Linear regression: robust heteroscedastic confidence bands that have some specified simultaneous probability coverage

  • Rand R. Wilcox
  • Journal of Applied Statistics, November 2016, Taylor & Francis
  • DOI: 10.1080/02664763.2016.1257591

Confidence band for a regression line.

What is it about?

Regression attempts to estimate the typical value of some variable Y, given some value for the variable X. The resulting estimate will be inaccurate, so a goal is compute an interval containing that contains the unknown typical value. Moreover, this is done in such a way that all of the intervals contain the true value with some specified probability.

Why is it important?

Extant techniques make highly restrictive assumptions. The new method deals with these concerns resulting in a better understanding of how well a regression line, based on observed data, contains the true regression line if that would result if all individuals in the population under study were measured.

Read Publication

The following have contributed to this page: Dr Rand Wilcox