Components of Experimentation

The "Classic" Approach:

Step 1:

Form research hypotheses.

Presented in "Introduction" of a journal paper.

Step 2:

Translation of hypotheses into a set of treatment conditions; selection of experiment design.

Presented in "Methods" section.

Step 3:

Conduct experiment; collect data. Use statistical analysis to determine the repeatability of any differences observed in an experiment.

Presented in "Results" section.

Step 4:

Assimilation of outcomes of statistical tests of research hypothesis into theory that generated hypotheses originally

or

creation of a theoretical explanation if none is available in the literature (reconstructive phase).

Presented in "Discussion" section.

Notes:

  1. Much computer science experimentation is reconstructive, because there are few theories for real-world systems.

  2. Theories in classic science are entrenched; thus when an experiment refutes a theory, the "Discussion" section usually questions the adequacy of the experiment, not the theory!

Navigation: Previous, Up, Next