dc.contributor.author | McDermott, JJ | |
dc.date.accessioned | 2013-07-25T08:47:55Z | |
dc.date.available | 2013-07-25T08:47:55Z | |
dc.date.issued | 1995 | |
dc.identifier.citation | Preventive Veterinary Medicine Volume 25, Issue 2, December 1995, Pages 121–133 Veterinary Epidemiology and Economics | en |
dc.identifier.uri | http://www.sciencedirect.com/science/article/pii/016758779500534X | |
dc.identifier.uri | http://erepository.uonbi.ac.ke:8080/xmlui/handle/123456789/51039 | |
dc.description.abstract | n recent years, there has been a rapid expansion in the use of quantitative methods in veterinary epidemiology. These developments have been in response to both more ambitious analytical objectives and the opportunities provided by the improved analytical capacity of computers. Some of these methods, for diagnostic tests, spatial analysis, modelling and decision support systems are very briefly reviewed.
However, the widespread use of these and other analytical developments has led to increased controversy over the value of statistical inference in observational studies. We need to be more critical of the methods we use. In many circumstances, better use of basic methods, such as descriptive statistics, can help to qualify the inherent uncertainty of decisions based on statistical models. In other circumstances more sophisticated statistical models that do not rely on unrealistic assumptions are required.
Four recommendations to improve the statistical assessment of individual observational studies are made: (1) incorporate a biological framework, (2) clearly state and test assumptions, (3) use more realistic (elaborate) models when assumptions fail and (4) present more and better descriptive statistics to supplement the traditional results of statistical models. More caution is needed in interpreting individual studies. This will require improvements in methods of comparison across studies. | en |
dc.language.iso | en | en |
dc.publisher | University of Nairobi | en |
dc.subject | Epidemiology; Data analysis; Statistical modelling | en |
dc.title | Progress in analytic methods — more sophistication or back to basics | en |
dc.type | Article | en |
local.publisher | Department of Public Health, Pharmacology and Toxicology | en |