THRIVING after the FALL
Book Two PAIRING after the FALL

PAIRING · Part II · The Campfire Contract

The numbers and sources

Chapter 6: Flooding

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.

  • 54 open
  • 19 to check first

Read about the chapter · Its 54 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.

  • 79 couples, 21 divorces: the sample behind the 93 per cent (CO-04). About 29 per cent positive predictive value when cross-validated, from SY-05 via the concept note (to be checked). Rows C06-11, C06-15.
  • 95 to 100 beats a minute: the flooding threshold, carried as a range (CO-01, CO-04). Row C06-03.
  • r = .360, .380, .392: demand-withdraw and dissatisfaction overall, wife-demand, husband-demand, from 74 studies and 14,255 participants (CO-04). Row C06-19.
  • 0.06 percentage points: the rise in erectile dysfunction prescriptions at the point where a Danish wife's income passed her husband's, in more than 200,000 couples, absent in cohabiting couples (CO-07). Row C06-37.
  • g = 0.69, confidence interval 0.58 to 0.81, about 73 per cent overlap: the sex difference in sex drive across 211 studies and 621,463 people, falling to g = 0.54 adjusted for response bias (PS-06 PDF, page 5). Row C06-47.

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
79 couples recruited 1980, followed 14 years, interbeat interval, pulse transmission time, skin conductance, somatic activity, 15-minute conflict discussionThe 1980 protocol and sampleGottman and Levenson 1985, J Pers Soc Psychol; Gottman and Levenson 1999, J Marriage Fam, "The Timing of Divorce", N = 79CO-04C06-01physiologyrobust
Resting pulse 60 to 80 beats a minute, 100 as the pulse of a brisk walkGeneral physiology for the openingNo document sourcenoneC06-54physiologyopen
Flooding threshold 100 beats a minute (CO-04) or 95, with 85 for conditioned athletes (CO-01)Where flooding beginsGottman's definition. CO-01 cites a clinical manualCO-01, CO-04C06-03physiologycontested
Adrenaline and cortisol release, raised blood pressure, reduced frontal control above the lineWhat flooding isNo primary source named for the frontal claimCO-01, CO-04C06-04physiologyopen
Men flooded faster, stayed flooded longer, took longer to return to baselineSex difference in floodingGottman and Levenson 1985, J Pers Soc PsycholCO-04, CO-01C06-05physiologyrobust
Stonewalling as escape conditioning, silence lowers the pulseMechanism of withdrawalNo primary source namedCO-04C06-06behaviouropen
About 85 per cent of stonewallers menSex of stonewallersGottman's clinical claim, traced only to a couples-app blogCO-01, PS-02, PS-03C06-07behaviourcontested
Four Horsemen: criticism, contempt, defensiveness, stonewalling, contempt most corrosivePredictors of marital distressGottman 1994, Why Marriages Succeed or Fail, not in the documentsCO-01C06-08behaviourrobust
5 to 1 positive to negative in conflict, 20 to 1 outside itGottman's ratioBlog-sourced, no paper namedCO-01C06-09behaviouropen
Up to 94 per cent (1998) and 93 per cent (1999) divorce prediction accuracyThe prediction claimGottman, Coan, Carrere and Swanson 1998, J Marriage Fam; Gottman and Levenson 1999, J Marriage FamCO-01, CO-03, CO-04C06-10behaviourfallen as prophecy, the constructs stand
N = 79 with 21 divorcesSample behind the 93 per centGottman and Levenson 1999CO-04 PDF page 9C06-11behaviourrobust
Extreme-group samples of 60 to 79 couples (CO-01) or 60 to 130 (CO-03)Sample sizes of the prediction studiesHeyman and Slep 2001CO-01 PDF page 4, CO-03C06-12behaviourcontested
Same-sample fitting, no cross-validationThe arithmetic objectionHeyman and Smith Slep 2001, J Marriage Fam 63:473, "The hazards of predicting divorce without crossvalidation"CO-01C06-13behaviourrobust
About 29 per cent cross-validated positive predictive valueOut-of-sample predictionSY-05 via the concept note. CO-03 says only that cross-validated accuracy "often failed to exceed chance"SY-05C06-15behaviouropen
Sex-typed sequences (husband rejects influence, wife's negative start-up, failure to de-escalate) failed to replicate on a community sample, constructs held, "equifinal"Replication of the choreographyHeyman and Slep 2002, journal not named in CO-01CO-01C06-16behaviourrobust
Demand-withdraw described 1988The patternChristensen 1988, chapter not named precisely in CO-04CO-04C06-17behaviourrobust
74 studies, 14,255 participants, r = .360 with dissatisfaction, wife-demand r = .380 (majority), husband-demand r = .392 (substantial minority), clinical samples r = .413Demand-withdraw and dissatisfactionSchrodt, Witt and Shimkowski 2014, Communication Monographs 81:28CO-04 (r = .360 from PDF page 9)C06-19behaviourrobust
Same asymmetry in Brazil, Italy, Taiwan and the United StatesDemand-withdraw across culturesChristensen, Eldridge, Catta-Preta, Lim and Santagata 2006, J Consult Clin Psychol 74:1029CO-04C06-20culturerobust
63 heterosexual, gay and lesbian couples: the demander is whoever wants more change, regardless of sexSex-neutral demandCO-04 attributes it to the same 2006 reference. Possibly Holley, Sturm and Levenson 2010. Primary source to be identifiedCO-04C06-21cultureopen
"Biology provides the mechanism (flooding), culture dictates who demands and who withdraws"Summary lineCO-04's own consensus line, no primaryCO-04C06-22cultureopen
21 gay and 21 lesbian couples, 12-year study (CO-04 gives 106, which is Garanzini 2017)Same-sex control sampleGottman, Levenson, Gross et al. 2003, J Homosexuality 45:23, N = 42CO-06, CO-04C06-23behaviourcontested on sample size
More affection and humour at start-up, less belligerence, domineering and fear, unhappy same-sex couples abnormally low arousalSame-sex conflict styleGottman et al. 2003CO-06C06-24physiology and behaviourrobust
106 gay and lesbian couples, uncontrolled therapy studyLater therapy sampleGaranzini et al. 2017, J Marital Fam TherCO-06C06-25behaviouropen
HCMST 2017 wave N = 3,510Size of Rosenfeld's panelHow Couples Meet and Stay Together surveyCO-06 PDF page 3C06-26behaviourrobust
Same-sex and different-sex break-up rates nearly identical once marital status is controlledDissolution ratesRosenfeld 2014, J Marriage Fam 76:905CO-06C06-27culturerobust
1996 paper asking whether men are really more physiologically affected than womenFraming of the endocrine questionKiecolt-Glaser et al. 1996, J Consult Clin Psychol 64:324CO-04C06-29physiologyrobust
Women show larger, more persistent rises in cortisol, adrenaline and noradrenaline after conflict, largest when husbands withdraw, 90 newlywed couplesEndocrine sex difference in conflictKiecolt-Glaser and Newton 2001, Psychol Bull 127:472CO-04C06-30physiologyrobust
42 married couples, suction blisters, 30-minute conflict, hostile couples healed at 60 per cent of the rate, IL-6 and TNF-alpha raisedConflict and wound healingKiecolt-Glaser et al. 2005, Arch Gen Psychiatry 62:1377CO-04C06-31physiologyrobust
47 couples (94 people), intranasal oxytocin, positive to negative ratio F = 4.18, p = .047, cortisol F = 7.14, p = .011Oxytocin and couple conflictDitzen et al. 2009, Biol Psychiatry 65:728CO-04 PDF page 3C06-33physiologycontested
Oxytocin lowered salivary alpha-amylase in women, raised it in men, rise in men correlated with positive communicationSex by directionDitzen et al. 2013, PsychoneuroendocrinologyCO-03C06-34physiologyopen
126 articles, more than 72,000 people, r = .07 to .21 (PS-07), mortality r = .11, cardiovascular reactivity r = −.13 (CO-04), comparable to diet and exerciseMarital quality and healthRobles, Slatcher, Trombello and McGinn 2014, Psychol Bull 140:140CO-01, CO-04 PDF page 3, PS-07C06-35physiologyrobust, figures differ between documents
Lonely married men have a flatter diurnal cortisol slopeLoneliness inside marriageKORA-Age cohort, paper not named in CO-01CO-01C06-36physiologyopen
More than 200,000 married Danish couples 1997 to 2006, erectile dysfunction prescriptions up 0.06 percentage points where wife's income passes husband's (p < .01), wives more insomnia and anxiety prescriptions, no discontinuity for cohabiting couples or where she out-earned him before marriageThe income line with a bodily readoutPierce, Dahl and Nielsen 2013, Pers Soc Psychol Bull 39:359, "In Sickness and in Wealth"CO-07C06-37culture, physiology readoutrobust
6,035 households, men's distress U-shaped against wife's share, minimum about 40 per cent, maximum at 100Male distress by relative incomeSyrda 2020, Pers Soc Psychol BullCO-07C06-38cultureopen
Distribution of couples falls off a cliff past the point where the wife earns half. Cliff vanishes once the roughly 0.25 per cent of exactly equal earners are removedRelative income distributionBertrand, Kamenica and Pan 2015, Q J Econ 130:571; Binder and Lam 2022, J Hum ResourCO-07C06-39culturefallen (the cliff)
15.9 percentage points more likely to misreport when wife just out-earns husband, men up 5.8, women down 5.1Misreporting of earningsMurray-Close and Heggeness 2018, DemographyCO-07C06-40culturerobust
NLSY, 2,750 people, fully dependent men about 15 per cent annual infidelity against about 5 per cent for dependent women, male minimum at about 70 per cent of income, 85 per cent of dependent men do not cheat in a yearInfidelity by economic dependenceMunsch 2015, Am Sociol Rev 80:469CO-07C06-41cultureopen
PSID, 46 years, more than 6,300 couples. Pre-1975 marriages with 50-50 housework 36 per cent more likely to divorce than wife-75 per cent. Post-1975 housework loses predictive power. Husband without full-time job 3.3 against 2.5 per cent annual divorce (33 per cent in CO-06 and CO-07, about 30 per cent in CO-02). Wife's income no effectPredictors of divorce in two erasKillewald 2016, Am Sociol Rev 81:696CO-06, CO-07, CO-02C06-42culturerobust, hazard figure differs between documents
29 countries: a husband's unemployment raises separation risk only where the breadwinner norm is strongNorm dependence of the breadwinner effectGonalons-Pons and Gangl 2021, Am Sociol RevCO-07C06-44culturerobust
Female breadwinning costs life satisfaction in West Germany, not in the EastNorm dependence within GermanySprengholz et al. 2022, journal not named in CO-07CO-07C06-45culturerobust
1,865 people, cross-sectional, women's desire declines with relationship duration, men's stableDesire and durationKlusmann 2002, Arch Sex Behav 31:275PS-06C06-46behaviouropen
211 studies, 621,463 people, g = 0.69 (95 per cent CI 0.58 to 0.81), about 73 per cent overlap, adjusted for response bias g = 0.54, about 79 per cent overlapSex difference in sex driveFrankenbach et al. 2022, Psychol Bull 148:621. Overlap figures computed by the author from the overlap coefficient, not in PS-06PS-06 PDF page 5C06-47behaviourcontested
Men's orgasm rates above 90 per cent against 35 to 65 per cent for women, more pain and fear for women. Frankenbach's reply concedes suppression of reported desireThe case against the desire gapConley and Yang 2024, Psychol Bull; Frankenbach et al. 2024 replyPS-06C06-48behaviourcontested
Well-being gain from frequency plateaus at about once a week. Total desire in a couple matters more than how well matched it isFrequency and desireMuise, Schimmack and Impett 2016, Soc Psychol Pers Sci 7:295; Muise et al. 2020PS-06C06-49behaviourrobust
113 samples, 66,132 people, men higher on avoidance and lower on anxiety, D = .28 in community samples (about 78 per cent overlap), .12 in student samples, −.07 in web samplesAttachment sex differenceDel Giudice 2011, Psychol Bull 137:193PS-07C06-50behaviourrobust
628 people over 59 years, anxiety peaks in young adulthood, avoidance declines linearly with ageAttachment across the lifeChopik, Edelstein and Grimm 2019, J Pers Soc PsycholPS-07C06-51behaviourrobust
43 datasets, more than 11,000 couples. Relationship variables about 45 per cent of current quality, individual variables about 21 per cent, change over time at most 15 to 20 per cent. Perceived partner responsiveness the strongest predictorWhat predicts relationship qualityJoel et al. 2020, Proc Natl Acad Sci 117:19061CO-01, PS-07C06-52behaviourrobust
20 to 30 minute time-out for adrenaline and cortisol to clearTime to come downBlog-sourced in both documents, no primary namedCO-01, CO-04C06-53physiologyopen
Affiliations: Gottman and Levenson at Illinois and Indiana in 1980, Heyman and Slep at Stony Brook, Christensen at UCLA, Rosenfeld at Stanford, Kiecolt-Glaser at Ohio State, Ditzen then at ZurichWhere the named people workedGeneral knowledge, not in the documentsnoneC06-02, C06-14, C06-18, C06-26, C06-28, C06-32noneopen

The case for and the case against

The case the chapter makes

  • Flooding is a measured bodily state above roughly 95 to 100 beats a minute, with adrenaline and cortisol release and reduced frontal control. Past it the body cannot listen (CO-01, CO-04, C06-03, C06-04).
  • Men flooded faster, stayed flooded longer and took longer to return to baseline in Gottman and Levenson's 1985 sample. Stonewalling is escape conditioning, the body reaching for silence to bring the pulse down (Gottman and Levenson 1985, CO-04, C06-05, C06-06).
  • Women show larger and more persistent rises in cortisol, adrenaline and noradrenaline after conflict, especially when the husband withdraws. The sex that reacts more depends on the measure (Kiecolt-Glaser and Newton 2001, CO-04, C06-30).
  • Hostile conflict slows wound healing to 60 per cent of the rate and raises inflammatory markers (Kiecolt-Glaser et al. 2005, CO-04, C06-31).
  • Demand-withdraw correlates with dissatisfaction at r = .360 across 74 studies and 14,255 people. The wife-demand form is the majority at r = .380, the husband-demand form a substantial minority at r = .392 (Schrodt, Witt and Shimkowski 2014, CO-04, C06-19).
  • The demander is whoever wants more change, regardless of sex. Biology provides the mechanism, culture decides who demands and who withdraws (Christensen et al. 2006 and the 63-couple comparison, CO-04, C06-20, C06-21, C06-22).
  • Gottman's predictors are risk factors, not prophecies. The 93 per cent rests on same-sample validation of 79 couples with 21 divorces. The sex-typed sequences failed to replicate (Heyman and Slep 2001 and 2002, CO-01, CO-03, CO-04, C06-10 to C06-16).
  • The income line bites married men who signed the old contract and not cohabiting men or men whose wives always earned more (Pierce, Dahl and Nielsen 2013, CO-07, C06-37). The breadwinner effect switches on and off with the local norm (Gonalons-Pons and Gangl 2021, Sprengholz et al. 2022, CO-07, C06-44, C06-45).
  • Relationship variables explain about 45 per cent of present quality, individual variables about 21 per cent, and perceived partner responsiveness is the strongest single predictor (Joel et al. 2020, CO-01, PS-07, C06-52).

The case against

  • Christensen's own comparison of heterosexual, gay and lesbian couples: the demander is whoever wants change, and sex drops out (CO-04, C06-21).
  • Heyman and Slep 2002: the sex-typed sequences failed to appear in a community sample while the sex-neutral constructs held (CO-01, C06-16).
  • Pierce 2013: the discontinuity vanished in cohabiting couples and in couples where the wife had always earned more. The men were responding to a promise, not to being out-earned as such (CO-07, C06-37).
  • Gonalons-Pons and Gangl 2021 across 29 countries, and Sprengholz et al. 2022 across the old inner German border: the breadwinner effect follows the norm (CO-07, C06-44, C06-45).
  • Rosenfeld 2014: same-sex and different-sex couples dissolve at nearly the same rate once marriage is controlled, although same-sex couples flood less (CO-06, C06-27).
  • Joel et al. 2020: everything about the individuals combined, sex included, explains about 21 per cent of present quality and at most 15 to 20 per cent of change over time (CO-01, C06-52).
  • The 85 per cent of stonewallers claim is blog-sourced and untraced to any paper (CO-01, PS-02, PS-03, C06-07).
  • Conley and Yang 2024: the measured sex difference in desire is partly a difference in what sex gives each partner, which depresses women's reported desire (PS-06, C06-48).

Where the sources disagree

  • The flooding threshold: 95 beats a minute (CO-01, citing a clinical manual, with 85 for conditioned athletes) against 100 (CO-04). The data edition carried the range. The narrated edition carries 100 (C06-03).
  • 85 per cent of stonewallers male: Gottman's clinical claim, traced in every document only to a couples-app blog (CO-01 ref 6, PS-02, PS-03). Carried as an impression, not a fact (C06-07).
  • Prediction accuracy: 94 per cent in the 1998 paper (CO-01) against 93 per cent in the 1999 paper (CO-03, CO-04). Two papers, both carried in the data edition, rounded to "better than nine times in ten" in the narrated edition (C06-10).
  • The about 29 per cent cross-validated positive predictive value: from SY-05 via the concept note, not in CO-01, CO-03 or CO-04, which say only that cross-validated accuracy "often failed to exceed chance" (C06-15).
  • Extreme-group sample sizes: 60 to 79 (CO-01 PDF page 4) against 60 to 130 (CO-03). Both carried in the data edition (C06-12).
  • Gottman and Levenson 2003 sample: 42 (CO-06) against 106 (CO-04). 106 is Garanzini 2017's uncontrolled therapy sample. 42 carried (C06-23, C06-25).
  • Whether Christensen et al. 2006 (J Consult Clin Psychol) covers both the four-country comparison and the 63-couple heterosexual, gay and lesbian comparison, or whether CO-04 has merged two papers (C06-20, C06-21).
  • Killewald's husband-unemployment hazard: 33 per cent (CO-06, CO-07) against about 30 per cent (CO-02). The narrated edition says "about a third" (C06-42).
  • Robles's effect sizes: r = .07 to .21 (PS-07) against r = .11 for mortality and r = −.13 for cardiovascular reactivity (CO-04 PDF page 3). Both carried in the data edition (C06-35).
  • Bertrand, Kamenica and Pan's cliff at 50 per cent (CO-07) against Binder and Lam's dissolution of it once exactly equal earners are removed (CO-07). The cliff is an artefact, the misreporting is not (C06-39, C06-40).
  • Frankenbach's g = 0.69 against Conley and Yang's reading of it as partly a difference in what sex gives each partner (PS-06, C06-47, C06-48).
  • Ditzen 2009's oxytocin effects at p = .047 against the replication record for intranasal oxytocin set out in the Preface (CO-04, C06-33).

Open questions

marks for checking in the data edition (9, of which three mark one claim):

  • Gottman's and Levenson's 1980 affiliations, Illinois and Indiana, general knowledge, not in the documents (C06-02).
  • Heyman and Slep at Stony Brook University, general knowledge (C06-14).
  • The about 29 per cent cross-validated figure, marked three times in the data edition (Story, In practice, Numbers to carry). From SY-05 via the concept note, not in the CO documents (C06-15).
  • Christensen at UCLA, general knowledge (C06-18).
  • Rosenfeld at Stanford, general knowledge (C06-26).
  • Kiecolt-Glaser at Ohio State University College of Medicine, general knowledge (C06-28).
  • Ditzen then at the University of Zurich, general knowledge (C06-32).

The data edition's open questions:

  • Whether Christensen et al. 2006 is one paper covering four countries and the 63-couple comparison, or whether CO-04 has merged two papers. The 63-couple figure needs a primary source, possibly Holley, Sturm and Levenson 2010 (C06-21).
  • The source of the about 29 per cent cross-validated positive predictive value (C06-15).
  • Which Gottman source the 95 beats a minute threshold comes from. CO-01 cites a clinical manual (C06-03).
  • Whether Robles's r = .11 (mortality) and r = −.13 (cardiovascular reactivity) in CO-04 are the paper's own or a secondary source's summary (C06-35).
  • Whether the oxytocin and alpha-amylase sex-by-direction result in Ditzen 2013 has been replicated (C06-34).
  • No primary source is named for the prefrontal claim in the definition of flooding (C06-04), for stonewalling as escape conditioning (C06-06), for the 5 to 1 and 20 to 1 ratios (C06-09), for the 20 to 30 minute time-out (C06-53) or for the KORA-Age cortisol finding (C06-36).
  • Joel et al. 2020: "85 colleagues" was the data edition's count of the author list and needs checking (C06-52).
  • The opening scene is a composite of the 1980 protocol as CO-04 describes it. The pulse readings in the scene are illustrative, not data (C06-01, C06-54).

The documents

  • CO-01: What makes heterosexual couples last. Gottman, Heyman and Slep, Joel 2020, Robles 2014, KORA-Age.
  • CO-03: The polarity literature. Ditzen 2013, the 60 to 130 range.
  • CO-04: Conflict. The 79 couples, flooding, Schrodt 2014, Kiecolt-Glaser, Ditzen 2009, Christensen.
  • CO-06: Control groups. Gottman and Levenson 2003, Garanzini 2017, Rosenfeld, Killewald 2016.
  • CO-07: Money and status. Pierce 2013, Syrda 2020, Bertrand 2015, Binder and Lam 2022, Murray-Close and Heggeness 2018, Munsch 2015, Gonalons-Pons and Gangl 2021, Sprengholz 2022.
  • PS-06: Desire. Klusmann 2002, Frankenbach 2022, Conley and Yang 2024, Muise 2016 and 2020.
  • PS-07: Attachment. Del Giudice 2011, Chopik 2019, Joel 2020, Robles 2014.
  • Fact sheets: the CO and FA digest and the OR and PS digest.

Its documents in the Library

201 202 205 206 208 209 210 211 213 214 238