Discussion: View Thread

Computer packages used by solo biostat consultants?

  • 1.  Computer packages used by solo biostat consultants?

    Posted 08-21-2011 14:43
    I have recently been asked a question that is best asked by the solo consultants out there: what are the computer packages that you use to do your biostatistical consulting?  Obviously I know about lots them, but I just don't know which ones the independent consultants find most useful.  

    Thanks,
    Kathy Monti

    -------------------------------------------
    Katherine Monti
    Rho, Inc.
    -------------------------------------------


  • 2.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 15:12

    I use SAS for analyses. I use EAST, ExpDesign Studio, SiZ, and/or nQuery for sample size, power and such.
    Occasionally I use Excel for quick-and-dirty.
    David
    -------------------------------------------
    David Bristol
    Statistical Consulting Services
    -------------------------------------------








  • 3.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 15:17
    Dave, Now that SAS has the functionality of EAST, is there any advantage to maintaining EAST in your arsenal? ------------------------------------------- Jeffrey Davidson VP Global Biometrics Octagon Research -------------------------------------------


  • 4.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 15:32

    I know that SAS through the glmpower and power procedures does just about anything that nQuery does but I don't think it has the capability to do the group sequential and adaptive design procedures available in EaSt.
    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 5.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 17:23
    Try looking at the SEQDESIGN procedure and SEQTEST as well in version 9.2. ------------------------------------------- Jeffrey Davidson VP Global Biometrics Octagon Research -------------------------------------------


  • 6.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 17:34
    Okay I will grant you that the seqdesign procedure provides group sequential design capability.  The claim that was made was that SAS has full EaSt capability which is not the case.  Vaious adaptive designs can not be constructed with these SAS procs. 

    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 7.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 10:44
    By the way, what is EaST ?

    THank you !

    -------------------------------------------
    Hsin-Yi Weng
    Statistician
    -------------------------------------------








  • 8.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 16:14
    Hi Katherine,

    For solo consultants, there are several factors that inform the choice of statistical software:  1) area of expertise,  2) industry standards,  3) software pricing,  4) project scope and execution time,  5) client preference, 6) consultant preference.  The list likely goes on and I am sure that other consultants would be happy to add to it based on their own experiences.

    I will give some examples below to illustrate how these factors come into play in my view.  

    Because one of my areas of expertise pertains to Mixed Treatment Comparisons, I use R and WinBUGS to tackle projects in this realm.  Both of these software packages are free.  Other consultants use different software packages that reflect their own area of expertise.  For instance, someone working a lot with SEMs would have to use AMOS, M-PLUS or LISREL, say.     

    Industry standars are also important - I believe that SAS is the undisputed (?) golden standard in biopharmaceutical
    research.  If a consultant works in that area, he/she would definitely need access to SAS. However, R is starting to make some inroads into this world as well. 

    Software pricing is another important factor.  Solo consultants are perhaps more likely to work with commercial software packages whose licensing costs are affordable or even free.  For instance, STATA is a lot cheaper than SAS, so why not get STATA if you can still use it to achieve the same results?     

    The project scope is important - if all you have to do is tabulate data, perhaps SAS is the best choice.  If you need to produce graphics, a blend of SAS and R may be the best option.  If you have to conduct specialized statistical computations or analyses, maybe you can refer to a highly specialized software package. 

    If the project has to be executed rapidly, the software that will get you there first while not breaking the bank will be the safest bet.  I recently had to switch from using Matlab to using Octave for one of my consulting projects, as Matlab don't allow you to download one of their trial versions online and you would have to wait for them to ship an installation CD to you. Octave is free, so I could start using it right away.

    Both client and consultant preference are important factors in deciding the choice of software.  If the client would like their project tackled via SPSS, there is no point in extolling the virtues of R, even though R is free.  If the consultant "grew up" with SPSS, they would feel less comfortable using R and recommend to the client that they too use SPSS. 

    For any given project, the above factors (and I am sure, many others) come into play so that the consultant chooses the optimal software or combination of software for that project.  This is part of the mystique of being an effective consultant. 

    Kind regards,
     
    Isabella


    -------------------------------------------
    Isabella Ghement
    Ghement Statistical Consulting Co.
    -------------------------------------------








  • 9.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 09:53
    Most of the respondents have mentioned SAS.  Have you found any way to get a discounted price on SAS?  At NIH we paid about $250 per license, so everyone used it, but now that I retired, SAS quoted me a price of $10k.  I tried to get a license thru the university where I am an adjunct, but had to be on their site to use it.  I have switched everything to R, but do miss the functionality of SAS PROC GLIMMIX.
    Any suggestions as to how I can buy a SAS license for less than $10k?

    -------------------------------------------
    Linda Pickle
    StatNet Consulting, LLC
    -------------------------------------------








  • 10.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:19

    No, SAS has a monopoly and is not willing to work with the small consultant.  I would love to find an alternative to SAS.  I have had a SAS license for over 15 years and wanted to add a 2nd license. They refused to give me a deal or a savings on the 2nd license and then said that if I tried to combine the 2 licenses (so I would have the same anniversary date) that I would have to pay for 2 brand-new licenses.  I also found out that the SAS bundle I have had with SAS for 15 years is obsolete, and several of the modules are no longer functional, but again, to modify the SAS license I would have to start back at the starting price.  It is a major scam and really hurts the small SAS consultant.  They told me they didn't care about us.  They primarily are looking for the big clients.  Again, I wish there was an alternative to SAS, but since most of my clients demand I have a SAS license, there isn't much I can do.  I just wish that we could band together and try to demand some changes from SAS to help out the small consultant.
    -------------------------------------------
    Rebecca Hoagland
    Consulting Statistician
    -------------------------------------------








  • 11.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:32

    There is one package that claims to be an alternative to SAS "WPS"
    http://www.teamwpc.co.uk/products/wps

    But there are news reports about litigation.

    http://www.kdnuggets.com/2010/08/f-sas-says-wps-infringed-on-copyrights.html
    -------------------------------------------
    Chris Barker, Ph.D.
    President - San Francisco Bay Area Chapter of the American Statistical Association
    www,barkerstats.com
    -------------------------------------------








  • 12.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:49
    Sounds like copyright infringement to me.  I would steer clear of this WPL company.  Seems SAS already has a partial victory and I think they will win the rest.

    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 13.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:56

    Maybe SAS will be available, and at a lower price, through cloud computing and  "software as a service"

    http://www.dmnews.com/sas-builds-to-accomodate-new-saas-offerings/article/129178/

    ... Then there will be the new class of small and mid-tier companies ...


    -------------------------------------------
    Chris Barker, Ph.D.
    President - San Francisco Bay Area Chapter of the American Statistical Association
    www,barkerstats.com
    -------------------------------------------





  • 14.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 12:05

    Does this mean having the SAS Institute provide a solution to a specific problem?  I don't understand what cloud computing means.  If it is not software available for the statistical consultant to use as he/she see fit I don't see it being something a consultant would be interested in.  It is like deferring the consulting job to SAS.
    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 15.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 13:22


    As I understand, it is possible to develop SAS code, and "point it" to a server that SAS maintains on the internet, and I presume pay for however many seconds/minutes of computing is needed. The SAS code would run and the output sent to your local drive

    There is a service like this, now by SAS, for K12 teachers.

    http://www.sas.com/govedu/edu/k12/evaas/index.html#s1=4

    ...SAS offers SAS EVAAS for K-12 as a hosted application. Schools can benefit from SAS EVAAS for K-12 analyses without having to invest in new hardware, software or IT staff. Instead, states or districts send electronic data directly to SAS, where the data is cleaned and analyzed. ...

    -------------------------------------------
    Chris Barker, Ph.D.
    President - San Francisco Bay Area Chapter of the American Statistical Association
    www,barkerstats.com
    -------------------------------------------








  • 16.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 14:10
    That sounds like an awkward way to use SAS.  Also if you need to debug your code you might go through some wasted iterations before you get a good solution.

    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 17.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 14:34

    For some of us, its just like the "old days" where you submitted a "job" at the computer center which ran remotely on a "timesharing" computer,maybe an IBM360  say in the next town or next state. I interned at the NCHS and the jobs ran in North Carolina (perhaps at SAS institute)

    And you paid real dollars ($$) for CPU time.

     I am sure a few of us at one time or another used up somebody else's computer budget on an overnight run of an infinite do loop. You learned quickly how to avoid those problems.



    -------------------------------------------
    Chris Barker, Ph.D.
    President - San Francisco Bay Area Chapter of the American Statistical Association
    www,barkerstats.com
    -------------------------------------------








  • 18.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 17:23
    I'm in that "old days" group. While it's a painful way to learn how to write good code (annoyed computer operators were also part of the scene), those experiences ended up being quite beneficial for me in learning about efficiency and resource usage that still help me do my work.

    -------------------------------------------
    Michele Burlew
    Episystems, Inc.
    -------------------------------------------








  • 19.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 17:32

    I am a very old days programmer.  I started programmer in 1965 (age 18).  We used punch cards and IBM 7090 main frame computers.  My statistical computing days didn't start until 1978 when I became a full fledged statistician.
    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 20.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 13:27
    Your statements are untrue - the legal issues are largely unrelated to copyright infringement and the lawsuit was quickly thrown out of court (http://minequest.com/WordPress/?p=571).  There is no reason to avoid WPS unless you want to spend tens of thousands on SAS.

    Hadley

    -------------------------------------------
    Michael Wickham
    Assistant Professor / John Dobelman Junior Chair
    Department of Statistics
    Rice University
    -------------------------------------------





  • 21.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 14:04
    I don't know how you can say that what I said was untrue.  I was just stating an opinion based on what I read from the links.  What SAS won was copyright infringement on the manual.  The other items were deferred to another court.  Apparently you have information about that courts ruling.  I wouldn't say that the court threw out the suit.  It seems that they ruled in favor of WPL about the interpreter being permissible. Aside from what I have read from the links I know nothing about the case.  The basis of the case was copyright infringement.  If the WPS system allows use of SAS without the licensing of SAS then I think there is something wrong and I don't understand the courts ruling.  If you are going to use SAS you should be paying SAS.  If you don't want to pay the 10K then find a different software product.

    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 22.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 14:35
    For data exploration, labeling and recodes, particularly for Market Research applications, I found base SPSS to be helpful in order to do most of the basics.  I find it easier to use than Excel.  It provides nice basic presentation graphs.  SPSS is also better on a PC, even for very large datasets, than SAS.  They have a perpetual license and modules can be added as needed. 

    SAS is for enterprise wide legacy systems.  I would recommend that a consultant be given access to a remote desktop that can work with whatever SAS products are licensed at the organization.  Most large organizations have data security measures in place so that access to data, even within the organization is limited.  As a W-2 contractor, this turned out to be the most efficient way to do work remotely.

    For graphics, spatial statistics, and the latest techniques, definitely R.

    -------------------------------------------
    Elizabeth Smith
    -------------------------------------------








  • 23.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:48
    Each year our companies (SoCalGas and SDG&E) review our collection of SAS licenses and make a concerted effort to reduce what we can.  We'd been a long-time SAS shop; however, I won't be surprised when that ends.  One particular experience I had several years ago involved trying to negotiate use of SAS on a server partition for a custom application--after several conference calls with SAS representaives (sales and techs) with me and one of our best IT staff representives we ended the discussion--we couldn't even get an idea from SAS what they'd charge exactly despite detailed information provided by our IT person on just what the server arrangement involved.

    This experience made me more convinced that, at least for our work which is primarily time series applications, we should look to EXCEL add-in packages and ultimately make a commitment to get smart about using R.

    SAS is nice for the folks who've been raised on it, but the corporate keepers of the budgets aren't convinced it's worth the expense.

    My own preferences are to use proceedural types of code--it's the only way I know of to be able to figure out what happened when questions get raised and this is an important feature to not loose in our "hurry up" pace of businees these days.

    I certainly can sympathize with software developers and problems they have with copyright and piracy,  SAS's business approach seems to have stimulated more and more competitve interest.  I suspect that the group that can get a really good product for the small individual consultant will have a lot of pent-up demand.  Good luck with your search and until then I guess you'll have to try to "recover your SAS license costs" on clinet fees.

    -------------------------------------------
    Bruce Wetzel
    Forecasting Analyst
    Southern California Gas Co.
    -------------------------------------------








  • 24.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 12:20
    Rebecca Hoagland writes:

    > but since most of my clients demand I have a SAS license, there isn't much I can do.

    You should consider charging a higher rate for any client who demands that you have a SAS license. The clients who make such demands are typically those with deeper pockets. You can't recoup the entire $10K expense this way, of course, and you can't exceed the market price for consultants, but it might help somewhat. It would also discourage clients from making a demand for SAS without assessing first whether this demand makes sense.

    -------------------------------------------
    Stephen Simon
    Independent Statistical Consultant
    P. Mean Consulting
    -------------------------------------------





  • 25.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 12:57

    Stephen:  Yes...I do make up the cost of the license from my clients.  What is difficult is trying to expand because it is a catch-22.  I need the SAS license to get the work, but the cost of a SAS license is prohibitive to get if you are not guaranteed a specific number of hours of work.  I think it is one area that keeps one-man shops from expanding.   I also find that alot of Stat consultants want the high hourly rate, but do not want to spend any of that on computers or software.  My clients do pay a fairly good rate, with the understanding that I provide all the equipment.  My problem is trying to find consultants at a reasonable rate to help when I get more work than I can do.  Most want the high hourly rate, but do not want to invest in computer or software.  I can't afford to provide the computer and software plus the high hourly rate.  One or the other has to give. 
    -------------------------------------------
    Rebecca Hoagland
    Consulting Statistician
    -------------------------------------------








  • 26.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:20
    Only discounts i know of for SAS are for academic institutions.  I have been told that Goodnight does this to get graduate students hooked on SAS so that they will continue to use it after they graduate.  Don't know if this is true but it sounds like a good business move that Jim Goodnight would this of.

    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 27.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:20


    -------------------------------------------
    William Grant
    Professor
    SUNY Upstate Medical University
    -------------------------------------------
    Sorry 10K seems to be the going rate.
    Although we are an academic center and like NIH enjoyed a discount apparently SAS has discontinued all discounts.  Our costs would have more than tripled per user.  SUNY is a state institution, while large, without many deep pockets.  So we have collectively gone to SPSS, R and other packages for specific uses (MedCalc, G*Power, etc.)
    IMHO SAS's focus on commercial accounts has the possibly unintended outcome of having a group of medical researchers who 'grow up' without exposure to SAS and its benefits.  But obviously their call.







  • 28.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:23

    I do more SAS programming than statistical consulting. Several years ago, my company became a SAS Alliance Affiliate Partner. Through that relationship, I license SAS for about $3K a year, but it's a restricted license. The Affiliate license allows me to prototype projects and develop code for clients that clients run in production using their SAS licenses. My SAS license does not allow any production work. For the couple of clients I have who do not have SAS, I use JMP. The more and more I use JMP 9, the more I like it, and I find it to be a very good substitute for SAS when the project does not require a macro solution.

    -------------------------------------------
    Michele Burlew
    Episystems, Inc.
    -------------------------------------------








  • 29.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:32

    I agree with Michele. I have not used JMP but I did recently partner as a consultant with Data Vision a statistical consulting firm that did all of its analysis using JMP.  They found JMP a much easier tool to demonstrate to clients.  It certainly is a lot cheaper than SAS and has much of the capabilities of SAS.  I think SAS introduced JMP for small markets like small consulting companies and individual consultants.  Didn't know that the academic discounts for SAS were dropped but it is interesting to know.  For the data miners SAS has produced one of the best packages but it too has a very high price tag to license.
    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 30.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:56
    Michele,

    Can you elaborate on what you mean that your SAS license is "restricted" and that you can't do "production" work? Does that mean you can't run SAS interactively? Or you can't output any datasets or use ODS outputs?

    I haven't heard about this restricted license before. What are you restricted from doing with it?

    Thanks,
    Gabe Farkas

    -------------------------------------------
    Gabriel Farkas
    -------------------------------------------








  • 31.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 12:16
    Gabe -

    The software is not restricted in any way. The Affiliate license is restricted in that I am not allowed to run programs that do production work for a client, ie my license cannot replace a client's license. I can use my license to learn, to prototype a project, and to develop code where I turn the code over to the client for them to run the analyses. Because I don't do interactive statistical exploration and I mostly write systems for clients to use in a production environment where there are plenty of SAS licenses and I write SAS books, this kind of licensing arrangement works well for me. The client must have a SAS license when I develop code for them. As expected, SAS is serious about this part of the Affilate agreeement. The prototyping part has been very useful because I can spec a project before I sign a contract and get access to the client's systems. Here's the link to the Alliance program: http://www.sas.com/partners/ 


    -------------------------------------------
    Michele Burlew
    Episystems, Inc.
    -------------------------------------------








  • 32.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 14:36
    SAS has a reseller program where you can lease SAS from other vendors.  Don't know the details myself but check out

    http://www.sas.com/partners/programs/reseller/index.html



    -------------------------------------------
    James Loughlin
    J2Logic/Loughlin Consulting
    -------------------------------------------








  • 33.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 14:47


    The $100,000 fee for the Platinum program  is a bit out of my range. :)

    http://www.sas.com/resources/asset/103944_0410.pdf


    -------------------------------------------
    Chris Barker, Ph.D.
    President - San Francisco Bay Area Chapter of the American Statistical Association
    www,barkerstats.com
    -------------------------------------------








  • 34.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 15:12
    Just went to yahoo search and typed 'sas resellers'.  Got

    http://sasreseller.com/

    I'm sure there are others.

    -------------------------------------------
    James Loughlin
    J2Logic/Loughlin Consulting
    -------------------------------------------








  • 35.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 15:17


    -------------------------------------------
    William Grant
    Professor, Emergency Medicine
    SUNY Upstate Medical University
    -------------------------------------------
    to quote from their web site:

    "Special pricing and software bundles may be
    available to Small and Mid-sized Businesses (SMB)
    with annual revenue less than $500 million."

    I didn't realize small business meant less than $500 million.....

    It's only a guess but I don't think single consultants are their market







  • 36.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 16:20

    You won't know unless you contact them.  I've never dealt with a SAS/Reseller before.  I just know they are out there and that the reseller program was created for clients that are too small for SAS to deal with. 


    -------------------------------------------
    James Loughlin
    J2Logic/Loughlin Consulting
    -------------------------------------------








  • 37.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 10:17


    -------------------------------------------
    John Bartko
    Consulting Biostatistician
    -------------------------------------------
    Hello Kathy,  I can't afford to use SAS.   In my solo practice I use and have used for about 15 years
    NCSS and PASS.  NCSS is a very powerful package and does all I need, multivariate analysis, graphics, repeated measure ANOVA, Mixed  Models.  It is very affordable and very powerful and very easy to use.  I noticed that it is not among the ones listed so far.  thanks.  John  







  • 38.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 10:22

    I'm not sure about the estimate for SAS below.  I have a yearly renewal for SAS for just one workstation.  My last renewal cost from SAS was $3,528, which includes BASE, STAT, and GRAPH.  If memory serves me correct, the first time user's fee was double the renewal fee.  Every year, the renewal fee has been going up from $50 to $200 (e.g., for the three packages renewal was $2,740 in 2002).


    Allen I. Fleishman, Ph.D. Allen Fleishman Biostatistics Inc. 141 Roundtop Road Marlborough, MA 01752-2774 508-485-9652 (office & FAX) 508-579-9924 (cell) allen-fleishman@comcast.net AllenFleishmanBiostatistics.com





  • 39.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 11:54
    I was trained in SPSS because my Ph.D is in sociology, not statistics, and I like it because it remains more verbal and redundant in explaining procedures and so is more oriented towards how many social science people think an interact in solving problems.  Later I worked for a small foundation where we used SAS, then a large corporation that was standardized on SAS.  Although SAS "procs" are more sparsely explained and technical compared with teh highly verbal SPSS, SAS is like a programming language with a rich set of embedded statistical "procs."  It is very easy with SAS to know exactly how your variables are being processed at each step of a long analysis chain and it is easy to take variables out of a statistical analysis stream to modify them and feed them back in so you can work at different levels of analysis in the same program.  You string programming steps and "procs" together.

    When I left the corporate world to become an independent consultant I wanted to use SAS and the price was higher than alternatives but workable.  When I went to renew and asked the price for renewal, SAS asked me who my clients were -- they wanted to give me a price based on the industrial sector I was serving.  Because I was serving mainly large corporate clients (having just left a large corporation, I naturally had the recent experience to win bids for similar large corporations requiring similar statistical analyses), they gave me an absurdly high dollar figure to renew.  So, I went back to SPSS which offered a flat fee to any non-academic user.

    I really like SAS better than SPSS and miss its greater capabilities and the ways in which it is easy to set up data flows and create programs with embedded "procs" that solve an analytic problems with multiple steps whereas SPSS is clunky and seems to be most usable sequentially.  This makes it easy to lose the chain of proof of analysis should you be asked to produce it at a hearing.  With SAS you whole record of analysis is present and transparent.

    Since then I have found a colleague with a SAS certified shop and co-bid with them to get around this limitation.  I am also aware of others who are independent consultants but arrange to teach one course at a university so are permitted to use the SAS package at the university computing center for free.

    This is just one person's experience but I think it reflects part of the context for the puzzle of getting access to the right tools to be able to practice.

    -------------------------------------------
    Hugh Peach
    H. Gil Peach & Associates, LLC
    -------------------------------------------








  • 40.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 12:45
    There is a GUI interface softaware by SAS called JMP. It Is SAS' answer to SPSS, but a lot more user friendly and intuitive. Also, it is more economical. It can interact with SAS (if you have SAS license), but it is a stand along fully functional software with beautiful graphs. I consider it a good alternative, specially when my customers want to learn some of the analysis processes. ---------------------------------- Anamaria Kazanis, PStat ASKSTATS Consulting ----------------------------------


  • 41.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 12:52

    Hugh reinforces my point about SAS being superior with traceability of the analysis.  I maintain that this is the single most important reason why the FDA prefers SAS.
    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 42.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 13:11

    I work in pharma, I'm not sure what you mean by "traceability" of analysis. SAS has the log file, but there's a lot that the programmers and statisticians do on their own for "audit trails". This is all part of 21 CFR Part II. There's a catchy motto "if it isn't written down it doesn't exist".
    Certainly, we must, for development, keep the "sas logs", put sourcelines on programs and outputs, track program and output date/time stamps, etc.. We use standard "validated" SAS scripts/macros, for creating reports.
    I've worked in companies where there was a (Unix) version control system to keep track of SAS programs in development and production, but that was a UNIX tool.

    -------------------------------------------
    Chris Barker, Ph.D.
    President - San Francisco Bay Area Chapter of the American Statistical Association
    www,barkerstats.com
    -------------------------------------------








  • 43.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 13:39
    Wait a second, how can the FDA prefer one software package over another?  When a government agency steers business to--I mean "prefers"--one commercial product over another without a clear set of criteria and without an evaluation process, that's cronyism and corruption.  Fortunately I don't think that the FDA in fact has a formal preference for SAS.

    It may well be the case that a majority of individual FDA statisticians prefer SAS (that may be the case of statisticians overall), but that doesn't mean submitters are formally obligated to use SAS.

    Also, if traceability of analysis is important, nothing approaches Sweave, usable with S-Plus and R.  It makes copy-and-paste errors impossible, and removes any doubt about how data was processed in a report.  Of course this is only one piece of a system that should include documentation, version control, testing, etc.  The "literate computing" approach motivating Sweave really should be adapted to other statistical languages, but to my knowledge it hasn't been so yet.

    -Jim

    -------------------------------------------
    James Garrett
    Manager, R&D Statistics
    Becton Dickinson
    -------------------------------------------








  • 44.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 14:25

    If you have been following the discussion I have said that the FDA has a preference for SAS in clinical trial analysis. Of course this is not a formal requirement.  Others have mentioned that analyses have been accepted using other software tools including R. Just about all the pharmaceutical companies use SAS for primary analysis for a variety of reasons that we have been discussing.  FDA reviewers all have SAS and I am sure that they prefer reviewing submissions in SAS.  I nthink this is a far cry from an endorsement of SAS and is nothing to get excited about.
    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 45.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 15:28
    I agree with Michael.  In my dealings with the FDA I recall them mentioning many times that they do not endorse one statistical package over another.  I'm sure that some of the newer FDA statisticians probably like R better than they like SAS.  But the FDA does require validation of the statistical package as well as the actual code.  Thus the issue with R. 

    -------------------------------------------
    Rocco Brunelle
    Senior Statistician
    Bowsher Brunelle Smith LLC
    -------------------------------------------








  • 46.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 15:37
    THIS HAS GOTTEN RIDICULOUS AND ANNOYING!!

    What started out as an innocent question about software has turned into an annoying set of
    private conversations that are stuffing up everyone's email boxes.

    Once the basic question has been answered it would be better to just us "Reply to Sender".

    Marty Lesser




    -------------------------------------------
    Martin Lesser
    Director, Biostatistics Unit
    -------------------------------------------








  • 47.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 15:46
    I don't think this is ridiculous or annoying to most of us.  We reply to the sender if we are specifically addressing a question that only the sender is interested in otherwise the information should be shared.  I get email from several egroups and I agree that it can sometimes be irritating.  But I think the right recourse is to just delete the email.  I also think that getting sidetracked off topic is not good.  Perhaps some of the discussion about FDA preferance for SAS is off topic.  If the email really is that irritating you can take yourself off the eGroup list.

    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 48.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 16:13
    It might be nicer if we could have shorter email storms.  Maybe each person could limit themselves to 2 or 3 comments?  It would be unfortunate to have to choose signing off as a way to avoid the interruptions.  If there was a way to read/scan a summary of the dialogue, say all at the end of the day, and shield ourselves from the real-time messages, that would be helfpul too.
    -------------------------------------------
    Daniel Jeske
    University of California Department of Statistics
    -------------------------------------------








  • 49.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 16:25

    In your Amstat profile, under "My community Subscriptions", select "Daily Digest" rather than "real time"

    -------------------------------------------
    Caleb Wright
    -------------------------------------------








  • 50.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 16:26

      You can turn off the emails. and on the right hand side of the email is a 'view thread". You can read it on the website, at the end of the day.
    Messages can be posted, though I haven't tested if that turns off sending them to all the subscribers.
    here's the options
    eGroup Home
    Reply to eGroup
    Reply to Sender
    Post Message
    Forward Message
    Print Message
    View Thread
    Author's Messages
    Search
    My Subscriptions
    Mark As Inappropriate


    -------------------------------------------
    Chris Barker, Ph.D.
    President - San Francisco Bay Area Chapter of the American Statistical Association
    www,barkerstats.com
    -------------------------------------------






  • 51.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 18:23
    You might want to read this document about R and the FDA - http://www.r-project.org/doc/R-FDA.pdf.  In a nutshell, there is no such thing as validated software, but you need to have a well-documented "validated" process.  There is no issue with using R in FDA submissions.

    -------------------------------------------
    Hadley Wickham
    Assistant Professor
    Rice University
    -------------------------------------------








  • 52.  RE:Computer packages used by solo biostat consultants?

    Posted 08-24-2011 15:56
    After I saw the note on the displeasure of our discussions on SAS and R I gave it some thought about why I enjoyed this discussion so much and it surprized me that someone else found it annoying.  I also thought it was interesting that there was a suggestion of limiting our replies.

    Presonally I usually take a peak at any new discussions and I may add my 2 cents.  As an independent statisical consultant (retired from big pharma) there are three things that definitely will peek my interest:
    1 - hourly rate and or cost of a project
    2 - dealing with clients that do not appreciate statistics
    3 - statistical software packages

    I would love to hear more about all three topics and that is why I enjoyed the discussions on which software package we use.  Personally I use SAS and I have gone a whole year without using it but I didn't want to stop paying the yearly lease since it would cost me more the following year.  Also, if I wasn't so lazy I'd learn R but I'm concerned that my clients might want me to use SAS anyways.

    I just want to thank everyone for the great discussions.

    -------------------------------------------
    Rocco Brunelle
    Senior Statistician
    Bowsher Brunelle Smith LLC
    -------------------------------------------








  • 53.  RE:Computer packages used by solo biostat consultants?

    Posted 08-25-2011 08:12


    -------------------------------------------
    Georgette Asherman
    -------------------------------------------
    I spend a lot of time using R now, but not in a project for regulatory review.   R could be used in some statistical applications, especially simulations where it is superior to SAS  Also it vector-based structure fits well with matrix-type research from small well-plates up to protein arrays.   Treating data in a mathematically sensible way is liberating.   Much of what I did in SAS involved subverting the data set to get numbers out of it. 

    However most of what is done in clinical trial submissions is descriptive.   So the major concern is ease of production, ability to combine files from different sources, and deal with exceptions.    The more R I do, the more I hit odd behavior.   Much of this has to do with inconsistencies in data structures and approaches to handling missing data.      For example I took the advice of a popular R reference to loop through summary functions.   I found that because of a documented quirk in the row appending function, rbind, the mininum value was placed in the missing (NA) column.   Let's not get started on the set of apply functions for grouping data. 
    For the unglamorous work of FDA submissions, the workhorses of SAS PROC MEANS, FREQ and now REPORT go far.

    I read through the posted document regarding R.  While the R team does an excellent job, I find it hard to believe that a dispersed team of university employees can be as a focused on testing as dedicated software quality and test staff.





  • 54.  RE:Computer packages used by solo biostat consultants?

    Posted 08-25-2011 13:57


    -------------------------------------------
    John Bartko
    Consulting Biostatistician
    -------------------------------------------
    Hello I am in agreement with Rocco Brunelle's three topics of interest.  They peak my interests as well.  Not that other topics do not of course.  To his 3 I would add psychological aspects of consulting.  John







  • 55.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 14:31
    Submission would not have cut and pasted analyses simply because an FDA reviewer armed with the SAS code can reproduce the output, and log file.  Perhaps a phony time stamp could be pasted without being able to be checked but that is about it.

    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 56.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 14:42

     phony timestamp? Possible in theory, monumentally difficult in practice.

     Outputs are generated in SAS, the sourceline date/time should match the file date/time stamp.
     Before outputs are sent to FDA, most companies I know use some type of commercial software
     to upload outputs into a regulatory document where changes of any kind are tracked (the audit trail)
      The whole process is QA"d by people from the QA department.
    There's a different incentive, in that FDA, at any time, unannounced (or announced) can audit a company.
    That can involve padlocks on filing cabinets, etc. And to prepare for audits, the QA department conducts "internal" or "mock" audits.  The audits can involve interviews of employees, say, about their SOP's and procedures.  European regulatory agencies can also do audits for companies that have offices in Europe.

    -------------------------------------------
    Chris Barker, Ph.D.
    President - San Francisco Bay Area Chapter of the American Statistical Association
    www,barkerstats.com
    -------------------------------------------








  • 57.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 15:02

    I wasn't being serious.  I was just trying to think of something that could be modified without the FDA being able to detect it.  Not sure that I can think of a good reason fopr a company to forge a time stamp anyway.
    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 58.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 15:21

    Every once a great while an investigator/vendor may forge data. But when caught they can go to jail and it makes the front page news.

    http://www.bioportfolio.com/news/article/754591/Fda-Inspections-Find-Cetero-Faked-Records-Manipulated-Samples.html


    ...medicines tested by contractor Cetero Research might have to re-evaluate results, U.S. regulators warned after the firm was found faking documents and manipulating samples, according to Reuters....


    http://www.pharmatimes.com/article/11-08-02/Cetero_claims_data_falsification_was_a_one-off.aspx

    -------------------------------------------
    Chris Barker, Ph.D.
    President - San Francisco Bay Area Chapter of the American Statistical Association
    www,barkerstats.com
    -------------------------------------------








  • 59.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 15:39
    I think you exaggerate about this traceability advantage. All SAS offers is a log file. Construction of an audit trail is much a more sophisticated and comprehensive process and is a more a function of the computing environment than the specific statistical software used. There are many version control programs that can be used in this respect.

    I think FDA does indeed 'prefer' SAS, but this is more a function of SAS' dominant position in the industry and the fact that just about everybody at the FDA and the industry is conversant in SAS. That does not make SAS better. In fact, if the submissions conform to CDISC standards, there is no particular reason to prefer one software over another. All that matters is the file structure.

    Big pharma uses SAS for many reasons; among others is that historically they have had huge statistical programming departments where everybody uses SAS. R and S-Plus are making inroads in generation of figures but they are still very far from challenging SAS in any other area. In fairness, SAS has earned their dominance over the years, but nowadays many other software products provide equal or better functionality in some aspects (Sweave and graphics in R for example) and not so good in others (data management).

    As for validation, it is not clear to me what it means to be '21 CFR part 11 compliant', but R has a document addressing this issue: http://www.r-project.org/ (click on Certification on the right hand-side). I have not had a chance to read it.

    Regards,

    -------------------------------------------
    Carlos Alzola
    President
    Data Insights Inc.
    -------------------------------------------








  • 60.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 15:16

    As a solo consultant I do everything I can in R.  It free, powerful and totally customizable for each client's individual needs.  You also can't beat it for professional graphics.  Occasionally I have a biostats project that requires me to use SAS.
    -------------------------------------------
    Heather Krause
    Datassist
    -------------------------------------------








  • 61.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 15:49
    I mostly use R. I also use PASS for power analysis. I occasionally use Stata (e.g. survey estimation), SAS (multiple imputations) and SPSS/PASW (if clients insist). Winbugs for Bayesian analysis.

    -------------------------------------------
    Moni Neradilek
    Statistical Consultant
    The Mountain-Whisper-Light Statistics
    -------------------------------------------








  • 62.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 16:03
    Since WinBugs is free there is no reason to abandon it.  However,hyou can now do MCMC with SAS.

    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 63.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 16:10
    I primarily use SAS since it is required of the industry I work in without a lot of extra validation.  I also do some graphics in R, use StatXact when requested by the Client and have used nQuery in the past for sample size calculations.  SAS is very expensive, but most of the companies I work with require it, so it is a cost that has to be absorbed for consulting. 

    -------------------------------------------
    Rebecca Hoagland
    Consulting Statistician
    -------------------------------------------








  • 64.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 16:39
    As the complexity of the analysis increases I move from Statistix, to JMP, to SAS.  I find the speed and ease of use decrease as one follows that sequence, but the cost increases.

    -------------------------------------------
    Jon Baldock
    Baldock Statistical Services (formerly Agstat)
    -------------------------------------------








  • 65.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 17:36


    -------------------------------------------
    Edith Zang
    Independent Consultant
    NYCASA
    -------------------------------------------
    STATA and BMDP are both reasonably priced and contain most of the common statistical procedures you might need.





  • 66.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 17:39

    You could add Minitab to that list.
    -------------------------------------------
    Michael Chernick
    Director of Biostatistical Services
    Lankenau Institute for Medical Research
    -------------------------------------------








  • 67.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 18:07

    While not specifically designed for biostats, JMP is a comprehensive statistical package with many analysis and visualization features that researchers, engineers, statisticians, and consultants use to solve complex problems. In particular, the design of experiments (DOE) platform is outstanding, allowing for D-, I-, and alias optimal designs and analysis. JMP has excelllent capabilities in basic analysis and graphing, data mining, reliability, SPC, linear and nonlinear regression, modeling, multivariate analysis, neural nets, and bootstrapping. Additionally, JMP has its own scripting language to create desired analysis reports. JMP is truly a statistical discovery software that is quickly learned and easy to use. Plus, JMP is a division of SAS, and so we know the results are reliable. A free 90 day trial can be downloaded at www.jmp.com. We use JMP extensively at my company.
    -------------------------------------------
    David Trindade
    Fellow
    Bloom Energy
    -------------------------------------------








  • 68.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 11:44
    I use Systat as my primary full featured package.  It has a very broad feature set and excellent graphics.  It will also import a lot of different data formats, so if you run into data that's been developed in, for example, JMP, its easy to do a direct import. The 64 bit version will handle very large data sets.  I also use Statistix, which has a more limited feature set, but is fast, inexpensive, and is nice for the sorts of problems that people often do in Excel.  I also note that Statistix has been around for 20 years or so and is numerically solid, which Excel procedures sometimes are not.  I also have a copy of STATA that I use primarily for working with survey data and R because its free and represents the cutting edge. 

    One thing to consider is that packages with "perpetual licenses" are a better deal for a solo practitioner than packages that require annual renewal.  If your cash flow is down in a given year you won't lose your software because you did not pay your annual license fee.   Having more than one package is a good idea for at least two reasons.  First, something little and quick like Statistix is a lot easier to use for initial examination of a data set than say Systat or R.  Second, being able to do the "same" analysis in different software packages can be a useful QA step in the sense that if one implements the analysis twice in different command syntax, and gets the same result, confidence that the analysis is implemented as specified is increases.  (Note that this does not prove that that an analysis is correct; It just suggests that the analyst did what they intended.) 

    Finally I am not sure something like SAS is necessary for many solo practitioners.  That is, a lot of my work is helping clients figure out what to do and getting an analysis set up to do it.  Once this is done the final analysis can be rerun in SAS (most large companies have SAS licenses), which, as noted above, is a useful QA step.  

    -------------------------------------------
    Michael Ginevan
    M.E. Ginevan & Associates
    -------------------------------------------








  • 69.  RE:Computer packages used by solo biostat consultants?

    Posted 08-21-2011 20:09
    I work in the clincial area and I use SAS.  I also use PASS for sample size estimates. 

    -------------------------------------------
    Rocco Brunelle
    Senior Statistician
    Bowsher Brunelle Smith LLC
    -------------------------------------------








  • 70.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 10:19
    I use JMP for virtually everything and am quite happy with it.  I used SAS before the SAS Institute came out with JMP.  A lot of things are more quickly and easily done in JMP and saving time means being able to deliver a project using fewer billable hours.  And the cost to buy or license JMP is much lower than the cost for SAS.
    I sometimes want to try sophisticated analyses that SAS does but JMP does not.  I often have to create a chain of data files to get from what the client supplied to something I can analyze.  Documenting that in SAS is automatic because it's code driven but writing code takes more time than using pull downs.  JMP has some rudimentary ability to track what I did, but I find it useful to write short word documents in case questions arise later.   

    A lot depends on the user.  I'm comfortable with pull-down products like JMP but I'm equally comfortable with writing code.  Some people may be more comfortable with one or the other.  I have not tried any other packages and it could be that the "most useful" package is whatever package one has the most experience with.

    Caveat:  Most of my biostatistical experience is non-clinical (stability and CMC).
    -------------------------------------------
    I do my consulting part time.  The views expressed on this Web site are mine alone and do not necessarily reflect the views of my day time employer.
    -------------------------------------------
    Emil M Friedman, PhD
    65 Cables - Unit #06
    Waterbury, CT 06710
    203-527-8260 (land line at home)
    216-287-0821 (cell)
    203-790-2507 (land line at work)
    emil.friedman@alum.mit.edu  (work)
    emilfrie@alumni.princeton.edu  (home)
    http://www.statisticalconsulting.org



  • 71.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 10:25
    Hi

    I consult with all sorts of people, not just in biostats. I use SAS and R.

    I like SAS for ease of data handling, nice textual output, outstanding tech support, and general familiarity.

    I like R for graphics and for cutting edge statistics.

    Peter

    -------------------------------------------
    Peter Flom
    -------------------------------------------








  • 72.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 13:18
    A note of caution regarding fitting mixed effects models using JMP.  JMP does not have, nor does it claim to have, the full statistical algorithm capabilities of SAS.  My colleagues and I have noticed that, in SOME cases with mixed effects models, JMP cannot duplicate a correct model fit that has been correctly fit in SAS. Hence, we rely on R or SAS when fitting mixed effects models.  JMP certainly produces nice graphics and has excelllent exploratory data capabilities and strong DOE capabilities.  JMP was initially marketed primarily to engineers (hence, strong DOE).  It later found application as a front end to JMP-Genomics, but without the statistical algorithms contained in the "Genomics" portion of JMP-Genomics, the standalone JMP software does not provide the full statistical capabilities for mixed modeling as does SAS.  

    -------------------------------------------
    Bryan Vinyard
    Director
    USDA/ARS/Biometrical Consulting Service
    -------------------------------------------








  • 73.  RE:Computer packages used by solo biostat consultants?

    Posted 08-22-2011 17:23


    -------------------------------------------
    John Bartko
    Consulting Biostatistician
    -------------------------------------------
    Kathy, I use NCSS and PASS.  I would never use SAS given my practice and its cost.  I have used NCSS and PASS for at least 15 years.  Very friendly and does all I need, mixed models, multivariate analyses, graphics, good regression, cluster analysis, factor analysis, etc.  John







  • 74.  RE:Computer packages used by solo biostat consultants?

    Posted 08-23-2011 19:54


    -------------------------------------------
    John Bartko
    Consulting Biostatistician
    -------------------------------------------

    I use NCSS and PASS.  Can't afford SAS.  I have used NCSS and PASS for the last 15 years,   NCSS does all I need, multivariate, repeated measure ANOVA, mixed models, graphics, regression models.  NCSS is reasonably priced and has no licensing structure.  Recommend them.  John Bartko