Gender

Document Sample
Gender Powered By Docstoc
					Studying who values whom in the
  U.S. House of Representatives

         Kristin Kanthak
               EITM
          July 11, 2007
           This talk is based on…*
             Group ‘Tokenism’ and Descriptive
              Representation within Political
                      Organizations:
         Gender and Colleague Valuation in the U.S.
                 House of Representatives

                                Kristin Kanthak †
                             University of Pittsburgh
                               George A. Krause ‡
                             University of Pittsburgh
                                         &
                               Amanda Driscoll #
                         Washington University – St. Louis


* Coauthors bear no responsibility for the “crimes of research” confessed herein.
              Operationalizing the
              “smoke-filled room”
   Real legislating happens behind closed
    doors
   Member-to-member contribution patterns
    allow us to glimpse member valuations
       Leadership PACs
            Leadership PAC data
   Donor/recipient dyads
       Each leadership PAC matched with each party
        member
   Values range from zero to $15,000
       91 percent of observations are zero
   Donations are a consumption good
       No effect on electoral outcomes
       Tiny percentage of total campaign receipts
      “Soaking and poking”
Proportion who received donation, 2000 cycle

                Men Donors     Women Donors


Democrats       Men: 0.10        Men: 0.11
               Women: 0.08      Women: 0.14

Republicans     Men: 0.12        Men: 0.07
               Women: 0.17      Women: 0.06
                 Tokenism

   E.g., Kanter (1977), Laws (1975)
   “Token” minorities receive special
    attention from majority, are suspicious of
    each other
   Non-“token” minorities worry the majority,
    can work together
         Why should we care?

   Descriptive representation critical to
    democracy (Mill 1861, Pitkin 1967,
    Mansbridge 1999)
   Representation of women increases
    legitimacy of government (Schwindt-Bayer
    and Mishler 2005, Lawless 2004)
        The paper, beta version

   Report of the empirical pattern
   Grounded in tokenism literature
   Reviewers didn’t buy it
             The paper, reloaded

   Need to think carefully about first
    principles
       What do utility functions look like?
   What is the role of ideology?
                  First principles

   Token minorities get “special” attention
       Pressure to “be one of the boys”
   Non-token minorities get ignored
       But “old girls’ network” starts
   But really, ideological proximity is most
    important
       Constrained optimization

   The idea: Members value their colleagues
    subject to some budget constraint
   The problem: What’s the constraint?
   The moral: Formal models can reveal
    assumptions that do not make sense
               Decision theory

   “Am I better off with or without another
    member of my own group?”
   Valuation is not a strategic choice
       A consumption good
    How to value the other group

   Small “other” groups are tokens
   Larger groups threaten, then take, the
    majority
   Very large “other” groups make you the
    token
        Marginal utility from the other
                     group
   First derivative provides the utility for each
    additional member of other group
       The valuation of each individual
   What does this look like?
                                                            Figure 1:
                                                     Group A Utility and
                                          Marginal Utility from Members of Group B


                        0.1

                             0

                 -0.1




Utility
                 -0.2

                 -0.3

                 -0.4
                                 0              0.2       0.4          0.6     0.8          1
                                                                                      Group B
                                                                                     Proportion




                             1.5

                                 1
          Marginal Utility




                             0.5

                                 0

                             -0.5

                                 -1

                             -1.5
                                      0            0.2     0.4         0.6     0.8          1
                                                                                      Group B
                                             w*-                 w**           w*+   Proportion
    How to value your own group

   Small token groups feel threatened by
    each other
   After some threshold, large enough to
    work together
   When very large, new token group is
    valued
    Marginal utility of own group

   Again, derivative gives utility for each
    additional member of own group
       The valuation of each individual
   What does this look like?
                                                     Figure 2:
                                             Group B Utility and
                                   Marginal Utility from Members of Group B


                      0.4

                      0.3

                      0.2

          Utility     0.1

                           0

                     -0.1
                               0             0.2    0.4         0.6     0.8             1
                                                                                Group B
                                                                               Proportion




                    1.5

                      1
Marginal Utility




                    0.5

                      0

                    -0.5

                     -1

                    -1.5
                           0                0.2     0.4         0.6     0.8             1
                                                                                  Group B
                                      w*-                 w**             w   *+
                                                                                 Proportion
    Valuation within the same larger
                 group
   Assumption: Valuations are symmetric
       If group sizes were switched, behavior would
        be the same (Young and James 2001)
   Therefore, inflection points are the same
   Behavior for different subgroups on
    different sides of the inflection points
    ought to be different
                           Figure 3:
Differences in Group A and Group B Marginal Utility Valuations
                Within the Same Larger Group




                    1.5

                      1
 Marginal Utility




                    0.5
                                                                          Group B
                      0
                                                                          Group A
                    -0.5

                     -1

                    -1.5
                           0     0.2   0.4        0.6   0.8        1
                                *            *                     Group B
                                                              *
                               w*           w*          w    *
                                                                   Proportion
               Gender is binary

   Gender is a dichotomous variable, W
       Not enough variation in w
   Parties are on either side of Kanter’s
    (arbitrary) 15 percent inflection point
       Republicans are 8-9 percent women
       Democrats are 18-19 percent women
           What about ideology?
                       (other group)
   Conservative theory: Ideology is what
    really matters
       At the extremes of ideological divergence,
        being in the other group has no effect



                                 
         MU A, D   D2  1  D2  (1  2 2 w  3 3 w2 ),
                                            Figure 4:
                          The Impact of Value Divergence (D) on Group A
                                   Marginal Utility Valuations




                   1.5

                     1
                                                                                              D=0
Marginal Utility




                   0.5                                                                        D=.25
                     0                                                                        D=.5

                   -0.5                                                                       D=.75
                                                                                              D=1
                    -1

                   -1.5
                           0   0.1   0.2   0.3   0.4   0.5   0.6   0.7    0.8   0.9   1
                                                                         Group B Proportion
                                wL    wH
           What about ideology?
                      (same group)
   The mirror image of other group valuation
       At the extreme of ideological divergence,
        being in one’s own group is all that matters




                                
         MU B,D   D 2  1  D 2  (1  22 w  33w2 ),
                                             Figure 5:
                          The Impact of Value Divergence (D) on Group B
                                   Marginal Utility Valuations




                    1.5
                      1
                    0.5
                                                                                              D=0
Marginal Utility




                      0
                   -0.5                                                                       D=.25
                     -1                                                                       D=.5
                   -1.5                                                                       D=.75
                     -2
                                                                                              D=1
                   -2.5
                     -3
                   -3.5
                           0   0.1   0.2   0.3   0.4   0.5   0.6   0.7   0.8 0.9    1
                                                                         Group B Proportion
                                wL wH
        Theoretical predictions

   Majority group (or fellow minority group)
    valuations of minority group members are
    negatively (or positively) related to the
    minority’s relative group size
   The size of the minority group influences
    how dramatically preference (ideological)
    divergence affects colleague valuation
              Nature of the data

   Donor/recipient dyads
       Every leadership PAC matched with every
        party member
   Donations range from $0 to $15,000
       Virtually all donations are $0
             Ancillary variables
   Donor and recipient from same state
   Donor and recipient from same region
   Donor and recipient on same committee
   Percent of vote recipient received in last election
   Number of terms recipient served
   Recipient in leadership, on power committee
   Size of PAC
                  Tobit model

   Workhorse of campaign contributions
    empirical models
   Allows truncated normal distribution
   But what does zero mean?
       Some unobserved latent value, NOT the value
        itself
               Heckman model

   Two stage Heckman model
       Do I donate?
       How much do I donate?
   Zero here means donor chose zero in the
    first stage
   But regressors cannot be the same in the
    selection and outcome equation
    Corner solution semi-logarithmic
              Tobit model
   Zero is a (meaningful) corner solution to
    an optimization problem
       Wooldridge 2002; Gordon, Hafer, and Landa
        2006
   (But they all give about the same results
    anyway)
             “Additive model”

   What is the effect of gender on valuation?
   Gender, plus ideological distance, plus
    ancillary variables
                    Results
Independent   Rep. Men    Rep.     Dem. Men     Dem.
  Variable               Women                 Women



  Woman        1.808*     1.819      -3.270*    -2.193*
 Recipient    (0.3326)   (1.265)    (0.5421)   (0.8763)

Ideological    -12.01*   -28.29*    -6.966*    -6.183*
 Distance      (1.425)   (7.202)    (1.120)    (1.979)
               Effect of gender

   Republican and Democratic men have
    significantly different donation patterns
       Republicans give 20 percent more to women
       Democrats give 26 percent less to women
   So do Republican and Democratic women
       Republicans give 10 percent more to women
       Democrats give 22 percent less to women
         “Multiplicative model”

   Do colleagues “punish” men and women
    equally for ideological divergence?
   Gender, plus ideological distance, plus
    gender X ideological distance, plus
    ancillary variables
                    Results
Independent   Rep. Men    Rep. Dem. Men        Dem.
  Variable               Women                Women


  Woman        1.424*    0.5995     -3.030*    -2.276*
 Recipient    (0.3596)   (1.265)   (0.5729)   (0.9236)
Ideological    -13.76*   -37.45*   -6.437*    -6.435*
 Distance      (1.864)   (9.646)   (1.193)    (2.267)
  Woman        7.425*    32.81*    -4.324*     1.575
Recipient x    (2.484)   (12.60)   (3.469)    (4.545)
Ideological
 Distance
                   Simulations

   What are the “real” effects of gender and
    ideological distance on donations?
   Two sets of deterministic simulations,
    setting ancillary variables at their means
       Decision to give
       How much to give, given a decision to give
                                              Figure 6:
                  Simulated probability of donation for differing ideological distance,
                                all other values held to their means




                                Republican Men Donors                                                                          Republican Women Donors

               0.08                                                                                       0.08




                                                                                           Pr(Donation)
Pr(Donation)




               0.06                                                                                       0.06

               0.04                                                                                       0.04

               0.02                                                                                       0.02

                 0                                                                                           0
                      0   0.1   0.2   0.3     0.4   0.5   0.6      0.7   0.8   0.9   1                               0   0.1    0.2   0.3     0.4    0.5   0.6     0.7   0.8   0.9   1
                                            Ideological distance                                                                            Ideological distance




                                Democratic Men Donors                                                                          Democratic Women Donors

               0.08                                                                                       0.08
Pr(Donation)




                                                                                         Pr(Donation)
               0.06                                                                                       0.06

               0.04                                                                                       0.04

               0.02                                                                                       0.02

                 0                                                                                          0
                      0   0.1   0.2   0.3     0.4   0.5   0.6      0.7   0.8   0.9   1                           0       0.1   0.2    0.3     0.4   0.5    0.6     0.7   0.8   0.9   1
                                            Ideological distance                                                                            Ideological distance




                                                             Men                                                         Women
                                                                              Figure 7:
                                                    Simulated amount donated (given that a donation is made)
                                                                 for differing ideological distance,
                                                               all other values held to their means




                                                             Republican Men Donors                                                                                                               Republican Women Donors


                                 2000                                                                                                           2000
            Amount of donation




                                                                                                                         Amount of donation
                                 1500                                                                                                           1500

                                 1000                                                                                                           1000

                                   500
                                                                                                                                                          500
                                       0
                                                                                                                                                                   0
                                               0     0.1     0.2      0.3     0.4    0.5     0.6   0.7   0.8   0.9   1
                                                                                                                                                                           0       0.1     0.2     0.3      0.4     0.5     0.6     0.7     0.8     0.9     1
                                                                            Ideological distance
                                                                                                                                                                                                           Ideological distance




                                                                   Democratic Men Donors                                                                                                           Democratic Women Donors

                      2000                                                                                                                                         2000


                                                                                                                                              Amount of donation
Amount donated




                      1500                                                                                                                                         1500

                      1000                                                                                                                                         1000

                                 500                                                                                                                               500

                                  0                                                                                                                                    0
                                           0       0.1     0.2      0.3      0.4    0.5     0.6    0.7   0.8   0.9   1                                                         0     0.1     0.2     0.3      0.4     0.5     0.6     0.7     0.8     0.9       1
                                                                            Ideological distance                                                                                                             Ideological distance




                                                                                                   Men                                                                                   Women
           Support for theory

   Republican men give more to women
   Democratic men give less to women
   Republicans punish more than Democrats
    for ideological divergence
          Divergence from theory
   Republican women do not give less to women
       No significant gender effect
   Democratic women do not give more to women
       They give significantly less to women
   Republicans punish women less for divergence
   Democratic men punish women more for
    divergence
                   Conclusions

   Tokenism exists
       Explains majority group behavior and
        ideology-based valuations
   Token valuations can trump ideology-
    based valuations
   Minorities may feel pressure to mimic
    majority group behavior

				
DOCUMENT INFO