It is not uncommon in designing research for an investigator to collect an array of variables representing characteristics on each observational unit. Some of these variables are central to the investigation, whereas others reflect preexisting differences in observational units and are not of interest per se. The latter are called concomitant variables, also referred to as covariates. Frequently in practice, these incidental variables represent undesired sources of variation influencing the dependent variable and are extraneous to the effects of manipulated (independent) variables, which are of primary interest. For designed experiments in which observational units are randomized to treatment conditions, failure to account for concomitant variables can exert a systematic influence (or bias) on the different treatment conditions. Alternatively, concomitant variables can increase the error variance, ...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles