The Importance of Estimates and Confidence Intervals Rather than P Values
Author:
- Margaret ElyMEMargaret Ely
- 1 February 1999
- journal article
- other
- Published by SAGE Publications in Sociology
- Vol. 33 (1) , 185-190
Abstract
The importance of reporting estimates and confidence intervals for statistical analyses has been well publicised in the arena of medical studies for some years now. The requirement to give confidence intervals for the main results of a study has been included in the statistical guidelines for contributors to medical journals since the 1980s and methodological points such as this are discussed in the Statistical Notes section of the British Medical Journal. If the use of quantitative methods in British sociology is to be encouraged, as Frank Bechhofer (1996) suggests is needed, it is important to have a forum for the dissemination of basic methodological issues which is accessible to researchers within the discipline. This note aims to achieve such dissemination by using an example from current research to illustrate this fundamental, but often overlooked, aspects of quantitative analysis.Keywords
This publication has 6 references indexed in Scilit:
- Teenage family life, lifestyles and life chances: associations with family structure, conflict with parents and joint family activityInternational Journal of Law, Policy and The Family, 1998
- Social and psychological wellbeing of children from divorced families: Australian research findingsAustralian Psychologist, 1996
- Quantitative Research in British Sociology: Has it Changed Since 1981?Sociology, 1996
- The Changing Effects of Lone Parent Families on the Educational Attainment of their Children in a European Welfare StateSociology, 1994
- Parental divorce and the well-being of children: A meta-analysis.Psychological Bulletin, 1991
- Confidence intervals rather than P values: estimation rather than hypothesis testing.BMJ, 1986