This final appendix outlines a formula that can be used to predict any individual's likelihood of supporting radical innovations. The formula builds on the base-rate probabilities given in Table 10. These predicted probabilities are derived from a logistic regression model for the 28 scientific debates in my study. The table specifies the likelihood that an individual will support a radical revolution such as Copernican theory or Darwinism. These probabilities are stratified by age, social attitudes, and birth order. [1]
To determine an individual's base-rate probability, find the applicable category in Table 10. For example, a firstborn who is over the age of sixty and socially conservative has a 3 percent likelihood of endorsing a radical innovation such as Darwinism. A lastborn who is under the age of thirty and who is socially liberal, has a likelihood of acceptance of 96 percent.
These base-rate probabilities may be adjusted to take into account seven additional biographical influences. To the appropriate predicted probability in Table 10, add or subtract the following percentages, depending on an individual's attributes:
1. Pronounced parent-offspring conflict. In the case of firstborns who have experienced pronounced conflict with a parent, add 30 percent to the relevant base-rate probability in Table 10. "Pronounced parent-offspring 'conflict" may be defined as more conflict than is found in four-fifths of the general population. For laterborns who have experienced pronounced parent-offspring conflict, add 10 percent to appropriate base-rate probability in Table 10. For cases in which parent-offspring conflict (or deidentification) is less extreme, but exceeds that observed in two-thirds of the general population, these two adjustments may be halved.
2. Pronounced
shyness. For firstborns who are shyer than four-fifths of the general
population, add 20 percent to the relevant probability in Table 10. For shy lastborns, subtract 15 percent from
the appropriate probability in Table 10. Make no adjustment for middle
children.
3. Age gaps between adjacent siblings. For firstborns whose age gap with their closest sibling is less than 2.0 years or more than 5.0 years, add 5 percent to the base rate probability in Table 10. Only children may consider themselves to be large-age-gap firstborns. For laterborns who are less than 2.0 years younger than their next older sibling, or more than 5.0 years younger, subtract 5 percent from the base-rate probability.
4. Parental loss and surrogate parenting by older siblings. If a firstborn has lost a parent before the age of ten and has acted as a surrogate parent toward younger siblings, subtract 15 percent from the base-rate probability in Table 10. If a laterborn has lost a parent before the age of ten and has experienced surrogate parenting by older siblings, add 10 percent to the base-rate probability in Table 10.
5. Gender. For females born prior to 1900, add 10 percent to the base-rate probability in Table 10. Recent studies do not show consistent gender differences on openness to ideas. Still, women are more politically liberal than men, and a modest upward adjustment— perhaps 5 percent— seems warranted for women in contemporary populations. [2]
6. Race. For individuals from minority races— especially those subject to discrimination (for example, blacks and Jews)— add 10 percent to the base-rate probability in Table 10.
7. Friendship. If an individual is a particularly good friend with the leader of a radical revolution, add 10 percent to the predicted probability in Table 10.
This ten-variable formula for openness to radical innovation mimics the results of a multivariate logistic regression model. One should use common sense in applying this model within any particular behavioral context. For example, during pre-Darwinian debates over evolution, less than a third of the scientific community endorsed this controversial idea. For the scientific population as a whole, this mean rate of acceptance for pre-Darwinian theories of evolution was 30 percent not 50 percent, as Table 10 assumes.
Applied to the participants of a radical controversy, the formula presented in this appendix will do a good job of rank ordering participants according to their likelihood of acceptance. On average, the base-rate formula in Table 10, Section A, should be 75 percent correct in classifying individuals as supporters or opponents of innovation, as long as the innovation is truly a Radical Ideological Revolution.[3] After making appropriate adjustments to this base-rate prediction in sections B— F, the formula's accuracy should approach 85 percent. Fudging on the details will decrease the formula's accuracy.
[1] The predicted probabilities listed in Table 10 are based on a logistic regression model that includes complete biographical data on 1,436 participants in 28 scientific controversies (1543-1967). The model includes four significant main effects: age, social attitudes, relative birth rank, and ideological implications of the innovation. The model also includes three significant two-way interaction effects: ideological implications by age; ideological' implications by social attitudes; and ideological implications by relative birth rank. All main effects are statistically significant at p<.0001, and the interaction effects are statistically significant at p<.005. The two-way interaction effects reflect the fact that age, social attitudes, and relative birth rank are all significantly better predictors of receptivity in proportion to the radicalism of the innovation.
Predicted probabilities in Table 10 are listed only for those scientific controversies that involved radical ideological implications. For this class of controversies, the model's predictors yield a multiple correlation with observed radicalism of .50 (df=l/582, t=13.46, p<\ in a billion; 75.2 percent correct classifications, based on the 2 X 2 table of observed versus predicted outcomes). For participants in Technical Revolutions and Controversial Innovations, the multiple correlation is .31 (rf/=l/482, f=6.38, p<\ in a billion; 69.0 percent correct classifications). For scientists who debated Conservative Theories, the multiple correlation is: .10 dy=l/366, f=1.88, p<.06-, 59.2 percent correct classifications). The chance rate for correct classifications is 50.4 percent.
[2] On studies of gender and openness to ideas, see Feingold 1994.
[3] For Technical Revolutions and Controversial Innovations, approximate probabilities derived from the model may be calculated as follows. For conservative firstborns, multiply the base-rate probability in Table 10 by 5.0. For socially moderate firstborns, multiply the base-rate probability by 1.8. For radical firstborns, multiply the base-rate probability by 1.2. For middleborns and lastborns who are social conservatives, multiply the base-rate probability by 1.8. For middleborns and lastborns who are social moderates or liberals, multiply the base-rate probability by 1.1. In debates over Conservative Theories, firstborns and social conservatives are more likely than their counterparts to endorse new ideas. Although young people are more likely to endorse Conservative Theories, the role of age is significantly less predictive than it is during liberal innovations.
If you, or someone else, is a clear-cut exception to the formula's prediction, ask yourself why. The most likely causes of exceptions will involve some of the influences, especially situational factors, that are discussed throughout this book. No formula can do complete justice to all of the contingencies that affect human behavior, although formulas can include appropriate adjustments for some of these contingencies.
Source: Born to Rebel by Frank Sulloway