From: IN%"margory@rcn.com" "margory cohen" 3-JAN-2004 09:13:19.08 To: IN%"Applied-ethology@sask.usask.ca" CC: Subj: RE: New Member with a problem ----- Original Message ----- From: To: Cc: Sent: Sunday, December 28, 2003 8:24 AM Subject: New Member with a problem > Good morning, > Our names are Sam & Lyn Bailey who live in southern Mississippi. We have been > a canine foster home for about a year dealing with many different social > problems with our rescues. > We presently have a canine, Rusty, abt 1 yr old, knew exactly where he came > from. We got him when he was about 6-7 months old with no issues to deal with. > We had him neutered and were getting him ready to be put up for adoption when > he started "tail chasing". He was always agressive towards his food, meaning > he would woof it down as if were his last meal. This did not bother us because > we have seen that trait in other rescues and they worked out of that. > However, Rusty, who weighs abt 50 # and is a lab mix (NO CHOW) has gone from > periodic tail chasing to compulsive tail & foot biting, self mutilation. He > spins & spins while growling severely, never loosing his balance. He will come > out of it, instantly, & then be the sweetest, kindest dog. He is kid,cat & dog > friendly. > We have recently had a thorough checkup, bloodwork, etc, all within "normal > limits". We have started him on 110 mg of Chlomiperamine & 5 mg of Diazepam, 1x > a day. > Does anyone have any suggestions? > Sam margory cohen: A late response to yours. If you still need help, I'd check for a good trainer - thru International Association of Canine Professionals -- www.dogpro.org == somebody near you who can see you and the dog and evaluate the dog's behaviour and set you all on a program a new owner would follow and commit to as well. So long as you've a clean bill of health physically for the dog. Lots of vets and theorists like to drug dogs before real training. Only hits the surface. Doesn't 'fix' anything. And then the drugs wreck havoc of their own. I wouldn't call it aggression toward food that makes a dog scarf up his food. Tail chasing can be from boredom, frustration, not knowing what else is possible. Self-chewing can be boredom as well. Training can open up a whole other word. Means lots of work for owner and dog but I've seen dogs get turned around and everybody stays safe -- the dog, the family, the community. Good luck. -margory cohen San Francisco From: IN%"olsson@ibmc.up.pt" "Anna Olsson" 7-JAN-2004 02:54:31.83 To: IN%"applied-ethology@sask.usask.ca" "applied ethology" CC: Subj: statistics question Dear all, What should one do when an ANOVA results in nearly significant interaction between different factors? When you get near significance on a comparison of the parameters of interest, you can report your findings directly in the paper and the reader can decide for him/herself whether or not to consider the results as different. But when I encounter a nearly significant value at this level of the analysis, I have to decide myself which line to follow in the continued analysis. We happened to get two such interactions in the results we work on right now, with the disturbing p-values 0.063 and 0.057. Because they are so close to 0.05 I would be inclined to consider the interaction and analyze the different groups separately, but I'm not sure if this is the formally correct approach. I would really appreciate your input on how to address the problem. Best wishes Anna O From: IN%"info@otheranimals.org" "Other Animals" 7-JAN-2004 07:31:53.25 To: IN%"Applied-ethology@sask.usask.ca" CC: Subj: Statistics help wanted Hi everyone, I am trying to do some VERY basic statistical analysis of some behavioural data from captive chimps, and would greatly appreciate it someone could give me a few pointers as to what to use. Anyone able to help? Chris Other Animals From: IN%"NabilBrandl@orangenet.dk" "Nabil Brandl" 8-JAN-2004 12:47:10.78 To: IN%"olsson@ibmc.up.pt" "Anna Olsson", IN%"applied-ethology@sask.usask.ca" "applied ethology" CC: Subj: RE: statistics question Dear Anna Olsson When you get significant interaction effect, this means that you are in a wrong direction and you are in a real trouble. You can not say that p-value=0.06 is nearly significant/not significant. The interaction effect shows that the main effect in the investigation gave no diffirance. You have to search for another main effect without interaction. It is very difficult and mis-leading results to explaine the difference caused by the main factor with interaction effect. You can allways explaine your results by using a very sofisticated statistical methods, but you will get a mis-leading explanation. The interaction effect is a good tool to show to scientists that something is wrong with the main effect. Best regards Cand. Agro.,MSc. Nabil Brandl mailto: NabilBrandl@orangenet.dk http://nabilnabil.homestead.com ----- Original Message ----- From: Anna Olsson To: applied ethology Sent: Wednesday, January 07, 2004 9:54 AM Subject: statistics question Dear all, What should one do when an ANOVA results in nearly significant interaction between different factors? When you get near significance on a comparison of the parameters of interest, you can report your findings directly in the paper and the reader can decide for him/herself whether or not to consider the results as different. But when I encounter a nearly significant value at this level of the analysis, I have to decide myself which line to follow in the continued analysis. We happened to get two such interactions in the results we work on right now, with the disturbing p-values 0.063 and 0.057. Because they are so close to 0.05 I would be inclined to consider the interaction and analyze the different groups separately, but I'm not sure if this is the formally correct approach. I would really appreciate your input on how to address the problem. Best wishes Anna O From: IN%"NabilBrandl@orangenet.dk" "Nabil Brandl" 8-JAN-2004 12:47:31.23 To: IN%"info@otheranimals.org" "Other Animals", IN%"Applied-ethology@sask.usask.ca" CC: Subj: RE: Statistics help wanted Try this software: Cand. Agro.,MSc. Nabil Brandl mailto: NabilBrandl@orangenet.dk http://nabilnabil.homestead.com Tutorial TUTORIAL CONTENTS Introduction Section 1 Design Designing a Study Using the Grid to Create a New Study Using an MS Word Document to Create a New Study Creating a Form for Data Entry Creating an Internet Survey Section 2 Data Management Data Entry and Editing Automatic Record Advance Viewing and Changing Records Deleting a Record Section 3 Analyses Designing and Running Analyses Running Procedures Using Graphics Multiple Procedures Customizing Analyses Frequently Asked Questions General Web Surveys Analyses StatPac Home ----- Original Message ----- From: "Other Animals" To: Sent: Wednesday, January 07, 2004 2:31 PM Subject: Statistics help wanted > Hi everyone, > > I am trying to do some VERY basic statistical analysis of some behavioural > data from captive chimps, and would greatly appreciate it someone could give > me a few pointers as to what to use. > > Anyone able to help? > > Chris > Other Animals From: IN%"Lorenz.Gygax@fat.admin.ch" 9-JAN-2004 01:05:54.37 To: IN%"olsson@ibmc.up.pt", IN%"applied-ethology@sask.usask.ca" CC: Subj: RE: statistics question Dear Anna, I think there is no absolute answer to your question and thus no = absulute formally correct approach. Statistics - in my eyes - is not as exact as = e.g. pure maths and thus some decisions must be taken based on what makes = sense (and not based on strict rules). I would like to add, that there is nothing wrong with data that show an interaction when analysed. It's just one of those cases where the interdependence of your explanatory and response variables are more = complex than what we ususally like (due to the ease/complication of explaining = and reporting our findings). Basically I see two approaches in your problem: (1) In an ANOVA, the p-values of the different effects (of your main = effects and the interaction) are actually multiple tests on the same data set. = Thus you could argue that, in principal, a correction would be needed and = thus even values close to 0.05 are not as significant as they seem on first sight. This would mean: forget about the interaction and concentrate on = the main effects. (2) On the other hand, p-values are continuous and the split-off at = 0.05 is a convention. This gives you a good argument to actually stick with the interaction even if it does not formally reach the significance level. I would also base this decision on the sample size that you = investigate. If you are faced with a large sample, you can argue that the effect = which is close to significance is usually rather small (and thus possibly not = very meaningful regarding the biological relevance). Thus you neglect it = because any difference however slight will become significant if the sample is = large enough. (It is also possible that your effect is indeed large but can = only be estimated with a large error. Then you would have a non-significant, large Effect. Thus, you might want to actually check your estimates to = get an idea on the size of your estimated interaction effect.) On the other hand if your sample is small, you could expect the = interaction to become even more pronounced if you had investigated a larger sample. = In this case, I would rather live with the interaction and report it or = split the analysis as you suggested. Finally, I think the most important aspect in such an analysis is = reporting what you did. Then, any reader can reflect your results based on how = you analysed your data. I hope this helps. Regards, Lorenz -=20 Lorenz Gygax, Dr. sc. nat. Tel: +41 (0)52 368 33 84 / lorenz.gygax@fat.admin.ch =20 Center for proper housing of ruminants and pigs Swiss Veterinary Office agroscope FAT T=E4nikon, CH-8356 Ettenhausen / Switzerland Fax : +41 (0)52 365 11 90 / Tel: +41 (0)52 368 31 31 > What should one do when an ANOVA results in nearly=20 > significant interaction between different factors? When you=20 > get near significance on a comparison of the parameters of=20 > interest, you can report your findings directly in the paper=20 > and the reader can decide for him/herself whether or not to=20 > consider the results as different. But when I encounter a=20 > nearly significant value at this level of the analysis, I=20 > have to decide myself which line to follow in the continued=20 > analysis. We happened to get two such interactions in the=20 > results we work on right now, with the disturbing p-values=20 > 0.063 and 0.057. Because they are so close to 0.05 I would be=20 > inclined to consider the interaction and analyze the=20 > different groups separately, but I'm not sure if this is the=20 > formally correct approach. From: IN%"thevirtualeditor@hotmail.com" "Andrea Herbert" 14-JAN-2004 15:15:47.01 To: IN%"Applied-ethology@sask.usask.ca" CC: Subj: Classic Study on Behaviorial Ecology of Hyenas Returns to Print I thought you would be interested to know that "Kalahari Hyenas: Comparative Behavioral Ecology of Two Species" by M. G. L. Mills, originally published in 1990 and since out of print, is now available again from The Blackburn Press. The book focuses on a comparison of the feeding ecology and social systems of the brown and spotted hyenas. Extracts from the author's field notes form a captivating feature of the text and convey all the thrill and anticipation of observing the animals in their own habitat. "It is a first-class book to which every ecologist should have access and which every behavioral ecologist and carnivore biologist should read." J. of Mammalogy, 74, 240 "Kalahari Hyenas: Comparative Behavioral Ecology of Two Species" By M. G. L. Mills, ISBN 1-930665-830, Paperback, 298 pages. For more information, see: http://www.blackburnpress.com/kahycobeecof.html The Blackburn Press is a relatively new publishing company, founded with the mission of keeping in print and available for purchase at reasonable prices book titles that larger publishers have lost interest in and have declared to be "out of print." It specializes in scientific and technical books and textbooks that are classics in their field. Andrea Herbert The Blackburn Press Publishers of classic scientific and technical books P.O.Box 287, Caldwell, N.J. 07006 973-228-7077 Fax: 973-228-7276 AHerbert@BlackburnPress.com Explore the latest additions to our list at http://www.BlackburnPress.com From: IN%"saluqi@ix.netcom.com" "John Burchard" 14-JAN-2004 15:49:53.31 To: IN%"applied-ethology@sask.usask.ca" "Applied ethology list" CC: Subj: RE: Statistics help wanted Other Animals wrote: > > Hi everyone, > > I am trying to do some VERY basic statistical analysis of some behavioural > data from captive chimps, and would greatly appreciate it someone could give > me a few pointers as to what to use. > > Anyone able to help? I think to give a meaningful answer to this question it would be necessary to know something about the nature and the structure of the data, as well as about the hypothesis or hypotheses it is desired to test. John -- Dr. John Burchard Tepe Gawra Salukis http://saluqi.home.netcom.com/ saluqi@ix.netcom.com From: IN%"olsson@ibmc.up.pt" "Anna Olsson" 15-JAN-2004 05:10:42.34 To: IN%"applied-ethology@sask.usask.ca" "applied ethology" CC: Subj: rats, cats and ecological balance Dear all, I'm following a lively discussion on another list about the role of cats as pest controllers versus wildlife destructors. The discussion is as I said lively and opinions diverge: whilst one argues that cats are too small and light-weight to be able to fight rats, another argues that cats might even kill roedeer kids... It would be nice to drop some scientific 'evidence' into the melting pot - can anyone help? Anna Olsson