Journal article
Abstruse Comparisons: The Problems of Numerical Contrasts of Two Groups
Journal of clinical epidemiology, Vol.52(1), pp.13-18
1999
DOI: 10.1016/S0895-4356(98)00133-4
PMID: 9973069
Abstract
The most common quantitative comparison in medical literature is a contrast of two numbers, such as two means or two rates. The two numbers, A and B, can be compared as a direct increment (A−B), ratio (A/B), relative change ([A−B]/B), or other index of contrast. To appreciate the quantitative distinction, a reader must know the “setting” reflected by the basic values of A and B. For example, a ratio of 2.0 does not distinguish comparisons between rates of 60% versus 30% and 0.006% versus 0.003%. Despite the frequency of published comparisons, they can be expressed with two types of abstrusity: quantitatives, if the basic values for A and B are not readily evident; and qualitative, if the component underlying variables are unfamiliar and not suitably explained. Among the published articles during the first six months of 1995 for
JAMA and
New England Journal of Medicine, 57 that satisfied inclusion criteria were reviewed for compliance with standards for avoiding the two types of abstrusity. The standards for quantitative abstrusity were applied to the published abstract-summary, because it is often the only “sound bite” that is read and remembered by most readers. The standards for qualitative abstrusity, however, could be fulfilled in the text, not just in the abstract-summaries of each article. Among the 57 abstract-summaries, 30% were abstruse quantitatively, and 11 (48%) of 23 pertinent papers were qualitatively abstruse. Abstrusity can be eliminated if authors and editors insist that quantitative contrasts cite the basic numbers being compared and the meaning of the associated variables and their rating scales.
Details
- Title: Subtitle
- Abstruse Comparisons: The Problems of Numerical Contrasts of Two Groups
- Creators
- Bevan Yueh - VA Puget Sound Health Care SystemAlvan R. Feinstein - Yale University
- Resource Type
- Journal article
- Publication Details
- Journal of clinical epidemiology, Vol.52(1), pp.13-18
- DOI
- 10.1016/S0895-4356(98)00133-4
- PMID
- 9973069
- NLM abbreviation
- J Clin Epidemiol
- ISSN
- 0895-4356
- eISSN
- 1878-5921
- Publisher
- Elsevier Inc
- Number of pages
- 6
- Language
- English
- Date published
- 1999
- Academic Unit
- Otolaryngology
- Record Identifier
- 9984966751702771
Metrics
4 Record Views