POLS 6386 MEASUREMENT THEORY
Tenth Assignment
Due 8 April 2003


  1. In this problem we are going to use Epsilon to create a file of paired candidate names that match the Thermometer "roll call" files we used in question 2 of homework 9. We are going to do this for the fourteen 2000 candidates/parties. To illustrate how this works, download the files below and bring them up in Epsilon so that your screen looks exactly like the image below:

    Listing of 1968 Feeling-Themometer Political Figures


    With the cursor in the top window at the beginning of the file (the WALLACE line), type, in turn:

    ALT-Xload-bufferyank_back_down.txt
    ALT-Xload-bufferntimesn-2.txt
    ALT-Xload-buffermacro_hw10.txt

    Be sure to enter the commands in this order or Epsilon will complain because the first macro (hw-10ybd) is called by the second (ntimesn-2) and the second is called by the third (hw-10test).

    Now type:

    ALT-Xhw-10test

    and you should see:



    1. Trace your way through these macros and modify them to work with the fourteen 2000 candidates/parties. Use the file below with the altered macro:

      Listing of 2000 Feeling-Themometer Political Figures/Parties

    2. OPTIONAL FOR EXTRA CREDIT (up to 50 POINTS)Write your own Epsilon command line macro that produces the unique pairs of names and turn it in along with instructions on how to use it. The number of extra credit points will be determined by how elegant your solution is!

  2. In this problem we are going to use my Optimal Classification (OC) program to analyze the thermometer scores as roll call votes like we did for question 2 of Homework 9 only now we are going to invert the role of the respondents and the groups! In this instance, the groups are the rows of the roll call matrix and the columns are unique pairs of respondents! Suppose respondent 1 gives a Thermometer score of 50 to Wallace and respondent 2 gives a Thermometer score of 60 to Wallace. Then Wallace "votes" Yea on this "roll call." Continuing, suppose respondent 1 rates Humphrey a 97 and respondent 2 rates Humphrey a 97. They Humphrey "votes" Abstain on this "roll call." This process is illustrated in the Table below:
                Respondent   
                   1   2      "Roll Call 1-2"
     WALLACE      50  60            6
     HUMPHREY     97  97            0
     NIXON        85  65            1
     MCCARTHY     60  70            6
     REAGAN       40  30            1
     ROCKEFELLER  40  40            0
     LBJ          75  97            6
     ROMNEY       30  50            6
     R.KENNEDY    80  97            6
     MUSKIE       60  97            6
     AGNEW        40  50            6
     LEMAY        30  60            6
    Continuing this process for all unique pairs of respondents produces a "roll call" matrix of 12 by 890,445 for the 1968 Thermometer data. For 2000, the "roll call" matrix is 14 by 1,069,453. This process is discussed in more detail in the assigned reading Non-Parametric Analysis of Binary Choice Data. Note that, ignoring the abstentions -- the "0"s -- there are only 212 or 4,096 unique patterns of "roll calls" for the 1968 Thermometer data. For 2000, there are only 214 or 16,384 unique patterns of "roll calls."

    We are going to analyze a subset -- 21,000 roll calls -- of these "roll call" matrices with a large version of the Optimal Classification (OC) program. Download the program, "control card" files, and two data files below:

    Maximum Classification Scaling Program (PERFLBIG)

    1968 Thermometer-Roll-Call File (VOTE_THERM_1968_RESP.ORD, 12 records, 11meg)

    2000 Thermometer-Roll-Call File (VOTE_THERM_2000_RESP.ORD, 14 records, 15meg)

    1. Run PERFLBIG on the 1968 data. Your PERF21.DAT file should look something like this:
       29 MARCH     2003  15.26.47.00.
       RANDOM NUMBER SEED     76200
      \OLDTIME\VOTE_THERM_1968_RESP.ORD                               
      NON-PARAMETRIC MULTIDIMENSIONAL UNFOLDING OF THERMOMETER DATA   
          221000   10   36    2
      (36A1,100(3900I1))                                              
      (I5,1X,36A1,2I5,50F8.3)                                         
       ******************************************************************************
        1 ROLL CALLS   2   16742  188987  0.08859  0.91141  0.73734           0.00000
          LEGISLATORS  2   16855  188987  0.08919  0.91081  0.73557  0.00000
        2 ROLL CALLS   2   16712  188987  0.08843  0.91157  0.73781           0.99719
          LEGISLATORS  2   16716  188987  0.08845  0.91155  0.73775  1.00000
        3 ROLL CALLS   2   16694  188987  0.08833  0.91167  0.73809           0.99775
          LEGISLATORS  2   16701  188987  0.08837  0.91163  0.73798  1.00000
        4 ROLL CALLS   2   16687  188987  0.08830  0.91170  0.73820           0.99833
          LEGISLATORS  2   16696  188987  0.08834  0.91166  0.73806  1.00000
        5 ROLL CALLS   2   16686  188987  0.08829  0.91171  0.73822           0.99844
          LEGISLATORS  2   16697  188987  0.08835  0.91165  0.73805  1.00000
        6 ROLL CALLS   2   16683  188987  0.08828  0.91172  0.73826           0.99859
          LEGISLATORS  2   16690  188987  0.08831  0.91169  0.73816  1.00000
        7 ROLL CALLS   2   16683  188987  0.08828  0.91172  0.73826           0.99876
          LEGISLATORS  2   16686  188987  0.08829  0.91171  0.73822  1.00000
        8 ROLL CALLS   2   16682  188987  0.08827  0.91173  0.73828           0.99830
          LEGISLATORS  2   16688  188987  0.08830  0.91170  0.73819  1.00000
        9 ROLL CALLS   2   16682  188987  0.08827  0.91173  0.73828           0.99846
          LEGISLATORS  2   16682  188987  0.08827  0.91173  0.73828  1.00000
       10 ROLL CALLS   2   16132  188987  0.08536  0.91464  0.74691           0.99795
          LEGISLATORS  2   16136  188987  0.08538  0.91462  0.74685  1.00000
       MEAN VOLUME LEG.   0.0020   0.0000   1.0000   0.0000   0.0000
       MACHINE PREC.   2   16133  188987   0.08537   0.91463
                            15.26.47.03.
       ELAPSED TIME OF JOB  15.34.16.51.
      Turn in your PERF21.DAT file.

    2. The first 20 lines or so of the coordinate output file PERF25.DAT should look something like this:
       29 MARCH     2003  15.26.47.00.
          1  WALLACE                             102917027   0.940   0.002   0.674  -0.297
          2  HUMPHREY                             99817534   0.943   0.002  -0.607   0.398
          3  NIXON                               193117642   0.891   0.002   0.389   0.263
          4  MCCARTHY                            128315966   0.920   0.002  -0.162  -0.936
          5  REAGAN                              199215712   0.873   0.002   0.392   0.122
          6  ROCKEFELLER                         106715510   0.931   0.002  -0.281  -0.959
          7  LBJ                                  80617529   0.954   0.002  -0.459   0.617
          8  ROMNEY                              115611186   0.897   0.002  -0.046   0.483
          9  R.KENNEDY                           214017710   0.879   0.002  -0.503   0.066
         10  MUSKIE                              137014832   0.908   0.002  -0.620  -0.073
         11  AGNEW                               131313696   0.904   0.002   0.514   0.311
         12  LEMAY                               105114643   0.928   0.002   0.708   0.005
          1    1   3   7   1  10   2  0.667  0.438  0.605  0.797
          2    2   4   5   2   8   4  0.500  0.089 -0.266 -0.964
          3    3   8   4   1   9   3  0.750  0.401  0.346  0.938
          4    4   3   8   3   1  11  0.000 -0.591  0.993  0.115
          5    5   5   7   2   3   9  0.600 -0.177  0.175  0.985
          6    6   3   9   0   3   9  1.000 -0.270  0.475  0.880
          7    7   4   8   0   8   4  1.000  0.292  0.884 -0.467
                            etc etc etc
      Use R to plot the 12 candidates in two dimensions. This plot should be very similar to the one you did for question 2 of homework 5 and question 3 of homework 7.

    3. Place hypothetical voters at Wallace's position, Humphrey's position, and Nixon's Position -- let them be voters 1, 2, and 3 respectively. Fill in the "roll call" matrix below that is implied by their positions.
      
                     "Roll Call 1-2"  "Roll Call 1-3"  "Roll Call 2-3"
       WALLACE              ?                ?                ?
       HUMPHREY             ?                ?                ?
       NIXON                ?                ?                ?
       MCCARTHY             ?                ?                ?
       REAGAN               ?                ?                ?
       ROCKEFELLER          ?                ?                ?
       LBJ                  ?                ?                ?
       ROMNEY               ?                ?                ?
       R.KENNEDY            ?                ?                ?
       MUSKIE               ?                ?                ?
       AGNEW                ?                ?                ?
       LEMAY                ?                ?                ?
    4. Although there are 212 or 4,096 possible voting patterns for the 1968 candidate configuration, not all of these are consistent with the configuration in two dimensions! Give an example of a pattern that is impossible given the 1968 configuration in two dimensions.

    5. Run PERFLBIG on the 2000 data. Turn in a copy of your PERF21.DAT file.

    6. Use R to plot the 14 candidates/parties in two dimensions. This plot should be very similar to the one you did for question 2 of homework 8.

    7. Place hypothetical voters at Bush's position, Gore's position, and Nader's Position -- let them be voters 1, 2, and 3 respectively. Fill in the "roll call" matrix below that is implied by their positions.
      
                     "Roll Call 1-2"  "Roll Call 1-3"  "Roll Call 2-3"
       CLINTON              ?                ?                ?
       GORE                 ?                ?                ?
       BUSH                 ?                ?                ?
       BUCHANAN             ?                ?                ?
       NADER                ?                ?                ?
       MCCAIN               ?                ?                ?
       BRADLEY              ?                ?                ?
       LIEBERMAN            ?                ?                ?
       CHENEY               ?                ?                ?
       HILLARY              ?                ?                ?
       DEMPARTY             ?                ?                ?
       REPUBPARTY           ?                ?                ?
       REFORMPTY            ?                ?                ?
       PARTIES              ?                ?                ?
    8. Although there are 214 or 16,384 possible voting patterns for the 2000 candidate configuration, not all of these are consistent with the configuration in two dimensions! Give an example of a pattern that is impossible given the 2000 configuration in two dimensions.

  3. In this problem we are going to look at the roll call vote in the House on 21 March 2003 that expressed "Support and Appreciation of the Nation for the President and the Members of the Armed Forces Who are Participating in Operation Iraqi Freedom". The vote division was 392 Yea, 11 Nay, 22 Present, and 10 Not Voting. We are going to create coordinates for 107th House using the Quadratic-Normal (QN) scaling program and then produce a smoothed histogram of the combined Nay and Present voters on the Iraqi resolution.

    Download the program, control card file, and the 107th roll call data and place them in the same directory.

    Quadratic-Normal Scaling Program (QUADMD00)

    Control Card File for QN Program (QUADSTRT.H107 -- rename to QUADSTRT.DAT!)

    107th House Roll Call Data

    1. Use R to make a smoothed histogram of the Republicans and Democrats in the 107th House using the estimated coordinates above. Be sure to label everything appropriately. The plot should look a lot like the one you did for question 3.a of Homework 8.

    2. Use R to make a smoothed histogram of 33 Nay and Present voters on the Iraqi resolution. Some of these 33 may not have served in the 107th but most of them did. Be sure to label everything appropriately.

    3. Use Epsilon to change the number of dimensions to "2" in QUADSTRT.DAT and run the program again (be sure to save QUAD0048.DAT from the one dimensional run as it will be overwritten). Use R to plot the legislators in two dimensions. Use "D" for Non-Southern Democrats, "S" for Southern Democrats, "R" for Republicans, and "P" for President Bush. This graph should be in the same format as the one you did for question 3.b of Homework 8. Turn in the R code and the plot. Be sure to label everything appropriately.