Statistical spin, linguistic obfuscation or how 14 per cent is turned into 50 per cent
Farhad Dalal
September 25, 2018 DOI: 10.4324/9780429457814-13 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Commentary |
|---|---|
| Sample size | 137 |
| Population | People receiving treatment for a condition with relapse risk, divided into an MBCT plus treatment-as-usual group and a treatment-as-usual-only group |
| Interventions | Mindfulness Based Cognitive Therapy (MBCT) Treatment-As-Usual (TAU) |
| Duration | 60 weeks |
| Key points | At 60 weeks, 44% of the MBCT plus usual-care group (31/71) had relapsed compared with 58% of the usual-care-only group (38/66), a 14 percentage-point difference. The author argues this corresponds to benefit for only about 1 or 2 out of 10 people, with 8 or 9 out of 10 not helped, and notes the original paper does not present the figures this way. |
Abstract
The participants were divided into two groups. One group, 71 people, received Mindfulness Based Cognitive Therapy (MBCT) as well as Treatment-As-Usual (TAU), and the other group, 66 people, received only TAU (sixth corruption). At the end of the 60 weeks, they found that out of the 71 people in the MBCT group, 31 had relapsed (44 per cent); and out of 66 in the TAU group, 38 people relapsed (58 per cent). This leaves us with the result that just 14 per cent fewer people having relapsed in the treatment group at the end of one year. This is the total benefit: 14 per cent additional people did not relapse. In other words, the treatment was of benefit to about 1 or 2 out of 10 people. It is always instructive to state the reverse: 8 or 9 out of 10 people were not helped by the treatment. Interestingly, these figures are never stated in this form in the paper.