Re: [amelia] ameliabind and zelig

Hi One, I am one of the Amelia authors, but I am also currently working on improving the latest release of Zelig. Zelig has had recent issues with multiply imputed data and I am currently working on addressing these systemically. There should be a new version of Zelig released next week that I can certify for multiply imputed data. I am sorry for the delay and hope you can be patient for this pause. If you would like me to use your dataset as part of my battery of tests and diagnostics as I am working towards this, please send it to me off list, and I will check that it works correctly (and delete it when the development cycle is completed), otherwise, I will be certain to simulate an example like you constructed. Feel free to contact me on or off list with any questions, James. -- James Honaker, Senior Research Scientist //// Institute for Quantitative Social Science, Harvard University ________________________________ From: amelia-bounces(a)lists.gking.harvard.edu [amelia-bounces(a)lists.gking.harvard.edu] on behalf of One Hwang [one(a)healingzoom.com] Sent: Wednesday, March 27, 2013 10:21 PM To: zelig(a)lists.gking.harvard.edu Cc: amelia(a)lists.gking.harvard.edu Subject: [amelia] ameliabind and zelig Hi, I want to combine multiple amelia runs, and use the zelig function on the entire set of amelia outputs. However, I noticed that when I run zelig on the combined set, the coefficients don't look right. Here are the script and the R outputs: ## Amelia II: Multiple Imputation ## (Version 1.7, built: 2013-02-10) Attaching package: ‘zoo’ The following object(s) are masked from ‘package:base’: as.Date, as.Date.numeric ZELIG (Versions 4.1-3, built: 2013-01-30) -- Imputation 1 -- 1 2 3 4 5 6 7 8 9 10 11 12 13 14 -- Imputation 1 -- 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Amelia output with 2 imputed datasets. Return code: 1 Message: Normal EM convergence Chain Lengths: -------------- Imputation 1: 14 Imputation 2: 15 Rows after Listwise Deletion: 96 Rows after Imputation: 171 Patterns of missingness in the data: 8 Fraction Missing for original variables: ----------------------------------------- Fraction Missing year 0.00000000 country 0.00000000 tariff 0.33918129 polity 0.01169591 pop 0.00000000 gdp.pc 0.00000000 intresmi 0.07602339 signed 0.01754386 fiveop 0.10526316 usheg 0.00000000 # run zelig on the first amelia run Call: lm(formula = formula, weights = weights, model = F, data = data) Residuals: Min 1Q Median 3Q Max -29.686 -11.067 -3.367 10.003 45.556 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 3.317e+01 1.891e+00 17.538 < 2e-16 *** polity -1.060e-01 2.434e-01 -0.435 0.664 pop 2.925e-08 5.405e-09 5.411 2.15e-07 *** gdp.pc -2.738e-03 5.235e-04 -5.231 4.99e-07 *** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 16.62 on 167 degrees of freedom Multiple R-squared: 0.3234, Adjusted R-squared: 0.3112 F-statistic: 26.6 on 3 and 167 DF, p-value: 4.07e-14 # run zelig on the second amelia run Call: lm(formula = formula, weights = weights, model = F, data = data) Residuals: Min 1Q Median 3Q Max -40.13 -11.72 -3.63 10.16 43.77 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 3.400e+01 1.984e+00 17.133 < 2e-16 *** polity -6.123e-02 2.545e-01 -0.241 0.81 pop 3.289e-08 5.689e-09 5.781 3.56e-08 *** gdp.pc -3.016e-03 5.501e-04 -5.484 1.51e-07 *** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 17.47 on 167 degrees of freedom Multiple R-squared: 0.3495, Adjusted R-squared: 0.3378 F-statistic: 29.91 on 3 and 167 DF, p-value: 1.574e-15 # run zelig on the combined amelia runs Call: lm(formula = formula, weights = weights, model = F, data = data) Residuals: Min 1Q Median 3Q Max -29.686 -11.067 -3.367 10.003 45.556 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 3.317e+01 1.891e+00 17.538 < 2e-16 *** polity -1.060e-01 2.434e-01 -0.435 0.664 pop 2.925e-08 5.405e-09 5.411 2.15e-07 *** gdp.pc -2.738e-03 5.235e-04 -5.231 4.99e-07 *** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 16.62 on 167 degrees of freedom Multiple R-squared: 0.3234, Adjusted R-squared: 0.3112 F-statistic: 26.6 on 3 and 167 DF, p-value: 4.07e-14 Can someone please tell me why summary(z.out3) is returning the exact same results as summary(z.out1)? How can I fix this problem so that the ameliabind output is treated the same as a direct output from the amelia function? Thanks.