Raw Data

Lisa:

I'm attaching the "raw" data set that I sent to Julie to work with in SAS. These are the behavior data and do not include the fight data.

Once you have looked these over for a while and feel comfortable with them I'll send you the real raw data from the Observer to this point to try to show you how I prepared the data. Jeff

MAINBEH.xls

 

Thanks for sending that data, I thought it looked BEAUTIFUL! I may be the only one who will view that way but all that time sitting on my numb butt watching those videos made for a beautiful data set - and I swear you are a whiz with Excel, and the colors in your notes were great! It looks like it took you forever to do all of that. Thanks so much!

I understand what you did with the frequency and percentage of the maintenance behaviors I think - you just took the data I sent you and every time a pig in that pen was lying counts as one frequency right? So there should be a lot more lying frequency counts in pens with 6 pigs in it than 3 pigs right?

Then, for percentage you just added up all the frequencies of those maintenance behaviors and divided each behavior by the total to get the percentage of each behavior?

Why did you transform the percentage?? (I tried to find info on transforming data from my workbook from class but unfortunately it isn't organized very well and has no glossary! any ideas under what topic it would be?)

So we aren't tracking any info yet on each individual pig for maintenance behaviors?

Also, please read below what I have so far from going back to my stats class on what this project's model is and let me know how right or wrong I am:

Experiment is RBD (randomized block design) with replication.

dependent variable: fight scores, amount of time in maintenance behaviors, N:L ratio, haptoglobin values, total cortisol, CBG, percent free cortisol, production measures (ADG, backfat, etc.)

treatments: 3, one is split marketed (SM, 63), control (C, 66), and a modified control (MC, 33)

experimental material: have 24 pens, divided into 4 blocks of 6 pens per block so each treatment is replicated twice within each block - therefore is a RBD with replication - 4 reps.

concerns: each block may perform differently due to arriving to farm at different times and possibly each experimental unit (each pen) could be different do to differences in barn temperature/drafts.

Thanks again for spliting up the data into by period, by day, and by taping. So we will run the data in those 3 ways to see if we can find significance?  Thanks a million Jeff! Lisa

 

> I understand what you did with the frequency and percentage of the maintenance
> behaviors I think - you just took the data I sent you and every time a pig in
> that pen was lying counts as one frequency right? So there should be a lot
> more lying frequency counts in pens with 6 pigs in it than 3 pigs right?
There are more FREQ in total #'s but the data are presented in % so there is no difference. Why are they in % and hence no difference? This is due to the fact that we need to have some common ground to compare the data.

> Then, for percentage you just added up all the frequencies of those
> maintenance behaviors and divided each behavior by the total to get the
> percentage of each behavior?
Correct.

> Why did you transform the percentage??
> (I tried to find info on transforming data from my workbook from class but
> unfortunately it isn't organized very well and has no glossary! any ideas
> under what topic it would be?)
To achieve normal distribution (Frank). See the orange book. The arcsine square root transformation is detailed in the classic book, Bakeman, R., & Gottman, J.M. (1997). Observing interaction: An introduction to sequential analysis. New York: Cambridge University Press. 2nd edition.

> So we aren't tracking any info yet on each individual pig for maintenance
> behaviors?
The experimental unit in the experiment is the pen. We always mark all the animals in the pen and collect individual data and then sum them together. Frank and I wonder why we do this, it makes no sense but we still do it. We are trying to develop ways to mark large numbers of cattle individually so we can sum them up. I wonder why?

Frank is German, we are going to AZ and he's being a pain, he's really going to be upset as we have to go and setup a OBS project right now. I hope this helps, sorry to be so short.  Jeff

LIRU Home Page
Send comments or suggestions to LIRU Webmaster
Revised: April 18, 2008