Dealing with discreteness: making 'exact' confidence intervals for proportions, differences of proportions, and odds ratios more exact

Agresti, A
January 2003
Statistical Methods in Medical Research;Jan2003, Vol. 12 Issue 1, p3
Academic Journal
journal article
'Exact' methods for categorical data are exact in terms of using probability distributions that do not depend on unknown parameters. However, they are conservative inferentially. The actual error probabilities for tests and confidence intervals are bounded above by the nominal level. This article examines the conservatism for interval estimation and describes ways of reducing it. We illustrate for confidence intervals for several basic parameters, including the binomial parameter, the difference between two binomial parameters for independent samples, and the odds ratio and relative risk. Less conservative behavior results from devices such as (1) inverting tests using statistics that are 'less discrete', (2) inverting a single two-sided test rather than two separate one-sided tests each having size at least half the nominal level, (3) using unconditional rather than conditional methods (where appropriate) and (4) inverting tests using alternative p-values. The article concludes with recommendations for selecting an interval in three situations-when one needs to guarantee a lower bound on a coverage probability, when it is sufficient to have actual coverage probability near the nominal level, and when teaching in a classroom or consulting environment.


Related Articles

  • Parent preferences regarding stimulant therapies for ADHD: a comparison across six European countries. Nafees, Beenish; Setyawan, Juliana; Lloyd, Andrew; Ali, Shehzad; Hearn, Sarah; Sasane, Rahul; Sonuga-Barke, Edmund; Hodgkins, Paul // European Child & Adolescent Psychiatry;Dec2014, Vol. 23 Issue 12, p1189 

    The objective is to identify attributes of ADHD stimulant medications that influence treatment preferences of parents of children and adolescents with ADHD across six European countries, using a discrete choice experiment (DCE). Different attributes (and associated levels) of stimulant therapies...

  • Development and psychometric evaluation of a context-based parental self-efficacy instrument for healthy dietary and physical activity behaviors in preschool children. Bohman, Benjamin; Rasmussen, Finn; Ghaderi, Ata // International Journal of Behavioral Nutrition & Physical Activit;10/20/2016, Vol. 13, p1 

    Background: Parental self-efficacy (PSE) refers to beliefs of parents to effectively engage in behaviors that result in desired outcomes for their children. There are several instruments of PSE for promoting healthy dietary or physical activity (PA) behaviors in children. These measures...

  • Changing non-participation in epidemiological studies of older people: evidence from the Cognitive Function and Ageing Study I and II. GAO, LU; GREEN, EMMA; BARNES, LINDA E.; BRAYNE, CAROL; MATTHEWS, FIONA E.; ROBINSON, LOUISE; ARTHUR, ANTONY // Age & Ageing;Sep2015, Vol. 44 Issue 5, p867 

    Background: non-participation in epidemiological studies threatens the generalisability of findings. Objective: to investigate the change in non-participation between the Medical Research Council Cognitive Function and Ageing Study (CFAS) I and II. Design: a comparison of two epidemiological...

  • Factors influencing sexual function of middle-aged married Korean women. YOUNGJU JEE; YOUNGHAE KIM // Journal of Physical Therapy Science;2015, Vol. 27 Issue 3, p819 

    [Purpose] This study investigated the status of women's sexual function and relevant factors given the fact that women's health is crucial to the national health, and in particular that women's sexual health has a significant impact on their overall health. [Subjects and Methods] This study...

  • Findings on Advance Care Plans among Cognitively Impaired Older Adults. Jeznach, Anna; Tuokko, Holly Anna; Garcia-Barrera, Mauricio A.; Stajduhar, Kelli // Canadian Journal on Aging;Jun2015, Vol. 34 Issue 2, p165 

    Cognitively impaired and cognitively intact older adults seemingly differ regarding engagement in aspects of advance care plans (ACPs). Through informant reports in the Canadian Study on Health and Aging, we examined differences between deceased cognitively impaired and intact older adults in...

  • Propensity Score Methods for Analyzing Observational Data Like Randomized Experiments: Challenges and Solutions for Rare Outcomes and Exposures. Ross, Michelle E.; Kreider, Amanda R.; Yuan-Shung Huang; Matone, Meredith; Rubin, David M.; Localio, A. Russell // American Journal of Epidemiology;6/15/2015, Vol. 181 Issue 12, p989 

    Randomized controlled trials are the "gold standard" for estimating the causal effects of treatments. However, it is often not feasible to conduct such a trial because of ethical concerns or budgetary constraints. We expand upon an approach to the analysis of observational data sets that mimics...

  • Students' understanding of "Women-Centred Care Philosophy" in midwifery care through Continuity of Care (CoC) learning model: a quasi-experimental study. Yanti Yanti; Mora Claramita; Ova Emilia; Mohammad Hakimi // BMC Nursing;2015, Vol. 14 Issue 1, p1 

    Background: The philosophy of midwifery education is based on the 'Women-centred care' model, which provides holistic care to women. Continuity of care (CoC) is integral to the concept of holistic women-centred care and fundamental to midwifery practice. The objective of this study was to...

  • Multilevel analysis with messy data. Longford, N.; Longford, N T // Statistical Methods in Medical Research;Dec2001, Vol. 10 Issue 6, p429 

    This paper reviews applications of the method of multiple imputation to dealing with multilevel data that have several kinds of imperfections. These are classified into two broad categories: missing values and imprecise measurement (corrupted recording). The role of the model describing the data...

  • Addressing missing outcome data in meta-analysis. Mavridis, Dimitris; Chaimani, Anna; Efthimiou, Orestis; Leucht, Stefan; Salanti, Georgia // Evidence Based Mental Health;Aug2014, Vol. 17 Issue 3, p85 

    Objective Missing outcome data are a common problem in clinical trials and systematic reviews, as it compromises inferences by reducing precision and potentially biasing the results. Systematic reviewers often assume that the missing outcome problem has been resolved at the trial level. However,...

  • Place of death in the Czech Republic and Slovakia: a population based comparative study using death certificates data. Loucka, Martin; Payne, Sheila A.; Brearley, Sarah G. // BMC Palliative Care;2014, Vol. 13 Issue 1, p13 

    Background: Place of death represents an important indicator for end-of-life care policy making and is related to the quality of life of patients and their families. The aim of the paper is to analyse the place of death in the Czech Republic and Slovakia in 2011. Research questions were focused...


Read the Article


Sorry, but this item is not currently available from your library.

Try another library?
Sign out of this library

Other Topics