THRIVING after the FALL
Book Two PAIRING after the FALL

PAIRING · Part I · Overlapping Bells

The numbers and sources

Chapter 5: Two Shelves and a Trap

The narrated edition tells this chapter with a handful of figures. Here is everything behind it: each figure with its study, its research document and its row in the log, the case for and the case against, where the sources disagree, and what is still open.

  • 59 open
  • 23 to check first

Read about the chapter · Its 59 rows in the log

Numbers to carry

The figures the chapter is built on, each with its research document and its row in the log.

  • Extra-pair paternity in the Low Countries over five centuries: about 1 to 2 per cent per generation, 0.4 in surveilled farming villages, 5.9 in the poorest industrial city districts (Larmuseau and colleagues, 2016 and 2019, per SY-02). Rows C05-10, C05-11, C05-12, C05-13.
  • Power posing: no effect on testosterone, cortisol or risk in a pre-registered replication of 200 people (Ranehill and colleagues, 2015, per SY-04). Row C05-20.
  • Stereotype threat: 55 per cent of replication designs reproduced it, half of those confounded, 30 per cent of clean designs worked (Stoet and Geary, 2012, per SY-01). Row C05-31.
  • The STEM paradox: 472,242 adolescents in 67 countries, under 25 per cent women among STEM graduates in Finland, Norway and Sweden against 35 to 40 in Algeria, Tunisia and Turkey (Stoet and Geary, Psychological Science, 2018, corrected after Richardson 2020, per HC-07). Another report gives over 300,000 in 73 countries. Rows C05-37, C05-38, C05-40.
  • No within-country divergence in values as equality rose: 235,339 people, 32 countries, eight survey rounds (Fors Connolly, Goossen and Hjerm, Sex Roles, 2020, per HC-07). Row C05-42.

Every figure

One row for each figure or finding in the data edition's chapter, whether or not the narrated edition kept it. Layer says whether it belongs to physiology, behaviour or culture. Status is how the finding stands in the research: robust, contested, open or fallen. It is not yet the log's verdict.

FigureWhat it measuresStudy (authors, year, venue, N)DocumentRowLayerStatus
Testosterone Rex, 2017The constructivist end of the shelf: one molecule running the male half of the species is a mythFine, C. (2017) Testosterone RexSY-03, concept note §4C05-01cultureopen (year unverified)
T, 2021The evolutionary end of the shelf: the molecule runs a great dealHooven, C. (2021) T: The Story of TestosteroneSY-03, concept note §4C05-02cultureopen (year unverified)
Each author has read the other's book and was not persuadedNarrator's inferenceNo primary sourceSY-03C05-03cultureopen
1739Hume notices the unannounced step from is to oughtHume, D. (1739) A Treatise of Human Nature, Book IIISY-02C05-04culturerobust
1903Moore names the naturalistic fallacyMoore, G. E. (1903) Principia EthicaSY-02C05-05culturerobust (Moore's own definition differs from the is-to-ought gloss, see Open questions)
1978, NatureDavis names the moralistic fallacyDavis, B. D. (1978) "The moralistic fallacy", Nature 272:390SY-02C05-06culturerobust (wording to confirm)
350 hours a yearThe German border's effect on mothers' paid hours, East against WestBoelmann, Raute and Schönberg, forward reference to the Preface and chapter 13Preface sourcesC05-07culturerobust
Rousseau's noble savage, Hobbes's nasty savageThe two shelves' creation myths, neither matches the archaeologyNo primary named, SY-02's summary of Fry and Söderberg, Bowles, NatarukSY-02C05-08cultureopen
Dual mating strategyEvolutionary psychology's hypothesis: provisioning from a reliable partner, better genes from more formidable menNo primary named (candidates: Gangestad and Thornhill)SY-02C05-09behaviourfallen on the paternity data
500 years of records, 2016 and 2019Y-chromosome analyses matched to genealogical records in the Low CountriesLarmuseau et al. (2016) Trends in Ecology & Evolution, Larmuseau et al. (2019) Current BiologySY-02C05-10behaviourrobust
1 to 2 per cent per generationExtra-pair paternity over five centuriesLarmuseau et al. 2019SY-02C05-11behaviourrobust
0.4 per centExtra-pair paternity in tightly knit farming villagesLarmuseau et al. 2019SY-02C05-12behaviourrobust
5.9 per centExtra-pair paternity among the poorest classes of industrial citiesLarmuseau et al. 2019SY-02C05-13behaviourrobust
"Given what we are, arrangement X under condition Y tends to produce Z"The conditional sentenceSY-02's own formulation, also in the PrefaceSY-02C05-14culturerobust (method, not finding)
Cooperative breedingHuman infants too costly for two parents alone, grandmothers and kin do much of the holdingHrdy, S. B. (2009) Mothers and OthersSY-02C05-15behaviourrobust
Threshold, not manufactureTestosterone lowers the threshold for whatever the culture rewards with statusSapolsky, R. (2017) Behave, ch. 4SY-02C05-16physiologyrobust
"Biology dictates the cost of a social arrangement, not its moral worth"The price-list principleSY-02's own phraseSY-02C05-17culturerobust (method)
The first yearThe father's body is still adjusting during the first year, forward reference to chapter 8Gettler 2011 and successorschapter 8 sourcesC05-18physiologyrobust
2010, Psychological ScienceTwo minutes of expansive posture raised testosterone, lowered cortisol, raised financial risk-takingCarney, D. R., Cuddy, A. J. C., Yap, A. J. (2010) Psychological Science 21(10):1363 to 1368SY-04C05-19physiologyfallen
200 participants, 2015, pre-registeredNo effect on testosterone, cortisol or risk, felt-power effect survivedRanehill, E. et al. (2015) Psychological Science 26(5):653 to 656SY-04 (N read from the PDF page image)C05-20physiologyrobust
Original sample "much smaller" than 200Carney 2010 sample sizeNot in the librarySY-04C05-21physiologyopen
Public disavowalCarney, lead author, disowned the findingCarney's faculty web page statement, 2016, date and wording to confirmSY-04C05-22cultureopen (undated in the library)
184 studies, 1.6 million pupils, 21 countries, 2014, g = 0.10 (girls), 0.06 (boys, mathematics)Single-sex schooling advantage after allowing for selectionPahlke, E., Hyde, J. S., Allison, C. M. (2014) Psychological Bulletin 140(4):1042 to 1072SY-04 (g values read from the PDF page image)C05-23culturefallen (as an advantage)
d about 2, women 60 to 75 per cent of men'sUpper-body strength across occupational cohortsNo single primary named, candidates Courtright et al. 2013, Leyk et al. 2007SY-04 (SY-02 gives d about 2.5)C05-24physiologyrobust
d = 0.93, 503,188 respondents, 47 inventories, 2009, engineering d = 1.11Things-versus-people interest gapSu, R., Rounds, J., Armstrong, P. I. (2009) Psychological Bulletin 135(6):859 to 884HC-07, SY-04C05-25behaviourrobust
394 men, five years, 2011Fatherhood lowers testosterone, replicated sinceGettler, L. T. et al. (2011) PNAS 108(39):16194 to 16199 (full cohort N = 624, 394 may be a subsample)SY-04C05-26physiologyrobust
78 per cent small (d < 0.35), 2005, multivariate D = 2.71, about 10 per cent overlap, 10,261 people, 2012Hyde's similarities hypothesis against Del Giudice's whole-profile distance, both reproduce, the choice of metric called ideologicalHyde, J. S. (2005) American Psychologist 60(6):581 to 592, Del Giudice, M., Booth, T., Irwing, P. (2012) PLoS ONE 7(1):e29265SY-01, SY-04C05-27behaviourcontested (metric)
More than 500,000 people, 2018, PNASWomen score higher on the Reading the Mind in the Eyes test across ages and countriesGreenberg, D. M. et al. (2018) PNAS 115(48):12152 to 12157SY-04 (effect size stripped, not recoverable)C05-28behaviourrobust as a pattern, mixed as a mechanism
Paid accuracy closes the gap, 2001Empathic accuracy gap disappears when men are paid to be accurateKlein, K. J. K., Hodges, S. D. (2001) Personality and Social Psychology Bulletin 27(6):720 to 730SY-04C05-29behaviourrobust
1999, Journal of Experimental Social PsychologyStereotype threat: able women underperform on a hard maths test when told it shows sex differences, match men when told it does notSpencer, S. J., Steele, C. M., Quinn, D. M. (1999) JESP 35(1):4 to 28SY-01C05-30culturecontested
55 per cent of designs reproduced it, half of those confounded, 30 per cent of clean replications worked, 2012Stereotype-threat replication recordStoet, G., Geary, D. C. (2012) Review of General Psychology 16(1):93 to 102SY-01C05-31culturecontested (what 30 per cent means)
10,888 participants, 2015, Psychological BulletinMen obtain better economic outcomes in negotiation, gap disappears or reverses when women negotiate for others or the range is transparentMazei, J. et al. (2015) Psychological Bulletin 141(1):85 to 104SY-01C05-32behaviourrobust and movable
23,031 people, 26 cultures, 2001Women higher in neuroticism and agreeableness, gaps widest where sex roles are loosestCosta, P. T., Terracciano, A., McCrae, R. R. (2001) JPSP 81(2):322 to 331HC-07C05-33behaviourrobust (pattern)
17,637 people, 55 nations, 2008Personality gap predicted by a nation's wealth, health and educationSchmitt, D. P., Realo, A., Voracek, M., Allik, J. (2008) JPSP 94(1):168 to 182SY-01, HC-07C05-34behaviourrobust (pattern)
347,192 people, 48 countries, emotionality d = 0.84, 2020Sex gap in HEXACO emotionalityLee, K., Ashton, M. C. (2020) Journal of Personality 88(6):1075 to 1090HC-07C05-35behaviourrobust
80,000 people, 76 countries, r = 0.67 with log GDP per head, 2018, ScienceGaps in risk, patience, altruism and trust near zero in poor countries, largest in rich onesFalk, A., Hermle, J. (2018) Science 362(6412):eaas9899HC-07C05-36behaviourrobust (pattern), contested (cause)
472,242 fifteen-year-olds, 67 countries, 2018, girls matched or beat boys in science in about two thirds of countriesPISA science scores by sexStoet, G., Geary, D. C. (2018) Psychological Science 29(4):581 to 593HC-07, SY-03 (SY-05 gives over 300,000 in 73 countries)C05-37behaviourcontested
Under 25 per cent women among STEM graduates in Finland, Norway, Sweden, 35 to 40 per cent in Algeria, Tunisia, TurkeyThe STEM graduate paradoxStoet and Geary 2018, Figure 3 and supplementary data (UNESCO graduate shares)HC-07C05-38culturecontested
Global Gender Gap IndexThe equality index usedWEF GGGI, per HC-07HC-07C05-39cultureopen (confirm)
2020, same journal, correction issuedThe STEM result rested on an undisclosed "propensity" measure, with the plain graduate share the correlation shrinks or vanishesRichardson, S. S. et al. (2020) Psychological Science 31(3):338 to 341, Stoet and Geary (2020) corrigendumHC-07, SY-01C05-40culturerobust (the correction is on record)
300,000 pupils, 64 countries, 2020, PNASThe "maths is not for girls" stereotype is stronger in richer, more equal countries and explains away the wealth and STEM-gap linkBreda, T., Jouini, E., Napp, C., Thebault, G. (2020) PNAS 117(49):31063 to 31069SY-01, HC-07 (N read from the SY-04 PDF page image)C05-41culturecontested
235,339 people, 32 countries, eight rounds, 2020, Sex RolesCross-section reproduced, no within-country widening of the values gap as equality roseFors Connolly, F., Goossen, M., Hjerm, M. (2020) Sex Roles 83:101 to 113HC-07C05-42culturerobust (HC-07 says "if anything, convergence")
27 large studies re-analysed (54 articles reviewed), 2024The paradox holds for personality and reverses for mathematics and partner preferencesHerlitz, A. et al. (2024), journal not given in HC-07HC-07C05-43behaviouropen (venue missing)
One report switched sidesSY-01, briefed to steelman the biological thesis, argued the constructivist caseSY-01 text and the prompt it was written fromSY-01C05-45culturerobust (method)
78 against 85.5 per centLibrary discrepancy: Hyde's share of small differences against Zell'sHyde 2005, Zell, Krizan, Teeter (2015) American Psychologist 70(1):10 to 20SY-01, SY-02, SY-04 against the SY distilled sheetC05-46behaviouropen (two different studies, not one figure)
About 10 per cent larger male brain, sex explains about 1 per cent of remaining variance, 2021Three decades of MRI, pooledEliot, L. et al. (2021) Neuroscience & Biobehavioral Reviews 125:667 to 697SY-01C05-47physiologyrobust
More than 300 studies reviewed, 2010Prenatal-testosterone brain-organisation theory: dose-response rule fails, CAH girls confounded by surgery and medical stressJordan-Young, R. (2010) Brain StormSY-01C05-48physiologycontested
2011, PsychoneuroendocrinologyTestosterone as a dependent variable: infant care lowers it in men, competition and authority raise it in both sexesvan Anders, S. M. et al. (2011) PsychoneuroendocrinologySY-01C05-49physiologycontested (cause or consequence)
More than twelve million people, mean d = 0.21, about 85 per cent overlap, random woman outscores random man 42 times in 100, 2015Average psychological sex difference across 106 meta-analysesZell, E., Krizan, Z., Teeter, S. R. (2015) American Psychologist 70(1):10 to 20SY-01C05-50behaviourrobust
d = 0.05, 1,286,350 people, 2010Mathematics gapLindberg, S. M., Hyde, J. S., Petersen, J. L., Linn, M. C. (2010) Psychological Bulletin 136(6):1123 to 1135SY-01C05-51behaviourrobust
d = 0.11 favouring women, 1988Verbal ability gapHyde, J. S., Linn, M. C. (1988) Psychological Bulletin 104(1):53 to 69SY-01C05-52behaviourrobust
71 per cent, 2023Share of meta-analytic sex differences that mirror experimental manipulations of powerEagly, A. H., Wood, W. (2023), journal not given in SY-01SY-01C05-53culturecontested
First fifteen months of lifeCaregivers embed sex roles before the child can sit upFausto-Sterling, A. et al., specific paper not givenSY-01C05-54cultureopen (paper missing)
2005, closing sectionInflated claims of difference carry a social costHyde 2005 American PsychologistSY-01C05-55culturerobust (argument)
14 of 16 domains, 1999Men take more riskByrnes, J. P., Miller, D. C., Schafer, W. D. (1999) Psychological Bulletin 125(3):367 to 383SY-01C05-56behaviourrobust
d = 0.2 means about 85 per cent shared area, d = 2 means two different instrumentsOverlap arithmeticStandard overlap arithmetic, Zell 2015 for the 85 per centSY-01C05-57behaviourrobust
Fell when hormones were measured rather than inferred from calendar daysThe ovulatory-shift literaturePreface's sourcesPrefaceC05-58physiologyfallen

The case for and the case against

The case the chapter makes

  • Both shelves sell a verdict and the evidence supports neither (the chapter's own argument, from SY-02, SY-04 and SY-01 read together).
  • The naturalistic fallacy (Hume 1739, Moore 1903) and the moralistic fallacy (Davis 1978) are two jaws of one trap, each shelf uses both (SY-02, rows C05-04 to C05-06).
  • A shelf-one hypothesis, the dual mating strategy, falls to the paternity record: 1 to 2 per cent per generation, 0.4 in surveilled villages, 5.9 in anonymous cities (Larmuseau 2016, 2019, rows C05-10 to C05-13).
  • The conditional sentence relocates the argument from verdicts to costs, biology dictates the cost of an arrangement, not its moral worth (SY-02, rows C05-14, C05-17, Hrdy 2009, row C05-15).
  • Findings fall in families: small, unregistered, hormone-measuring studies (Carney 2010 against Ranehill 2015, rows C05-19 to C05-22) and findings that never asked who was in the room (Pahlke 2014, row C05-23). Survivors are large, multi-country, big to begin with and physiology-backed (rows C05-24 to C05-26).
  • Robust and movable are not opposites: stereotype threat (Spencer 1999, Stoet and Geary 2012, rows C05-30, C05-31) and negotiation (Mazei 2015, row C05-32).
  • The gender-equality paradox is a robust pattern with an unknown mechanism: pattern from Costa 2001, Schmitt 2008, Lee and Ashton 2020, Falk and Hermle 2018, Stoet and Geary 2018 (rows C05-33 to C05-38), challenge from Richardson 2020, Breda 2020, Fors Connolly 2020, Herlitz 2024 (rows C05-40 to C05-43).
  • The useful question is which layer, and what it would cost to change (the chapter's own formulation of the three-layer method).

The case against

  • Sex explains about 1 per cent of brain structure once head size is allowed for (Eliot 2021, row C05-47).
  • The prenatal-organisation theory fails its dose-response test and its flagship natural experiment is confounded (Jordan-Young 2010, row C05-48).
  • Adult testosterone is a dependent variable, so hormone levels in a stratified society are a product of the stratification (van Anders 2011, row C05-49).
  • Most psychological sex differences are small: Hyde 2005 (78 per cent), Zell 2015 (d = 0.21, 85 per cent overlap, 42 in 100), Lindberg 2010 (maths 0.05), Hyde and Linn 1988 (verbal 0.11) (rows C05-27, C05-50 to C05-52).
  • 71 per cent of meta-analytic sex differences mirror what experimental manipulations of power produce (Eagly and Wood 2023, row C05-53), roles are embedded by caregivers in the first fifteen months (Fausto-Sterling, row C05-54).
  • Inflated claims of difference carry a social cost because they become expectations, so a "price list" is the naturalistic fallacy with better manners (Hyde 2005, row C05-55).
  • The narrator's residue: Del Giudice's 2.71, Byrnes's 14 of 16 risk domains, the 30 per cent of clean stereotype-threat replications, and the wall not moving the fathers (rows C05-27, C05-56, C05-31, Preface). A standoff, not a win.

Where the sources disagree

The data edition's Contested list.

  • Whether the multivariate distance is a finding or an artefact: Del Giudice 2012 (D = 2.71, about 10 per cent overlap) against Hyde 2005 (78 per cent small). Both reproduce, SY-04 calls the choice of metric ideological. Row C05-27.
  • Whether the paradox reflects freedom or stereotype: Schmitt 2008, Falk and Hermle 2018, Stoet and Geary 2018 against Breda 2020, Richardson 2020, Fors Connolly 2020, Herlitz 2024. Rows C05-33 to C05-43.
  • Whether adult hormone differences are cause or consequence: van Anders 2011. Row C05-49.
  • Whether 30 per cent of clean stereotype-threat replications is "real" or "failed", which depends on the shelf. Row C05-31.

Report-to-report discrepancies from the data edition's working notes.

  • Stoet and Geary 2018 sample: 472,242 adolescents in 67 countries (HC-07, SY-03) against over 300,000 in 73 countries (SY-05, per the SY distilled sheet). The data edition carried both, the narrated edition carries neither, saying "fifteen-year-olds in dozens of countries". Row C05-37.
  • Breda et al. 2020 (300,000 pupils, 64 countries) and Breda and Napp 2019 (300,000 students, 64 countries) are given identical samples in HC-07. Possibly the same PISA wave, possibly conflated. Row C05-41.
  • Hyde 2005: 78 per cent "small or close to zero" (SY-01, SY-02, SY-04) against Zell 2015: 85.5 per cent small (SY distilled sheet). Two different studies, the data edition carried 78 for Hyde only, the narrated edition carries neither figure. Row C05-46.
  • Grip strength: d about 2.5 (SY-02) against d greater than 2.0 (SY-04). The data edition said "two standard deviations" and "60 to 75 per cent" from SY-04, the narrated edition refers to the Preface's "different piece of equipment" without a figure. Row C05-24.
  • Mental rotation d (0.57 to 1.0 across reports) and Del Giudice 2016 classification accuracy (up to 77 against 69 to 77): not used in either edition.
  • Fors Connolly 2020: HC-07 says "if anything, convergence", the data edition's "no divergence" and the narrated edition's "the within-country answer was no" should not overstate it. Row C05-42.

Figures stripped in extraction and recovered from the SY-04 PDF page image (page 7): Ranehill N = 200, Pahlke g = 0.10 (girls) and 0.06 (boys, mathematics), Breda N = 300,000. Recovered by rendering the page as an image and reading it, not by guessing. Row C05-59.

Figures stripped and not recoverable from the PDFs (none used in either edition): Walter et al. 2020 sample size, Manson et al. 2022 samples, UK Biobank imaging N and median effect, RMET effect size (Greenberg 2018), Carney et al. 2010 original N (not in the library at all), Su et al. 2009 Realistic, Investigative, Social and Artistic d values (only Things against People 0.93 and engineering 1.11 are given, in HC-07).

Open questions

The data edition's marks for checking (5), removed from the narrated prose and carried here.

  1. Larmuseau: first name and affiliation (SY-02 gives neither). Row C05-10.
  2. Carney, Cuddy and Yap: first names (SY-04 gives surnames only). Row C05-19. The narrated edition names none of them.
  3. Carney, Cuddy and Yap 2010: original sample size (not in the library, SY-04 says only "much larger" of Ranehill). Row C05-21.
  4. Greenberg, Klein and Hodges: first names (not given). Rows C05-28, C05-29. The narrated edition names none of them.
  5. Costa, Terracciano, McCrae, Lee, Ashton, Falk, Hermle, Stoet, Geary: first names (not given). Rows C05-33 to C05-37. Also uncompleted, without inline markers in the data edition: Spencer, Steele, Quinn, Mazei, Richardson, Jouini, Napp, Thebault, Fors Connolly, Goossen, Hjerm, Herlitz, Ranehill, Lindberg, Linn, Krizan, Teeter. The narrated edition names Stoet, Geary, Richardson, Fors Connolly, Goossen, Hjerm and Herlitz by surname only, as the data edition does, no first name has been added from outside the library.

The data edition's open questions from its working notes.

  • Whether Carney's disavowal should be dated and placed (SY-04 gives no date or venue). Row C05-22.
  • Whether chapter 13's "350 hours" belongs this early, or should soften to "a whole population of mothers". The narrated edition keeps the figure, as the Preface carries it. Row C05-07.
  • Whether the three closing lists counted towards the data edition's 4,000-word target (moot in the narrated edition, the lists are here).

Other open points from the verification rows.

  • Moore's own definition of the naturalistic fallacy differs from the is-to-ought reading the chapter gives, check that the gloss is fair. Row C05-05.
  • Davis 1978: confirm volume, page and the exact wording of his definition. Row C05-06.
  • Whether the "dual mating strategy" is characterised fairly (primary not named in SY-02). Row C05-09.
  • Whether the 2016 Larmuseau paper is a review rather than the primary data paper. Row C05-10.
  • Whether the "expect the bill" reading is Hrdy's or the narrator's. Row C05-15.
  • Gettler 2011: whether 394 is a subsample of the 624-man cohort. Row C05-26.
  • Pahlke 2014: which outcomes the g = 0.10 and 0.06 refer to. Row C05-23.
  • Stoet and Geary 2012: confirm the three percentages and the "adjusted for prior scores" characterisation. Row C05-31.
  • Stoet and Geary 2018: confirm the country graduate shares and the year of the graduate data, confirm the index is the WEF Global Gender Gap Index. Rows C05-38, C05-39.
  • Herlitz 2024 and Eagly and Wood 2023: journals not given in the library. Rows C05-43, C05-53.
  • Fausto-Sterling: specific paper not given. Row C05-54.
  • Larmuseau 2019: confirm the 1 to 2 per cent range and the "per generation" framing, the 0.4 and the "surveillance" reading, the 5.9 and its class definition and period. Rows C05-11 to C05-13.

The documents

  • SY-01, Steelman the opposite thesis: the constructivist case in full. The case against rests on it.
  • SY-02, Nature, prehistory and the naturalistic fallacy: Hume, Moore, Davis, the conditional sentence, Hrdy, Sapolsky, Larmuseau.
  • SY-04, Replication map: what holds and what fell: the graded map, power posing, single-sex schooling, the metric verdict on Hyde against Del Giudice.
  • HC-07, The gender-equality paradox: the worked example, from Costa 2001 to Herlitz 2024.
  • the digest of SY-01, SY-03 and SY-05 and the HC digest: the shelf map and the report-to-report discrepancies.

Its documents in the Library

199 234 235 236 237 238