Assumptions need to be made about four things: the prices paid by drug users; the average amount consumed on a day in which consumption occurs; the average purity of each drug; and the average annual number of days’ consumption for each category of consumption frequency. In addition to these mean values, assumptions also need to be made about their degree of variation between individuals. Tables 4.10 and 4.11 set out these assumptions. Given the thin evidence available, they necessarily contain a large element of guesswork. The assumptions about average prices are based loosely on the NCIS national averages for the period April 2003-April 2004 (Table 4.5). That source gives little basis for an assumption about the margin of error associated with the NCIS average price. Evidence from elsewhere (for example the analysis by Reuter and Caulkins, 2004, of the US STRIDE data) suggests a wide range of price variation between individual deals, so this suggests a large element of sampling error in the NCIS average, quite apart from other, more systematic, error. In Table 4.9 the authors therefore specify rather wide error margins.
Mean quantities per day of use are based on the range of fragmentary evidence summarised in the previous section. Mean purity is based very loosely on figures summarised in Table 4.6 above. The ranges of variation given in Table 4.9 below are to be interpreted as analogous to 90 per cent confidence intervals. However, they are arbitrary and are used only to give a rough indication of the likely range of estimation uncertainty.
Table 4.9: Assumed values for mean price, quantity and purity, with assumed margins of error
Drug Mean price a Mean quantity per day of
use for intensive usersa,b Mean purity
c,d
Cannabis £2.50 ± £0.75 1.2 ± 0.4 (intensive)
0.55 ± 0.4 (non-intensive) -
Amphetamines £8.50 ± £2.50 1.0 ± 0.2 (all users) 11% ± 4%
Ecstasy £4.50 ± £1.50 2.0 ± 0.5 (intensive) 1.5 ± 0.5 (non-intensive) 26% ± 10% (65±25mg per tab) Cocaine £55.00 ± £10.00 0.8 ± 0.2 (intensive) 0.55 ± 0.2 (non-intensive) 50 ± 6% Crack £95.00 ± £15.00 0.7 ± 0.2 (intensive) 0.4 ± 0.2 (non-intensive) 65 ± 7% Heroin £60.00 ± £10.00 0.48 ± 0.1 (intensive) 0.34 ± 0.1 (non-intensive) 40% ± 6% Notes: (a) all quantities in grams, except for ecstasy in tabs; (b) intensive users defined as using at least three times a week; (c) there is no purity factor for cannabis, since it is not usual to measure cannabis quantities in terms of the active constituent (THC); (d) estimate of the mean MDMA content of ecstasy tabs supplied by NCIS
A further important issue is the impact of treatment on consumption rates. Addicts often spend a large (and recently increasing) amount of time in treatment, so this is a potentially important factor. Table 4.10 shows that the AS gives little evidence of a difference in consumption levels per use-day associated with participation in a drug treatment
programme.33 However, there are differences in reported frequency of use, which are significantly lower for arrestees in treatment. This finding suggests that use of reported frequency as a predictor will automatically account for the impact that treatment has in suppressing drug use, so no adjustment is made for the impact of treatment on the rate of use per use-day.
Table 4.10 Distribution of self-reported quantities consumed from AS data
(weighted for survey design and non-response; sample trimmed to give reported quantity between 0.1 and 2 grams; standard errors in parentheses)
Mean consumption per use-day by treatment status
Heroin Powder
cocaine
Crack
currently in treatment 0.95 (0.04) 1.34 (0.36) 0.97 (0.11)
not currently in treatment 0.82 (0.04) 1.11 (0.03) 0.84 (0.03) Proportion of users consuming at least
three times a week
Heroin Powder
cocaine
Crack
currently in treatment 0.09 (0.01) 0.11 (0.01) 0.11 (0.01)
not currently in treatment 0.48 (0.02) 0.28 (0.09) 0.32 (0.07)
A further set of assumptions is needed to translate the verbal descriptions of frequency of use, which are embedded in the OCJS and AS survey questions, into numerical ranges for the number of days’ consumption per year. There is an important issue about the interpretation of reported frequencies of use. Survey recall over a period as long as a year and the assessment of “usual” frequencies of consumption are both known to be potentially unreliable. In general specific questions about actual behaviour in a relatively short, recent reference period are to be preferred. This has not been investigated very fully for surveys of drug use. However, Johnson et al. (1985, Tables 3-2 and 3-3) give some comparisons between claimed expenditures over the past year and monitored expenditures over a shorter reference period for a group of US heroin users. They found that reported annual consumption was seriously overstated: by a factor of two on average, with the effect being particularly strong for irregular users. Daily users were found to overstate their rate of expenditure by around 47 per cent. The discrepancies were considerably reduced when reported usual daily expenditure rates were compared with reported expenditures per use- day. In their enquiry, heroin-using subjects appeared to be liable to “forget about days without heroin use and, perhaps, days with low heroin use”. Thus, the reported AS and OCJS frequencies may be biased upwards. In the absence of relevant UK evidence on this source of bias, reported frequencies are treated as essentially accurate for the baseline assumptions, which are given in Table 4.11. The potential bias arising from this should be an urgent issue for future research.
33
A regression of quantity on frequency of use and treatment status gives an insignificant treatment coefficient in every case.
Table 4.11: Numerical interpretation of frequency-of-use questions (baseline assumptions)
OCJS: frequency in last year AS: usual frequency
Description Assumed range days per year Description Assumed range days per year
Once only 0 – 2 Less often than once a year 0 - 1
More than once, less than
once every couple of months 2 – 4 A few times a year 1 – 9
Once every couple of months 4.5 – 9 One or two days a month 9 – 30
Once a month 9 – 21 One or two days a week 30 – 130
Two or three times a month 21-54 Three or four days a week 130 – 234
Once or twice a week 54 – 156 Five or more days a week 234 - 320
Most days 156 – 320