Hi Brian,
That error message means the the algorithm is not converging, probably
because there is very little information for ei within your dataset. The
best fix is to change the value on the prior for rho, which is erho in eiR.
For example, changing to from erho=.5 to erho=1 will probably fix this.
Thanks for the e-mail, we will make sure to add a more informative error
message for this for next time! Also, let me know if for some reason this
doesn't fix this.
Best,
Molly
On Fri, Jun 8, 2012 at 12:47 AM, Brian Reed <brianreedbyu(a)gmail.com> wrote:
To whom it may concern,
I am trying estimate the percentage of Mormons in Utah who voted
republican across time in gubernatorial, senatorial, presidential, and
attorney general elections. I have data sets for nearly every election
dating back to 1960. These data sets report the proportion of Mormons out
of the total population at the county level for every county in Utah. The
data sets also report the proportion of voters who voted republican at the
same county level and the total number of voters in each county. For the
majority of the data sets, the "ei" program has worked fine. However, for
two of the data sets, the "ei" program reports an error message that I
haven't figured out. The error message appears after the "dbuf<-ei()"
command. Here is the exact text (the first three lines appear just before
the error message):
[1] "Running 2x2 ei"
Maximizing likelihood
Importance Sampling..
Error in if (any(lower > upper)) stop("lower>upper integration limits")
:
missing value where TRUE/FALSE needed
To further clarify my problem, I've attached a data set for which "ei"
worked fine--"2000Attgen.csv"--and a data set for which "ei" reported
the
above error message--"2004Attgen.csv". You can upload these files into R
and then run the commands in the attached "Commands.R" script file. The
"Printout from commands.R" script file presents the printout I see when I
run the same commands I've attached.
My questions are: What exactly is the error message telling me? And how do
I fix the error?
Thank you in advance for your assistance. Please feel free to respond to
this e-mail address if you have further questions.
Regards,
Brian Reed
Research Fellow
Center for the Study of Elections and Democracy (CSED)
Brigham Young University
-
---
EI mailing list served by HUIT
Subscribe/Unsubscribe:
http://lists.gking.harvard.edu/mailman/listinfo/ei
More info on ei and eiR:
http://GKing.Harvard.edu/eiR
EI mailing list
EI(a)lists.gking.harvard.edu
https://lists.gking.harvard.edu/mailman/listinfo/ei