This study formed part of the South African National HIV, Incidence, Behaviour and Communication (SABSSM) 2008 survey, which included questions assessing the extent of alcohol use and problem drinking among South Africans. Method: A multistage random population sample of 15 828 persons aged 15 or older (56.3% women) was included in the survey. Alcohol use was assessed using the Alcohol Use Identification Test (AUDIT). Tabulation of data for different age groups, geolocality, educational level, income, and population group produced the estimates and associated confidence intervals. The odds ratios for these variables in relation to hazardous or harmful drinking were also computed. Results: Current alcohol use was reported by 41.5% of the men and 17.1% of women. White men (69.8%) were most likely and Indian/Asian women (15.2%) least likely to be current drinkers. Urban residents (33.4 %) were more likely than rural dwellers (18.3%) to report current drinking. Risky or hazardous or harmful drinking was reported by 9%: 17% among men and 2.9% among women. In men, risky drinking was associated with: the 20-54 year age group; the Coloured population group; lower economic status; and lower education. Among women, risky drinking was associated with: urban residence; the Coloured population group; lower education; and higher income. Conclusion: An increase in current, binge drinking and hazardous or harmful drinking prevalence rates was observed from 2005 to 2008 in South Africa. Multilevel interventions are required to target high-risk drinkers and to create awareness in the general population of the problems associated with harmful drinking. Future prospective studies are needed to assess the impact of problem drinking.

No abstract

A general theory for a case where a factor has both fixed and random effect levels is developed under one-way treatment structure model. Estimation procedures for the fixed effects and variance components are consider for the model. The testing of fixed effects is considered when the variance-covariance matrix is known and unknown. Confidence intervals for estimable functions and prediction intervals for predictable functions are constructed. The computational procedures are illustrated using data from an on-farm trial.

Illicit drug use negatively affects development of human and physical capital of any nation. Huge financial resources are allocated to prevent and curb illicit drug use. The use of these drugs continue to spread across race and age groups, despite application of various control measures. The information provided in this paper contributes towards understanding the extent and influence of illicit drugs use in South Africa. A population-based national HIV prevalence, behaviour and health survey conducted in 2008, incorporated questions on the extent and use of illicit drugs. A multistage random population sample of 15 845 persons aged 15 years or older (58% women and 42% men) was included in the survey. The use of combined illicit drugs excluding cannabis was reported by 1.7% of the 13 119 participants, and including cannabis by 4.3 % of the 13 128 participants. The Coloured men (14.3%) were the most likely to use cannabis, where as the Indian women (0.6%) were the least likely. The urban residents (5.4%) were more likely to report use of any illicit drug including cannabis than rural dwellers (2.5%). Illicit drug use has a high association with illnesses thus call for interventions to address this serious problem.

No abstract

The construction of an exact confidence interval (CI) for a single coefficient of variation (CV) is computationally cumbersome, but a number of approximation methods exist. The existing methods of CI construction are not appropriate for the agricultural experiments involving a single crop grown over several locations. There is a need to assess and identify an appropriate approximate method from the existing methods, and provide a new approach for multiple experiments. Simulation and real data were used in the evaluation process. The Vangel (1996) approximation method was computationally easier and produced an approximate length of CI close to that obtained by the exact method. For multiple experiments, the bootstrapping method performed better than other methods in the construction of CIs. Expected lower and upper confidence limits for coefficient of variation for various crop types were established using data from 513 trials conducted in Ethiopia. These CIs are used to monitor variability of new trials conducted on same locations.

No abstract

The estimator of effect size, the sample mean difference divided by the sample standard error of the difference is studied in the context of mixed models and is related to the analysis of on-farm trials. A single treatment is compared against possibly different controls using a completely randomized design on each farm. A lower (l-u)l00% confidence limit on mean difference of the treatment and the average control is obtained. The best linear unbiased predictors (BLUPs) of the mean difference of the treatment and the individual controls as well as the lower {1-a)100% prediction limits are provided. The effect of omitting or not omitting the farm-bytreatment interaction variance component in the weighting process is assessed using two numerical examples.