CoChat

Correlated measurement error--implications for nutritional epidemiology

Day, N. · 2004 · International Journal of Epidemiology
Cited by 94

The DOI matches its Crossref record; 2 of the 2 references it cites were found in a registry or an index.

Abstract

BACKGROUND: In nutritional epidemiology, it is common to fit models in which several dietary variables are included. However, with standard instruments for dietary assessment, not only are the intakes of many nutrients often highly correlated, but the errors in the estimation of the intake of different nutrients are also correlated. The effect of this error correlation on the results of observational studies has been little investigated. This paper describes the effect on multivariate regression coefficients of different levels of correlation, both between the variables themselves and between the errors of estimation of these variables. METHODS: Using a simple model for the multivariate error structure, we examine the effect on the estimates of bivariate linear regression coefficients of (1) differential precision of measurement of the two independent variables, (2) differing levels of correlation between the true values of the two variables, and (3) differing levels of correlation between the errors of measurement of the two variables. As an example, the prediction of plasma vitamin C levels by dietary intake variables is considered, using data from the European Prospective Investigation of Cancer (EPIC) Norfolk study in which dietary intake was estimated using both a food frequency questionnaire (FFQ) and a 7-day diary (7DD). The dietary variables considered are vitamin C, fat, and energy, with different approaches taken to energy adjustment. RESULTS: When the error correlation is zero, the estimates of the bivariate regression coefficients reflect the precision of measurement of the two variables and mutual confounding. The sum of the observed regression coefficients is biased towards the null as in univariate regression. When the error correlation is non-zero but below about 0.7, the effect is minor. However, as the error correlation increases beyond 0.8 the effect becomes large and highly dependent on the relative precision with which the two variables are measured. At the extreme, the bivariate estimates can become indefinitely large. In the example, the error correlation between fat and energy using the FFQ appears to be over 0.9, the corresponding value for the 7DD being approximately 0.85. The error correlation between vitamin C and fat, and vitamin C and energy, appears to be below 0.5 and smaller for the 7DD than for the FFQ. The impact of these error correlations on bivariate regression coefficients is large. The effect of energy adjustment differs widely between vitamin C and fat. CONCLUSION: High levels of error correlation can have a large effect on bivariate regression estimates, varying widely depending on which two variables are considered. In particular, the effect of energy adjustment will vary widely. For vitamin C, the effect of energy adjustment appears negligible, whereas for fat the effect is large indicating that error correlation close to one can partially remove regression dilution due to measurement error. If, for fat intake, energy adjustment is performed by using energy density, the partial removal of regression dilution is achieved at the expense of substantial reduction in the true variance.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1093/ije/dyh138

Cites

2 of 2 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Cited by

See all 94

Metadata

Authors
Day, N.
Venue
International Journal of Epidemiology
Published
2004-12-01
Volume
33
Issue
6
Pages
1373-1381
ISSN
  • 0300-5771 (ISSN-L)
  • 1464-3685
OpenAlex
W2163499887
Cited by
94
References
2
FWCI
4.26
Type
journal article
Language
en
Open access
closed

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.