Skip to content
SA Gaming Machine Statistics

Analysis · Councils

Which council areas really differ from the state rate?

A council with 17 machines can swing a long way from one year to the next; one with 1,100 can’t. Ranking areas by NGR per machine treats both the same and over-reads the small ones. This page compares each area with the state rate of the same year and asks whether the difference is bigger than the area’s size would explain.

Gambling Help Line 1800 858 858 or gamblinghelponline.org.au.

State rate, council basis, FY 2024/25

$85,936

NGR ÷ machines at 30 June across the 48 published areas; the Statewide page divides by the year’s mean machines ($85,858)

Year-to-year scale c

1.05

SD of the log ratio × √machines; 437 degrees of freedom

Size check (slope)

−0.51

95% CI −0.68 to −0.34; the funnel assumes −0.50

Areas consistently above or below

14 / 34

of 55 areas with three or more years; 7 unclear (15 / 37 if years were independent)

Section 1

Funnel plot against the state rate

Each point is one area CBS published that year (combined groups stay whole). The lines fan out around the state rate: narrow for areas with many machines, wide for small ones. Most areas sit outside the year-to-year limits, which says the differences between councils are real and lasting, not small-number noise. Switch to the second set of limits to see which areas stand out even after allowing for that spread.

NGR per machine by council area

Nominal dollars of each year, so every area is compared with the state rate of the same year: the published areas’ NGR divided by their machines at 30 June, the same basis as each area’s own rate.
Control limits

Limits show how far an area of a given size usually moves from year to year by chance. Points outside them differ from the state rate by more than that noise.

  • Above the 99.8% limit (8)
  • Above the 95% limit (1)
  • Within the limits (9)
  • Below the 95% limit (3)
  • Below the 99.8% limit (27)
  • 95% limits
  • 99.8% limits

39 of 48 areas (81%) fall outside the 95% limits in FY 2024/25. If every area shared the state rate and the variance model held, about 2.4 (5%) would by chance; 39 or more would happen with probability < 0.001 (binomial, areas treated as independent). The areas are every published area, not a sample, so the count gets no confidence interval. Diamonds are combined CBS groups. Scale: c = 1.05; 2024/25 over-dispersion φ = 34.4.

Limits: state rate × exp(± z · c / √machines), with c pooled from every area’s year-to-year variation (FY 2013/14 to FY 2024/25, without FY 2019/20, which has no machine counts). The between-council spread τ² follows Spiegelhalter (2005), with z winsorised at 10%. Smallest area in FY 2024/25: Karoonda East Murray, Southern Mallee, 17 machines.

Why the limits use the log scale

Year-to-year swings are proportional: a rural area at a third of the state rate moves by a third as many dollars. The standard deviation of an area’s log ratio falls with size at a slope of −0.51 (95% CI −0.68 to −0.34, n = 44 areas with four or more years), close to the −0.5 that a 1/√machines variance implies. That check is what licenses the funnel’s shape.

What this is not

A council outside the limits is different from the state rate, not worse or better run. Venue mix (large metropolitan hotels against small country clubs), tourism and population all differ between areas, and none of them is in these data.

Section 2

Each area’s typical ratio, with year-to-year uncertainty

For every area published in three or more years, the geometric mean of its NGR per machine divided by the state rate, with a 95% t interval across its years. A ratio of 1.5× means the area’s machines typically took half as much again as the state average. Terracotta: consistently above the state rate; teal: consistently below; grey: the interval includes 1.

An area’s years are not independent: a council above the state rate one year tends to be above it the next. The pooled lag-1 autocorrelation of the log ratios within areas is 0.43 (365 pairs of consecutive years in 44 areas with four or more years), so every interval’s standard error is widened by √((1 + ρ) / (1 − ρ)) = 1.58. Treating the years as independent would call 15 areas above and 37 below instead of 14 and 34.

Area (as published by CBS) and yearsRatio (95% CI)
  1. Campbelltown3 years, FY 2022/23 to 2024/251.83×1.63 to 2.06
  2. Campbelltown, Tea Tree Gullygroup8 years, FY 2013/14 to 2021/221.77×1.69 to 1.86
  3. Playford11 years, FY 2013/14 to 2024/251.73×1.64 to 1.82
  4. Salisbury11 years, FY 2013/14 to 2024/251.70×1.67 to 1.73
  5. Tea Tree Gully3 years, FY 2022/23 to 2024/251.60×1.51 to 1.69
  6. West Torrens11 years, FY 2013/14 to 2024/251.50×1.42 to 1.59
  7. Marion11 years, FY 2013/14 to 2024/251.42×1.30 to 1.54
  8. Onkaparinga11 years, FY 2013/14 to 2024/251.38×1.34 to 1.43
  9. Charles Sturt11 years, FY 2013/14 to 2024/251.29×1.26 to 1.32
  10. Prospect, Walkervillegroup11 years, FY 2013/14 to 2024/251.25×1.19 to 1.32
  11. Gawler11 years, FY 2013/14 to 2024/251.14×1.09 to 1.19
  12. Whyalla11 years, FY 2013/14 to 2024/251.11×1.04 to 1.19
  13. Burnside, Mitchamgroup11 years, FY 2013/14 to 2024/251.11×1.04 to 1.18
  14. Port Adelaide Enfield11 years, FY 2013/14 to 2024/251.09×1.05 to 1.13
  15. Victor Harbor3 years, FY 2022/23 to 2024/251.07×0.96 to 1.19
  16. Murray Bridge3 years, FY 2022/23 to 2024/251.06×0.98 to 1.16
  17. Karoonda East Murray, Murray Bridge, Southern Malleegroup8 years, FY 2013/14 to 2021/220.99×0.92 to 1.07
  18. Norwood Payneham and St Peters11 years, FY 2013/14 to 2024/250.98×0.91 to 1.05
  19. Holdfast Bay11 years, FY 2013/14 to 2024/250.94×0.88 to 1.00
  20. Unley11 years, FY 2013/14 to 2024/250.91×0.87 to 0.95
  21. Renmark Paringa11 years, FY 2013/14 to 2024/250.85×0.77 to 0.94
  22. Grant, Mount Gambiergroup10 years, FY 2013/14 to 2024/250.83×0.80 to 0.85
  23. Port Pirie11 years, FY 2013/14 to 2024/250.78×0.74 to 0.82
  24. Port Lincoln11 years, FY 2013/14 to 2024/250.77×0.72 to 0.81
  25. Kangaroo Island, Victor Harbor, Yankalillagroup8 years, FY 2013/14 to 2021/220.74×0.68 to 0.82
  26. Mount Barker11 years, FY 2013/14 to 2024/250.74×0.70 to 0.78
  27. Port Augusta11 years, FY 2013/14 to 2024/250.73×0.68 to 0.79
  28. Adelaide Plains3 years, FY 2022/23 to 2024/250.72×0.48 to 1.07
  29. Barunga West, Copper Coastgroup11 years, FY 2013/14 to 2024/250.66×0.60 to 0.73
  30. Berri Barmera11 years, FY 2013/14 to 2024/250.61×0.59 to 0.64
  31. Kingston, Naracoorte Lucindale, Robegroup9 years, FY 2013/14 to 2024/250.59×0.49 to 0.71
  32. Adelaide Plains, Lightgroup8 years, FY 2013/14 to 2021/220.57×0.53 to 0.62
  33. Kangaroo Island3 years, FY 2022/23 to 2024/250.57×0.24 to 1.32
  34. Loxton Waikerie9 years, FY 2013/14 to 2024/250.56×0.53 to 0.59
  35. Adelaide11 years, FY 2013/14 to 2024/250.54×0.49 to 0.59
  36. Alexandrina11 years, FY 2013/14 to 2024/250.54×0.52 to 0.56
  37. Coober Pedy, Flinders Ranges, Roxby Downsgroup8 years, FY 2013/14 to 2021/220.53×0.44 to 0.64
  38. Light3 years, FY 2022/23 to 2024/250.52×0.34 to 0.80
  39. Coober Pedy, Flinders Ranges, Roxby Downs, Unincorporated Far Northgroup3 years, FY 2022/23 to 2024/250.51×0.42 to 0.60
  40. Ceduna, Elliston, Lower Eyre Peninsula, Streaky Bay, Wudinnagroup9 years, FY 2013/14 to 2022/230.49×0.43 to 0.56
  41. Barossa11 years, FY 2013/14 to 2024/250.47×0.43 to 0.52
  42. Clare and Gilbert Valleys11 years, FY 2013/14 to 2024/250.46×0.42 to 0.51
  43. Wattle Range11 years, FY 2013/14 to 2024/250.45×0.42 to 0.48
  44. Yorke Peninsula11 years, FY 2013/14 to 2024/250.41×0.37 to 0.46
  45. Coorong, Tatiaragroup11 years, FY 2013/14 to 2024/250.39×0.36 to 0.43
  46. Mid Murray11 years, FY 2013/14 to 2024/250.38×0.35 to 0.40
  47. Wakefield11 years, FY 2013/14 to 2024/250.38×0.31 to 0.46
  48. Yankalilla3 years, FY 2022/23 to 2024/250.34×0.22 to 0.51
  49. Adelaide Hills11 years, FY 2013/14 to 2024/250.33×0.29 to 0.37
  50. Mount Remarkable, Orroroo Carrieton, Peterboroughgroup8 years, FY 2013/14 to 2021/220.29×0.22 to 0.39
  51. Goyder, Northern Areasgroup11 years, FY 2013/14 to 2024/250.27×0.25 to 0.29
  52. Karoonda East Murray, Southern Malleegroup3 years, FY 2022/23 to 2024/250.27×0.13 to 0.54
  53. Cleve, Franklin Harbour, Kimba, Tumby Baygroup11 years, FY 2013/14 to 2024/250.26×0.22 to 0.31
  54. Mount Remarkable, Peterboroughgroup3 years, FY 2022/23 to 2024/250.23×0.20 to 0.26
  55. Unincorporated Far North, Unincorporated West Coastgroup8 years, FY 2013/14 to 2021/220.14×0.11 to 0.17

Ratios remove everything that moves the whole state in a year (inflation, the COVID-19 years, statewide policy), so the interval reflects the area’s own year-to-year variation. A combined group is one unit for as long as its membership is unchanged; when CBS changes a group, the new composition starts a new row. With three to eleven years per area, the t interval is used rather than a bootstrap. The widening uses the large-sample AR(1) factor and one pooled ρ for every area; the estimate from short series is biased towards zero, so the adjustment is more likely too small than too large. See DR-001 on combined groups.