Skip to the content.

Summary of Statistics Terms

In this post, I just want to summarize statistics terms, that might be used when analyzing data or reading papers.

Hypothesis Testing

In general, hypothesis testing is: comparing means!

*** run away ***


Well it’s oversimplification, but most of them tries to prove whether or not means between two segments of population are indeed different. There are a couple of ways on how to do this, I will simply summarize them for now:

Terms Short Definition
ANOVA Analysis of Variance
ANCOVA Analysis of Covariance
t-test Student test
MANOVA Multivariate Analysis of Variance
MANCOVA Multivariate Analysis of Covariance


Is a way to test difference in variance between two or more levels/factors/variants/treatments.


ANOVA with 1 independent variable and 2 treatments.


ANOVA with 2 or more independent variables and 2 or more treatments.


Is an extension of ANOVA, by taking into account another variable that might covariated/correlated (see difference between covariance and correlation) with dependent variable.

Another way to explain7: ANCOVA evaluates if the means of dependent variable are equal across levels of treatment, while controling continuous nuisance variable.

When To Use?

If you believe that the effect of some variables depends on other variables8.


3 general application for using ANCOVA6:

  1. Increasing power of F-test.
  2. Equating Non-Equivalent Groups.
  3. Means Adjustment for Multiple Dependent Variables.

How to Conduct ANCOVA?

know the assumptions in ANCOVA!

5 Assumption in using ANCOVA7:

  1. Linear relationship between dependent variable and nuisance variable (or concomitant variable).
  2. Homogenuous Error Variances.

    The error is random

  3. Independence of Error Terms.

    Errors is uncorrelated or in other words covariance matrix in diagonal.

  4. Normality of Error.

    The error should be normally distributed,

  5. Homogeneity of regression slopes

    In other words regression lines should be parallel among groups.

How to conduct ANCOVA:

  1. Test multicollinearity.
  2. Test the homogeneity of variance assumption: use Levene’s Test.
  3. Test the homogeneity of regression slopes assumption.
  4. Run ANCOVA analysis.










Written on September 13, 2018