1) All columns have a name;
    
    2) All Type 3's  --  those supervisors could not rate as Type 
        1 or Type 2 but about whom they could only say 
        "uncertain"  --  have been moved to the bottom of the 
        file to facilitate using just the Type 1's and Type 2's 
        as the training set;

    3) A new column has been added (titled "Supr") with a value 
        (1, 2 or 3) differentiating the three supervisors one 
        from another;

    4) All *'s as far as I know have been removed;

    5) Three rows each of which had close to half their values 
        missing have been deleted; all remaining rows have I 
        believe at most 1 or 2 missing values; and lastly

    6) At the very bottom of the file are separately (from the 
        rest of the observations) set out two observations  --  
        56 and 58  -- which rep the answers from two people who 
        it is strongly believed are both Type 1's  -- altruistic.
        Despite this supposed commonality, they only agreed on 82 
        answers vs. 67 which I believe they would have agreed on 
        if the second person had been choosing at random. I will 
        attempt some regression to see if this is a significant 
        difference (82 vs 67, out of 200).

    
    Another thought I had:

        It might be interesting to count the number of 1's ("agree"), 
        2's ("disagree") and 3's ("unsure") for altruists vs venals 
       and see if there was a greater tendency for those supervisor-rated 
       "career-oriented/self-interest-focused" to choose "agree"; 
       that alone might be a tip-off.