code.m

% Data documentation at the end.

clear all
v8_assumption = 100;
v9_assumption = 5;
age_range = [18 30];

years_v = [1991, 1993, 1994, 1996, 1998, 2000, 2002, 2004, 2006, 2008, 2010, 2012, 2014, 2016, 2018];
GSS = importdata('GSS.dat');
GSS = GSS(GSS(:,3) >= age_range(1) & GSS(:,3) <= age_range(2) ,:); % only people in the desired age range
GSS = GSS(:,[1 4 5 6]); % keep only year(1), sex(2), partners last year(3) and partners last 5 years(4) as the variables
GSS = GSS(GSS(:,3)~=-1 & GSS(:,3)~=98 & GSS(:,3)~=99 & GSS(:,3)~=95,:); % remove bad answers (last 1 years)
GSS = GSS(GSS(:,4)~=-1 & GSS(:,4)~=98 & GSS(:,4)~=99 & GSS(:,3)~=95,:); % remove bad answers (last 5 years)

m1_data = []; m5_data = [];
f1_data = []; f5_data = [];
m1_gini = []; m5_gini = [];
f1_gini = []; f5_gini = [];

for y = years_v
    G_y = GSS(GSS(:,1) == y ,:); % choose year
    G_m = G_y(G_y(:,2) == 1 ,:); G_f = G_y(G_y(:,2) == 2 ,:); % choose sex
    G_m1 = G_m(:,3); G_m5 = G_m(:,4);
    G_f1 = G_f(:,3); G_f5 = G_f(:,4); % data for male/female 1/5 year partners.

    % This is [year, total respondents, v0-v9 as given above]
    m1_data = [m1_data; y, length(G_m1), grab_vars(G_m1)]; m5_data = [m5_data; y, length(G_m5) grab_vars(G_m5)];
    f1_data = [f1_data; y, length(G_f1), grab_vars(G_f1)]; f5_data = [f5_data; y, length(G_f5) grab_vars(G_f5)];

    % Convert to a population of people and the number of times they've had
    % sex, then calculate a gini coefficient based off that.
    m1_gini = [m1_gini; y, gini_coefficient(convert_to_population(y,m1_data,v8_assumption,v9_assumption))];
    m5_gini = [m5_gini; y, gini_coefficient(convert_to_population(y,m5_data,v8_assumption,v9_assumption))];
    f1_gini = [f1_gini; y, gini_coefficient(convert_to_population(y,f1_data,v8_assumption,v9_assumption))];
    f5_gini = [f5_gini; y, gini_coefficient(convert_to_population(y,f5_data,v8_assumption,v9_assumption))];
end

figure(1)
hold on
plot(m5_data(:,1) , 100 * m5_data(:,3)./m5_data(:,2) ,'color',[0.47 0.81 0.94],'LineWidth',4)
plot(f5_data(:,1) , 100 * f5_data(:,3)./f5_data(:,2) ,'color',[0.96 0.76 0.76],'LineWidth',4)
set(gca,'FontSize', 15)
title('Share of 18-30 year olds sexless in the last 5 years')
legend('Males', 'Females','FontSize',20)

figure (2)
hold on
plot(m1_data(:,1) , 100 * m1_data(:,3)./m1_data(:,2) ,'color',[0.47 0.81 0.94],'LineWidth',4)
plot(f1_data(:,1) , 100 * f1_data(:,3)./f1_data(:,2) ,'color',[0.96 0.76 0.76],'LineWidth',4)
set(gca,'FontSize', 15)
title('Share of 18-30 year olds sexless in the last 1 year')
legend('Males', 'Females','FontSize',20)

figure(3)
hold on
plot(m1_gini(:,1) , m1_gini(:,2) ,'color',[0.47 0.81 0.94],'LineWidth',3)
plot(f1_gini(:,1) , f1_gini(:,2),'color',[0.96 0.76 0.76],'LineWidth',3)
set(gca,'FontSize', 15)
title('Gini Coefficient facing males and females 18-30 in the last 1 year of sex','FontSize',13)
legend('Males', 'Females','FontSize',20)

figure(4)
hold on
plot(m5_gini(:,1) , m5_gini(:,2) ,'color',[0.47 0.81 0.94],'LineWidth',3)
plot(f5_gini(:,1) , f5_gini(:,2),'color',[0.96 0.76 0.76],'LineWidth',3)
set(gca,'FontSize', 15)
title('Gini Coefficient facing males and females 18-30 in the last 5 years of sex','FontSize',13)
legend('Males', 'Females','FontSize',20)



function v_args = grab_vars(G)
    v_args = [];
    for i=0:9
        v_args = [v_args length(G(G==i))];
    end
end

function vec = convert_to_population(year,data,v8_assumption,v9_assumption)
    n = [0 1 2 3 4 sqrt(5*10) sqrt(11*20) sqrt(21*100) v8_assumption v9_assumption];

    x = data(data(:,1)==year ,:);
    vec = [];
    for i = 0:9
        vec = [vec, zeros(1,x(i+3))+n(i+1)];
    end
end

function g = gini_coefficient(population)
    s = 0;
    for i = 1:length(population)
        for j = 1:length(population)
            s = s + abs(population(i) - population(j));
        end
    end
    g = s/(2 * length(population).^2 * mean(population));
end

%{
1 label variable year     "Gss year for this respondent                       ";
2 label variable id_      "Respondent id number";
3 label variable age      "Age of respondent";
4 label variable sex      "Respondents sex";
5 label variable partners "How many sex partners r had in last year";
6 label variable partnrs5 "How many sex partners r had in last 5 years";
7 label variable ballot   "Ballot used for interview";

field 1: YEAR
field 2: SEX
   2        "Female"
   1        "Male"
field 3: AGE
   99       "No answer"
   98       "Don't know"
   89       "89 or older"
field 3: SEX PARTNERS IN LAST 1 YEAR
   99       "No answer"
   98       "Don't know"
   95       "Several" <-- IGNORE, NO SUCH RESPONSE
   9        "1 or more, # unknown"
   8        "More than 100 partners"
   7        "21-100 partners"
   6        "11-20 partners"
   5        "5-10 partners"
   4        "4 partners"
   3        "3 partners"
   2        "2 partners"
   1        "1 partner"
   0        "No partners"
   -1       "Not applicable"
field 4: SEX PARTNERS IN LAST 5 YEARS
   same as above.
%}