I think statistics should help make sense of data, and that it shouldn't be presented as too difficult to understand. Frank Sulloway's 1996 book "BORN TO REBEL: Birth Order, Family Dynamics and Creative Lives" is deep enough to make a semester class, with extraordinary statistical explanations as clear as glass. ($30, Pantheon Books) Michael Shermer describes it as "the most rigorous scientific work of history ever written." Yet he makes the statistics natural and easy to understand.

Sulloway received an award for extraordinary achievement from the American Academy of Achievement for the book. Previous honorees have been E. O. Wilson, Francis Crick, Stephen Jay Gould, Murray Gellman, and Jane Goodall.

Sulloway reanalyzes old birth order information, and takes a fresh look at why siblings have so many differences in personality and outlook. Their very similar genetic and environmental heritage would predict similar personality outcomes. Yet most studies find similarities in firstborns, only children, middle children and last borns.

Rebels turn out to be laterborn, 3.1 times as likely to accept revolutionary ideas and 4.7 times as likely to support radical revolutions. The process of proving that took quite a few steps, with hypotheses and statistics at each step. Sulloway eventually considered many developmental steps, and tried to develop factors for the strength of each. The death of an older sibling could move a child up in birth order behavior if it happened early. Parental conflict usually contributed to rebelliousness for any birth position.

So WHY are firstborns likely to be more conservative and influenced by authority? WHY are laterborn children likely to grow up more liberal and receptive to ideological change? Sure, firstborns usually get more attention, have greater responsibilities and parental expectations. But Sulloway develops a Darwinian sibling-competition model to explain his findings. He sees children competing for limited parental favors and resources. Firstborns have the opportunity to nail down an ecological niche, forcing laterborn siblings to diversify into new areas.

Reviewing history, Sulloway found evolution supported by younger siblings immediately, and older siblings a generation later. He found no correlation of belief with social class. Copernicus was believed by only 1 out of 4 firstborn contemporaries, but 3 out of four laterborn contemporaries. Republicans tend to nominate firstborn justices, while democrats tend to nominate laterborn justices. 32 political revolutions, the Protestant Reformation, and 61 reform movements in American history show the same trends. (Remember, we're talking correlation, not absolute predestination!) Even animal studies showed similar sibling effects.

Creativity in science leads to new theories and progress. Who excels? In dozens of scientific revolutions most of the leaders were laterborn or had significant parental conflict. Firstborns were studying vitalism, spiritism and eugenics while Darwin, Lavoisier, Lister, Harvey, Newton, and others led the way to new science. Sulloway believes that even though firstborn children score higher on IQ tests, laterborn children are more likely to break with established traditions and show creative ability in a new way.

"Tomboys" and "sissies," dreaded by conformist parents, are reported from another study: both are possible with a firstborn girl and second born boy. Ratings were based on conventional behaviors. Younger sisters of sisters were rated most feminine, and older brothers of brothers were rated most masculine.

Firstborns tend to be dominant, aggressive, ambitious, jealous, and conservative. Laterborns often show rebellious streaks, but usually use sociable and cooperative strategies. Siblings go out of their way to be different from each other, "exploiting unoccupied niches" to gain more parental attention. Laterborns take more risks, firstborns are more conscientious, etc. Data like this has been reported for eons. Sulloway utilized this huge data resource along with multiple regression correlation to test various hypotheses. He included 5 pages of appendix to explain correlation, including:

Medical Statistics

A correlation of r=0.05 explains <1% of variance, but that level of benefit with a new medication improves your odds of survival by 11%. Even low correlations are significant!

r      % Survive-no med     % Odds improve
0.01           49.5                1
0.10           45                 22
0.30           35                 86
0.50           25                300
0.80           10                900