Internet Sources
Standards for citing online materials are still evolving. I have followed the Chicago style with a few simplifying adaptations. Major institutional websites such as the College Board or the Bureau of Labor Statistics are more easily found by Googling than by typing in a URL. I do not give URLs for the specific page I used unless finding it required significant searching once I reached the website (and even then, websites are so constantly in flux that you will often find a “page not found” message when you enter the URL that worked for me). Regarding books and other documents in the public domain that have been accessed online, you can find the context for any specific quote by going to the institutional website giving access to that book, searching for the book, then entering a short phrase from the quotation into the search function. In accordance with Chicago’s guidelines, I do not include the date when I accessed the website. If it no longer exists when you read this book, knowing that it did at some date in the past does not seem helpful.
Frequently Used Abbreviations
Prologue: November 21, 1963
1. Deadline Hollywood website, http://www.deadline.com/2010/05/full-series-rankings-for-the-2009-10-broadcast-season/. Televisionista website, http://televisionista.blogspot.com/2008/06/tv-ratings-2007-2008-season-top-200.html.
2. The Current Population Survey didn’t yet ask about children in the house, but we know from later years that wives without young children are more than twice as likely to work full time than wives who do have young children.
3. The Motion Picture Association of America Production Code website, http://productioncode.dhwritings.com/multipleframes_productioncode.php.
4. Time magazine website, http://www.time.com/time/magazine/article/0,9171,898033,00.html.
5. Author’s analysis of data from Gallup poll #1963-0678, obtained from the Roper Center for Public Opinion Research website.
6. FBI, Crime in the United States 1963.
7. Ross, 1987.
Part I: The Formation of a New Upper Class
1. Reich, 1991.
2. Herrnstein and Murray, 1994.
3. Brooks, 2000, 10.
4. Florida, 2002, xxvii.
5. SAUS-2011, table 509.
6. The age range of 25 and older is chosen to be coordinate with the available census breakdown for zip codes. But it is also convenient as an age range for embracing the new upper class. No one except in the entertainment industry or sports rises to the top 5 percent of an occupation under the age of 25, and many people in the new upper class who are still employed after age 65 remain in their prominent positions.
7. The calculation was as follows: BLS statistics for 2010 indicate that 121,987,000 Americans ages 25 and older were employed (Employment and Earnings Online, table 8, January 2011). The March CPS for 2010 indicates that 23.4 percent were in the professions or managerial positions. That leads to a top 5 percent of employed persons in those occupations in 2010 consisting of 1,427,248 persons (Occupations and Earnings 2010, table A3, available online at the Bureau of Labor Statistics’ website).
8. This number needs a correction that I won’t try to make. In 2010, 16 percent of the spouses of people in the professions and managerial positions ages 25 and older were also in one of those occupations. But trying to estimate what percentage of these couples consisted of two people in the most successful 5 percent would be pushing this attempt to develop a ballpark figure much too far.
9. In Eisenhower’s time, the cabinet consisted of ten posts: the heads of the Departments of State; Treasury; Defense; Justice; Interior; Agriculture; Commerce; Health, Education, and Welfare; Labor; and the Postmaster General. Postmaster General has not been a cabinet position since 1971, so I used the nine cabinet departments as the basis for comparing the Eisenhower and Kennedy cabinets with those of George W. Bush and Barack Obama (as of 2011). Forty-six percent of the cabinet members in the Eisenhower and Kennedy years (1953–63) grew up in lower-middle-class or working-class families compared to 27 percent in the Bush and Obama cabinets (2001–10). Thirty-two percent of the cabinet members in the Eisenhower and Kennedy years grew up in upper-middle-class, rich, or politically influential families, compared to 54 percent in the Bush and Obama cabinets.
1: Our Kind of People
1. The campuses were the University of California at Berkeley, Stanford, the University of Chicago, the University of Michigan, Michigan State, Ohio State, and Swarthmore. Qualifying census tracts are limited to those with at least 500 adults ages 25 and older and 250 families or more.
2. Department of Housing and Urban Development website, http://www.huduser.org/periodicals/USHMC/spring2001/histdat2.html.
3. Los Angeles Public Library website, http://dbase1.lapl.org/.
4. The estimate of 80,000 is for 1961–62, in Phillips, 1991, appendix A.
5. The full account of how this apocryphal exchange entered literary history is told in a letter to the editor of the New York Times Book Review by Eddy Dow for November 13, 1988. Obtained from the archives of the New York Times website.
6. I am indebted to the current owner of Topridge for access to materials about Mrs. Post’s lifestyle at Topridge and her other homes.
7. By the 2000 census, 66 percent of the residents of Cambridge had college degrees and the median income for Cambridge as a whole stood well above the national average at $80,565.
8. Brooks, 2000, 55–57.
9. In 2010, 55 percent of all new cars sold in the United States were foreign (www.goodcarbadcar.net). In the parking lot of the Walmart in predominantly working-class Newton, Iowa (centile score 47), on June 10, 2011, 83 percent of the cars were American makes (n=200). In the shopping center and surrounding streets of predominantly working-class Brunswick, Maryland (centile score 41), on the afternoon of May 27, 2011, 64 percent of the cars were American makes (n=200). In the parking lots of Wildwood Shopping Center and Georgetown Square in affluent Bethesda, Maryland (centile 99), on the morning of June 27, 2011, 23 percent (n=171) and 17 percent (n=150) of the cars were American makes, respectively.
10. Author’s analysis of data from the Centers for Disease Control’s Behavioral Risk Factor Surveillance System (BRFSS), annual survey for 2009. Data downloaded from the Centers for Disease Control website.
11. The data are available from the Saguaro Seminar’s Civic Engagement in America website, Bowling Alone, http://www.bowlingalone.com/data.htm.
12. Author’s analysis of data for the population as a whole. Data are from the Pew Research Center’s Diet/Gambling/Movies survey, released November 13, 2007. Data available from the Pew Research Center website, http://pewsocialtrends.org/category/data-sets/.
13. In the Centers for Disease Control’s Behavioral Risk Factor Surveillance System for 2009, 35 percent of the respondents said that they smoked some days or every day.
14. Pew Research Center, “Americans Spending More Time Following the News,” September 12, 2010, available online at the Pew Research Center for the People & the Press website, http://people-press.org/2010/09/12/americans-spending-more-time-following-the-news/.
15. For a review of the studies relating socioeconomic status to television viewing, see Gorely, Marshall, and Biddle, 2004.
16. This number is based on Nielsen data for the first two quarters of 2010, continuing a shallow long-term upward trend: http://blog.nielsen.com/nielsenwire/media_entertainment/state-of-the-media-tv-usage-trends-q2-2010/.
17. For the relationship of socioeconomic status to breast-feeding, see Heck, 2006.
18. See, for example, Aronson and Huston, 2004; Mcloyd, 1998; Parcel and Menaghan, 1989.
19. For a sampling of resources for New York parents applying to preschool, see http://blogs.urbanbaby.com/newyork/2010/08/17/a-league-of-your-own-for-school-admissions/.
20. An excellent summary of the technical literature for the general reader is Bronson, 2009.
21. U.S. News & World Report website, http://colleges.usnews.rankingsandreviews.com/best-colleges.
22. Brooks, 2000, chapter 3; Florida, 2002, chapters 5–9.
23. Florida, 2002, quoting office architect Don Carter, 123.
2. The Foundations of the New Upper Class
1. Quoted in Karlgaard, 2005.
2. Herrnstein and Murray, 1994, chapters 2 and 3. For a more recent survey of the literature on this topic, see Gottfredson, 2003.
3. Goldberg, 2003, 51–52.
4. All the Fortune 500s since 1955 can be found on the CNN Money website, http://money.cnn.com/magazines/fortune/fortune500/2011/. The corporations ranked 100 in 1960 and 2010 were McDonnell Douglas and Amazon, respectively. The 500th-ranked corporations were Masonite and Blockbuster.
5. Eberstadt, 2008. Broda and Weinstein, 2008, make the case that the Consumer Price Index has systematically understated real gains in purchasing power.
6. For an examination of whether a change in CPS methodology could have produced the jump in 1994–95, see Raffalovich, Monnat, and Hui-shien, 2009, which concludes that it probably did not. For the analysis using IRS data, see Piketty and Saez, 2006. Data in Figure 2.1 use one income figure per family unit.
7. Brooks, 2000, 178–85.
8. For a full presentation of the data on college stratification as of the early 1990s, see Herrnstein and Murray, 1994, chapter 1.
9. Herrnstein and Murray, 1994, 38. The schools were Brown, Bryn Mawr, Columbia, Harvard, Mount Holyoke, Princeton, Radcliffe, Smith, University of Pennsylvania, Vassar, Wellesley, Williams, and Yale.
10. Herrnstein and Murray, 1994, 30.
11. Bender, 1960, 4.
12. Soares, 2007, 38.
13. In absolute numbers, the four largest concentrations of students with admissions test scores in the top 5 percent were all in public universities—the state universities of California at Berkeley, Illinois at Urbana-Champaign, Michigan at Ann Arbor, and Wisconsin at Madison (Geiger, 2002, table 2). It would seem that the high-scoring students in these schools would be exposed to a much more diverse set of classmates than those in the Ivies. But taking such students as a percentage of the entire entering class in those universities is misleading. All of the public universities that were part of Geiger’s 105 schools have honors programs, and all of them are described in the same way: They try to replicate the experience of the small liberal arts college within the framework of the large state university. Students in honors colleges have access to special courses, with small classes filled with other honors students and taught by specially selected faculty. Many of the programs also have housing set aside for the honors students. The University of Michigan recently opened the Perlman Honors Commons—in effect, a separate student union for honors students. These programs go a long way toward replicating the cognitive profile and much of the social interaction found at elite private schools. They are also increasingly competitive. As of 2010, getting into the honors programs of the top public universities required credentials similar to those required for many elite private colleges.
14. The website for the U.S. News & World Report rankings is http://colleges.usnews.rankingsandreviews.com/best-colleges.
15. Espenshade and Radford, 2009, chapter 4.
16. Soares, 2007, tables 1.1 and 6.1.
17. Ibid., 3.
18. Ibid., table 6.6.
19. The figure for private schools is specifically for non-Catholic private schools.
20. Golden, 2006.
21. Soares sees four factors as evidence that admissions committees are still admitting the “right” kind of people: (1) having a parent with a postgraduate or professional degree, (2) graduating from a non-Catholic private school, (3) coming from a family in which both parent and student have visited an art museum, and (4) being an officer in high school government. The first three lend themselves to interpretations involving academic merit. A parent with a PhD, medical degree, or law degree from a reputable university is virtually guaranteed to have a high IQ (obtaining a postgraduate or professional degree screens very effectively for high IQ). Students who have obtained entrance to competitive private schools have undergone a screen for IQ, and students who graduate from private schools are likely to be academically better prepared than students from public schools. The young person who has gone to an art museum with a parent may have done so in blind obedience to the parent, but there is likely to be a correlation with the child’s IQ and whether, being given an opportunity to go to the art museum, he actually did. As for the fourth variable, being an officer in high school government, an admissions office that takes such nonacademic achievements into account would seem to be giving points for the student’s actual accomplishments, not a penumbra of cultural capital.
22. I continue to use “math” and “verbal” for what are now called the critical reading and math reasoning tests. The percentages on parental education are unpublished figures provided to me courtesy of the College Board.
23. Since so many people reading this book, especially parents with children nearing college, assume that coaching can raise their children’s SAT scores by large amounts, discussion of this issue is warranted. From 1981 to 1990, three separate analyses of all the prior studies were published in peer-reviewed journals. They found a coaching effect of 9 to 25 points on the SAT verbal and of 15 to 25 points on the SAT math. See Herrnstein and Murray, 1994, 400–402. Derek Briggs, using the National Education Longitudinal Study of 1988, found effects of 3 to 20 points for the SAT verbal and 10 to 28 points for the SAT math (Briggs, 2004). Donald Powers and Donald Rock, using a nationally representative sample of students who took the SAT after its revisions in the mid-1990s, found an average coaching effect of 6 to 12 points on the SAT verbal and 13 to 18 points on the SAT math (Powers and Rock, 1999). These effects are not large enough to sway many college admissions decisions.
No study published in a peer-reviewed journal shows average gains approaching the fabled 100-point and 200-point jumps you hear about in anecdotes. When investigating this issue in 2007, I asked Kaplan and Princeton Review for such evidence. Kaplan replied that it chooses not to release data for proprietary reasons. Princeton Review did not respond (Murray, 2007).
The illusion of large gains arises mainly from two artifacts. The first is self-selection. The students who seem to profit from a coaching course tend to be those who, if the course had not been available, would have worked hard on their own to prepare for the test. The second is the conflation of the effect of coaching with the effect of preparation that students can do on their own. No student should walk into the SAT cold. It makes sense for students to practice some sample items and to review their algebra textbook if it has been a few years since they have taken algebra. But once a few hours have been spent on these routine steps, most of the juice has been squeezed out of preparation for the SAT. Combine self-selection artifacts with the role of basic preparation, and you have the reason that independent studies using control groups show such small average gains from formal coaching.
24. The transmission works through both genes and environment, but the distinction is blurred because cognitive ability in the parents is associated with parenting practices that promote the child’s cognitive ability. In addition, it has been found that the shared environment among siblings—which includes the things that parents do to promote cognitive development in their children—has a small long-term role independent of genes. See Plomin et al., 2001; Rowe, Vazsonyi, and Flannery, 1994; Rowe, 2003.
25. The correlation of spousal IQ has been in the region of +0.4 since spouses have been tested (Jensen, 1998, 183), indicating an underlying role of cognitive ability in mate selection that probably has always existed. But a correlation alone is not enough for understanding the kind of phenomenon discussed in the text, in which people with high IQs marry each other. A positive correlation reflects the degree to which two phenomena vary together, but nothing more. If every woman married a man whose IQ score was exactly equal to hers, the correlation of spousal IQ would be +1.0, and it would also be +1.0 if every woman married a man whose IQ score was exactly 20 points higher than hers—but the implications for the IQ of offspring would be radically different.
26. Schwartz and Mare, 2005.
27. This statement is true for public universities and unselective private colleges, where gaining admission is easier than graduating. It is not necessarily true of selective colleges. “The hardest thing about Harvard is getting in” was already a commonplace when I was there in the early 1960s.
28. Murray, 2008, chapter 3.
29. The original Coleman Report is Coleman, Campbell, and Hobson, 1966. For a collection of reanalyses of the Coleman Report, see Mosteller and Moynihan, 1972.
30. In the NLSY-79, the means for whites obtaining bachelors, masters, and PhDs or professional degrees by the year 2000 (when the NLSY-79 subjects were ages 36–43) were 113.3, 116.9, and 125.6, respectively. For blacks, the comparable means were 99.1, 101.7, and 112.2. The Latino means were 106.7, 106.4, and 115.2. See Murray, 2009, for a full discussion of the stability of IQ scores for various degree levels.
31. The whiteness of the broad elite is discussed in chapter 3, this page-this page, relative to the racial composition of the most affluent and best-educated zip codes. The narrow elite in the private sector is also still overwhelmingly white. A few examples: Among the Fortune 500 CEOs as of 2011, 98 percent were white (as always, meaning non-Latino whites). Among the 51 directors nominated for Academy Awards from 2000 to 2011, 92 percent were white. Among the 123 syndicated columnists in 2008 with the largest number of outlets, 95 percent were white. My search on senior executives in major television networks, both news and entertainment, did not produce examples of nonwhite executives in jobs that shape content, but the ethnicity of many of those executives could not be identified.
The whiteness of the narrow elite in government jobs varies. Statewide offices are still overwhelmingly held by whites. For example, as of the end of 2010, 45 of the 50 governors and 96 out of 100 senators were whites of European origin. Elections at the district and municipal level are more likely to produce ethnically diverse officeholders (the House of Representatives as of the end of 2010 was 83 percent white), but they also contribute few members of the narrow elite. Presidential appointments are also ethnically diverse. For example, federal judges serving as of the end of 2010 were 78 percent white. Data for these statements were collected from a large number of websites, including person-by-person web searches. The data on federal judges were obtained from www.uscourts.gov/JudgesAndJudgeships/BiographicalDirectoryOfJudges.aspx.
32. For a discussion of the psychometric properties of the AFQT, see Herrnstein and Murray, 1994, appendix 2. The scores used here are normed by age using comparable procedures for the 1979 and 1997 cohorts of the NLSY.
33. Kalmijn, 1994; Kalmijn, 1998.
34. Arum, Roksa, and Budig, 2008.
35. The standard linear regression equation
predicts the magnitude of regression to the mean independently of whatever causal mechanism may be involved (Humphreys, 1978). In the case of parental-child regression to the mean in IQ, is a given child’s expected IQ, X is the midpoint parental IQ for a given pair of parents,
is the sample mean for midpoint parental IQ,
is the sample mean for offspring IQ, rxy is the sample correlation of midpoint parental and offspring IQ, and sxand sy are the sample standard deviations of midpoint parental IQ and offspring IQ, respectively.
To fill in these parameters, I use a white mean of 103 and standard deviation of 14.5. These are based on the mean of the array of sample means produced by standardizations for the Stanford-Binet (version 5, subjects ages 12–23, 2001); the Wechsler Adult Intelligence Scale (version III, subjects ages 16–64, 1995); the Wechsler Intelligence Scale for Children (version IV, subjects ages 14–16, 2002); and the Armed Services Vocational Aptitude Battery (NLSY-97, subjects ages 13–17, 1997). The mean standard deviation of those same data sets was 14.5. Data for the Stanford-Binet and Wechsler standardizations were provided courtesy of William Dickens, and reported in Dickens and Flynn, 2006. In the calculations of the variance of midpoint parental IQ (the equation for doing so is given in the appendix of Humphreys, 1978), I specified a correlation of spousal IQ of +0.5, which, given a standard deviation of 14.5 for the white population standard deviation, produced an expected standard deviation of midpoint parental IQ of 12.6. For a review of the literature on familial IQ correlations that leads to these specifications, see Bouchard, 1981.
36. An IQ of 135 assumes that the average graduate of an elite college is at the 99th centile of IQ of the entire population of seventeen-year-olds. This is consistent with the median combined Critical Reading and Mathematics scores of 1400 or more among the top dozen schools in the most recent U.S. News & World Report rankings (http://colleges.usnews.rankingsandreviews.com/best-colleges). In 2010, a combined score of 1400 put a student at about the 97th percentile of all students who took the SAT (based on the distribution produced by the known means and standard deviations for the two tests and a correlation of +0.7 between them). But the number of test-takers in 2010 represented only 36 percent of the seventeen-year-olds in the country. Any plausible assumptions about the proportion of the 62 percent of seventeen-year-olds who didn’t take the SAT who could have gotten a combined score of 1400 or more puts a student who actually does score 1400 well into the 99th centile of the seventeen-year-old population. For 2010 SAT test data, see College-Bound Seniors 2010, available at the College Board website. For a discussion of estimating SAT scores for those who don’t take the test, see Murray, 2008, 70, and the associated notes.
37. Murray, 2009, 102.
38. Gottfredson, 2003.
39. The following numbers are not statistically derived, but represent the results of a simulation that used Stata’s DRAWNORM command to create a sample of 10 million normally distributed values of two variables with means of 103, standard deviations of 14.5, and a correlation of +0.5.
3: A New Kind of Segregation
1. Massey, 2009.
2. Ibid., figure 5.
3. Ibid., figure 8.
4. Ibid., 85.
5. Four times the median poverty threshold for the CPS (based on persons of all ages and races) in 1999, the income year for the 2000 census, was $67,824, cutting off the 58th centile of family income in the CPS for 1999.
6. In the comparisons of 1960 and 2000, I use census tracts for 1960 and zip codes for 2000. The 1960 census tract data are taken from the Elizabeth Mullen Bogue file (hereafter Bogue), named for the woman who did much of the keypunching of the data published in the printed publications of the Bureau of the Census. Those published data in 1960 included 175 metropolitan areas with 104,010,696 people, or 58.0 percent of the resident population. Zip codes, which didn’t yet exist in 1960, are a much more easily understood unit than census tracts, so I use zip codes for the 2000 census, using complete national data downloadable from the American FactFinder tool on the Bureau of the Census’s website. In those instances when I directly compare 1960 and 2000 data, I restrict the comparison to the metropolitan areas covered by the Bogue file.
7. The figure in the text is the median of the median family income in the four census tracts, weighted by the number of families in each census tract. Parallel weighting is used for other statistics that aggregate across census tracts or zip codes. The $60,700 threshold would be passed by an Austin teacher with a bachelor’s degree, eighteen years of experience, working a 230-day year. Salary schedules are from http://www.austinisd.org.
8. Weighted mean based on the population ages 25 and older.
9. Not all of the high-education zip codes were rich. The zip code for the University of Texas campus (78705) had the third-highest proportion of BAs, 73 percent, but a median income of only $46,480 dollars, reflecting the presence of lots of grad students who had BAs but hardly any income. Otherwise, however, education and wealth went together.
10. Moll, 1985. The others were William and Mary, Miami University of Ohio, University of California (all campuses), University of Michigan, University of North Carolina at Chapel Hill, University of Vermont, and University of Virginia.
11. For the 2000 census, what I am calling for convenience “north of Central Park” consisted of all the zip codes from Ninety-Fourth Street northward on the West Side and from Ninety-Sixth Street northward on the East Side. The Upper East Side included the zip codes that encompassed East Sixtieth to Ninety-Sixth streets and Fifth Avenue to the East River. The 1960 census tracts used for the analysis were consistent with these zip code borders within two blocks.
12. The salary schedule for the New York City Department of Education can be found at http://schools.nyc.gov/NR/rdonlyres/EDDB658C-BE7F-4314-85C0-03F5A00B8A0B/0/salary.pdf.
13. The three-block radius also included my family in one direction and my wife’s family in another. Our fathers were both mid-level executives at Maytag.
14. The other 4 percent of the SuperZip population consists of Native Americans, Americans with origins in the Pacific islands, and people classified as mixed race.
15. In 2011, for example, Asian applicants made up 18 percent of the acceptances at Harvard. Harvard Gazette, May 11, 2011, http://news.harvard.edu/gazette/story/2011/03/an-unprecedented-admissions-year/. For a complete analysis of trends in minority admissions since the 1980s, see Espenshade, 2009. The most detailed presentation of the evidence that Asian students have an admissions disadvantage—Asian applicants have to have higher SAT scores than other students (including whites) to have an equal probability of being admitted—is in Espenshade, Chung, and Walling, 2004.
16. The phrase “honorary whites” is associated with Hacker, 1992, but it does not occur in that book. Although Hacker has used the phrase in many talks, he cannot recall ever consigning it to print (Andrew Hacker, personal communication, May 14, 2011).
17. Census Bureau, http://2010.census.gov/2010census/data/index.php.
18. The American FactFinder tool was transitioning to a new version in 2011, but the link to it appears on the home page of the Census Bureau, www.census.gov. The ethnic profiles of the six zip codes in the 2000 census were as follows:
Zip code | % White | % Black | % Hispanic | % Asian |
02461 | 82.0 | 1.0 | 2.1 | 11.8 |
10583 | 81.8 | 2.1 | 2.7 | 11.4 |
20007 | 82.5 | 4.0 | 4.2 | 5.7 |
60657 | 82.4 | 3.4 | 4.4 | 5.0 |
90212 | 82.4 | 1.7 | 2.2 | 8.3 |
94301 | 81.7 | 1.9 | 2.1 | 9.4 |
19. This number is based on the zip code classifications as of 2000, and does not include zip codes associated exclusively with a post office.
20. The class numbered 776 students. As of the twenty-fifth reunion, information was available for 743 of them (96 percent). Fifteen were deceased and 135 were living abroad. For those living in the United States, the twenty-fifth reunion profiles showed the town or city in which they lived but not the zip code. Many of the smaller towns had a single zip code. For those with multiple zip codes, I used online white pages to determine the home address and its zip code. Since the names almost always included a middle initial and the name of the spouse or partner, I was able to determine home zip codes for all but 45 of those living in the United States. The numbers in the text are based on the 547 who were known to be living in the United States and for whom a home zip code could be determined. The zip codes of those for whom data could not be obtained were probably even more heavily concentrated in the elite zip codes than those whose zip codes were obtained—almost all of those 45 names could be found in online white pages, and they had work addresses in the most exclusive zip codes of New York City, Los Angeles, San Francisco, Chicago, or Boston, but their home phone numbers and addresses were unlisted.
21. It is worth noting that the reviewers of Bobos in Paradise for the New York Times, Wall Street Journal, and Washington Post were Janet Maslin, Emily Prager, and Jonathan Yardley, respectively, all of whom had spent their adult lives intimately familiar with members of the new upper class, and all of them thought that Brooks nailed it.
22. For details of the sample, see appendix B.
23. You may be wondering about the three isolated SuperZips: two little ones in the top left quadrant of the map and one large one on the right center. The two little ones are the zip code that contain the tiny Maryland towns of Barnesville, with a population of 138 adults, and Beallsville, with 76. The large one is zip code 20721, adult population 14,451, part of Bowie, Maryland, one of three SuperZips in the nation with a majority African American population (82 percent black). The other two are the tiny zip code 45384, adult population 123 (95 percent black) of Wilberforce, Ohio, near Dayton; and zip code 60461, adult population 3,347 (55 percent black), the Chicago suburb of Olympia Fields, Illinois.
24. In a few cases, a SuperZip was separated from a cluster of other SuperZips by a single zip code with a centile score of 90 or higher. I included such SuperZips in the cluster.
25. Including San Francisco with New York, Washington, and Los Angeles is a judgment call based on the enormous influence that the information technology sector has acquired in the last three decades, not just technologically and economically but culturally. To the CEOs of multibillion-dollar businesses who do not live in the cities I listed and are incensed at being omitted from the narrow elite, I can only observe that lots of large corporations could go bust without making a ripple on the national scene.
26. For literature reviews and original data, see Cardiff, 2005, and Mariani, 2008. The Left’s faculty dominance varies widely by type of both school and department. The humanities and social sciences have the most drastic tilt. Here are examples of Left:Right ratios from Mariani, 2008, table 3, going from highest to lowest: English 7.4:1, history/political science 6.2:1, social sciences 5.8:1, humanities 5.4:1, physical sciences 4.1:1, biological sciences 4.0:1, engineering 1.5:1, health sciences 1.0:1, business 0.8:1. Faculties of highly selective schools are even further to the Left (3.7:1) than schools that are not highly selective (2.7:1).
27. In a survey of five hundred journalists that the Pew Project for Excellence in Journalism conducted in 2007, Pew reported the results for three levels of the national media: executives (CEOs, general managers, and publishers), senior editors and producers, and working journalists and editors. The least liberal tilt was found for executives, who had a Left:Right ratio of 1.6:1. For senior editors and producers, the ratio was 2.1:1. For working journalists and editors, it was 6.7:1. Among the latter group, 12 percent described themselves as very liberal, 28 percent as liberal, 3 percent as conservative, and 3 percent as very conservative (Project for Excellence in Journalism, 2008, 55). The PDF of the report is available at http://www.stateofthemedia.org/2008/Journalists%20topline.pdf. See also Groseclose, 2005.
28. Source: almost any Academy Awards show.
29. Callahan, 2010, chapter 1.
30. The percentages in the text represent town-level results when a SuperZip represented the only zip code for the town, or when SuperZips represented at least half of the zip codes when a town had more than one. Data were obtained by going to the websites of state boards of election. Some of them present data by town; some of them don’t.
31. Bishop, 2008, 1–8.
32. Data downloaded from the ADA website.
4: How Thick Is Your Bubble?
1. Tocqueville, 1840, vol. 2, Google Books.
2. The idea for creating a test and items 7 and 11 (and perhaps a few others I’ve forgotton) came from reading Brooks, 2000.
3. “Chief breadwinner” is defined as the person with the higher-rated occupational category in a household headed by a married couple.
4. Author’s analysis, based on persons in the NLSY-79 sample followed from 1979 through 2006 with no more than two missing interview waves.
5. The Pew Forum on Religious and Public Life website, http://religions.pewforum.org/affiliations.
6. Bishop, 2008.
7. Murray, 2008, chapter 2.
8. These assume that the standard deviation for a school with a mean IQ of 115 is 12 instead of the national standard deviation of 15, consistent with what is empirically observed with subgroups that score substantially higher or lower than a national mean.
9. Centers for Disease Control, http://www.cdc.gov/BRFSS/.
10. Chinni, 2010, introduction, Kindle edition.
11. Pickuptrucks.com, http://news.pickuptrucks.com/2011/01/2010-year-end-top-10-pickup-truck-sales.html.
12. In the DDB Life Style data for 1995–98: If you did not have a college degree and were anywhere under $100,000 per year in income, you had a 14 percent chance of fishing five or more times per year. With a college degree and an income greater than $100,000, you had a 4 percent chance. Extrapolate that relationship to people who are in the top few centiles of socioeconomic status, and the percentage presumably drops accordingly.
13. My basic source was http://en.wikipedia.org/wiki/List_of_casual_dining_restaurant_chains. I went to the specific websites of restaurants with worldwide outlets to estimate the number of outlets in the United States.
14. Some of the chains are privately held, and revenues must be estimated. Twelve billion dollars is an extremely conservative estimate.
15. Box Office Mojo website, http://www.boxofficemojo.com/yearly/chart/?yr=2009&p=.htm.
16. Deadline Hollywood website, http://www.deadline.com/2010/05/full-series-rankings-for-the-2009-10-broadcast-season/.
17. SBJNet website, http://sbj.net/main.asp?SectionID=18&SubSectionID=23&ArticleID=86519.
5: The Bright Side of the New Upper Class
1. Herrnstein, Bekle, and Taylor, 1990.
2. Herrnstein and Murray, 1994, 34.
Part II: The Formation of a New Lower Class
1. U.S. Bureau of the Census, 1975, vol. 1, series D, table nos. 182–232. These figures refer to workers of both sexes, all ages, and all races.
6: The Founding Virtues
1. Grund, 1837, Google Books.
2. Quoted in Adams, 1889, Google Books.
3. Quoted in ibid.
4. Quoted in ibid.
5. Quoted in ibid.
6. Kurland, 1986, vol. 1, chapter 13, document 36, http://press-pubs.uchicago.edu/founders/documents/v1ch13s36.html.
7. Benjamin Franklin to William Strahan, February 16, 1784, in Murphy, 1906, Google Books.
8. Quoted in Spalding, 1996, 30.
9. Adams, 1889, Google Books.
10. Tocqueville, 1840, vol. 2, Google Books.
11. Grund, 1837, Google Books.
12. Adams, 1889, Google Books.
13. Hamilton, 1833, Google Books.
14. Grund, 1837, Google Books.
15. Thomas Jefferson to Nathaniel Macon, in The Works of Thomas Jefferson, federal ed., vol. 12 (New York and London: G. P. Putnam’s Sons, 1904–5), Online Library of Liberty.
16. The other virtues were piety, philanthropy, industry, and economy in one list and harmony, industry, and frugality in the other. George Washington to the General Assembly of Presbyterian Churches, May 1789, in Allen, 1988, 181; and George Washington to the Marquis de Lafayette, January 29, 1789, in Allen, 1988, 161, Online Library of Liberty.
17. John Adams to Secretary Jay, September 23, 1787, in Adams, 1856, vol. 8, Online Library of Liberty.
18. Thomas Jefferson to William Duane, August 4, 1812, in Ford, 1904, Online Library of Liberty.
19. For historical crime data, see Gurr, 1989.
20. The data on prosecutions for theft come from Nelson, 1967. The data on the population of Middlesex County come from Chickering, 1846, Google Books.
21. By way of comparison: The rate of reported larceny-thefts in the United States in 2008 was 319 per 10,000 people, a number that doesn’t add in all the other forms of property crime or consider all the larceny-thefts that go unreported. Uniform Crime Reports for 2008.
22. In the cities, there is also the peculiar role of mobs in the nineteenth century. This is a rich topic unto itself, but, like frontier fighting, the activities of mobs only occasionally fell into the categories of crime or dishonesty as we normally think of it.
23. Tocqueville, 1840, vol. 1, Google Books.
24. Hamilton, 1833, Google Books.
25. Grund, 1837, Google Books.
26. James Wilson, “Of the Natural Rights of Individuals,” in Collected Works of James Wilson, vol. 2, ed. Kermit L. Hall and Mark David Hall (Indianapolis: Liberty Fund, 2007), Online Library of Liberty.
27. John Adams, Diary, June 2, 1778.
28. John Adams, “John Adams to the Young Men of the City of New York,” in Charles Francis Adams, ed., The Works of John Adams, Second President of the United States: With a Life of the Author, Notes and Illustrations, vol. 9 (Boston: Little, Brown and Co., 1856), Online Library of Liberty. Even Benjamin Franklin, whose extramarital liaisons were many and enthusiastic, observed that “a bachelor is not a complete human being. He is like the odd half of a pair of scissors, which has not yet found its fellow, and is therefore not even half so useful as they might be together.” Benjamin Franklin to Thomas Jordan, London, May 18, 1787, in Murphy, 1906, Google Books.
29. Martineau, 1837, part 2, Google Books.
30. Tocqueville, 1840, vol. 2, Google Books.
31. Ibid.
32. Ibid.
33. Grund, 1837, Google Books.
34. Novak, 2002, 34.
35. Washington’s Farewell Address, Online Library of Liberty.
36. John Adams to the officers of the first brigade of the third division of the militia of Massachusetts in The Works of John Adams.
37. John Adams to F. A. Vanderkemp, quoted in Novak, 2002, epigraph.
38. James Madison to Frederick Beasley, November 20, 1825, quoted in Novak, 2002, 33.
39. Thomas Jefferson, Notes on the State of Virginia, Online Library of Liberty.
40. The anecdote is given in the diary of the Reverend Ethan Allen, now held by the Library of Congress, and is quoted in full in Novak, 2002, 31. Its authenticity is unverified. Allen was a child during Jefferson’s presidency, so it is probably a secondhand account at best. But this passage from a letter written in 1807 reflects a sensibility consistent with the anecdote: “The practice of morality being necessary for the well-being of society, [our Creator] has taken care to impress its precepts so indelibly on our hearts that they shall not be effaced by the subtleties of our brain. We all agree in the obligation of the moral precepts of Jesus, and nowhere will they be found delivered in greater purity than in His discourses. It is, then, a matter of principle with me to avoid disturbing the tranquility of others by the expression of any opinion on the innocent questions in which we schisamatize.” Thomas Jefferson to James Fishback, in Foley, 1900, Google Books.
41. Thomas Jefferson to William Canby, September 18, 1813, http://www.beliefnet.com/resourcelib/docs/57/Letter_from_Thomas_Jefferson_to_William_Canby_1.html.
42. Quoted in Clark, 1983, 413.
43. Quoted in Novak, 2002, 37. See the rest of Novak’s chapter 2 for examples of links between Christianity and the needs of a self-governing society.
44. Tocqueville, 1840, vol. 1, Google Books.
45. New York Times, April 22, 1910, p. 1, New York Times Archives.
1. For an excellent treatise on that proposition, see Hymowitz, 2006.
2. Roper Center for Public Opinion Research website, http://www.ropercenter.uconn.edu/. USGALLUP.556POS.R137M.
3. Roper Center for Public Opinion Research website.
4. This calculation multiplies the “happily married among those married” percentage from the GSS with the percentage of persons married in the CPS, with its much larger and nationally representative sample.
5. Aronson and Huston, 2004.
6. Fomby and Cherlin, 2007; Cavenagh and Huston, 2006.
7. Bronte-Tinkew et al., 2006; Harper and McLanahan, 1998.
8. Sourander et al., 2006.
9. Bauman, Silver, and Stein, 2006; Denise et al., 2005.
10. Warner and Hayward, 2006.
11. Pearson, Muller, and Frisco, 2006.
12. Carlson, 2006.
13. Brown, 2006.
14. The citations of specific journal articles are only illustrative of a large literature. Some major review sources are McLanahan and Sandefur, 1994; Mayer, 1997; McLanahan, 2001; Aronson and Huston, 2004; and Hymowitz, 2006.
15. In making this calculation, I exclude children living with a widowed parent.
16. Malinowski, 1930, Google Books.
17. Laslett, Oosterveen, and Smith, 1980.
18. Brown and Manning, 2009.
19. Bumpass and Lu, 2000.
20. Ibid.
21. Aronson and Huston, 2004, table 1.
22. Ibid., table 2.
23. Brown, 2004, table 1.
24. Summarized in Bumpass and Lu, 2000.
9: Industriousness
1. For whites ages 16 and older, the unemployment rate was 5.1 percent in 1960–64 and 4.5 percent in 2008. I use 2004–8 instead of the most recent five-year period, 2006–10, to avoid clouding the comparison with the high unemployment rates of 2009 and 2010. Bureau of Labor Statistics website.
2. The slight declines in labor force dropouts between the 1960 measure, based on the decennial census, and 1968, based on the CPS, should be ignored. Overall, we know that white male labor force participation among prime-age males during the decade remained flat, and the slight decline is prudently attributed to the difference in the sources. There could be some incomparability, despite the identical labor force question used in both surveys, because the CPS data all come from the March survey, whereas the census data are collected over a broader span of time. Another issue is the assignment of occupations for people who are out of the labor force. The census of 1960 was significantly more likely than the CPS surveys to identify someone out of the labor force with an occupation.
3. Another peculiarity of the graph is the sudden jump for men with no more than a high school education in 1993–94. I have satisfied myself that it is not a result of miscoding or other data errors, but I have no explanation for it.
4. Herrnstein and Murray, 1994, chapters 7–8.
5. I use the national unemployment rate for the civilian noninstitutional population ages 16 and older as reported by the Bureau of Labor Statistics to reflect the overall state of the labor market.
6. Logit analysis regressing a binary variable (employed or unemployed) on year and the national unemployment rate.
7. The CPS reports hours in intervals (1–14, 15–29, 30–34, 35–39, 40, 40–48, 49–59, and 60+). I used the midpoint of each interval, and 65 for those in the 60+ group, to reach my estimate of hours per week.
8. Sundstrom, 1999, presents evidence that time-diary estimates of hours per week show smaller weekly totals than the CPS estimates, and that the college-educated show the greatest discrepancy. Whether this discrepancy represents a real overestimate of time spent working or differences in the kinds of work captured by the two measures is not known.
9. Bureau of Labor Statistics website, www.bls.gov. Occupational Employment Statistics for 2009, http://www.bls.gov/oes/current/oes_nat.htm. Converted to 2010 dollars.
10. Bureau of Labor Statistics website.
11. The leisure hours were 104.3 for college men and 101.9 for men without a high school diploma, a 2 percent difference. Aguiar and Hurst, 2009, table 2-2.
12. Ibid., 2.
13. Ibid., tables 3-2B, 3-3B, 3-3C.
14. The 1985 study did not have a race variable. These results apply to men of all races.
15. Nakosteen and Zimmer, 1987.
16. Becker, 1981.
17. Gilder, 1973. Gilder, 1986, is an expanded and revised version.
18. The seminal article arguing for an increase in male productivity is Korenman and Neumark, 1991. Some ingenious evidence supporting this position is Ginther and Zavodny, 2001, who uses shotgun weddings (men marrying pregnant women whom they might not have married absent the pregnancy) as a way of diminishing selection effects. Arguing against the increase in male productivity are Cornwell and Rupert, 1997; Krashinsky, 2004; and Dougherty, 2006.
19. The results are based on a logit analysis regressing a binary variable (in or out of the labor force) on year, marital status (binary), the unemployment rate for white males ages 30–49, categorical variables for education (college degree or more, no more than a high school diploma, and in between), and an interaction term for education and marriage. The fitted values for 1960 and 2010 set the unemployment rate for prime-age white males at the 1960/1968–2010 mean of 4.1 percent.
20. Author’s analysis, IPUMS based on white women ages 18–64 not in school.
21. I set the minimum size for computing a percentage at 100. The CPS data for Belmont had small numbers of single women in their thirties and forties through the mid-1970s. This means that the Belmont percentage for 1969 is based on 1968–70, the percentage for 1972 is based on 1971–72, the percentage for 1974 is based on 1973–74, and the percentage for 1976 is based on 1975–76.
22. From 1960 to 2008, the mean for all employed white women ages 30–49 was 36.5 hours, showing an upward trend from about 35 hours at the end of the 1960s to more than 37 hours for all but one of the years from 1988 through 2008.
23. More formally, the trendline shows the percentage of homes in the CPS in which, for married households, one of the spouses had worked at least forty hours during the week preceding the interview or, for unmarried households, in which the person designated as head of household had worked at least forty hours during the week preceding the interview. The sample is restricted to persons coded as either head of household or spouse of head of household.
10: Honesty
1. High school dropouts with blue-collar occupations ages 30–49 usually live in working-class neighborhoods even if they spent their childhoods in Belmont. The reason for the Belmont-Fishtown classification is to have a way of characterizing the existing population of a neighborhood, not the socioeconomic class of their parents. But crime is exceptional, since so much of it is committed by young men in their teens or twenties, hence I inquire into the probable socioeconomic backgrounds of prison inmates, as described in appendix E. There is no basis for thinking that a substantial number of prisoners who qualify for Fishtown as adults actually were born to middle-class or upper-middle-class parents.
2. The data on inmates’ occupations were sparse. The survey asked several different questions about the job that the probationer had held at the time of the interview or before the arrest, but even combining all of those answers produced occupational data for fewer than half of the respondents. Among those who did give occupations, 79 percent were in occupations that qualified them for Fishtown, while only 7 percent were in occupations that qualified them for Belmont. Among all white males ages 20–49, the comparable proportions were 59 percent and 25 percent.
3. Joe Nocera, “Still Stuck in Denial on Wall St.,” New York Times, October 1, 2010.
4. Declarations of bankruptcy under Chapter 13 include a repayment plan for some or all of the debt.
5. From 1972 through 2005, the figure is based on “nonbusiness” filings. In the SAUS prior to 1981, filings were reported in terms of the occupation of the debtor. Combining all the published data, we have both measures from 1972 to 1980. The total for lines for “employees” and “other, not in business” in the pre-1981 coding were within a few hundred cases of the number for “nonbusiness” during those years, so I used that total as a proxy for nonbusiness cases for 1960–71.
6. Domowitz and Eovaldi, 1993, lists thirteen “prodebtor” provisions of the 1978 act, including, among others, an expansive list of exemptions (property that the bankrupt can keep) and restrictions on the rights of creditors.
7. Domowitz and Eovaldi, 1993, tested multivariate regression models using data from 1961 through 1985 and concluded the effect of the bill in its first years was not significant.
8. Michelle White (White, 1998) describes how this works in terms of two types of people: Type A, who would file for bankruptcy only if misfortune creates unmanageable financial distress, and Type B, who “plan in advance to take advantage of the possibility of bankruptcy in the same way that many households plan in advance to reduce their tax liability” (p. 693). She then works through the financial calculations for bankruptcy laws with different exemption levels, and demonstrates that it is indeed possible under American bankruptcy law to plan for bankruptcy to pay—if you are a person who doesn’t much care if you stiff your creditors.
9. Sullivan, Warren, and Westbrook, 2000.
10. Zhu, 2011.
11: Religiosity
1. Putnam, 2000, 66.
2. Ibid., 67.
3. Leege and Kellstedt, 1993; Verba, Schlozman, and Brady, 1995; McKenzie, 2001.
4. Levin, 1994.
5. Hummer et al., 1999.
6. Idler and Kasl, 1992.
7. Lehrer and Chiswick, 1993.
8. Koenig, McCollough, and Larson, 2001.
9. See, for example, Donahue and Benson, 1995; Muller and Ellison, 2001; Regnerus, 2000.
10. Hutchinson, 1986.
11. Hoge, Johnson, and Luidens, 1994, 1.
12. Ibid., 2–6.
13. Ibid., 1994, 1–4. The percentage of people who said they had attended a worship service in the last seven days had moved within the 47–52 percent range from the mid-1950s through 1963, then dropped to the 40–42 percent range in the first half of the 1970s.
14. Hadaway and Marier, 1998.
15. International Social Survey Program: Religion 2, 1998, cited in Hunsberger and Altemeyer, 2006, 13, table 1.
16. Larson and Witham, 1998.
12: The Real Fishtown
1. Milano, 2008, 76–77. Milano, a lifelong resident of Fishtown, has also written two other histories about Fishtown.
2. Upon reading that twenty people in Fishtown in 1960 were not white, Ken Milano wrote, “Twenty? Wow, sounds like a lot. I’m surprised.” Data for 1960 are based on Philadelphia census tracts 18A and 18B. For subsequent censuses through 2000, they were Philadelphia census tracts 143 and 158. The borders of the census tracts correspond closely with the local definition of the boundary of Fishtown—the Delaware River, Frankford Avenue, and halfway between Norris Street and York Street.
3. Rossi, 1955.
4. Binzen, 1970.
5. Quoted in ibid., 103.
6. Quoted in ibid., 103.
7. Smallacombe’s research also covered Fishtown as locally defined, but her center of activity was in the adjacent area to the north.
8. A literature on white working-class culture exists that I have not tried to review here. Much of it—Binzen’s book on Fishtown is a journalistic example—was written in the 1960s and 1970s, prompted by the white working-class backlash against what was seen as government favoritism toward African Americans at the expense of white working-class Americans. Examples are Sennett and Cobb, 1972, and Rubin, 1976. Much of the literature on the white working class since the 1960s has dwelt on racial issues, but other useful descriptions of white working-class society independently of race are Kornblum, 1974; Hirsch, 1983; Halle, 1984; and MacLeod, 1987. Smallacombe’s work is uniquely valuable for this book partly because it serendipitously uses the Kensington District as its locale (when I chose Fishtown as the name for my fictional working-class community, I had no idea that Smallacombe’s dissertation existed) and partly because it describes life in the late 1990s, after the trends I describe in part 2 had taken hold.
9. Smallacombe, 2002, 206.
10. Ibid., 209.
11. Ibid.
12. Ibid., 208.
13. Ibid., 210.
14. Ibid., 165.
15. Ibid., 220.
16. The census tract data for 1960 show labor force participation for males ages 14 and older, while the census tract data for 2000 use the age range of 16 and older. I chose 20–64 as an age range at which Fishtown males would be expected to be in the labor force both in 1960 and in 2000 (college attendance for Fishtown men ages 20 and over was still very low as of 2000). I applied the national labor force participation rates in 1960 for white males ages 14–18 and those ages 65 and older to the number of actual Fishtown males in those age ranges to obtain the estimate of labor force participation among males ages 20–64. For 2000, I followed a parallel procedure, except that the age range for the younger males was 16–19 instead of 14–19. I double-checked the extremely high 2000 figure (30 percent) by replicating the national statistics limited to white males who came from working-class families. That exercise produced an estimate of 29 percent.
17. Smallacombe, 2002, 194.
18. Ibid.
19. This and following quotations from Ken Milano are taken from conversations and e-mails in the spring of 2011.
20. Smallacombe, 2002, 214.
21. Ibid., 85.
22. Ibid., 166.
23. Ibid., 239
24. Ibid., 238
25. Ibid., 227.
26. Ibid., 227–28.
27. Ibid., 233
28. Ibid., 254.
29. Ibid., 259.
30. Ibid., 271–72.
31. Ibid., 264.
32. Ibid., 148.
33. Ibid., 147.
34. For an account set in Oelwein, Iowa, see Reding, 2009.
13: The Size of the New Lower Class
1. Income is defined as money from private sources in the form of wages, income from a business, dividends, interest, rent, or other income that does not come from government benefits. I am using the poverty threshold for a household consisting of just two adults, with the household head under the age of 65 as of the 2010 CPS, expressed in constant dollars for calendar 1959–2009. I use this procedure instead of the actual thresholds for each year because of changes in the reporting and calculation of the poverty threshold that make the actual thresholds not quite comparable across time. But the differences are minor. For example, the actual threshold for a two-adult household headed by a male (a distinction no longer made in the calculation of poverty thresholds) in 1959 was $1,965. The threshold based on the 2010 CPS is $1,960 (1959 dollars).
2. Six percent overstates the prevalence of minimum-wage jobs, since many of those are jobs such as waitperson or dealer in a casino where tips are the major source of net income. Bureau of Labor Statistics website, http://www.bls.gov/cps/minwage2010.htm.
3. Bureau of Labor Statistics website, http://www.bls.gov/oes/current/oes_nat.htm#00-0000. Occupation 37-1012, converted to 2010 dollars.
4. I also am not saying that authentic stay-at-home full-time fathers are not as honorably occupied as stay-at-home full-time mothers. But there are still too few of them to affect the statistics.
5. The source of income is not given in the 1960 census. The calculation for 1959 thus assumes that none of the income for males ages 30–49 came from government benefits. This is not technically correct, but the sources and amounts of government assistance for males were still extremely rare (mostly veteran’s and disability benefits) and small as of 1959, so the degree of error is unlikely to be more than a percentage point.
6. Ross, Danziger, and Smolensky, 1987. The reductions in poverty in the 1940s and 1950s continued but did not accelerate in the 1960s. See Murray, 1984, for the details about the indictment of the policies of the 1960s.
7. This figure is based on total income including government benefits, because the GSS doesn’t split out that category—one of the technical issues that prevent trustworthy calculation of the nonoverlapping part of the community isolate population.
8. Figure 13.4 extrapolates the trendline for isolates observed in the GSS data from 1974 to 2004 back to 1960. The assumption behind this is based on Robert Putnam’s trendlines for social capital from the 1950s onward in Putnam, 2000, which shows steeply falling organizational membership from the early 1960s onward. Putnam’s findings are discussed in more detail in chapter 14.
1. Quoted in Alfred L. Malabre Jr., Lost Prophets: An Insider’s History of the Modern Economists (Cambridge, MA: Harvard Business School Press, 1994), 220.
14: The Selective Collapse of American Community
1. Banfield, 1958, 85.
2. Tocqueville, 1840, 514.
3. Olasky, 1992, 86. Chapters 5 and 6 of The Tragedy of American Compassion have a wide range of similar data.
4. Pollock, 1923, in Skocpol, 2003, 63–64.
5. Skocpol, 2003, especially chapters 2 and 3.
6. Ibid., 108–9.
7. Ibid., 110–11.
8. When Robert and Helen Lynd conducted their classic study of Muncie, Indiana, in the mid-1920s, they reported the memberships in organizations among their samples of “business-class” and “working class” respondents, effectively representing the white-collar and blue-collar occupations. Fifty-seven percent of working-class men and 36 percent of working-class wives (all of the respondents were married) belonged to at least one organization, numbers that are higher than any observed for Fishtown in the data present, but they pale in comparison to the percentages for the business class: 97 percent among the men and 92 percent among their wives. B. Lynd and H. Lynd, 1929, appendix table 19.
9. Putnam, 2000.
10. Ibid., chapters 2, 3, 6, 7.
11. This actually understates the real decrease in participation, Putnam points out, because voting in the South after 1965 represents many black votes that do not reflect a new propensity to participate in the election, but the ability to do so in the aftermath of the civil rights movement and the Voting Rights Act of 1965. Putnam, 2000, 31–33.
12. Brooks, 2000, 106.
13. See, for example, Nie, 2001; Wellman et al., 2001; Hampton and Wellman, 2003; Bargh, 2004; Williams, 2006.
14. This and the next two examples are ones that have directly involved members of my family, and they are only a few of the many examples our family has experienced. If you reach blindfolded into a bowl of marbles and the first three you pick at random are purple, chances are high that the bowl has a lot of purple marbles. When the examples of social capital via the Internet are so plentiful, both from members of my own family and from the families of friends, it seems extremely likely to me that we are witnessing a transformation of traditional social capital that goes far beyond anything that the scholarly literature has yet documented.
15. Jim Jansen, “Use of the Internet in Higher-Income Households,” November 24, 2010, Pew Research Center’s Internet & American Life Project, www.pewinternet.org.
16. Keith N. Hampton et al., “Social Networking Sites and Our Lives,” June 16, 2011, Pew Research Center’s Internet & American Life Project, www.pewinternet.org.
17. Unlike most graphs using GSS data, this one shows data points for the individual election years, since sample sizes were adequate (in all but the 1968 and 2008 elections, at least two GSS surveys could be combined to produce the samples for Belmont and Fishtown).
18. Approximately 72 percent of all adults of all races without a high school diploma voted in the 1952 and 1956 election samples used for the analyses in The American Voter (Campbell, 1960, table 15-1, p. 252). Interpolating the white voting turnout among those who hadn’t completed high school is imprecise, but it could not have been lower than 75 percent and may have approached 80 percent.
19. Fukuyama, 1995.
20. Putnam, 2007, 134.
21. Putnam used a variety of data sources to obtain comparable measures of pre-1972 trust. For adults as a whole, all races, the proportion of trusting people had dropped from about 53 percent to 49 percent from 1960 to 1972. Putnam, 2007, table 38, p. 140.
22. Ibid., 149–50.
23. Murray, 2003, chapter 19.
24. Field, 2003; Putnam, 2000.
15: The Founding Virtues and the Stuff of Life
1. I cannot retrieve the name of the show or the date of the interview—I saw it many years ago, and Google has been no help—but that sad smile on Geffen’s face made his words stick.
2. The relationships of self-reported happiness to family, vocation, community, and faith in the surveys from 1990 to 1998 and 2000 to 2010 were examined separately to see if they had changed. They had not, except for minor variations. Combining the surveys expands the sample sizes and provides greater stability for the multivariate analyses. For a more extensive discussion of the relationship of the quantitative measure of happiness to work, marriage, religion, income, and a variety of other topics, see Brooks, 2008.
3. For a review of the literature and evidence showing an independent effect of marriage on happiness, see Stutzer, 2006.
4. The happiness item offered four alternatives instead of the three offered by the GSS, and the wording of the alternatives is not quite the same as the wording of the GSS alternatives: “not happy at all,” “not very happy,” “happy,” and “very happy.” Accordingly, the absolute percentages of people who identified themselves as “very happy” in the Social Capital Benchmark Survey cannot be compared to those in the GSS data. To give a sense of the difference in the results: In the GSS data for the 2000s, 32 percent of prime-age whites said they were “very happy” compared to 42 percent in the Social Capital Benchmark Survey. In the GSS data for the 2000s, 9 percent said they were “not too happy” compared to a combined total of 4 percent in the Social Capital Benchmark Survey who said they were “not happy at all” or “not very happy.”
5. The SCBS also had a “protest activities index” that combined measures of nonelectoral forms of political participation—signing petitions, attending political meeting or rallies, membership in political groups or labor unions, and engaging in a demonstration, protest, boycott, or march. But 46 percent of the sample had a score of zero, making it impossible to establish “very low” and “very high” categories that resembled the cutoffs for the other index. For the record, 39 percent of those who had a protest activities index score of zero answered that they were “very happy,” compared to 46 percent of those who got the top scores on that index.
6. All cutoffs were calculated using sample weights applied to the entire SCBS sample (all races, all ages).
7. I should note that the giving and volunteering index includes an indicator based on religion-based volunteering and charity, which might have tapped into the high levels of happiness already discussed among the very religious. You may be wondering whether all these indexes are catching the same people in the “very low” and “very high” categories, which would account for the similarity in results. The answer is no. The correlations among the indexes are moderate, mostly in the +0.3 to +0.5 range, which means that people fall into different parts of the range for different indexes. Take, for example, the pair of indexes with the highest correlation (+0.54) among prime-age whites, the group involvement index and the giving and volunteering index. For the giving and volunteering index, 1,219 people fell into the highest category, but only 413 of them also fell into the highest category on the group involvement index. And those results are for the most highly correlated pair. It seems fair to conclude that different types of community engagement are each (largely) independently associated with higher levels of self-reported happiness.
8. The simple version of a multivariate analysis considers each independent variable separately (an independent variable is a hypothesized cause of the dependent variable). In the type of analysis used here, known as logit, the simple version shows you the size of boost given to the probability of answering “very happy” by every category of every independent variable. Take, for example, the work satisfaction variable, which has four categories: “very dissatisfied,” “a little dissatisfied,” “moderately satisfied,” and “very satisfied.” The analysis gives you a separate number—the size of the boost—for each of the four categories.
The more complicated version of multivariate analysis asks what happens when we consider how permutations interact—a happy marriage with no religion, high social trust with high job satisfaction, and so on. The problem is that the number of permutations grows exponentially with the addition of variables. It is easily possible to calculate analyses that contain every permutation of several variables—the computer doesn’t get tired—but the great majority of the numbers associated with the interaction terms are not only going to be statistically nonsignificant, they are going to be so small that they have no discernible effect on the probability of responding “very happy.” Furthermore, you must remember that the computer is not worried about whether there is a good reason to expect that an interaction effect may exist or whether the sample size for a given permutation is large enough to be interpretable; it just blindly follows its instructions. In doing so, it assigns an “effect” to every interaction no matter what. The program has no capacity for saying, “This is just noise obscuring real relationships,” so the analyst has to make that judgment. In the analysis reported in the text, only marriage and vocation had nontrivial interaction effects. The results are thus based on categorical variables for marriage, work satisfaction, social trust, and strength of religious involvement, and the interactions between marriage and work satisfaction. The equation also includes age as a control variable.
9. See Brooks, 2008, chapter 5, for a recent review of the literature on happiness and income and chapter 6 for a review of the literature on happiness and income inequality.
10. The analysis was conducted using interaction terms of income with the categorical independent variables, but all of the interaction effects were substantively tiny and did not approach statistical significance. The results reported in table 15.2 replicated the one reported in Figure 15.6 with the addition of a continuous variable expressing family income in constant dollars.
11. These figures are fitted to age 40 and $50,500, the rounded median income ($50,499) of the sample used in the multivariate analyses.
17: Alternative Futures
1. Vol. 2 of the Glasgow Edition of the Works and Correspondence of Adam Smith (Indianapolis: Liberty Fund, 1981). Chapter: [IV.vii.c] part third: Of the Advantages which Europe has derived from the Discovery of America, and from that of a Passage to the East Indies by the Cape of Good Hope. Online Library of Liberty.
2. Rifkin, 2004.
3. Zuckerman, 2008; Layard, 2005.
4. Fischer, 2010.
5. Fischer and Hout, 2006.
6. Many libertarians would quarrel with that statement, arguing that the New Deal was the beginning of the end. I agree that it was the thin edge of the wedge that rationalized the later expansions of federal power, but, for practical purposes, the American project was still alive and well until the next inflection point after Kennedy’s assassination.
7. Toynbee’s vogue came about because of D. C. Somervell’s abridgment of volumes 1–6, published in 1946. The six volumes themselves were published from 1934 to 1939. In 1957, Somervell published his abridgment of volumes 7–10, which Toynbee had published between the end of World War II and 1954.
8. Toynbee, 1946, chapter 19.
9. Ibid., 439.
10. Himmelfarb, 1984.
11. I ignore the quasi-aristocratic code that might be said to have existed among the very small northeastern elite in the late nineteenth century.
12. “True Manliness,” in The New McGuffey Fourth Reader, 1901, 42–47. Available online at Google Books.
13. Nonjudgmentalism is even more extreme in western Europe than it is in the United States, but I am not sufficiently familiar with the data from western Europe to be confident that the discrepancy between the behavior and the words of the European upper class are as great as they are in the United States.
14. The Oxford English Dictionary gives a secondary meaning that is now archaic: “uncomely, unhandsome.”
15. Regarding the Spelling mansion: Jeannine Stein, “The House of Spelling: Massive Construction Project in Holmby Hills Flusters Some Neighbors,” Los Angeles Times, April 8, 1988. Regarding Henry McKinnell’s departure from Pfizer: CNBC News, December 22, 2006, “Pfizer’s McKinnell—The $200 Million Man,” CNBC News website; “Golden Parachutes: Bosses Who Walked Away with Large Payouts,” the online version of The Economist, July 27, 2010.
16. Both the Murphy and the Forbes data include bonuses, stock options, and other forms of compensation along with salary. The Forbes chart used for Figure 17.1, from http://www.forbes.com/lists/2011/12/ceo-pay-20-year-historical-chart.html, purports to show pay from 1989 to 2011, but those years refer to the year of publication, not the year of compensation. I have moved all of the numbers back one year, so that, for example, I assign to 2010 the $9.026 million mean that Forbes assigns to 2011.
17. I think an ethical issue does arise for CEOs who want a deal that will pay them handsome separation packages even if they drive the company over a cliff, but that’s peripheral to the discussion here.
18. Bizjak, 2011.
19. In the advanced countries of the West, the private sector accounts for either all the production of wealth or all but a trivial proportion. The taxes paid by government employees in advanced countries amount, with the rarest exceptions, to a partial clawback of their salaries, not a contribution to the financing of the welfare state.
20. The Federalist, no. 51, http://www.foundingfathers.info/federalistpapers/.
21. The slogan was first stated by Marx in Critique of the Gotha Programme, April or early May 1875. Available online at http://www.marxists.org/archive/marx/works/download/Marx_Critque_of_the_Gotha_Programme.pdf.
22. Wilson, 1998.
23. Murray, 2005.
24. Murray, 2006, 21.
25. Fogel, 2000, 17.
26. Ibid., 25.
27. Ibid., 176–77.
Appendix A: Data Sources and Presentation
1. Cleveland, 1979.
2. Fischer and Hout, 2006, 253.
Appendix C: Supplemental Material for the Chapter on Belmont and Fishtown
1. “Blue-collar professions” refer to high-skill blue-collar jobs that lend themselves to self-employment or have a natural career path to supervisory positions on blue-collar work sites.
2. This stipulation means that a person whose spouse has an AA degree or higher is not assigned to Fishtown even if both the husband and wife are in Fishtown occupations.
3. For databases that show a person’s years of completed education rather than highest degree completed, persons who have completed a thirteenth year of schooling are categorized as “no more than a high school diploma.” Persons who have completed fourteen years of schooling are considered equivalent to those who have achieved an associate’s degree.
4. Married couples in which both had a college degree, the head of household was ages 30–49, but neither had an occupation were not assigned to a neighborhood, because of the likelihood of missing data or some other problems with the data (there are few circumstances in which neither person in such a couple would not even have an occupation). In the CPS database, there were only 471 such couples among the 691,942 married white couples in which the head of household was ages 30–49.
5. The rule in the text includes married persons who are living in households in which neither is the head—for example, married children still living with one of the spouse’s parents. People in such circumstances are identified only in the CPS or census data, and constitute a minuscule portion of the population.
6. Age 21 is chosen for the cutoff because of its traditional standing as the age of majority.
7. Nakao and Treas, 1994.
8. Hauser and Warren, 1997.
9. Unpublished data provided courtesy of Earl Hunt.
10. In practice, using one of the prestige scales would have produced about the same results as using the measure of cognitive demands. The correlation of the g-loadings with the Nakao and Treas occupational prestige index was +0.74 and the correlation with the Hauser and Warren index was +0.76. Based on the 1990 occupational coding used by the Census Bureau.
11. For a discussion of the literature on job productivity and cognitive ability as of the early 1990s, see Herrnstein and Murray, 1994, chapter 3. For an update on the literature as of the early 2000s, see Gottfredson, 2003.
12. The bottom-ranked occupations were loggers, graders and sorters of agricultural products, operators of construction equipment, miners, stevedores and other materials movers, stock handlers, packers and wrappers, packagers, and equipment cleaners.
Appendix D: Supplemental Material for the Marriage Chapter
1. Bramlett and Mosher, 2002, table 35.
2. These results are produced by logit analyses in which a binary variable (whether the child experienced a divorce by the time his mother was age 40) was regressed on neighborhood in the first model and on neighborhood, mother’s age at marriage, and mother’s age at giving birth in the second model.
Appendix E: Supplemental Material for the Honesty Chapter
1. The results for the original cohort born in 1945 were published in Wolfgang, Figlio, and Sellin, 1972. The study was replicated with the 1958 Philadelphia birth cohort; see Tracy, 1990.
2. These results are produced by regressing the number of prior arrests on age, years of education, occupational class, and the interaction of education and occupational class for white male prisoners ages 20–49. Age was fitted to 30 for both neighborhoods. Educational values were fitted based on the average education attainment among prisoners from the two neighborhoods—seventeen and ten years for Belmont and Fishtown, respectively. The fitted numbers of prior arrests resulting from this procedure were 2.46 and 5.92 for Belmont and Fishtown, respectively.
3. Levitt, 1996; Dilulio and Piehl, 1991; Dilulio and Piehl, 1995.
4. Dilulio, 1991.
5. Greenwood and Abrahamse, 1982, xiii.
6. In this statistic and the others that follow, I do not use UCR arrest data from 1974 to 1980. During those years, a significant number of law enforcement agencies reported results that did not cover the full twelve months, and those were included in the published UCR volumes.
7. Bloomberg LP.
8. For the record, the incidence of SEC Accounting and Auditing Enforcement Releases rose from 1983 to 2003 and subsequently declined.
9. For the record, the incidence of IRS fraud cases declined from the first data in 1978 through 2009.