Zweifel et al
- Zweifel B.
- Procter E.
- Techel F.
- Strapazzon G.
- Boutellier R.
Risk of avalanche involvement in winter backcountry recreation: the advantage of small groups.
investigate the relationship between avalanche risk and group size. They find that groups of 4 or more have a higher relative risk of an avalanche accident than individuals or groups of 2. Multiple websites popular among backcountry skiers have interpreted this finding as evidence that solo backcountry skiing might be safer than being in a group., ,
These commentators seem to have overlooked that Zweifel et al focus on the relative risk of a group being an involved in an accident. This is not the same as the relative risk of an individual in a group being involved in an accident. In fact, the empirical findings of Zweifel et al indicate that the risk of an individual being the victim of an avalanche accident is approximately independent of group size.
Suppose that the probability that an individual is involved in an avalanche is independent of group size. If we say that a group was involved in an avalanche if 1 or more individuals in the group were involved, then we would expect the probability that a group has an accident to be roughly proportional to group size. To be exact, if the probability of an individual accident is p, then the probability that a group of size n has an accident is 1−(1−p)n≈np. The relative risk of a group of size n compared with a group of size 2 is 1(1p)n/1(1p)2≈2n. The relative risk estimates of Zweifel et al very closely follow this pattern. For almost all group sizes and data sources, the 95% confidence interval for relative risk contains the relative group risk that one would expect if individual accident risk were independent of group size (the 2 exceptions are Davos self-registration with a group of size 4 and Tyrol with a group of 1).
Given that this article has been of interest to many backcountry skiers, it is important to communicate the results in manner that is easy for anyone to understand and incorporate into their decision making. I would emphasize the following: The results suggest that the probability that an individual gets in an accident is approximately independent of group size. The probability that someone else in a group is involved in an accident increases with group size, but not any more than would be expected from adding independent events. It is approximately equally likely that a person among 3 groups of 2 will have an accident as it is that a person in 1 group of 6 will. This is somewhat counterintuitive in that the 3 groups of 2 are likely to choose different routes, and so would be exposed to more potential avalanche terrain than a group of 6 that travels together. In this sense, there is some evidence in favor of the conventional wisdom that larger groups make poorer decisions. On the other hand, individual accident risk appears to be about the same regardless of group size, and a larger group would likely be more effective for companion rescue.
Published online: July 05, 2016
All rights reserved.