Difference between revisions of "Recording behavioral data"

Revision as of 18:00, 18 November 2011

Behavioral Data should be recorded in a tab delimited file in a 'redundant' way, which will allow the most flexibility for later analysis, for example in statistical packages like R.

Namely, every variable that is ever manipulated in the experiment or related experiments should be given a column in a tab-delimited file. Each row should represent a single trial. Redundancy will occur because some variables will be the same for most trials.

Some examples are as follows.

Euphilomedes predation trials

The behavioral data for the Lampe Speiser RRH paper is available on Google Docs [1]

In this experiment, we mainly had 3 treatments - normal, blindfolded, and control. We put them in the presence of a predator to and counted how many ostracods survived, and how many were lost (presumed eaten). It might be tempting to just do the experiments and tally up how many ostracods survived from each trial.

However, doing that would limit the flexibility of statistical analyses that we could perform down the line. The redundancy listed above increases flexibility a lot. So instead, we recorded the data with columns for each variable, and each row was the result from a single trial (looking at the file linked above will clarify this).

Nematocyte firing trials

As another example, Federico conducted experiments on nematocyte firing. He varied light level

The spreadsheet is here [2]