Understanding statistical hypothesis testingBMJ 2014; 348 doi: https://doi.org/10.1136/bmj.g3557 (Published 30 May 2014) Cite this as: BMJ 2014;348:g3557
- Philip Sedgwick, reader in medical statistics and medical education
- 1Centre for Medical and Healthcare Education, St George’s, University of London, London, UK
Researchers assessed the efficacy of varenicline (a licensed cigarette smoking cessation aid) in helping users of smokeless tobacco to quit. A double blind, placebo controlled, parallel group, randomised controlled trial study design was used. The intervention was varenicline 1 mg twice daily. Treatment lasted for 12 weeks, with 14 weeks’ follow-up. All participants were aged 18 years or more. They had been using smokeless tobacco for at least one year before recruitment, with no abstinence from smoking of longer than three months, but wished to quit. A total of 431 participants were recruited and randomised to varenicline (n=213) or placebo (n=218). All participants were offered brief behavioural support or counselling at the discretion of the investigators.1
The primary endpoint was continuous abstinence from smoking for four weeks at the end of treatment (weeks 9-12), confirmed by cotinine concentration. Statistical hypothesis testing was two sided, with a critical level of significance of 0.05 (5%). The rate of abstinence in the varenicline group was significantly higher than in the placebo group (59% v 39%; relative risk 1.6, 95% confidence interval 1.32 to 1.87; P<0.001).
Which of the following statements, if any, are true?
a) The alternative hypothesis states that, in the population sampled, treatment with varenicline is inferior or superior to placebo with regard to the primary endpoint
b) The research hypothesis states that, in the population sampled, treatment with varenicline is superior to placebo with regard to the primary endpoint
c) It can be inferred that …