Social mediators of relationships between childhood reading difficulties, behaviour problems, and secondary school noncompletion

Figures & data

Table 1. Group and ATP cohort means, SDs (in brackets), and significant differences between groups in RDs and BPs at 7–8 years (raw data)

	Neither	BPs‐only	RDs‐only	RD + BP	ATP cohort
Parent reported behaviour problems^2
Hostile‐aggressive^3 ^,^4	0.10 (0.12)	0.72 (0.42)	0.16 (0.20)	0.70 (0.50)	0.36 (0.35)
Anxious‐fearful^3 ^,^4 ^,^5	0.13 (0.16)	0.81 (0.45)	0.23 (0.26)	0.58 (0.38)	0.44 (0.38)
Hyperactive ^3 ^,^4 ^,^6	0.07 (0.14)	0.69 (0.57)	0.21 (0.24)	0.91 (0.53)	0.43 (0.49)
Teacher reported behaviour problems^2
Hostile‐aggressive^3 ^,^4	0.01 (0.03)	0.52 (0.38)	0.02 (0.05)	0.47 (0.44)	0.18 (0.32)
Anxious‐fearful^3 ^,^4	0.08 (0.12)	0.79 (0.49)	0.12 (0.15)	0.67 (0.42)	0.38 (0.40)
Hyperactive^3 ^,^4 ^,^6	0.00 (0.00)	0.59 (0.55)	0.02 (0.08)	0.87 (0.65)	0.30 (0.48)
ACER word knowledge reading test^7 ^,^8	17.72 (1.61)	17.03 (1.42)	4.54 (3.04)	4.39 (3.10)	14.08 (4.97)
N	125	66	38	38	1,601 (P) 1,264 (T) 1,224 (WT)

ATP = Australian Temperament Project; BP = behaviour problems; P = parent report; RD = reading difficulties; T = teacher report; WT = Word Knowledge Reading Test.

Possible range of 0–2.

Omnibus ANOVA results comparing the RD + BP, RD, BP and Neither groups: p < .001.

Scheffé comparisons: the BP and RD + BP groups were significantly higher than the RD and Neither groups by parent and teacher reports (p < .001).

Scheffé comparisons: the BP group was significantly higher than the RD + BP group by parent report (p < .001).

Scheffé comparisons: the RD + BP group was significantly higher than the BP group (p < .03 by parent report; p < .003 by teacher report).

Possible range of 0–20.

Scheffé comparisons: the RD and RD + BP groups were significantly lower than the BP and Neither groups (p < .001).

Table 2. Group and ATP cohort means⁹, SDs (in brackets), and significant differences in school engagement, and family and peer attachment. Pooled estimates for the childhood groups, raw data for the ATP cohort

	Neither	RDs‐only	BPs‐only	RD + BP	ATP cohort
School involvement^10	1.30 (0.29)	1.42 (0.31)	1.65 (0.46)	1.87 (0.57)	1.48 (0.44)
Teacher relationships^11	1.36 (0.42)	1.33 (0.35)	1.71 (0.59)	1.66 (0.68)	1.51 (0.53)
Family attachment^12	1.93 (0.69)	1.88 (0.59)	2.27 (0.86)	2.20 (0.78)	2.05 (0.74)
Peer attachment^13	2.02 (0.60)	1.96 (0.57)	2.38 (0.74)	2.26 (0.61)	2.18 (0.67)

Possible range on school involvement and teacher relationships = 1–4; on family and peer attachment = 1–5. Higher scores reflect poorer school involvement, teacher relationships, and family and peer attachment. ATP = Australian Temperament Project; BP = behaviour problems; RD = reading difficulties.

F (3, 13,593) = 1,334.41, p < .0001.

F (3, 13,593) = 502.76, p < .0001.

F (3, 13,580) = 214.86, p < .0001.

F (3, 13,581) = 320.03, p < .0001.

Table 3. Logistic regression results showing unadjusted and adjusted associations between childhood RDs, BPs, and secondary school noncompletion (pooled estimates)

	Unadjusted model 1		Adjusted model 2 – with covariates		Adjusted model 3 – school involvement		Adjusted model 4 – teacher relationships		Adjusted model 5 – family attachment		Adjusted model 6 – peer attachment		Adjusted model 7 – full model
	OR	95% CIs	OR	95% CIs	OR	95% CIs	OR	95% CIs	OR	95% CIs	OR	95% CIs	OR	95% CIs
Neither (reference group)	1.00	—	—	—	—	—	—	—	—	—	—	—	—	—
BPs‐only	4.91^**	1.77–13.65	3.97^*	1.29–11.32	2.18	0.67–7.01	3.01	0.97–9.30	3.38^*	1.12–10.72	3.25^*	1.06–10.00	2.18	0.67–7.08
RDs‐only	4.37^*	1.30–14.68	3.61^*	0.90–10.94	2.71	0.74–9.90	3.59	0.99–12.92	3.28	0.93–11.58	3.22	0.92–11.32	2.79	0.75–10.40
RD + BP	22.99^***	8.03–65.80	15.98^***	4.50–42.95	6.76^**	1.95–23.49	13.47^***	4.21–43.07	13.31^***	4.24–41.72	13.74^***	4.41–42.86	7.02^**	1.97–25.07
Mediators					School involvement		Teacher relationships		Family attachment		Peer attachment		School involvement
—	—	—	—	—	6.15^***	2.35–16.12	2.72^**	1.29–5.73	1.49	0.89–2.48	1.62	0.78–3.39	5.54^**	1.59–19.33
													Teacher relationships
—	—	—	—	—	—	—	—	—	—	—	—	—	1.14	0.42–3.15
Covariates
Child gender	—	—	0.51	0.23–1.14	0.67	0.28–1.61	0.61	0.26–1.42	0.48	0.21–1.08	0.63	0.26–1.55	0.67	0.28–1.62
Mother age	—	—	1.35	0.83–2.19	1.40	0.82–2.41	1.42	0.84–2.39	1.40	0.85–2.32	1.38	0.83–2.28	1.41	0.82–2.41
Mother education	—	—	2.25^***	1.41–3.58	2.14^**	1.31–3.51	2.32^***	1.44–3.75	2.30^***	1.43–3.70	2.33^***	1.44–3.76	2.14^**	1.31–3.51
Mother NESB	—	—	1.23	0.13–11.28	1.11	0.12–10.03	1.09	0.13–9.51	1.05	0.12–9.01	1.19	0.14–10.09	1.11	0.12–10.02
Single parent	—	—	1.30	0.31–5.37	1.29	0.29–5.73	1.38	0.33–5.85	1.16	0.29–4.65	1.31	0.33–5.30	1.31	0.29–5.86
Number children	—	—	1.07	0.66–1.74	1.11	0.66–1.01	1.14	0.69–1.89	1.10	0.68–1.81	1.08	0.66–1.76	1.11	0.66–1.87

BP = behaviour problems; CIs = confidence intervals; NESB = non‐English speaking background; OR = odds ratio; RD = reading difficulties.

* p < .05.

** p < .01.

*** p < .001.

Table 4. Mediation of RD + BP, BPs‐only, and RDs‐only group effects on secondary school noncompletion

	Lowest 95% bias‐corrected bootstrapped CI across the 50 imputed datasets	Highest 95% bias‐corrected bootstrapped CI across the 50 imputed datasets
School involvement
BPs‐only^19	0.13	1.23
RDs‐only	−0.09	0.63
RD + BP^19	0.28	0.43
Teacher relationships
BPs‐only^19	0.03	0.74
RDs‐only	−0.32	0.21
Rd+BP	−0.06	0.62
Family attachment
BPs‐only	−0.11	0.60
RDs‐only	−0.35	0.21
RD + BP	−0.09	0.56
Peer attachment
BPs‐only	−0.15	0.58
RDs‐only	−0.47	0.14
RD + BP	−0.30	0.43

BP = behaviour problems; CIs = confidence interval; RD = reading difficulties.

As the 95% bias‐corrected bootstrapped CIs do not cross zero, a significant mediation effect was observed.