Different Results

Comments

14 comments

  • Avatar
    DougO
    Are the dates on your odf files (the data files) the same? Are the total direct effects the same in the output impacts? Are you viewing the results in the same year of dollars (dropdown box to the immediate left of the impact results windows). When you say they are different - are the differences large in the 7th significant digeit and beyond. Different cpu chips use different math co-processors resulting in differences in how numbers get handled. The 4-byte reals used in IMPLAN only carry 7-significant digits. In user preferences (under file) make sure you are both using "Trade Flows" and using the same default year. Let me know if this doesn't clear up the differences.
    0
    Comment actions Permalink
  • Avatar
    katelll
    The Total Effect in the Total Impact Summary for Employment 383.1 for me and 373.0 for him. We are using the same years and the same dollar amounts (though that doesn't change the employment numbers). We are both using trade flows and the same default year.
    0
    Comment actions Permalink
  • Avatar
    DougO
    What are the comparable total [u]direct[/u] [u]output[/u] impacts? What are the dates of your .odf data files? on the appliance:/implan/data files then year and state of the data files directory.
    0
    Comment actions Permalink
  • Avatar
    katelll
    The differences in the output are in the induced effects. The date on one data file is 12/9/2010 5:41PM and the date on the other one is 01/14/2011 5:09PM.
    0
    Comment actions Permalink
  • Avatar
    DougO
    Curious, the date on the current 2009 data is 12/9/2010, which makes me believe that the machine holding the 01/14/2011 data, has outdated software as one of the updates maintains the date of the file (rather than changing the date to the date written to the appliance). Make sure that both machines have the current updates and create new models (you can import the impact activity from the old model) and retry. The current software is 3.0.10.2. One of the updates reloads the 2009 structural matrices which changes some deflators which, I think, may be the source of your differences.
    0
    Comment actions Permalink
  • Avatar
    katelll
    My IMPLAN says it's up to date. I've even just finished updating windows as well. Should I try to re-upload the data files to my IMPLAN? If you think so, how exactly would I go about doing that?
    0
    Comment actions Permalink
  • Avatar
    DougO
    The confirmation e-mail from your order would have the order number. You would go to Help?Download Data Order to re-download. If you no longer have the receipt you can call MIG (651-439-4421) to have them look you up.
    0
    Comment actions Permalink
  • Avatar
    katelll
    I just re-uploaded the data, but that still didn't correct my differences. I don't know what else to try. Is there a reset button of sorts or anything?
    0
    Comment actions Permalink
  • Avatar
    DougO
    Two things to check: 1) Both sets of odf data files now have the same file date. 2) In the direct effect - do you have 2009 as the event year in both? If not, do they both display the same deflators?
    0
    Comment actions Permalink
  • Avatar
    katelll
    The date on my data files is now 3/28/2011, it's still not the 12/9/2010 like the other one. My IMPLAN also still says that it's up-to-date. In addition, 2009 is the event year in both, so the deflators are the same.
    0
    Comment actions Permalink
  • Avatar
    DougO
    The software might say that update is complete and not have done anything. Did you confirm that both machines have the current software version (3.0.10.2)? Version number can be found under Help>About.
    0
    Comment actions Permalink
  • Avatar
    DougO
    If the update is not updating call MIG and they can get the new updater installed.
    0
    Comment actions Permalink
  • Avatar
    DougO
    One of the techs tells me that it does sound like you have one of the origianl November 2009 devices which has the old updater and you will need to call in. Just re-read and saw that the difference is in the induced effect. One other possibility (if the update doesn't fix it) is that the preferences may have been changed on one of the devices (File>User Preferences). The default is internalizing only the household institutions. If one of the installations has checked some of the other institutions, then that would increase the induced effect.
    0
    Comment actions Permalink
  • Avatar
    katelll
    I finally got the same numbers. The preferences were different. Two other institutions had been checked. I just ran a few other models, and as far as I can tell, it is fixed. Thank you so much for your help.
    0
    Comment actions Permalink

Please sign in to leave a comment.