# Errors, Variances and Correlations

We are so accustomed to thinking in terms of raw scores, with their apparent pin-point precision, that we find the concept of measurement error at first abhorrent and later perplexing. We understand, finally, that measurement error always exists, but we don't know how to handle it when reporting our results.

Measurement error always affects our findings. When reporting our results we need to decide whether the result is supposed to describe the local details of this data collection, or to infer some more general "reality". Local results, such as an individual's performance on a test and the test's reliability, must contain their measurement error. General inferences, however, such as the math level of third graders, are intended to be estimated without bias from measurement error. To ascertain what, if anything, of global import we have discovered, we need to deduct the effect of measurement error from our final estimates.

Consider person n with measure bn of standard error sn. The more observations we make, the more we know about this person, and the smaller sn will be. What about the dispersion of a sample of N people? When measurement error is uncorrelated with what is measured, the variance of the sample of observed measures contains the sample variance and also the measures' error variance. But it is only the sample variance that estimates the population sampled.

The estimation for this corrected sample variance is:

s^2 = sum( bn^2 - b.^2 - sn^2 ) / (N-1)

where b. = sum(bn)/N

Correlation coefficients are also affected. Measurement error reduces the maximum observable correlation. Adjusting observed correlation coefficients for measurement error isolates how much sample variance is left unexplained.

Consider the typical regression analysis, in which dependent measures are mistaken for points without error. Let R^2 be the observed multiple squared correlation (proportion of explained variance), E be the observed RMS residual from the regression, and S be the observed standard deviation of the measures. Then E contains both unexplained variance and measurement error, and R^2 has been reduced accordingly. R^2 is defined as

R^2 = (S^2 -E^2) / S^2

If r^2 is the squared multiple correlation corrected for measurement error in the dependent variable and e is the RMS measurement error, then

r^2 = (S^2 - E^2) / (S^2 - e^2)

= R^2 / [ 1 - (e/E)^2 (1 - R^2)]

e<=E; r>=R; when e=0, then r=R; when e=E, then r=1.

It is this larger correlation r, and not R, which tells us how successfully our independent variables have explained the variance of the dependent measure.

Errors, Variances and Correlations, B Wright … Rasch Measurement Transactions, 1991, 5:2 p. 147

Rasch Publications
Rasch Measurement Transactions (free, online) Rasch Measurement research papers (free, online) Probabilistic Models for Some Intelligence and Attainment Tests, Georg Rasch Applying the Rasch Model 3rd. Ed., Bond & Fox Best Test Design, Wright & Stone
Rating Scale Analysis, Wright & Masters Introduction to Rasch Measurement, E. Smith & R. Smith Introduction to Many-Facet Rasch Measurement, Thomas Eckes Invariant Measurement: Using Rasch Models in the Social, Behavioral, and Health Sciences, George Engelhard, Jr. Statistical Analyses for Language Testers, Rita Green
Rasch Models: Foundations, Recent Developments, and Applications, Fischer & Molenaar Journal of Applied Measurement Rasch models for measurement, David Andrich Constructing Measures, Mark Wilson Rasch Analysis in the Human Sciences, Boone, Stave, Yale
in Spanish: Análisis de Rasch para todos, Agustín Tristán Mediciones, Posicionamientos y Diagnósticos Competitivos, Juan Ramón Oreja Rodríguez

 Forum Rasch Measurement Forum to discuss any Rasch-related topic

Go to Top of Page
Go to index of all Rasch Measurement Transactions
AERA members: Join the Rasch Measurement SIG and receive the printed version of RMT
Some back issues of RMT are available as bound volumes
Subscribe to Journal of Applied Measurement

Go to Institute for Objective Measurement Home Page. The Rasch Measurement SIG (AERA) thanks the Institute for Objective Measurement for inviting the publication of Rasch Measurement Transactions on the Institute's website, www.rasch.org.

Coming Rasch-related Events
Oct. 9 - Nov. 6, 2020, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Core Topics (E. Smith, Winsteps), www.statistics.com
Jan. 22 -Feb. 19, 2021, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Core Topics (E. Smith, Winsteps), www.statistics.com
May 21 -June 18, 2021, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Core Topics (E. Smith, Winsteps), www.statistics.com
June 25 - July 23, 2021, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Further Topics (E. Smith, Winsteps), www.statistics.com
Aug. 13 - Sept. 10, 2021, Fri.-Fri. On-line workshop: Many-Facet Rasch Measurement (E. Smith,Facets), www.statistics.com
June 24 - July 22, 2022, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Further Topics (E. Smith, Winsteps), www.statistics.com