How did the ‘whistle-to-whistle’ ban affect gambling advertising on TV? A live football matching study

Figures & data

Table 1. Game sections.

Section	Description	Categorisation by minute
1	Pre-game	Up to 5 min before kickoff
2	5-min before	5 min before kickoff to 12 min after kickoff.
3	First-half	12 to 44 min after kickoff.
4	Half-time	45 to 74 min after kickoff.
5	Second-half	75 to 100 min after kickoff.
6	Post-game	Greater than 100 min after kickoff.

Table 2. Descriptive statistics.

Variable		2018		2019		Total
		Freq	%	Freq	%	Freq	%
Total number of matches		468	45%	581	55%	1049	100%
Total number of matches by channel
	ITV	1	0.2%	7	1%	8	1%
	Sky	238	51%	249	43%	487	46%
	TNT	190	41%	198	34%	388	37%
	Other	39	8%	127	22%	166	16%
Total number of matches by day of the week
	Monday	33	7%	46	8%	79	8%
	Tuesday	56	12%	56	10%	112	11%
	Wednesday	45	10%	53	9%	98	9%
	Thursday	46	10%	47	8%	93	9%
	Friday	60	13%	62	11%	122	12%
	Saturday	109	23%	128	22%	237	23%
	Sunday	119	25%	189	33%	308	29%
Total number of matches by time of day^a
	Midday	74	16%	114	20%	188	18%
	Early afternoon	70	15%	100	17%	170	16%
	Early evening	107	23%	126	22%	233	22%
	Late evening	217	46%	241	41%	458	44%
Total frequency of adverts		2634	62%	1620	38%	4254	100%
Total frequency of adverts by channel
	ITV	18	1%	20	1%	38	1%
	Sky	2023	77%	1157	71%	3180	75%
	TNT Sports	580	22%	395	24%	975	23%
	Other	13	0%	48	3%	61	1%
		Mean	SD	Mean	SD	Mean	SD
Average adverts		5.8	5.30	2.9	3.50	4.2	4.60
Average programme length (mins)		154.90	32.70	151.60	33.10	152.9	32.70

^aMidday (up to 12:59); Early afternoon (13:00 to 16:59); Early evening (17:00 to 18:59); Late evening (19:00 onwards).

Figure 1. Average frequency of television gambling advertisements by game section and year.

Table 3. Inverse probability weighted (IPW) matching model results.

	Pre-game	5-min before	Half-time	Post-game	Total Programme
Post-ban (2019)	0.34***	−0.39***	−2.18***	−0.01	−2.30***
	[0.09,0.59]	[−0.44, −0.33]	[−2.32, −2.04]	[−0.22,0.20]	[−2.75, −1.84]
Pre-ban (2018)	1.74***	0.45***	2.43***	1.21***	5.29***
	[1.55,1.93]	[0.40,0.50]	[2.30,2.57]	[1.05,1.37]	[4.88,5.70]
ITV (comparator)
Sky	−1.93*	−2.03*	−2.03*	−2.03*	−2.02*
	[−3.98,0.12]	[−4.10,0.04]	[−4.11,0.05]	[−4.11,0.05]	[−4.09,0.05]
TNT Sports	−1.95*	−2.01*	−2.01*	−2.01*	−2.00*
	[−4.01,0.10]	[−4.08,0.07]	[−4.09,0.07]	[−4.09,0.07]	[−4.07,0.07]
Other	−2.27**	−0.87	−0.89	−0.89	−0.89
	[−4.39, −0.15]	[−2.97,1.22]	[−3.00,1.21]	[−3.00,1.21]	[−2.99,1.20]
Sunday (comparator)
Monday	0.18	0.21	0.21	0.21	0.18
	[−0.53,0.88]	[−0.36,0.78]	[−0.36,0.78]	[−0.36,0.78]	[−0.39,0.75]
Tuesday	−0.18	−0.21	−0.21	−0.21	−0.23
	[−0.78,0.42]	[−0.71,0.30]	[−0.71,0.30]	[−0.71,0.30]	[−0.73,0.28]
Wednesday	−0.23	−0.06	−0.07	−0.07	−0.08
	[−0.87,0.41]	[−0.58,0.45]	[−0.58,0.45]	[−0.58,0.45]	[−0.60,0.44]
Thursday	−0.06	−0.28	−0.30	−0.30	−0.31
	[−0.70,0.58]	[−0.81,0.25]	[−0.83,0.24]	[−0.83,0.24]	[−0.84,0.23]
Friday	−0.30	−0.26	−0.23	−0.23	−0.28
	[−0.91,0.32]	[−0.76,0.25]	[−0.74,0.28]	[−0.74,0.28]	[−0.78,0.23]
Saturday	−0.07	−0.26	−0.30	−0.30	−0.28
	[−0.54,0.39]	[−0.64,0.12]	[−0.68,0.08]	[−0.68,0.08]	[−0.66,0.10]
Midday (comparator)^a
Early afternoon	−0.19	−0.14	−0.16	−0.16	−0.14
	[−0.78,0.40]	[−0.58,0.31]	[−0.61,0.29]	[−0.61,0.29]	[−0.59,0.31]
Early evening	−0.21	−0.22	−0.24	−0.24	−0.22
	[−0.74,0.31]	[−0.62,0.18]	[−0.64,0.17]	[−0.64,0.17]	[−0.62,0.18]
Late evening	−0.30	−0.37*	−0.40*	−0.40*	−0.35*
	[−0.86,0.25]	[−0.77,0.04]	[−0.81,0.02]	[−0.81,0.02]	[−0.76,0.06]
Constant	2.33**	2.42**	2.45**	2.45**	2.42**
	[0.24,4.42]	[0.33,4.51]	[0.35,4.54]	[0.35,4.54]	[0.33,4.51]
Observations	736	1049	1042	1042	1045

Models report unstandardized coefficients; 95% confidence intervals in brackets; ^aMidday (up to 12:59); Early afternoon (13:00 to 16:59); Early evening (17:00 to 18:59); Late evening (19:00 onwards).

*p < 0.1, **p < 0.05, ***p < 0.01.

Figure 2. Density plot of covariate balance (channel the game was televised on) in the IPW models.

Figure 3. Density plot of covariate balance (day of the week the game was televised) in the IPW models.

Figure 4. Density plot of covariate balance (time the game was televised) in the IPW models.

Table 4. Balance of covariates in IPW models.

	Observations^a				Balance Test^b
	Raw		Weighted
	Treated	Control	Treated	Control	p-value
Pre-game	381	355	368.2	367.8	0.98
5-min before	581	468	527	522	0.21
Half-time	576	466	523.4	518.6	0.21
Post-game	576	466	523.4	518.6	0.21
Total programme	578	467	524.9	520.1	0.22

^aThe number of observations is the number of live games; ^bBalance test is a Chi-Squared test where H0:Covariates are balanced between treatment and control groups in the IPW model.

Data sharing agreement

Advertising data used in this study is not available for sharing due to licensing restrictions. However, STATA do-files and log-files are available on request from the corresponding author (EM), in addition to the scraped kickoff dataset.