Re: Input of non-normal variables into GLM models
Highland Statistics Ltd. wrote: On Tue, 13 Nov 2007 20:07:33 +0200, Anon. [EMAIL PROTECTED] wrote: Sami Ullah wrote: Hey Ecologers: I have a various variables for running multiple linear regression model using GLM. Some of my predictor variables are non-normally distributed. Using multiple linear regression, I use proc-univariate to check if the residuals in the regression model met the normality criteria, which the model did. Now I am wondering if it is advisable if I can keep the skewed predictor variables in the model or have to go for non-parametric analysis? The distribution of the predictor variables is irrelevant, so you can happily keep them in. Well, the distribution is almost irrelevant. You can get problems if they are co-linear (i.e. highly correlated), or if you have outliers (which can have a large influence on the fit). Agree. One extra thingI would argue that normality of explanatory variables (predictors) is actually bad. It means that most observations have the same (or similar) value for that explanatory variable, which may (!) make it more difficult to find a significant effect. Bad experimental design. Perhaps a histogram shaped like the uniform distribution would be the best. It means that you have similar number of observations for each part of you sampled gradient...for that explanatory variable. Theoretically, the best distribution (in terms of power) is to have a bimodal distribution, with values either at their maximum or minimum. However, this design makes it impossible to check whether the relationship is linear or not. I mention this because I forgot to mention that linearity is assumed, and this is more important. It's also easy to check - plot the residuals against the predictor. If they look curved, then it suggests that the relationship is not linear, and that would be a reason to transform. Bob -- Bob O'Hara Dept. of Mathematics and Statistics P.O. Box 68 (Gustaf Hällströmin katu 2b) FIN-00014 University of Helsinki Finland Telephone: +358-9-191 51479 Mobile: +358 50 599 0540 Fax: +358-9-191 51400 WWW: http://www.RNI.Helsinki.FI/~boh/ Journal of Negative Results - EEB: http://www.jnr-eeb.org
Postdoc: Economics and the evolution of mutualism, Harvard
Subject: Postdoc: Economics and the evolution of mutualism A 24-month post-doctoral position, potentially extendable to 36 months, is available starting in January, 2008. The position is a joint appointment between the Department of Organismic and Evolutionary Biology and the Department of Economics at Harvard University, and is funded by a National Science Foundation grant provided in partnership with the European Science Foundation EUROCORES program The Evolution of Cooperation and Trading (TECT, http://www.esf.org/tectwww.esf.org/tect). Under TECT, eight PIs in six countries have organized a collaborative research project, called BIOCONTRACT, to develop and test theories regarding the evolution of mutualism. The aim of the postdoctoral position at Harvard is to adapt economic theory to the evolutionary theory of mutualism. In particular, we are interested in the branch of economics that models transactions in which participants have private information, raising the possibility of cheating. Contract theory and other kinds of economic game theory have the potential to link biological market models with the body of mutualism theory that is largely concerned with the evolution of cheating. We have assembled a team of economists and biologists to adapt economic models to a variety of mutualisms for which we have extensive empirical data, including ants and plants, ants and lycaenid butterflies, figs and wasps, and plasmids and bacteria. Research topics that can be addressed include the evolution of interspecific signaling and deception, asymmetries between hosts and symbionts, adverse selection and market segmentation as applied to partner choice mechanisms, tailored models of specific mutualisms for which we have detailed behavioral data, and general models outlining the kinds of contracts that can exist between species, with the overall aim of understanding the factors promoting or deterring the evolution of mutualism. The post-doctoral researcher will work with Professor Naomi E. Pierce in the Department of Organismic and Evolutionary Biology and Professors Jerry Green and Drew Fudenberg in the Department of Economics. Regular meetings and opportunities for extended interactions with the other members of the BIOCONTRACT team and their research groups will be available: Prof. Ulf Dieckmann (IIASA, Austria), Dr. Francisco Dionisio (Univ. Lisbon, Portugal), Dr. Jérôme Orivel (CNRS Toulouse, France), Dr. István Scheuring (Eötvös Univ., Hungary), and Dr. Douglas Yu (Univ. East Anglia, UK). The applicant should have a PhD in Biology, Economics, Mathematics, or related area. Proficiency in mathematical modeling is required, and familiarity with biological models of cooperation, evolutionary game theory, or economic modeling is desirable. Salary is commensurate with experience. Applicants should submit a curriculum vitae, statement of research interests and representative publications, and should arrange for three letters of reference to be sent to Naomi Pierce, Department of Organismic and Evolutionary Biology, 26 Oxford Street, Cambridge, MA 02138, USA. Review of applications will begin December 15, 2007. The start date is flexible: the position could start as early as January 1, or as late as July 1, 2008. Further information about the Department is available at its website: http://www.oeb.harvard.edu. Harvard University is an Affirmative Action/Equal Opportunity Employer. For informal inquiries: Naomi Pierce ([EMAIL PROTECTED]) Jerry Green ([EMAIL PROTECTED]) Drew Fudenberg ([EMAIL PROTECTED])
Re: Input of non-normal variables into GLM models
Hi Meaghan, Linear regression models assume that the residuals of the response variable are normally distributed. Predictor variables iinstead are not assumed to have a normal distribution. Not even to be random variables. In fact, there assumed to take valueswhich are fixed a-priori. Although this assumtion is often false, it can be relaxed if the error in the determination of the predictor values is small relative to the range of values included in the model. These things are nicely explained in Draper Smith 1981 Applied Regression Analysis. Wiley w At 03:28 p.m. 13/11/2007, you wrote: can you normalize this variable through one of the various methods of normalization? On Tue, November 13, 2007 10:46 am, Sami Ullah wrote: Hey Ecologers: I have a various variables for running multiple linear regression model using GLM. Some of my predictor variables are non-normally distributed. Using multiple linear regression, I use proc-univariate to check if the residuals in the regression model met the normality criteria, which the model did. Now I am wondering if it is advisable if I can keep the skewed predictor variables in the model or have to go for non-parametric analysis? Thanks Meaghan Murphy, Ph.D. Candidate Malcolm L. McCallum Assistant Professor of Biology Editor Herpetological Conservation and Biology http://www.herpconbio.org [EMAIL PROTECTED] [EMAIL PROTECTED] -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.503 / Virus Database: 269.15.31/1130 - Release Date: 14/11/2007 09:27 a.m. -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.503 / Virus Database: 269.15.31/1130 - Release Date: 14/11/2007 09:27 a.m. William B. Batista Facultad de Agronomía UBA Av. San Martín 4453 1417DSE Buenos Aires Argentina Tel: +54-11-4524-8077 www.agro.uba.ar/users/batista -- No virus found in this outgoing message. Checked by AVG Free Edition. Version: 7.5.503 / Virus Database: 269.15.31/1130 - Release Date: 14/11/2007 09:27 a.m.
Positions at ORNL: GIS Analyst (Post Masters)
Please Post or Distribute Geographical Information Systems (GIS) Analyst Environmental Sciences Division Oak Ridge National Laboratory Oak Ridge, Tennessee ORNL08-24-ESD Project Description: The Environmental Sciences Division (ESD), (http://www.esd.ornl.gov ) at the Oak Ridge National Laboratory (ORNL), (http://www.ornl.gov ) has several openings for research associate positions beginning immediately. These positions involve the use of geographic information systems in hydrological modeling, and ecological indicators and regional simulation of ecological impacts. These projects are funded by NASA or the Department of Energy and are described on the Internet. Modeling and Synthesis Thematic Data Center (Robert Cook, PI) http://nacp.ornl.gov/mast-dc/ Integrated Terrestrial Carbon Cycle Modeling (Mac Post, PI) http://www.nacarbon.org/cgi-nacp/wcb/web/investigations/inv_pgp.pl?pgid- 288 Carbon Cycle Modeling of Agro Ecosystems (Tris West, PI) http://nacarbon.org/cgi-nacp/web/investigations/inv_pgp.pl?pgid=136 Candidates should have an M.S. in geography, ecology, soil science, forestry, or a related field. We seek a candidate with superior oral and written communication skills, excellent organizational skills, and strong personal motivation. The candidate must have knowledge and practical experience in using GIS software such as ESRI (workstation Arc/Info, ArcGIS, ArcView and associated modules such as GRID and Spatial Analyst) or other GIS/processing package, and knowledge and experience in using image processing software such as ERDAS Imagine, INVI, or IDRSI. Familiarity with visualization tools like IDV, NCL (NCAR Command Language), FERRET, GLODAP, LAS, GrADS, IDL, CDAT or others is an advantage. In addition, the candidate must have experience in acquiring and managing spatial data, familiarity with applying basic ecological and geographical principles within a spatial modeling framework. The candidate should have familiarity with UNIX and Windows operating systems, familiarity with environmentally relevant data layers, and programming experience, specifically in the context of developing GIS applications. The candidate should be willing to work closely with other research staff in multiple capacities including contributing to a team designed to develop integrated ecological models and supporting the GIS and data needs of research team members. Responsibilities include managing and updating multiple spatial data sets, acquiring and assimilating new relevant spatial data sets, using GIS data and principles of landscape ecology to characterize landscape patterns at multiple scales, providing documentation for collected and newly created spatial data sets, and data integration. Annual salary depends on experience. The position will be based at ORNL in Oak Ridge, Tennessee. All travel expenses are provided. The position is for one year. Qualifications: Master's degree required in Geography, Ecology, Soil Science, Forestry or related field. Required discipline: Other Physical Sciences Applicants cannot have received the most recent degree more than five years prior to the date of application and must complete all degree requirements before starting their appointment. How to Apply: Qualified applicants may apply online at (https://www2.orau.gov/ORNL_POST/ ). All applicants will need to register before they can begin the online application. For complete instructions, on how to apply, please see the instructions at (http://www.orau.gov/orise/edu/ornl/ornl-pdpm/application.htm ). When applying for this position, please reference the position title and number (ORNL08-24-ESD). This appointment is offered through the ORNL Postmaster's Research Associates Program and is administered by Oak Ridge Associated Universities (ORAU). This appointment is open to all qualified U.S. and non-U.S. citizens without regard to race, color, age, religion, sex, national origin, physical or mental disability, or status as a veteran or disabled veteran.
Re: Input of non-normal variables into GLM models
On Tue, 13 Nov 2007 20:07:33 +0200, Anon. [EMAIL PROTECTED] wrote: Sami Ullah wrote: Hey Ecologers: I have a various variables for running multiple linear regression model using GLM. Some of my predictor variables are non-normally distributed. Using multiple linear regression, I use proc-univariate to check if the residuals in the regression model met the normality criteria, which the model did. Now I am wondering if it is advisable if I can keep the skewed predictor variables in the model or have to go for non-parametric analysis? The distribution of the predictor variables is irrelevant, so you can happily keep them in. Well, the distribution is almost irrelevant. You can get problems if they are co-linear (i.e. highly correlated), or if you have outliers (which can have a large influence on the fit). Agree. One extra thingI would argue that normality of explanatory variables (predictors) is actually bad. It means that most observations have the same (or similar) value for that explanatory variable, which may (!) make it more difficult to find a significant effect. Bad experimental design. Perhaps a histogram shaped like the uniform distribution would be the best. It means that you have similar number of observations for each part of you sampled gradient...for that explanatory variable. My suggestion is not to transform anything (dependent and independent), unless you have outliers in your explanatory variables. Or if you have something trivial like weight-length data. I've come across the impression that the predictors have to be normally distributed a few times, but I don't know where it originates from - certainly not from statistical theory. It is a fairy tail that many people seem to believe in. Also the normality of the response variable, the raw data, is such a fairy tail. The assumtpion is that you have normality of your response (dependent) variable at EACH X value, and without 30 replicates or so, you cannot check this. And who has 30 replicates? Kind regards, Alain Dr. Alain F. Zuur First author of: 1. Analysing Ecological Data (2007). Zuur, AF, Ieno, EN and Smith, GM. Springer. 680 p. URL: www.springer.com/0-387-45967-7 2. Analysing Ecological data using GLMM and GAMM in R. (2008). Zuur, AF, Ieno, EN, Walker, N and Smith, GM. Springer. 3. An introduction to R for the life scientists: - With a paper submission guide - (2008). Zuur, AF, Ieno, EN and Meesters, EHGW. Springer Other books: http://www.brodgar.com/books.htm Statistical consultancy, courses, data analysis and software Highland Statistics Ltd. 6 Laverock road UK - AB41 6FN Newburgh Tel: 0044 1358 788177 Email: [EMAIL PROTECTED] URL: www.highstat.com URL: www.brodgar.com Bob -- Bob O'Hara Department of Mathematics and Statistics P.O. Box 68 (Gustaf Hällströmin katu 2b) FIN-00014 University of Helsinki Finland Telephone: +358-9-191 51479 Mobile: +358 50 599 0540 Fax: +358-9-191 51400 WWW: http://www.RNI.Helsinki.FI/~boh/ Blog: http://deepthoughtsandsilliness.blogspot.com/ Journal of Negative Results - EEB: www.jnr-eeb.org =
PostDoc: Population-level effects of Wolbachia
Position Announcement Postdoctoral Research Fellow Vector Research Group University of Kentucky * Lexington, KY Applications are invited for a full-time postdoctoral fellow. The position is available immediately. Wolbachia cause a form of sterility in insects known as cytoplasmic incompatibility (CI), which results in karyogamy failure and arrested embryonic development. In populations that include both uninfected and infected individuals, unidirectional CI can drive the replacement of the uninfected cytotype with the infected cytotype, resulting in the conversion of an uninfected population into an infected population (termed: population replacement). In populations where individuals are infected with different Wolbachia types, bi-directional CI can occur: sterility results in both cross directions between mates infected with different Wolbachia types. Models predict that in natural populations, sterility resulting from bi-directional CI is a transient event, since one infection will predominate and replace the other cytotype. While substantial effort has been devoted to examining the effect of Wolbachia on individuals (e.g., Wolbachia effects on egg hatch, longevity, etc ), relatively little is known about population-level effects of Wolbachia. We seek a collaborator with interest and experience in mathematical modeling, statistics, demography and ecology. The successful applicant will lead the analysis of existing data sets of insect populations that are either infected with Wolbachia or are uninfected. The recruit will also be encouraged to develop additional modeling and/or empirical projects. Applicants should have a Ph.D. The ideal candidate will have a competitive publication record and enjoy working in a multidisciplinary environment. Funding is available for up to four years with annual renewal contingent upon satisfactory performance. Funds are available for participation in professional meetings. Applicants are requested to send a detailed curriculum vita, statement of career goals and research interests, reprints of recent papers, and the names and contact details of three referees. Applications, informal enquiries, and requests for additional information should be addressed to: Stephen Dobson or Charles Fox Department of Entomology University of Kentucky S-225 Agricultural Science Center North Lexington, KY 40546 [EMAIL PROTECTED] http://www.uky.edu/~sdobson http://www.uky.edu/~cfox * University of Kentucky recognized as top ten overall institution for postdocs, The Scientist Survey: Best Places to Work for Postdocs, vol.17.3, Feb. 10, 2003
Job @ UCS: Program Assistant, Invasive Species/Climate Change
The Union of Concerned Scientists is seeking a program assistant for invasive species and climate change. The program assistant helps with many aspects of UCS*s work on invasive species and climate change. These include research; outreach to activists, scientists, Capitol Hill, and the public; creating and preparing materials for the website and other audiences; database management; and other special projects as needed. The program assistant also provides administrative support, and specific support for the scientific staff, as needed. View a full description and application instructions here: http://www.ucsusa.org/ucs/about/jobs-at-ucs.html#Program_Assistant_Climate_Program While the job is likely to close on November 16, 2007, applications are encouraged until the position is filled and the announcement is removed from the UCS website. Apologies for the late notice to this listserve. -Michael Michael Halpern Outreach Coordinator Scientific Integrity Program Union of Concerned Scientists 1707 H Street, NW, Suite 600 Washington, DC 20006-3962 (202) 331-5452 fax: (202) 223-6162 http://www.ucsusa.org/scientific_integrity
Summary of responses about videoconferencing
Here is a summary of responses I received after requesting information about videoconferencing, as a possibility for future meetings of the Governing Board of the Ecological Society of America, or for Board members who can't travel to a particular meeting to participate from their home institutions. The Board meets three times/year, once at the annual meeting and twice in Washington, D.C. (where the Society's office is located). Thanks to those people who contributed suggestions. It looks like there may be some inexpensive (even free) options for voice-only, and maybe for one person to use a video link to the meeting. David Inouye ** Assuming ESA isn't going to spend *big* money for a corporate-type system, Apple's iChat does a nice job, and is compatible with AIM on the PC side. You do need a fast machine and network connection for whoever is hosting the conference. ** A Skypecast might work for voice-only conferencing: https://skypecasts.skype.com/skypecasts/home ** The LTER Network uses PolyCom conferencing units for meetings of its Executive Board. We have had up to 10 conferees at once, and will be attempting a 26 unit conference soon. This technology requires hardware which UNM purchased that allows up to 48 participants at once, or multiple simultaneous conferences. The key issue with any videoconferencing technology is bandwidth. The second issue is firewall interference. ** We use a Polycom (http://www.polycom.com/usa/en/products/video/video.html) video system to communicate between the University of Montana's main campus and UM's Flathead Lake Biological Station. There is also a colleague who connects from his home office in Atlanta. We've had instances of the colleague e-mailing his PowerPoint presentation to the Biological Station and then giving his presentation remotely. The video is good and allows remote control (direction and zoom) of the camera. The communication is via Internet so not appropriate for low bandwidth connections -- when the system is on at the Biological Station, everyone else's Internet slows to a crawl. I don't know much about the specifics, including cost or installation, since it was here before I started my postdoc in June. But I'm impressed with the system -- it is easy to use and works very well. Hope this helps. We also use Skype (voice-only) for daily personal communications and quick conference calls. ** If you hear of any good suggestions, I would love to hear about it myself. We do frequent meetings, mostly teleconference. But when it comes to big ones like Board meetings, many stakeholders feel the need to be right there, burning up untold tons of carbon. For me to get to a meeting at our San Francisco HQ, it requires fully a years worth of my own carbon footprint per trip. And yet, sometimes one's input really does make an important difference. Video conferencing would really make a difference. ** I attended a virtual video interactive powerpoint presentation that worked quite well. After the live slideshow presentation questions were entertained from the audience from across the USA. Unfortunately, I do not know how this was accomplished or the software involved, only that it worked quite well. It was administered by the EPA. So basically, it is very possible to have virtual meetings that are interactive of high quality. Of course, I can not afford to travel to meetings and this would make attendance possible for many people. Also, I do not feel that these national and international meetings are sustainable, and this is a method to have high quality virtual meetings that are sustainable. ** PC Magazine review of video conferencing software 9/18/07 liked GoToMeeting GoToMeeting: https://www.gotomeeting.com/. Says it works with up to 15 attendees. 30-day free trial available. and recommended http://www.sightspeed.com/ as a less expensive alternative. 30-day free trial available. ** TNC often uses WebEx, which presents audio and presentation slides from the meeting location. Outsiders can call in and ask questions over the phone or through a chat feature. This is probably not exactly what you're looking for, but it is a step above a conference call. ** I'm writing in response to your question on Ecolog about video conference recommendations. I am a postdoctoral fellow at the National Center for Earth-Surface Dynamics (NCED). Our main locations are on 8 different campuses and we regularly use videoconferencing to bring together members at 5-6 locations for meetings and presentations. The H.323 based system NCED uses allows both video and sound and also simultaneous
