprint/18199/1/EPPE_-_Promoting_Equality_in_the_Early_Years.pdf); age-14 briefing: https://discovery.ucl.ac.uk/id/eprint/10018680/1/RB65_EPPSE_endKeystage3.pdf — [B]
3. Leach, P., Barnes, J., Malmberg, L.-E., Sylva, K., Stein, A., & the FCCC team (2008). The quality of different types of child care at 10 and 18 months. Early Child Development and Care, 178(2), 177–209. https://eprints.bbk.ac.uk/id/eprint/2289/ — [B]
4. NICHD Early Child Care Research Network (2002). Early child care and children's development prior to school entry. American Educational Research Journal, 39(1), 133–164. https://journals.sagepub.com/doi/10.3102/00028312039001133 — [B]
5. Vandell, D. L., Belsky, J., Burchinal, M., Steinberg, L., Vandergrift, N., & NICHD ECCRN (2010). Do effects of early child care extend to age 15 years? Child Development, 81(3), 737–756. https://doi.org/10.1111/j.1467-8624.2010.01431.x — [B]
6. McCartney, K., Burchinal, M., Clarke-Stewart, A., Bub, K. L., Owen, M. T., Belsky, J., & NICHD ECCRN (2010). Testing a series of causal propositions relating time in child care to children's externalizing behavior. Developmental Psychology, 46(1), 1–17. https://doi.org/10.1037/a0017886 — [B]
7. Mathers, S., Sylva, K., & Joshi, H. (2007). Quality of Childcare Settings in the Millennium Cohort Study (Research Report SSU/2007/FR/025). Department for Education and Skills. https://dera.ioe.ac.uk/id/eprint/8088/ — [C]
8. NICHD Early Child Care Research Network (1999). Child outcomes when child care center classes meet recommended standards for quality. American Journal of Public Health, 89(7), 1072–1077. https://doi.org/10.2105/AJPH.89.7.1072 — [B]
9. Datta Gupta, N., & Simonsen, M. (2016). Academic performance and type of early childhood care. Economics of Education Review, 53, 217–229. https://doi.org/10.1016/j.econedurev.2016.03.013 — [B]
10. Loeb, S., Fuller, B., Kagan, S. L., & Carrol, B. (2004). Child care in poor communities: Early learning effects of type, quality, and stability. Child Development, 75(1), 47–65. https://pubmed.ncbi.nlm.nih.gov/15015674/ — [C]
11. Melhuish, E., & Gardiner, J. (2018). Study of Early Education and Development (SEED): Impact Study on Early Education Use and Child Outcomes up to age four years (DFE-RR843). https://assets.publishing.service.gov.uk/media/61964e66d3bf7f055eb9b5fd/SEED-Age_4_RESEARCH_BRIEF.pdf — [B]
12. Melhuish, E., & Gardiner, J. (2021). Study of Early Education and Development (SEED): Impact Study on Early Education Use and Child Outcomes up to age seven years. https://assets.publishing.service.gov.uk/media/617a9b79e90e0719751282e4/SEED_Age_7_Impact_Report.pdf — [B]
13. Mathers, S., Singler, R., & Karemaker, A. (2012). Improving quality in the early years: A comparison of perspectives and measures. Daycare Trust / A+ Education / University of Oxford. https://cep.lse.ac.uk/pubs/download/dp1468.pdf — [C]
14. Coram Family and Childcare (2025). Childcare Survey 2025. https://www.coram.org.uk/wp-content/uploads/2025/03/ChildcareSurvey2025-1.pdf — [survey]
15. Nannytax, Employer Salary Guide 2025 (2024/25 average gross live-out rates: Greater London £19.48/hr — £50,648/yr; rest of UK £15.77/hr — £41,002/yr). https://www.nannytax.co.uk/wp-content/uploads/2025/08/NT_Employer_Salary_Guide-2025.pdf — [survey]
16. National Children's Bureau, Research Summary 10 (qualitative study of parents' views on childminders vs nurseries: flexibility, swapping days, holiday cover; single-provider fragility). https://www.ncb.org.uk/sites/default/files/uploads/files/NO33%2520-NextBestThing%2520ncb_rsch_summ_10_final.pdf — [qualitative]
17. Doctoral dissertation, University of North Carolina at Greensboro (2010): intensive mothering beliefs, employment-specific guilt, childcare satisfaction and workplace flexibility (n=210 full-time employed mothers of infants; longitudinal survey with structural equation modelling). https://eric.ed.gov/?id=ED516667 — [C]
18. The Early Years Foundation Stage (Amendment) Regulations 2023 — explanatory note (ratios: 1:3 under-2; 1:5 for 2-year-olds from September 2023; childminder maximum three young children with sibling/own-child flexibilities). https://www.legislation.gov.uk/uksi/2023/780/note/made — [statutory guidance]
19. Department for Education (2018, revised 2021). Survey of Childcare and Early Years Providers: Main Summary — SEND prevalence by provider type. https://assets.publishing.service.gov.uk/media/61b116918fa8f50385f7ed85/Survey_of_Childcare_and_Early_Years_Providers_2018_Main_Summary_revised_December2021.pdf — [survey]
20. Sylva, K., Melhuish, E., Sammons, P., Siraj-Blatchford, I., & Taggart, B. (2011). Pre-school quality and educational outcomes at age 11: Low quality has little benefit. Journal of Early Childhood Research, 9(2), 109–124. https://eric.ed.gov/?id=EJ929702 — [B]

Changelog

• 2026-09-08: Draft created. Evidence gathered from EPPE/EPPSE, NICHD SECCYD, SEED, FCCC infant study, Danish IV study, Loeb et al., Coram 2025, MoneyHelper/Nannytax, EYFS regulations. Overall rating B. Status: draft, awaiting independent review. (Superseded 2026-09-09: author-review model, no external review before v1.)
• 2026-09-09: Delegated author review (Guille, reviewer of record; agent executed under delegation). Ofsted section corrected to "broad, imperfect proxy" (weak/incomplete associations; few significant associations for under-30-month quality); removed "you can probably judge a childminder yourself"; nanny costs refreshed to 2025 Nannytax averages (ref 15) with employer-cost warning; NICHD 1999 citation verified with DOI; long-hours synthesis aligned across chapter (McCartney fragility + 2023 within-child null, no "concentrated in low-quality care" overstatement); guilt claim verified against Walls (2010) ERIC record. Status: approved.