WEAKNESS #3 TESTED — does the shackle show up in MATURE democracies as fragmentation + governance erosion?
Sample: "established democracies" = 79 countries with Freedom House score >=80 in 2005. Sources: V-Dem/V-Party (parties), FH sub-components.

[A] FRAGMENTATION (effective number of parties, Laakso-Taagepera), within-country, established democracies:
    trend +0.035/yr, p<1e-6 (N=756). Decade means: 1970-89 = 2.75 -> 1990-04 = 3.63 -> 2005-19 = 4.03.
    => party systems fragmented ~+47% since the 1970s. SUPPORTS "smaller, more numerous, friction-full creeds."

[B] PARLIAMENTARY PARTY POPULISM (seat-weighted V-Party populism index), established democracies:
    trend +0.0031/yr, p<1e-7. Mean 0.315 (1990-2004) -> 0.362 (2005-2019).
    => the parties actually winning seats became more populist. SUPPORTS "extreme creeds pulling into democracies."

[C] THE REFRAME (damage hides below the headline), established democracies, 2012-2025 within-country:
    Headline TOTAL score:           -0.216/yr (p<1e-4)   [modest decline]
    Functioning of Government (C):  -0.328/yr (p<4e-4)   [erodes ~50% FASTER than the headline]
    => in mature democracies governance/capacity erodes faster than the overall democracy score.

[D] DIRECT LINK fragmentation -> governance (Functioning of Govt ~ ENP, within established democracies, 2012-19 overlap):
    coefficient -0.41 (right sign: more parties -> weaker governance) BUT p=0.71, N=123 -> NOT significant.
    => the two co-occur; the DIRECT causal link is unproven in the short overlap window.

VERDICT: Weakness #3 reframe is largely SUPPORTED descriptively. Mature democracies ARE fragmenting, their parties ARE
more populist, and their GOVERNANCE erodes faster than their headline score -- so the flat-ish headline is NOT evidence the
shackle is absent; the symptom is just below the regime-level index. Honest limits: (1) descriptive, not causal -- the direct
fragmentation->paralysis link is right-signed but n.s.; (2) rising ENP could be healthy representation (Madison) -- the
"dysfunction" reading rests on the parallel governance decline, not proven causation; (3) V-Party ends 2019; (4) established
democracies DO show a modest headline decline too (-0.22/yr), so "flat" is "slow," not "none." For mature democracies the
honest claim stays: shackle + fragmentation + governance erosion OBSERVED; regime-level de-democratization PROJECTED.

==================================================================
FOLLOW-UP: can the fragmentation -> governance LINK be made significant? (tested 5 specs, more data)
DV = FH Functioning of Government (C). IVs = fragmentation (ENP), party populism. NOT p-hacking: all specs reported.
  [A] all countries, 2-way FE:        ENP +0.63 (p0.61)  -- null, wrong sign
  [B] + populism:                     ENP p0.60, populism p0.76 -- null
  [C] democracies only:               null
  [D] pooled LEVELS (no FE):          ENP +3.16 (p0.008) -- SIGNIFICANT but WRONG SIGN (more parties->better govt) & confounded (development)
  [E] change ΔC~ΔENP (N=110):         p0.93, corr +0.01 -- flat
VERDICT: the fragmentation->governance-paralysis CAUSAL link is NOT supported (null across specs; only sig result is
confounded + reversed). The three descriptive trends still stand separately (fragmentation rising; parties more populist;
governance erodes faster than headline). HONEST CLAIM: mature democracies are simultaneously fragmenting AND losing governing
capacity = CO-OCCURRENCE, not causation. Do NOT assert fragmentation causes the governance decline. Causal weight for the
mature-democracy case rests on the foreign-policy shackle (weakness #1: Ukraine 6-month gap, Iran domestic-cost), not this regression.
NOTE: refused to keep swapping IVs to manufacture significance (p-hacking) — reporting the null protects the book's credibility.
