6
$\begingroup$

About crossed factorial design, two factors A and B are crossed, if every level of A can occur in every level of B.

A full factorial/crossed experiment is an experiment whose experimental units take on all possible combinations of these levels across all such factors.

If I am correct about the above, are crossed factorial design and full factorial design the same concept?

$\endgroup$
  • $\begingroup$ How do Graeco-Latin squares meet your definitions? $\endgroup$ – Henry Jun 25 '13 at 6:52
  • 2
    $\begingroup$ "Crossed" means that the factors are formally independent, not nested. "Full factorial" = "all main effects + all interaction effects" between factors (plus covariate main effects, if any) $\endgroup$ – ttnphns Jun 25 '13 at 7:24
  • 2
    $\begingroup$ You might also have a "fractional factorial design", for instance a $2^{3-1}$ design, meaning three variables but only four runs so inteactions cannot be estimated. Such a design is also "crossed" but it is not "full factorial". $\endgroup$ – kjetil b halvorsen Jun 25 '13 at 14:52
  • $\begingroup$ @kjetilbhalvorsen; Thanks! why isn't " crossed design = full factorial design", given that "two factors A and B are crossed, if every level of A can occur in every level of B"? $\endgroup$ – Tim Jun 25 '13 at 19:43
2
$\begingroup$

(Crossed) factorial design is not always equal to Full factorial design.

In this context, the word "crossed" is a synonym of "factorial" (see here). Both indicates the presence of interactions between factors.

A factorial design can be either:

  • Full (all possible interactions) or,
  • Fractional (without all interactions, but some).

Therefore, a "factorial design" does not necessarily implies it is "fully crossed".

In the OP's case, where:

two factors A and B are crossed, if every level of A can occur in every level of B

It will be a full factorial design.

$\endgroup$
  • $\begingroup$ Thanks! "two factors A and B are crossed, if every level of A can occur in every level of B." So isn't a crossed design full factorial? $\endgroup$ – Tim Jun 25 '13 at 16:47
  • $\begingroup$ Yes, it is but you did not write the word "full" on question. And there is no need to use "crossed" and "factorial" together, because they are synonyms. $\endgroup$ – Andre Silva Jun 25 '13 at 16:49
  • $\begingroup$ I meant why isn't " crossed design = full factorial design", given that "two factors A and B are crossed, if every level of A can occur in every level of B"? $\endgroup$ – Tim Jun 25 '13 at 16:53
  • $\begingroup$ Then, What is the difference between a crossed factorial design and a full factorial design? Why are not igual designs? I don't understand. From a crossed factorial design is it possible to get the analysis of all interactions of the factors A, B, AB, AA, BB? Thanks. $\endgroup$ – cabita Jul 6 '14 at 14:47
  • $\begingroup$ @cabita, the terminology crossed factorial design is redundant, because crossed and factorial means the same thing, i.e., an experiment has factors w/ usually 2 levels/factor. E.g., factor A = I (levels: with irrigation, without irrigation), factor B = F (with fertilization, without fert.). A full factorial design has all possible combinations between factors and levels (in this case $2^2$ --> I+F, no_I+no_F, I+no_F, no_I+F). Yet, a factorial design can be fractional, where there are interactions, but not all possible ones. $\endgroup$ – Andre Silva Jul 6 '14 at 15:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.