5D Library › Early Experience, Epigenetic Methylation and the Limits of Intergenerational Transmission · 中文
Chapter 6 Breaking Negative Cycles: Partners, Mentors and Structured Interventions as Channel Filters
How strong is the intergenerational cycle?
“Children who were abused grow up to abuse their own children” is a widely repeated claim. It contains a real association, but the association is much weaker than people think, and the way it is measured often inflates it.
Evidence
About a third, not most. Reviewing the research of their time, Kaufman and Zigler concluded that the best estimate of the rate of intergenerational transmission was about 30 % (±5 %): about a third of people who were abused go on to abuse their children, and about two-thirds do not (Kaufman and Zigler 1987). A meta-analysis by Madigan and colleagues reviewed the methodological quality of 142 studies (149 samples, 227,918 parent–child dyads); across the 80 samples that yielded an overall estimate the effect was (95 % CI 0.37–0.54), which the authors took to confirm the cycle-of-maltreatment hypothesis while calling the effect modest; methodological quality moderated only the transmission of physical abuse, with smaller effects in more rigorous studies (Madigan et al. 2019).
The effect of the information source. Widom and colleagues followed people with court-documented childhood abuse or neglect and matched controls, and compared three sources of information: child-protection agency records, the person’s own report and reports by their children. People with an abuse history were more often reported to child-protection services, but they did not report more physical or sexual abuse of their children than the controls did; their children did report more sexual abuse and neglect. The authors argue that detection or surveillance bias can explain the higher rates in records, and that the extent of transmission depends largely on the source of information (Widom et al. 2015).
Calculation
A simulation of detection bias. Of 100,000 parents, 15 % have a documented history of childhood maltreatment; the true rate of maltreatment of their children is set at 12 %, and 9 % for other parents (illustrative). In families already known to child-protection services a true case is recorded with probability 0.60, in other families 0.30. A confidential self-report survey detects 70 % of true cases in both groups, with a 2 % false-positive rate in both.
# Detection bias in official records (the question of Widom et al. 2015).
# 100,000 parents, 15 % with a documented maltreatment history. TRUE rates of
# maltreatment of their children 12 % vs 9 % (ILLUSTRATIVE). Records catch a
# true case with probability 0.60 if the family is already known to child
# protection, 0.30 otherwise; a confidential survey catches 70 % in both
# groups, with 2 % false positives.
import random
rng = random.Random(1967)
def odds_ratio(a, n1, b, n0):
return (a / (n1 - a)) / (b / (n0 - b))
N, share = 100000, 0.15
hist = [rng.random() < share for _ in range(N)]
true = [rng.random() < (0.12 if h else 0.09) for h in hist]
rec = [t and rng.random() < (0.60 if h else 0.30) for h, t in zip(hist, true)]
rep = [(t and rng.random() < 0.70) or (not t and rng.random() < 0.02)
for t in true]
n1 = sum(hist); n0 = N - n1
for name, v in (("true maltreatment", true), ("official records", rec),
("self-report", rep)):
a = sum(x for x, h in zip(v, hist) if h)
b = sum(x for x, h in zip(v, hist) if not h)
print(f"{name:18s}: {a / n1:5.1%} vs {b / n0:5.1%} "
f"odds ratio {odds_ratio(a, n1, b, n0):.2f}")
print("records alone would suggest a 'cycle' twice as strong as the true one")
# true maltreatment : 11.7% vs 9.0% odds ratio 1.34
# official records : 7.0% vs 2.7% odds ratio 2.74
# self-report : 9.9% vs 8.0% odds ratio 1.26
# records alone would suggest a 'cycle' twice as strong as the true oneIn this simulation the true odds ratio is about 1.34; official records give an odds ratio of about 2.74, more than twice the true value; self-report misses some cases, but because it misses them equally in both groups, its odds ratio (about 1.26) is close to the truth. This does not mean the intergenerational association does not exist. It means that the strength of a “cycle” based on official records may count differences in surveillance as transmission.
What can interrupt the cycle?
Evidence
A partner’s warmth. Conger and colleagues analysed the Family Transitions Project, a 22-year, three-generation study that followed more than 500 adolescents into adulthood: the warmth and positive communication of participants’ romantic partners were associated with less harsh parenting by the participants towards their eldest child, and when partners showed high levels of these behaviours there was no evidence that harsh parenting passed from the participants’ parents to the participants (Conger et al. 2013). The authors also found that harsh parenting by one’s own parents slightly reduced the likelihood of choosing a supportive partner.
Safe, stable, nurturing relationships. In the nationally representative British E-Risk study, Jaffee and colleagues analysed 1,116 families: compared with families in which both mother and child had been maltreated, “cycle-breaking” families, in which the mother had been maltreated as a child but her child had not, showed more support and trust between the mother and her partner, more maternal warmth towards the child and less violence between the adults (Jaffee et al. 2013). Both studies are observational, and the authors note that further research is needed to establish whether these relationships play a causal role.
The state of the field. The introduction to the 2009 Developmental Psychology special section on the intergenerational transmission of parenting concluded that existing research shows significant but modest continuity across generations, and that high-quality work on the conditions under which parenting is not transmitted was still lacking (Belsky et al. 2009).
Evidence
Structured interventions. In the randomised controlled trial of Olds and colleagues, nurses visited first-time mothers (mostly young, unmarried or low-income) during pregnancy and infancy; a 15-year follow-up showed fewer verified reports of child abuse and neglect among the visited mothers than among controls (Olds et al. 1997). A systematic assessment of youth mentoring programmes found positive but modest average effects, with large differences between programmes and an important role for the quality and duration of the mentoring relationship (DuBois et al. 2011).
These studies point in the same direction: a secure partner, a highly responsive mentor and structured parenting support appearing in adulthood may weaken continuity from one generation to the next. Randomised trials are few and most of the evidence comes from longitudinal observational studies, so these should be seen as promising buffers, not as proven “antidotes”.
A Markov model with attenuation and filtering
Calculation
Simplify each generation’s parenting to two states: harsh or abusive (H) and not harsh (N). Let (borrowing Kaufman and Zigler’s rough “about 30 %” as an example) and (an illustrative background rate). A share of adults raised in H families have a buffer (a supportive partner, a mentor or a parenting programme), which removes a share (illustrative) of the excess risk. The lineage’s excess risk is multiplied each generation by .
# Two-state Markov chain for harsh parenting (H) across generations.
# P(H | parent H) = 0.30 (Kaufman and Zigler's rough "about 30 %", as an
# example); P(H | parent N) = 0.10 (ILLUSTRATIVE). A buffer (partner,
# mentor, programme) reaches a share phi of adults raised in H families and
# removes a share k = 0.7 (ILLUSTRATIVE) of the excess risk.
def chain(phi, k=0.7, p_hh=0.30, p_nh=0.10, gens=4):
eff = (1 - phi) * p_hh + phi * (p_nh + (p_hh - p_nh) * (1 - k))
lam = eff - p_nh; x, path = 1.0, []
for _ in range(gens):
x = x * eff + (1 - x) * p_nh; path.append(x)
return eff, lam, path, p_nh / (1 - lam)
print("lineage that starts with a harsh parent (generation 0)")
print(" buffer | P(H|H) | lam | P(H) in generations 1-4 | long-run P(H)")
for phi in (0.0, 0.3, 0.6, 0.9):
eff, lam, path, ss = chain(phi)
print(f" {phi:.1f} | {eff:.3f} | {lam:.3f} |"
f" {', '.join(f'{p:.3f}' for p in path)} | {ss:.3f}")
print("with no continuity at all (P(H|H) = P(H|N) = 0.10): long-run P(H) = "
f"{chain(0.0, p_hh=0.10)[3]:.3f}")
print("even with no buffer, 70 % of children of harsh parents do not become")
print("harsh parents")
# lineage that starts with a harsh parent (generation 0)
# buffer | P(H|H) | lam | P(H) in generations 1-4 | long-run P(H)
# 0.0 | 0.300 | 0.200 | 0.300, 0.160, 0.132, 0.126 | 0.125
# 0.3 | 0.258 | 0.158 | 0.258, 0.141, 0.122, 0.119 | 0.119
# 0.6 | 0.216 | 0.116 | 0.216, 0.125, 0.115, 0.113 | 0.113
# 0.9 | 0.174 | 0.074 | 0.174, 0.113, 0.108, 0.108 | 0.108
# with no continuity at all (P(H|H) = P(H|N) = 0.10): long-run P(H) = 0.100
# even with no buffer, 70 % of children of harsh parents do not become
# harsh parentsThe model shows several things. First, even without any buffer the cycle attenuates by itself: starting from a harsh parent, the risk is 0.30 in the next generation, 0.16 in the second and 0.13 in the third, quickly approaching the long-run level of 0.125, because only of the excess risk is kept in each generation. Second, the higher the buffer coverage, the lower the first generation’s risk (0.174 at a coverage of 0.9) and the faster the attenuation, and the long-run prevalence falls from 0.125 to 0.108. Third, even without a buffer, 70 % of the children of harsh parents do not become harsh parents.
The “filter” here is an analogy: a buffering relationship works like a filter that attenuates the negative signal and lets other signals through. It does not change what the child has already experienced; it changes the probability that the experience is “replayed” in the parenting of the next generation.
Note
The model’s simplifications. Real parenting is not binary; buffers are not randomly assigned (Chapter 3 and the Conger study in this chapter show that people’s choice of partners and environments is itself influenced by their experiences and characteristics); and the excess risk may come partly from genes and shared social circumstances (poverty, community violence), not only from the “copying” of parenting. The model’s parameters are illustrative; it shows only where buffers act and how their effect accumulates over generations.
Hypothesis · not established
Buffers act on the “automatic reaction” link (not yet established). One possible reason partners, mentors and parenting interventions weaken intergenerational continuity is that they give parents a model of, and support for, emotion regulation under stress that differs from their family of origin, so that the parents’ automatic reaction in moments of high pressure is replaced by another. A testable prediction: in longitudinal studies, the buffering effect of partner support on the continuity of harsh parenting should be partly mediated by parents’ observed reactions in stressful situations (rather than by their parenting knowledge). If the buffering effect is wholly mediated by parenting knowledge or economic conditions, or if parents’ stress reactions do not change with partner support, the hypothesis is weakened.
Chapter summary
Negative intergenerational cycles are real but weaker than the popular version: about a third, a modest meta-analytic effect, smaller in more rigorous studies; official records may also count surveillance bias as transmission. A partner’s warmth, support between mother and partner, warmth towards the child and less adult violence are associated with breaking the cycle, and home visiting and mentoring are supported by trials or systematic assessments. The Markov model shows that the cycle attenuates by itself, and faster the wider the buffers reach. The next chapter turns to classical family instruction.
References
Fan, Y. (2026). Early Experience, Epigenetic Methylation and the Limits of Intergenerational Transmission: From Glucocorticoid-Receptor Programming and Twin Variance Decomposition to Social-Learning Channels, Read Beside Classical Chinese Family Instruction [早期经验、表观遗传甲基化与跨代行为先验的传递边界] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23761337 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).