Analysis of Covariance
1. Industrial situations that show the usefulness of analysis of covariance
Analysis of covariance (ANCOVA) gives evaluation of whether the population means on the dependent variable (DV) adjusted for differences on the covariate(s), are different across the independent variable (IV) levels. The variability in the DV due to the control variable (concomitant variable or a covariate) is removed.
The covariate increases the opportunity to find statistical significance for the factors, fixed or random. Simply put, the covariate independent variable reduces the variance in the dependent variable that is to be explained by the factors.
An example of ANCOVA is testing the effect of computer experience on the use of e-shopping, making attitude towards e-shopping as a covariate. The covariate removes its effect from the e-shopping measure, since in general those with positive attitudes shop more.
2. Real life industrial research questions that could be analyzed by partial correlation Below are some of the research questions that can be analyzed by partial correlation:
What is the effect of job satisfaction on salary, type of education, gender, and smoker, factoring out age?
Is there an affect of electrical appliances on contraceptive use, removing the effect of education?
Is there an effect on prep hours and SAT scores, factoring the effect of GPA? 3. Variables
From the research questions above, the first one has IVs to be salary, type of education, gender, and smoker, and the dependent variable DV is job satisfaction. The covariate is age. The second one has electrical appliances as IV, education as covariate, and the DV is contraceptive use. The third one has IV as prep hours, GPA as covariate, and DV as SAT scores.
4. Nature of data
In ANCOVA, the level of measurement for the IVs can be continuous, dichotomous, nominal, ordinal, or grouped interval and the DV and covariate are required to be interval level and categorical. If the DV or
Analysis of covariance (ANCOVA) gives evaluation of whether the population means on the dependent variable (DV) adjusted for differences on the covariate(s), are different across the independent variable (IV) levels. The variability in the DV due to the control variable (concomitant variable or a covariate) is removed.
The covariate increases the opportunity to find statistical significance for the factors, fixed or random. Simply put, the covariate independent variable reduces the variance in the dependent variable that is to be explained by the factors.
An example of ANCOVA is testing the effect of computer experience on the use of e-shopping, making attitude towards e-shopping as a covariate. The covariate removes its effect from the e-shopping measure, since in general those with positive attitudes shop more.
2. Real life industrial research questions that could be analyzed by partial correlation Below are some of the research questions that can be analyzed by partial correlation:
What is the effect of job satisfaction on salary, type of education, gender, and smoker, factoring out age?
Is there an affect of electrical appliances on contraceptive use, removing the effect of education?
Is there an effect on prep hours and SAT scores, factoring the effect of GPA? 3. Variables
From the research questions above, the first one has IVs to be salary, type of education, gender, and smoker, and the dependent variable DV is job satisfaction. The covariate is age. The second one has electrical appliances as IV, education as covariate, and the DV is contraceptive use. The third one has IV as prep hours, GPA as covariate, and DV as SAT scores.
4. Nature of data
In ANCOVA, the level of measurement for the IVs can be continuous, dichotomous, nominal, ordinal, or grouped interval and the DV and covariate are required to be interval level and categorical. If the DV or