A LITTLE (MORE) KNOWLEDGE: COMMENTS ON WOOD <i>ET AL</i>. (2006)

Thomas Huxley responded to Pope's famous aphorism, ‘A little knowledge is a dangerous thing’ by asking who has so much as to be out of danger. Following concerns raised in Wood (2005) regarding claims about the causes of the decline in the Australian heroin supply, Wood et al. (2006) presents evidence that, they claim, is incompatible with the view that heroin seizures might have been instrumental in precipitating the reduction of heroin supply in Australia. This new evidence is a welcome contribution to the debates on this and related issues. A thoughtful review of candidate explanations for the Australian heroin ‘shortage’ in Degenhardt et al. (2005) was predicated in part on the belief that no similar decline had occurred at the same time in other relevant heroin markets (specifically, Canada or Hong Kong). Assuming that the indicators presented by Wood et al. (2006) are pointing to a decline in that country's supply around the same time as the downturn in Australia, even if no such decline occurred in Hong Kong, a reconsideration of explanations for the downturn in Australian heroin supply is merited. More specifically, granting that assumption raises the following questions: Do the Australian and Canadian downturns coincide? Do they have similar magnitude and dynamics? Are they likely to have been produced by identical causes? What implications may be drawn regarding the relationship between drug seizures and supply? What kinds of additional evidence would be most useful for advancing our understanding of heroin market dynamics and influences on them? Starting with the first question, the simple before–after comparisons made by the authors provide a somewhat overly consensual picture of how the trends have played out. Panel A (overdose deaths in British Columbia) and Panel C (kilograms of heroin seized in Canada per year) show a marked decline from 1998 to 1999, before the estimated start of the Australian decline in heroin supply. Panel A's decline appears to precede Panel B’s, and Panel C's trend is non-monotonic. Without finer-grained series than annual summaries, it is difficult to ascertain the extent of overlap between the Australian and Canadian supply downturns. Regarding the second question, indicators of the decline in Australia (Smithson et al. 2004, 2005) are large compared to the Canadian indicators. The decline in purity was about 65% and the decline in heroin-related ambulance call-outs about 82%. Without time-series analysis of the Canadian data, however, it is impossible to make any comparisons regarding dynamic properties. The third and fourth questions are more difficult to address. As pointed out in Degenhardt et al. (2005), a number of factors can contribute to a decline in heroin supply. The Canadian findings directly contradict the Australian findings only if we assume the relationship between law enforcement and heroin supply to be biconditional. While it is possible that both the Australian and Canadian downturns had identical causes, it is also arguable that at least some contributing factors were unique to each case. For instance, the larger magnitudes of declines in the Australian supply indicators could reflect greater noise in the Canadian indicators, but it could also be due to a contribution to the Australian downturn from increased heroin seizures. Finally, the status of the finding in Smithson et al. (2005) of a short-term feedback relationship between heroin seizures and purity is unaffected by the arguments in Wood et al. (2006) and still needs explanation. As for the fifth question, there are two short-term enhancements and a longer-term suggestion that spring to mind. First, if earlier data are available for any of the four indicators (i.e. prior to 1997) in Wood et al. (2006) then their inclusion in analyses would greatly enhance the assessment of the apparent declines in those indicators. For instance, Smithson et al. (2004, 2005) presented the entire available series for all relevant data such as seizures (from 1987 onward) or crime rates (e.g. burglary and robbery from 1993 onward), and indeed they provided crucial contextual information. Secondly, Wood et al. (2006) states that the naxolone utilization statistics were limited to fiscal-year summaries, but it is not clear whether similar limitations applied to all the other indicators. Before–after comparisons are not an effective way of analysing time-series data. They obscure the timing of the onset and/or cessation of a trend, non-monotonic fluctuations therein, and the autocorrelation structure in the series. If any indicator data are available in more disaggregated form than annual totals, a re-analysis of those data using time-series methods would yield two advantages: we would gain an understanding of trends and autocorrelation structures in the data, and short-term predictive relationships between the indicators could be identified. In the longer term, when and if time-series data on aspects of heroin and other drug markets in Australia and Canada are available in sufficiently disaggregated form, an excellent initiative would be to combine them in an international database. That would enable researchers to directly (and collaboratively) address questions about how these markets are interrelated and whether their internal dynamics are driven by the same influences. We may then know a little more.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.