1: %\documentclass[12pt,twoside]{article}
2: \documentclass[12pt]{article}
3: \usepackage{epsfig,times} % 17 dec. 2001: pbs avec times
4: \usepackage{color}
5:
6: %%%%% Margins for American paper format (JeanL)
7: % \topmargin= 0.2cm \oddsidemargin = +0.4cm \evensidemargin = +0.4cm
8: %%%%% Margins for European paper format (JeanL)
9: \topmargin=-0.7cm \oddsidemargin = -0.4cm \evensidemargin = -0.4cm
10:
11: %\textheight=237mm \textwidth=170mm \baselineskip 5mm
12: \textheight=247mm \textwidth=170mm \baselineskip 5mm
13:
14:
15: \parindent=0pt
16:
17: \begin{document}
18: \thispagestyle{empty}
19:
20:
21: %%%%%%%% Pour changer les valeurs par defaut pour taille figure,
22: %%%%%%%% sinon au-dela d'une hauteur de 134 mm = 70% on est rejete a la fin
23: \renewcommand{\topfraction}{.99}
24: \renewcommand{\bottomfraction}{.99}
25: %\renewcommand{\floatpagefraction}{.99}
26: \renewcommand{\textfraction}{.0}
27:
28:
29: %%%%% Definitions
30:
31: \newcommand{\nc}{\newcommand}
32:
33: \nc{\qI}[1]{\section{{#1}}}
34: \nc{\qA}[1]{\subsection{{#1}}}
35: \nc{\qun}[1]{\subsubsection{{#1}}}
36: \nc{\qa}[1]{\paragraph{{#1}}}
37:
38: % Enumerations
39: \def\qbu{\hfill \par \hskip 6mm $ \bullet $ \hskip 2mm}
40: \def\qee#1{\hfill \par \hskip 6mm #1 \hskip 2 mm}
41:
42: \nc{\qfoot}[1]{\footnote{{#1}}}
43: \def\qL{\hfill \break}
44: \def\qpar{\vskip 2mm plus 0.2mm minus 0.2mm}
45: %\def\tvi{\vrule height 12pt depth 5pt width 0pt}
46: \def\tvi{\vrule height 12pt depth 5pt width 0pt}
47: \def\qtvi{\vrule height 2pt depth 5pt width 0pt}
48: \def\qth{\vrule height 12pt depth 0pt width 0pt}
49: \def\qtb{\vrule height 0pt depth 5pt width 0pt}
50:
51: \def\qparr{ \vskip 1.0mm plus 0.2mm minus 0.2mm \hangindent=10mm
52: \hangafter=1}
53:
54: % Decale UN paragraphe
55: % Attention! La double accolade est vitale, sinon tout le
56: % est decale (cf TEX p.199)
57: % On peut aller a la ligne avec \qL=\hfill \break
58: % Par contre ne supporte pas les lignes blanches
59: \def\qdec#1{\par {\leftskip=2cm {#1} \par}}
60:
61:
62: %% Defs specifiques
63: \def\qdpt{\partial_t}
64: \def\qdpx{\partial_x}
65: \def\qddpt{\partial^{2}_{t^2}}
66: \def\qddpx{\partial^{2}_{x^2}}
67: %
68: \def\qn#1{\eqno \hbox{(#1)}}
69: \def\qds{\displaystyle}
70: \def\qw{\widetilde}
71: %
72: \def\qmax{\mathop{\rm Max}} % Petit livre Tex (p.167)
73: \def\qmin{\mathop{\rm Min}} % Petit livre Tex (p.167)
74:
75: % Interligne plus large
76: \baselineskip=25pt % le choix de plain est 12pt (Petit Livre p.227)
77:
78: %%%%% End of definitions
79:
80: \def\qci#1{\parindent=0mm \par \small \parshape=1 1cm 15cm #1 \par
81: \normalsize}
82:
83: \null
84: % {\large \it To appear in Physica A}
85: \vskip 2mm
86:
87: \centerline{\bf \Large A physicist's view of the notion of ``racism''}
88: %\vskip 5mm
89: %\centerline{\bf \Large }
90:
91: \vskip 1cm
92: \centerline{Charles Jego $ ^{\# 1} $,\quad Bertrand M. Roehner $ ^{+2} $ }
93: \vskip 3mm
94: \centerline{\#: Ecole Polytechnique,\quad +: Institute for
95: Theoretical and High Energy Physics}
96:
97:
98: \vskip 15mm
99:
100: {\bf Abstract}\quad
101: It is not uncommon (e.g. in the media)
102: that specific groups are categorized as being racist.
103: Based on an extensive dataset of intermarriage statistics
104: our study questions the
105: legitimacy of such characterizations. It suggests that, far from being
106: group-dependent, segregation mechanisms are instead situation-dependent.
107: More precisely, the degree of integration of a minority in terms of
108: the frequency of
109: intermarriage is seen to crucially depend upon the
110: the proportion $ p $ of the minority. Thus, a
111: population may have
112: a segregative behavior with respect to a high-$ p $ ($ p > 20\% $) minority $ A $
113: and
114: at the same time a tolerant attitude toward a low-$ p $ ($ p<2\% $) minority $ B $.
115: This remains true even when $ A $ and $ B $ represent the same minority;
116: for instance Black-White intermarriage is much more frequent in
117: Montana than it is in South Carolina. In short,
118: the nature of minority groups is largely irrelevant, the key factor being their
119: proportion in a given area.
120:
121:
122: \vskip 5mm
123: \centerline{April 21, 2007}
124:
125: %\vskip 8mm
126: %\centerline{\it Preliminary version, comments are welcome}
127:
128:
129: \vskip 1cm
130: Key-words: intermarriage, minority, ethnically mixed couples, integration,
131: segregation, interaction.
132: \vskip 1cm
133:
134: 1: Charles Jego, Centre de Physique Th\'eorique,
135: Ecole polytechnique, CNRS, 91128 Palaiseau, France
136: \qL
137: \phantom{1: }E-mail: jego@clipper.ens.fr
138: \vskip 5mm
139:
140: 2: Bertrand Roehner, LPTHE, University of Paris 6, 4 place Jussieu,
141: F-75005 Paris, France.
142: \qL
143: \phantom{2: }E-mail: roehner@lpthe.jussieu.fr
144: \qL
145: %\phantom{2: }FAX: 33 1 44 27 79 90
146:
147: \vfill \eject
148:
149: \qI{Introduction}
150:
151: In the {\it New York Times} of 24 February 1980 one reads the following
152: title ``Swedes discover their dark side: racism''. This is by no means
153: an isolated example; the medias frequently apply
154: the terms ``racism'' or ``racist'' to populations or peoples.
155: Over the period 1971-2005 {\it New York Times} articles featuring these words
156: appeared with a frequency of 57 articles per year.
157: For a scientist this raises
158: the question of how these notions can be defined objectively and whether it is
159: legitimate to apply them to groups of people or even to whole
160: nations as in the example above. Naturally, it is
161: well known that there is no scientific definition whatsoever
162: of the concept of race, but one can rely on the
163: self-identification definition used in U.S. censuses.
164: Through that procedure one can define (at least for statistical purposes)
165: populations and groups composed of
166: ``Whites''
167: \qfoot{Throughout this paper, ``White'' (W) means ``White non Hispanic''.}
168: ,
169: ``Blacks'' (or ``Afro-Americans''), ``American-Indians'' and so on.
170: A commonly held belief is that American states belonging to
171: the Deep South (Alabama, Louisiana, Mississippi, etc.) are more ``racist''
172: than northern states. Can such a claim be supported by quantitative evidence
173: in a way which is consistent with the {\it ceteris paribus}
174: (i.e. ``all other things being equal'') requirement?
175: Econophysics was founded on
176: the claim that ideas from physics can help us understand social phenomena.
177: This paper hopes to be an illustration of this claim.
178: \qpar
179:
180: In the social science literature
181: the question of segregation
182: is most often considered from an anthropomorphic
183: perspective, by which we mean that most studies single out
184: specific populations and rely on factors such as religion, socioeconomic
185: status, dating circumstances and so on
186: (Clark-Ibanez et al. 2004, Houston et al. 2004,
187: Kalminjn 1998, Pagnini et al. 1990. Tatum 1997).
188: In contrast, from a physicist's perspective the interaction between ethnic
189: groups is naturally seen as
190: a case of forming bonds between two types of units, a point of view
191: which naturally leads to comparative investigations. It is fair to say
192: that the comparative perspective was also adopted by some sociologists such as
193: Blau et al. (1984), Duncan et al. (1959), Lieberson et al. (1959), although
194: it was not developed in a systematic way.
195: \qpar
196:
197: To summarize the gist of our argument
198: by a quick example let us consider the case of Louisiana.
199: This state belongs to the
200: Deep South belt which until the mid-1960s had a well established tradition
201: of segregation; moreover the Katrina disaster of 2005 revealed
202: that inter-ethnic tension between Blacks and Whites is
203: just beneath the surface.
204: Apart from its substantial Black minority, Louisiana
205: also has a small minority of American Indians.
206: But whereas it
207: has one of the smallest proportions%
208: \qfoot{In a normalized sense which will be explained below.}
209: of Black-White (B-W)
210: couples among all US states,
211: it has one of the
212: largest proportions of American Indian - White (I-W) couples.
213: This example suggests that
214: speaking of Louisiana Whites as being a group prone to segregationist
215: attitudes without further qualification
216: is not consistent with observation.
217: The low rate of B-W intermarriage in Louisiana is mainly brought about by
218: the fact that
219: Blacks represent a proportion of 32\% in the total population, whereas
220: the proportion of American Indians is only 0.56\%.
221: Naturally, this effect is by no means specific to the United States.
222: Back in 1893, people of Italian descent made up 17\% of the
223: population of Marseilles in the south of France and it can be recalled
224: that on 16-17 August serious clashes between
225: French and Italian workers near Aigues-Mortes resulted in the death
226: of 18 people%
227: \qfoot{More details can be found in Roehner 2004, p. 197-198.}%
228: .
229: \qpar
230:
231: There are several ways of defining ethnic segregation/integration quantitatively,
232: namely: (i) Residential integration (ii) School integration (iii) Marriage integration
233: (iv) Economic integration.
234: The first two criteria are closely related for the obvious reason that residential
235: segregation at block or county level results in {\it de facto}
236: school segregation simply because pupils attend school in the area in which
237: they live. Residential segregation has been measured by several sociologists%
238: \qfoot{See for instance: Sorensen et al. (1975), Lieberson (1980),
239: and Iceland et al. (2002).}
240: while the second and third criteria have been less studied.
241: In the present paper, we use the criterion of marriage integration.
242: The conclusions drawn from this criterion are to a large extent
243: consistent with results based on residential segregation and
244: school integration (more on this below).
245: One advantage of the inter-marriage criterion is that one would expect it to be
246: less dependent on economic
247: conditions than the residential criterion because it seems possible for
248: two persons to meet one another (and possibly to get married) even if they
249: live in segregated areas; workplaces, dance halls, stadiums, holiday resorts
250: provide contact opportunities which
251: to some extent are independent of housing location
252: (see Houston et al. 2005). The fourth criterion would lead us to consider
253: segregation in the jobmarket and workforce. As census data contain much
254: information on occupations they would allow us to carry out such an
255: investigation but we will leave it to a subsequent paper.
256: \qpar
257:
258: The paper is organized as follows.
259: First we explain the methodology and test it on what we
260: call a ``null-experiment''. Then we describe our results for
261: ethnically-mixed couples.
262:
263: \qI{Methodology}
264:
265: Individual microdata from American censuses are available online on
266: the website of the Minnesota Population Center%
267: \qfoot{http://usa.ipums.org/uta/redirect-landing.shtml; IPUMS means
268: Integrated Public Use Microdata Series.
269: From a practical perspective, it must be
270: emphasized that the IPUMS website is very user friendly in the sense that
271: the steps of selecting, downloading, uncompressing and reading
272: the data through an appropriate software can be performed within a few minutes.}%
273: .
274: Fifteen federal censuses ranging from 1850 to 2000 are accessible through
275: 1\% samples; in addition, 5\% samples are available for some years%
276: \qfoot{While most of the samples are composed of randomly selected
277: households, some samples are not random in the sense that specific
278: categories are over-represented. In the present study we used only
279: random samples.}%
280: .
281: Once the data have been selected,
282: we count the
283: number of ethnically-mixed couples in each state. For instance, using an
284: unweighted random 1\% sample
285: of the 2000 census
286: we find 3,400 Black-White
287: couples in Alabama and 400 in New Hampshire.
288: These counts include married couples
289: (identification code 0201) as well
290: as unmarried male-female partners (identification code 1114).
291: To be compared in a meaningful way, these numbers must be normalized in
292: two ways.
293: %
294: \qbu A first natural normalization is to
295: compute the number of mixed couples with respect to total number of married
296: couples.
297: In 2000 Alabama and New Hampshire had 0.906 million and
298: 0.262 million married couples respectively.
299: Thus, one obtains proportions of $ 3,753 $ and $ 1,527 $ B-W couples
300: per million couples respectively.
301: However,
302: this comparison is still meaningless because it fails to take into account
303: the respective numbers of Black people in each state, namely
304: 259,000 in Alabama versus
305: 7,300 in New Hampshire. To take this difference into account we need a second
306: normalization.
307: %
308: \qbu Let us denote by $ p $ the proportion of a minority $ B $ in a population $ A $.
309: Then, it can be shown by a combinatorial argument
310: that if
311: male-female pairs are formed randomly in a population of size $ n $ the expected
312: proportion of mixed couples (for large $ n $) is:
313: $$ e_{A-B}= 2p(1-p) \qn{1} $$
314: It can be noticed that
315: if $ p=0.5 $ formula (1) gives $ e_{A-B}=0.5 $ as expected. The case
316: $ p \ll 1 $ which corresponds to a small population immersed in a much larger
317: population is of special interest because
318: it corresponds to most of the minorities to be found
319: in the United States (American Indians, Chinese, Japanese); in this
320: case, (1) leads to $ e_{A-B} \simeq 2p $.
321: Two crucial assumptions are made in the derivation of (1):
322: (i) selection of husband and wife occurs randomly which means
323: in particular that it is not subject to any distance limitation; in other words
324: the probability of a marriage is the same whether both people live
325: in southern California or in different parts of California.
326: (ii) there are no
327: institutional or social restrictions in the pairing of $ A $ and $ B $ people.
328: In real life, these assumptions are usually not fulfilled. Indeed, because of
329: housing segregation, the vicinity of $ B $ individuals comprises a
330: proportion of $ B $ people which may be much larger than the proportion
331: in the total population. Secondly, even once $ A-B $ contacts have been
332: established, marriage may not follow due to the ``barrier''
333: of social conventions. As a result of these restrictions, actual
334: rates of mixed couples show a discrepancy with respect
335: to the rate given by (1) and the magnitude of this discrepancy can serve
336: to measure the lack of integration. In short, the rationale
337: of our normalization procedure is that
338: equation (1) will be used not as a model but as a yardstick.
339: \qpar
340:
341: The normalization procedure can be summarized through the following
342: formula giving the normalized frequency $ f_{A-B}(S) $ of $ A-B $ couples
343: in state $ S $:
344: $$ f_{A-B}(S)= { c_{A-B} \over C }{ 1 \over 2p(1-p)} \qn{2} $$
345:
346: where
347: $ c_{A-B}= $ number of mixed couples living in state $ S $, $ C=$ number of
348: married couples living in state $ S $ and $ p= $ proportion of the minority
349: $ B $ in the total population of state $ S $.
350: $ f_{A-B}(S) $ defines a propensity for integration through marriage.
351: For the sake of brevity, we
352: subsequently refer to it as a marriage integration index
353: and express it in percent. In a perfectly
354: integrated society $ f_{A-B} $ would be equal to 100\%, as illustrated by
355: the case of people born in California considered below.
356: In a society with a strong propensity for endogamy, $ f_{A-B} $ will be
357: much smaller; on the contrary, in
358: a society with
359: a strong inclination for exogamy, $ f_{A-B} $ will be larger than 100\% %
360: \qfoot{The writings of ethologists , ethnologists and anthropologists
361: (e.g. Frazer 1910, Karandikar 1929, Makarius 1961, Kortmulder 1968)
362: suggest the existence of a ``natural'' tendency to exogamy.
363: If this
364: were not the case the very notion of species would have little meaning; for
365: instance elephants of Eastern Africa would in time have become markedly different
366: from those of Central Africa. In an other context, the tradition for royal heirs to
367: contract exogamic marriages by taking their wives in foreign courts which probably
368: goes beyond the desire to establish diplomatic ties.
369: On the contrary, diversification within a given
370: species can be created
371: artificially by suppressing interbreeding; in this way
372: large numbers of different races can be generated in a relatively short
373: time span and maintained as long as isolation is enforced, a process
374: illustrated by the numerous (over one hundred) races of cats and dogs,
375: all of which are
376: characterized by markedly different phenotypes. Diversification can also
377: arise slowly by genetic drift in situations of isolation
378: as for instance on islands. }%
379: .
380: \qpar
381:
382: Returning to our previous example and noting that
383: in Alabama, $ p=26\%$
384: whereas in New Hampshire $ p=0.73\% $ one gets
385: expected proportions $ e_{W-B}=38\% $ and $ e_{W-B}=1.4\% $ respectively.
386: Thus,
387: the B-W marriage integration indexes are
388: $ f_{W-B}(\hbox{\small Alabama})=0.37/38=0.010 $
389: and $ f_{W-B}(\hbox{\small New Hampshire})=0.15/1.4=0.11 $, an integration index
390: that is about 10 times larger than in Alabama.
391: \qpar
392:
393: {\bf Null experiment}\quad
394: Before giving complete results for all 50 states we wish to test the
395: normalization procedure
396: through a ``null-experiment'', by which we mean a test-observation
397: of a situation in which
398: one does not expect any segregation effect.
399: To this end,
400: we consider the minority formed in all states (except California)
401: by the people who were born in California%
402: \qfoot{This state was selected because it has the largest population, but
403: similar tests carried out for New York State and Illinois
404: led to comparable results.}%
405: .
406: In addition, in order to eliminate all effects that may be related to ethnicity
407: we restrict the sample to White non Hispanic people. In this experiment we count
408: as mixed couples any couple in which only one of the spouses is born in
409: California. The results are summarized in Fig. 1.
410: %%-----------------------------------------------
411: %%%% Fig. 1
412: \begin{figure}[tb]
413: \centerline{\psfig{width=12cm,figure=fvi1.eps}}
414: \vskip 2mm
415: {\bf Fig. 1: Normalized frequency of couples (born in Cal.) -- (not born in Cal.),
416: 2000 Census.}
417: {\small Horizontal scale: percentage of residents of the state who were
418: born in California (thereafter called the minority).
419: Vertical scale: frequency of couples in which one of the partners
420: (not both) belongs to the minority; this frequency has been normalized
421: with respect to the number of married couples in
422: the state and with respect to the importance
423: of the minority (details can be found in the text in the paragraph about the
424: definition of the marriage integration index).
425: Each label refers to one of the states
426: plus Washington, DC (California has been excluded). The sample has been
427: restricted to White non Hispanic people.
428: This is a situation in which
429: one does not expect any ethnic segregation effect in other words one expects
430: an horizontal scatter plot
431: at a level close to 100\% , which is indeed what is observed.
432: The slope of the
433: linear regression is $ 0.032\pm 0.04 $. Similar results hold for couples in
434: which one of the partners is born in Illinois or in New York State.}
435: {\small \it Source: The data are from a $ 1\% $, random
436: sample of the 2000 Census, available online on the website of the
437: Minnesota Population Center.}
438: \end{figure}
439: %% --------------------------------------------------
440: The graph suggest
441: two comments: (i) For most of the states, the frequency of mixed couples is
442: close to 100\% which is in conformity with randomly formed pairs.
443: (ii) As expected on account of the lack of ethnic identification, the
444: slope of the regression line is consistent with a zero value,
445: $ a=-0.032 \pm 0.04 $.
446:
447: \qI{Inter-marriage}
448: We now repeat the previous procedure for ethnically mixed couples. Fig. 2a
449: corresponds to the case of Black-White (B-W)
450: couples; it shows that the frequency
451: of mixed couples is at least 10 times smaller than in Fig. 1.
452: %%-----------------------------------------------
453: %%%% Fig. 2a
454: \begin{figure}[tb]
455: \centerline{\psfig{width=12cm,figure=fvi2a.eps}}
456: \vskip 2mm
457: {\bf Fig. 2a: Normalized frequency of Black -- White couples, 2000 Census.}
458: {\small Horizontal scale: percentage of the Black population.
459: Each label refers to one of the states plus Washington, DC. Circles are
460: drawn around Georgia and South Carolina (states with high proportions of Blacks),
461: squares are drawn around Montana and South Dakota (states with low
462: proportions of Blacks), diamonds are drawn around New Hampshire and
463: Vermont (states with low proportions of Blacks).
464: The correlation is $ -0.89 $ and the slope of the
465: linear regression is $ -0.62\pm 0.09 $}
466: {\small \it Source: The data are taken from an unweighted $ 1\% $
467: sample of the 2000 Census.}
468: \end{figure}
469: %% --------------------------------------------------
470: The
471: frequency of B-W couples has tripled in the period 1970-2000 but it still
472: remains at a low level. In addition, there is a marked negative slope
473: $ a=-0.62\pm 0.09 $. The pattern for American Indian - White (I-W) couples is
474: similar but the frequency is about 4 times higher and the slope about one half
475: of the previous one: $ a=-0.36\pm 0.12 $, see Fig. 2b.
476: %%-----------------------------------------------
477: %%%% Fig. 2b
478: \begin{figure}[tb]
479: \centerline{\psfig{width=12cm,figure=fvi2b.eps}}
480: \vskip 2mm
481: {\bf Fig. 2b: Normalized frequency
482: of American-Indian -- White couples, 2000 Census.}
483: {\small Horizontal scale: percentage of the American Indian population
484: with respect to total state population.
485: It can be seen that Georgia and South Carolina (circled) which had a small
486: mixed couple frequency in Fig. 2a have a high frequency here,
487: whereas Montana and
488: South Dakota (squares)
489: which had a high frequency in Fig. 2a have a low one here due to their
490: substantial proportion of American Indians.
491: Vermont and New Hampshire (diamonds) which
492: have only few minority residents of both kinds have a high frequency in
493: both graphs. Similar conclusions can be drawn as well from the data
494: for other states.
495: The correlation is $ -0.66 $
496: (confidence interval for probability $ 0.95 $ is $ -0.79 $ to $ -0.47 $);
497: the slope of the regression line is $ -0.36\pm 0.12 $.}
498: {\small \it Source: The data are taken from an unweighted $ 1\% $,
499: sample of the 2000 Census.}
500: \end{figure}
501: %% --------------------------------------------------
502: \qpar
503:
504: The most interesting observation
505: is the fact that the states of the Deep South (e.g. Alabama, Arkansas, Louisiana,
506: South Carolina) which have low B-W frequencies are at top frequency levels
507: in Fig. 2b. Similarly, states such as Arizona, Montana or South Dakota
508: have low I-W frequencies but high B-W frequencies. In addition
509: states such as New Hampshire or Vermont which do not have any substantial
510: ethnic minority whatsoever are at the top of the scatter plots in both graphs.
511: \qpar
512:
513: The observation that low
514: minority percentages are associated with higher integration levels
515: can be made
516: for other minorities as well. Let us give some examples.
517: %
518: \qbu Alabama, Arkansas and Florida have small percentages
519: ($ p_{\hbox{J}}< 0.1 \% $)
520: of Japanese Americans and some of the highest frequencies of
521: Japanese Americans - White (J-W) couples. In contrast, Hawaii has both
522: the highest percentage of ethnic Japanese
523: ($ p{\hbox{J}} \simeq 10\% $) and the lowest
524: frequency of J-W couples; there is a ratio of about 10 between the frequencies
525: in Florida and Hawaii.
526: %
527: \qbu The frequency of mixed Hispanic - non Hispanic couples is about
528: three times higher in Louisiana
529: ($ p_{\hbox{Hisp}}\simeq 2\% $) than in Texas or
530: California ($ p_{\hbox{Hisp}}\simeq 20\% $).
531: \qpar
532:
533: \qI{Other integration characterizations}
534:
535: \qA{Alternative criterion}
536: The previous observations can be confirmed by
537: using an alternative criterion which
538: does not require a renormalization procedure (at least for small $ p $). We
539: gauge the exogamous versus endogamous character of a minority by the
540: ratio:
541: $$ \Gamma _A=\hbox{Number of exogamous couples A-X}/\hbox{Number of
542: endogamous couples A-A} $$
543:
544: The notation $ X $ instead of $ B $ (as above) refers to the fact
545: that in this definition {\it all} exogamous couples of $ A $ with any other group
546: are summed up in the numerator.
547: Typical orders of magnitude of $ \Gamma $ are given in Table 1.
548: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
549: % TABLE 1
550:
551: \begin{table}[tb]
552:
553: \small
554:
555: \centerline{\bf Table 1\quad Variation of the exogenous/endogenous ratio
556: with respect to minority percentage}
557:
558: \vskip 3mm
559: \hrule
560: \vskip 0.5mm
561: \hrule
562: \vskip 2mm
563:
564: $$ \matrix{
565: \hbox{} & \hbox{Black} & \Gamma_{\hbox{\small Black}} & \qquad &
566: \hbox{Am. Ind.} & \Gamma_{\hbox{\small Am. Ind.}} \cr
567: \qtb
568: \hbox{} & \hbox{\%} & & &\hbox{\%} & \cr
569: \noalign{\hrule}
570: \qth
571: \hbox{\color{green} Montana+South Dakota} \hfill & {\color{green}
572: 0.41} & {\color{green} 1.2\phantom{00} } & &{\color{green} 7.2\phantom{0} } &
573: {\color{green}0.79} \cr
574: \hbox{\color{red} Georgia+South Carolina} \hfill& {\color{red} 29\phantom{.00} }&
575: {\color{red} 0.010 } & &{\color{red} 0.28 }&
576: {\color{red} 5.1\phantom{0} } \cr
577: \hbox{} \hfill & & & & \cr
578: \qtb
579: \hbox{\color{blue} Vermont} \hfill & {\color{blue} 0.49} &
580: {\color{blue} 1.0\phantom{00} }& & {\color{blue} 0.34 }&
581: {\color{blue}7.0\phantom{0} }\cr
582: \noalign{\hrule}
583: } $$
584:
585: \vskip 1.5mm
586: Notes: A large $ \Gamma $ indicates a high degree of integration whereas
587: a $ \Gamma $ close to zero suggests a high level of segregation.
588: For each minority the first column gives its population percentage.
589: The first two lines correspond to two contrasting situations in term of
590: minority proportion; Montana+South Dakota (these states have been
591: lumped together to increase the number of marriages)
592: has a sizable proportion of American
593: Indians whereas Georgia+South Carolina has a substantial Black population.
594: In the case of Vermont both minorities have a small
595: percentage. The table suggests that integration decreases when
596: the population percentage of the minority increases. It is particularly
597: striking that the integration of the Black populations in Montana+South Dakota
598: and in Vermont is higher than the integration of American Indians in
599: Montana+South Dakota because usually the integration of the Afro-American
600: population
601: is fairly low due to a long historical legacy of segregation.
602: \qL
603: Source: The data are from a $ 5\% $ random sample of the 1980 Census.
604: \vskip 2mm
605:
606: \hrule
607: \vskip 0.5mm
608: \hrule
609:
610: \normalsize
611:
612: \end{table}
613:
614: %% --------------------------------------------------------------
615: It shows that:
616: $$ \Gamma _{\hbox{\small Am.Ind.}}(\hbox{\small Low proportion of minority}) =6.4\
617: \Gamma _{\hbox{\small Am.Ind.}}(\hbox{\small High proportion of minority}) $$
618: %
619: $$ \Gamma _{\hbox{Black}}(\hbox{Low proportion of minority})=118\
620: \Gamma _{\hbox{Black}}(\hbox{High proportion of minority}) $$
621:
622: Note that the factors $ 6.4 $ and $ 118 $ cannot be really compared because
623: what we
624: call a ``high proportion'' is not the same in the two cases: for
625: American Indians ``high'' means $ 7.2 \% $, whereas for Blacks it means
626: $ 29 \% $. In addition there may be reinforcing and cumulating effects
627: due to high proportions persisting over long periods of time;
628: this historical aspect we leave for a subsequent study.
629:
630: \qA{Residential segregation}
631: At the beginning of the paper we said that our findings are consistent
632: with observations based on residential segregation. Let us shortly illustrate
633: this statement by a few examples based on a study published
634: by the Bureau of the Census (Iceland et al. 2002):
635: %
636: \qbu The most segregated
637: Metropolitan Area for Blacks in 2000 was Milwaukee-Waukesha in Wisconsin
638: (segregation index $ \delta=0.89 $
639: \qfoot{More precisely, this segregation index is related to the fraction
640: of the Black population that would have to move across blocks
641: to achieve an uniform
642: minority density.}
643: )
644: and it had a Black population percentage of $ p=25\% $;
645: the least segregated Metropolitan Area for Blacks was Orange county
646: in California ($ \delta =0.52 $) with a Black population representing
647: $ p=2.0\% $.
648: %
649: \qbu For Asians and Pacific Islanders the most segregated Metropolitan
650: Area was San Francisco (California):
651: $ \delta =0.83,\ p=33\% $ whereas the least segregated
652: was the Nassau-Suffolk area (New York): $ \delta=0.55,\ p=4.0\% $.
653: %
654: \qpar
655:
656: \qA{School integration}
657: The third characterization of ethnic integration that we mentioned is
658: school integration. In the late 1950s and early 1960s the {\it New York Times}
659: published annual maps showing the (fairly slow) progress of school integration.
660: For instance the map published on 12 May 1963 shows that
661: the percentage of Black pupils who were in class with Whites was
662: close to zero ($ <0.6\% $) in 7 states of the Deep South (Alabama, Arkansas,
663: Florida, Georgia, Louisiana, Mississippi, South Carolina); moreover, for the
664: 16 Southern states for which the {\it New York Times} gives data
665: there is a significant correlation ($ r=0.82 $)
666: between the lack of school integration and the
667: proportion of the Black population.
668: \qpar
669:
670: \qA{Hate crimes}
671: Is racial violence in the form of what the Federal Bureau of Investigation
672: calls {\it hate crimes} directed against minority members
673: also increasing with the minority's proportion? As hate crimes are a form
674: of rejection one would expect that their frequency decreases for any given
675: minority as this minority becomes better integrated. Such a
676: relaxation process suggests
677: that the historical background is of importance. That is why we restrict
678: our comparison to two communities which have been present in the United
679: States at least since the end of the War of Independence, namely Blacks
680: and American Indians.
681: In 2000 there were 104 hate crimes against Blacks
682: per million of their population as compared to a rate of 27 against American
683: Indians%
684: \qfoot{Statistical Abstract of the United States 2000, p. 188.}%
685: .
686: These figures are consistent with our previous finding that marriage integration
687: is substantially higher for American Indians.
688:
689:
690: \qI{Conclusion}
691: Using an analysis based on the number of intermarriages in the United States
692: we have seen that the proportion of minorities in the total population is
693: a key parameter in order to understand segregation patterns. In the light
694: of this finding
695: the title of the {\it New York Times} article
696: mentioned in the introduction can now be reinterpreted.
697: Did Swedes really
698: reveal a facet of their nature
699: which had not been apparent so far? One should recall that prior to 1980
700: there were almost no sizable
701: ethnic minorities in Sweden; even in 2006 they represented
702: less than $ 5\% $ of the population.
703: Thus, Sweden was in a situation similar to New Hampshire
704: or Vermont where tolerance is a natural consequence of small values of $ p $.
705: As $ p $ increased, Sweden faced the kind of situations
706: experienced by
707: U.S. states with comparable $ p $ values in Fig. 2a,b.
708: Thus, it is not surprising to see that Swedes reacted more
709: or less in the same way as residents of those states.
710: \qpar
711:
712: {\bf Acknowledgments} Many thanks to Peter Richmond who
713: offered useful comments and attracted our attention to two attempts in
714: the same direction by J\"urgen Mimkes (2006, no date).
715:
716:
717: \vfill\eject
718:
719:
720: %\appendix
721:
722:
723: \vfill\eject
724:
725: {\bf \large References}
726:
727: \qparr
728: Blau (P.), Becker (C.), Fitzpatrick (K.) 1984: Intersecting social affiliations and
729: intermarriage.
730: Social Forces 62, 585-606.
731:
732: \qparr
733: Clark-Ibanez (M.), Felmlee (D.) 2004: Interethnic relationships: the role of
734: social network diversity.
735: Journal of Marriage and Family 66, 293-305.
736:
737: \qparr
738: Ducan (O.D.), Lieberson (S.) 1959: Ethnic segregation and assimilation.
739: American Journal of Sociology 64, 4, 364-374.
740:
741: \qparr
742: Frazer (J.G.) 1910: Totemism and exogamy: a treatise on certain early
743: forms of superstition and society. Macmillan, London.
744:
745: \qparr
746: Houston (S.), Wright (R.), Ellis (M.), Holloway (S.), Hudson (M.) 2005: Places
747: of possibility: where mixed-race partners meet.
748: Progress in Human Geography 29, 6, 700-717.
749:
750: \qparr
751: Iceland (J.), Weinberg (D.H.), Steinmetz (E.) 2002: Racial and ethnic residential
752: segregation in the United States: 1980-2000.
753: U.S. Census Bureau. Census 2000 Special Report.
754:
755: \qparr
756: Jacobs (J.), Labov (T.) 2002: Gender differentials in intermarriage among
757: sixteen race and ethnic groups.
758: Sociological Forum 17, 621-646.
759:
760: \qparr
761: Kalmijn (M.) 1998: Intermarriage and homogamy: causes, patterns, trends.
762: Annual Review of Sociology 24, 395-421.
763:
764: \qparr
765: Karandikar (S.V.) 1929: Hindu exogamy. D.B. Taraporevala, Bombay.
766:
767: \qparr
768: Kortmulder (K.) 1968: An ethological theory of incest taboo and exogamy:
769: with special reference to the views of Claude Levi-Strauss.
770: Current Anthropology 9, 5, 437-449.
771:
772: \qparr
773: Lieberson (S.) 1980: A piece of the pie: Blacks and White immigrants since
774: 1880. University of California Press. Berkeley.
775:
776: \qparr
777: Lieberson (S.), Waters (M.C.) 1988: From many strands: ethnic and
778: racial groups in contemporary America.
779: Russel Sage Foundation, New York.
780:
781: \qparr
782: Makarius (R.) 1961: L'origine de l'exogamie et du tot\'emisme. Gallimard, Paris.
783:
784: \qparr
785: Mimkes (J.) 2006: A thermodynamic formulation of social science. In
786: ``Econophysics and sociophysics: trends and perspectives''
787: edited by B.K. Chakrabarti, A. Chakraborti, and A. Chaterjee, p. 279-310.
788: Wiley-VCH, Weinheim.
789:
790: \qparr
791: Mimkes (J.) no date: Intermarriage as ``thermometer'' of social systems.
792: Unpublished, personal communication.
793:
794: \qparr
795: Pagnini (D.L.), Morgan (S.P.) 1990: Intermarriage and social distance
796: among U.S. immigrants at the turn of the century.
797: American Journal of Sociology 96, 2, 405-432.
798:
799: \qparr
800: Peach (C.) 1980: Ethnic segregation and intermarriage.
801: Annals of the Association of American Geographers 70, 371-381.
802:
803: \qparr
804: Roehner (B.) 2004: Coh\'esion sociale.
805: Odile Jacob. Paris.
806:
807: \qparr
808: Roehner (B.M.) 2007: Driving forces of physical, biological and social
809: systems. A network theory investigation of social bonds and interactions.
810: To appear in July 2007. Cambridge University Press. Cambridge.
811:
812: \qparr
813: Sorensen (A.), Taeuber (K.E.), Hollingworth (L.J. Jr.) 1975: Indexes of
814: racial residential segregation for 109 cities in the United States 1940 to 1970.
815: Sociological Focus 8, 128-130.
816:
817: \qparr
818: Tatum (B.) 1997: ``Why are all the Black kids sitting together in the cafeteria?''
819: and other conversations about race.
820: Basic Books. New York.
821:
822:
823: \end{document}
824: