GETTING STARTED: Considerations in Computerized Research [22680]

Follow

Comments

1 comment

  • Avatar
    David McFarlane

    Thanks to James St. James for this extensive and valuable article on managing computerized research!  I do, however, take issue with one small point.

    Under the section "Ordering of Trials within a block", this article states, "E-Prime permits setting a maximum on the number of repetitions of a single trial type ..."  But based on my deep knowledge of E-Prime, this statement is at the very least misleading.  E-Prime most certainly does not contain any straightforward mechanism for generally setting an arbitrary maximum on the number of repetitions of a single trial type.  But if I have missed something, I would love for somebody to show me!

    E-Prime 2 and later, indeed, do include a setting to prevent re-using the same List level as the first level after a reset (i.e., reshuffle) -- sort of like dealing a deck of cards, taking note of the last card dealt, then reshuffling the deck, looking at the top card, and if that card matches the last card dealt then moving that card to some other random place in the deck before dealing out any more cards.  Indeed, E-Prime meaningfully terms this "No Repeat After Reset", nothing more than that.  This mechanism does not in any way, however, allow for setting an arbitrary maximum on the number of repetitions of a single trial type.

    That said, one may, with enough programming skill, write code to enforce any arbitrary randomization constraint they like, so the statement in the article is true in that fringe sense.  The PST website contains some example code for how to do that, although using an inefficient sort of "Bogosort" algorithm (look that up on Wikipedia).  Typically, in order to enforce specific sorts of randomization constraints, one may more simply generate suitable sequences outside of E-Prime, and then run those sequences in sequential order from E-Prime.

    For more discussion on this topic, please go to the E-Prime Google Group (groups.google.com/group/e-prime) and run searches using the terms "random", "pseudorandom", "pseudo-random", "constraint", and "constrain".

    0
    Comment actions Permalink

Please sign in to leave a comment.

Powered by Zendesk