Table 4.

Series of regression models explaining/predicting ‘violence perpetration’

Covariates from individual block sub-models	Covariate category	Any violence perpetration % For covariate category	Model 1 (“Children” covariate only) n = 164		Model 2 (+ Socio-demographics) n = 162		Model 3 (+ Socio-demographics + Psycho-social factors) n = 155		Model 4 (+ Socio-demographics + Psycho-social factors + Gambling) n = 155
			Odds ratio (95% CI)	p-value	Odds ratio (95% CI)	p-value	Odds ratio (95% CI)	p-value	Odds ratio (95% CI)	p-value
Children aged < 18 years usually living in household by Gender of Gambler	Male, no children (n = 65)	30.77	1 (reference)	0.01	1 (reference)	0.03	1 (reference)	0.26	1 (reference)	0.12
	Female, no children (n = 39)	43.59	1.74 (0.76–3.96)		1.58 (0.66–3.79)		1.24 (0.40–3.86)		1.35 (0.41–4.48)
	Female, with children (n = 31)	67.74	4.72 (1.89–11.84)		4.20 (1.59–11.07)		3.20 (0.99–10.39)		4.36 (1.28–14.87)
	Male, with children (n = 29)	44.83	1.83 (0.74–4.50)		2.47 (0.93–6.52)		1.76 (0.51–6.09)		1.77 (0.49–6.46)
Socio-demographics (block 1)
Asian	No (n = 133)	49.62			1 (reference)	0.02	1 (reference)	0.001	1 (reference)	0.001
	Yes (n = 29)	17.24	–		0.28 (0.096–0.81)		0.12 (0.03–0.42)		0.12 (0.03–0.44)
Maori	No (n = 131)	37.40			1 (reference)	0.01
	Yes (n = 31)	70.97	–		3.30 (1.33–8.16)		(Not significant here)		(Not significant here)
Psycho-social factors (block 2)
Aggression questionnaire— Anger subscale (> 3.5, <=6 vs <=3.5)	≤ 3.5 (n = 41)	17.07					1 (reference)	0.001	1 (reference)	0.002
	>3.5, ≤ 6 (n = 44)	40.91	–		–		2.84 (0.81–9.96)		2.40 (0.64–8.92)
	> 6, ≤ 10 (n = 47)	51.06	–		–		8.85 (2.29–34.26)		7.54 (1.86–30.59)
	> 10 (n = 32)	68.75	–		–		20.50 (4.32–97.24)		20.69 (4.11–104.01)
Difficulties in emotion regulation—Strategies	≤ 10 (n = 46)	28.26					1 (reference)	0.002	1 (reference)	0.003
	> 10, ≤ 15 (n = 38)	60.53	–		–		3.76 (1.11–12.76)		3.30 (0.93–11.70)
	> 15, ≤ 22 (n = 44)	29.55	–		–		0.30 (0.09–1.05)		0.25 (0.07–0.93)
	> 22 (n = 36)	61.11	–		–		1.43 (0.37–5.55)		1.28 (0.32–5.22)
Drugs issue in family/whanau in last 12 months	No (n = 137)	37.96					1 (reference)	0.02	1 (reference)	0.03
	Yes (n = 23)	69.57	–		–		4.53 (1.27–16.24)		4.58 (1.20–17.44)
Interpersonal Support Evaluation List—Tangible subscale	≥ 10, < 12 (n = 32)	53.13					1 (reference)	0.01	1 (reference)	0.02
	< 7 (n = 33)	54.55	–		–		0.53 (0.13–2.14)		0.66 (0.16–2.81)
	≥ 7, < 10 (n = 33)	30.30	–		–		0.09 (0.02–0.40)		0.10 (0.02–0.44)
	≥ 12 (n = 63)	38.10	–		–		0.33 (0.11–1.01)		0.35 (0.11–1.11)
Gambling (block 3)
Received counselling or medication for gambling in last 12 months	No (n = 127)	36.22							1 (reference)	0.03
	Yes but not now (n = 21)	66.67	–		–		–		3.81 (1.07–13.59)
	Yes currently (n = 16)	68.75	–		–		–		4.61 (0.92–23.10)
Adjusted R-squared			0.09		0.22		0.49		0.53

The gambling “block” model for violence perpetration (which just had covariates “Children aged < 18 years usually living in household by Gender of Gambler”, “Received Counselling or medication for gambling in last 12 months”, and “Pub or club EGMs main type”) had an odds ratio for violence perpetration comparing “Pub or club EGMs is the main type” to “Pub or club EGMs is NOT the main type” of 2.25 (1.09–4.64) with a significant p-value of 0.03. Had we included “Pub or club EGMs main type” in Model 4 for violence perpetration, which takes into account socio-demographic and psychosocial factors, we would have reduced the “Pub or club EGMs main type” odds ratio to an insignificant (with p = 0.9) 1.07 (0.39–2.93); it was therefore not included in Model 4