Data Analysis Australia Pty Ltd
How Often Does a Night Out Lead to an Assault
When considering how safe our society might be and the publicity given to alcohol-related assaults, it is worth asking the question of what the risk might be of assault on a typical night out.
In one sense this is a simple calculation of the ratio:
Number of alcohol-related assaultsNumber of nights out
However this calculation is beset with complications:
1. What is the information for the numerator, the number of alcohol-related assaults;
2. What is the information for the denominator, the number of “person night-outs”; and
3. How can this estimate be used?
I now discuss these in turn.
Number of Alcohol Related Assaults
A commonly quoted reference for the number of alcohol related assaults in Australia is Laslett et al 2010.[1] This estimates the number Australia wide to be around 70,000. Some comments can be made about this estimate:
· Several information sources are used by Laslett et al, namely:
○ The 2007 National Drug Strategy Household Survey conducted by the Australian Institute of Health and Welfare (AIHW);
○ The 1995 Personal Safety Survey conducted by the Australian Bureau of Statistics (ABS); and
○ Various Police data sources.
· These sources do not use consistent definitions and are hence not strictly comparable. For example, the National Drug Strategy Household Survey asked persons whether in the past 12 months any person affected by alcohol verbally abused them; physically abused them, or put them in fear. Respondents could respond yes or no to each of these. In contrast, the Personal Safety Survey asked whether in the past 12 months they had experienced various types of alcohol-related violence, including physical assault, physical threats, sexual assault and sexual threats. The Police data essentially counted assaults using a legal definition.
· Both surveys have been updated, with the personal safety survey repeated in 2012 and the national Drug strategy Survey in 2010. However the findings are largely similar.
· There is a fundamental difference between how the surveys collected information and how the Police did. The surveys estimated the number of people who had experienced such an event at least once, while the Police estimated the number of such events. Since a person may experience more than one such event, the surveys can be expected to give lower estimates.
· The events themselves are fundamentally different. It is unlikely that the criteria used by trained Police – both for an assault and for it being alcohol-related – are the same as for untrained members of the public.
In this context it is worth referring to a caution expressed by the ABS regarding such statistics:
These statistics are not designed to provide counts of either the total number of victims, nor the total number of individual offences that come to the attention of police as:
· The same victim may be counted more than once in incidents involving multiple offences where these offences are of different types (i.e. belong to different Australian and New Zealand Standard Offence Classification (ANZSOC) divisions) or if the same person is a victim on more than one occasion in the same reference year and reports these incidents to police on separate occasions.
· Conversely, for an incident involving multiple offences that belong to the same ANZSOC division offence category, only the most serious offence within that ANZSOC division will be counted. For further information refer to paragraphs 26-32.[2]
The same ABS publication provides a number of examples of how the counting of offences (and victims) can be complex.
Given the above issues, it is still necessary to consider how many relevant assaults may have taken place in a typical year. The 2007 National Drug Strategy Household Survey and the 1995 Personal Safety Survey provide estimates of the rate of experiencing alcohol-related physical assault in a twelve month period as 4% and 3.3% respectively. Given the population of Australia aged 18 or above is approximately 19,000,000, this suggests between 630,000 and 760,000 people would be the victim of an alcohol related incident in a given year.
This estimate is very much larger than the estimates provided by the Police data actually used in Laslett et al. Laslett uses an estimate of 166,507 assaults of all types in 2005, and suggests that between 41.7% and 44.1% are alcohol-related.[3] It is not clear in Laslett et al exactly what source was used for the total number of assaults beyond that it was from the ABS. Given the potential for confusion between assaults (the count of victim-perpetrator pairs), victims or offenders, it is not an unreasonable figure for reported assaults.
Laslett et al have some concerns regarding the proportion of alcohol related assaults based upon the victim surveys and the Police data from New South Wales and Western Australia. I believe that this discrepancy is likely to be explained by the difference in what is being measured. In the surveys the number of people who had at least one assault was compared with the number who had at least one alcohol related assault in the 12 month period. Since a person with at least one alcohol related assault may also have been the victim of some non-alcohol related assaults, I would expect that the ratio from the survey would be higher than the ratio from the Police data where actual assaults are directly counted.
There remains a factor of ten difference in the two types of estimate. Some but not all of this might be explained by the fact that not all assaults are reported (the surveys suggest that perhaps 40% are not reported). Of those reported, some are not proven or result in a conviction, and depending upon Police practice that is not consistent across Australia, might not be recorded as an unsolved offence. It is likely that a number of the events reported in the surveys (particularly the “physical abuse” in the National Drug Strategy Household Survey) might not pass the threshold be to an assault. However it is reasonable to expect that the number of alcohol related assaults is greater than the 70,000 suggested by Laslett et al, I would suggest that it might be between 150,000 and 600,000.
Not all assaults relate to “a night out”. In fact approximately half of all assaults appear to take place in a residential setting.[4] This suggests the number of assaults away from a residential setting to be between 75,000 and 300,000. Hence taking these factors into account, the Laslett et al estimate of 70,000 is reasonable for ‘alcohol related assaults away from residential premises”, although the actual figure could be somewhat higher. The number of assaults that might be considered serious could be much lower and since they are more likely to be reported and followed up by Police, the Police data may well be a better guide to these. In that case, an estimate of 35,000 such assaults away from a residential setting may be appropriate.
Number of Nights Out
Obtaining an estimate of how many “nights out” across Australia is not straightforward. This type of activity is not regularly monitored and escapes many statistical collections:
· The Census records where people reside and, with less precision, where they work. It does not record where they may spend their time.
· A night out might be considered to be a form of tourism activity, but standard definitions of a daytrip used in surveys such as the National Visitor Survey (NVS) and the International Visitor Survey (IVS) specifically exclude trips that do not include “four hours during daylight”.
· Transport surveys usually track activities, but the emphasis it typically on activities that contribute to transport demand peaks. While some information may be contained in such data, it is not readily available.
Collectively these do not provide good coverage. For example, transport surveys rarely cover visitors from outside the area being surveyed, and visitor surveys rarely collect fine detail on where people go.
In addition to these problems, the behaviour of Australians has been rapidly changing in recent decades. Australians are more likely to eat out and in most states there has been an increase in small bars and other entertainment venues. To some extent the need to plan for such changes does provide some sources of relevant information. The City of Melbourne has provided a series of reports that are often quoted in this respect. These suggest that on an average night there are 305,000 people in the City, and of these 84,000 are residents and 97,000 are visitors.[5] This suggests that there are 35.4 million night time visits (97,000 × 365) to the City each year, excluding people who go there for work or study at night.
The problem comes in how to scale up from such an estimate for the City of Melbourne to Australia. The City draws most of its visitors from the Melbourne metropolitan area, but it is not the only destination for night time entertainment and other activities in metropolitan Melbourne. The reality is that it probably amounts to less than half in a large metropolis such as Melbourne where many people will attend venues closer to home or where they work. Hence an estimate at 70 million might be reasonable, but it could well be somewhat more.
Scaling up from Melbourne to Australia might be done conservatively by using the state population.[6] This suggests a total number of visits of at least 280 million. Again this is a very conservative figure, corresponding to about 15 nights out for each Australian adult.
The Ratio
Overall this suggests a ratio of around 0.000125 or 0.0125% (corresponding to 35,000 alcohol-related assaults in a non-residential setting) to 0.0011 or 0.11% (corresponding to 300,000 assaults).
It is worth considering how these estimates might be used. What it does say is that the vast majority of nights out do not lead to assault for the individual.
Clearly the perceptions of the danger are sometimes greater than this. This is probably due to the group effect (“someone in the group I was in thought they were assaulted”) as well as stories spreading.
Naturally people who have many nights out will have a greater probability of being sampled at some stage, and other factors not considered here will have an impact upon these probabilities, making them higher for some, lower for others.
~ Page 5 ~ April 2014
[1] Laslett, A-M., Catalano, P., Chikritzhs, Y., Dale, C., Doran, C., Ferris, J., Jainullabudeen, T., Livingston, M, Matthews, S., Mugavin, J., Room, R., Schlotterlein, M. and Wilkinson, C. (2010) The Range and Magnitude of Alcohol’s Harm to Others. Fitzroy, Victoria: AER Centre for Alcohol Policy Research, Turning Point Alcohol and Drug Centre, Eastern Health.
[2] Australian Bureau of Statistics, 4510.0 - Recorded Crime - Victims, Australia, 2012 Explanatory notes.
[3] The Australian Institute of Criminology provides a similar estimate of “recorded assaults”, 176,427 in 2007.
[4] ABS 4510.0 Recorded Crime - Victims, Australia, 2012
[5] City of Melbourne, (2009). Melbourne City user Estimates and Forecasts, 2004-2020.
[6] ABS Catalogue 3102.1, Demographic Statistics, Sep 2013, indicates that Victoria had a population of 5,768,600 out of an Australian population of 23,235,800, that is 24.8%