Every Foundations session our Singapore chapter has run since August 2024 was recorded, transcribed and coded. This guide reports what each of the eight sessions actually produces in the room, which questions were put, who speaks, and which of the things convenors worry about turn out not to matter. Every recommendation carries the measurement behind it.
Most advice about running groups is precept. This is what twenty-four tapes actually show.
The sequence is the one feature of a Foundations evening with a measurable effect, and it should be run whole. The chapter's own practice sets the scale: a mean of 98 minutes, from 75 to 131, with 8 to 27 voices in the room and four to eight questions put, a median of six.
They are not interchangeable. Introduction rounds are where calling and the church come up; discussion is where ethics, identity, risk and suffering rise. Across the corpus, calling appears in 49% of introduction turns and falls to 23 in discussion, while ethics and excellence rise from 4% to 34, Spirit-guidance from 10 to 34, identity from 8 to 29, and both risk and suffering from 6 to 21.
An introduction round cannot do the work of a discussion, and a discussion without one starts cold. In all 24 recorded sessions the discussion ran deeper than the round preceding it.
The comparison is between segments of a fixed format rather than between conditions, so it shows the sequence doing what it is designed to do. It does not show that an alternative format would do less.
Each page carries the three recorded runs side by side, which film was shown, every question put, the subjects the session produces against the corpus baseline, and how the room divides by sex.
The opening night, and for most people the only night. It is the one session where people tell you how they came to faith, and the one where the room is most likely to contain someone who has not.
101 contributions · 102–125 min · 19–21 voices · 37% female
Most over-represented: Faith-journey stories at 25% against 10%
The money session, and the one where the room talks about ownership, funding and what it means to hold a business loosely.
78 contributions · 88–103 min · 15–23 voices · 38% female
Most over-represented: Wealth and stewardship at 29% against 20%
Rest, overwork and the idol of the business. The session that produces the most talk about Sabbath, by a wide margin.
90 contributions · 75–114 min · 10–27 voices · 29% female
Most over-represented: Rest and Sabbath at 9% against 5%
Craft, standards and the reputation of Christian businesses. The most sharply focused session in the corpus.
62 contributions · 77–82 min · 12–18 voices · 40% female
Most over-represented: Ethics and excellence at 45% against 20%
Ambition, identity and what success is measured against. Run in Cohort 2 under the title 'I Sacrifice, We Win'.
73 contributions · 91–106 min · 9–20 voices · 42% female
Most over-represented: Rest and Sabbath at 16% against 5%
Speaking about faith at work. The session that produces the most talk about suffering and mental health.
75 contributions · 88–127 min · 9–23 voices · 32% female
Most over-represented: Cross-cultural mission at 20% against 12%
Serving employees, customers and neighbours through the operations of the business rather than through words.
66 contributions · 91–96 min · 8–20 voices · 52% female
Most over-represented: The church as institution at 23% against 15%
The closing night: mission, the wider movement and what the group does next.
73 contributions · 88–131 min · 9–23 voices · 40% female
Most over-represented: The great commission at 11% against 3%
These are findings about the theology itself rather than about how to run an evening. They are the reason the protocol below is worth trusting.
Of 181 Pentecostal markers, 83 name the Spirit's direction and 40 expect material blessing. Only 9 name the work itself as something the Spirit gifts, voiced by nine different people, so no one develops the thought twice. Prosperity organises 3.6% of all talk. The room is fluent about being led and nearly mute about being provided for.
Weighted doctrinal material runs Evangelical-Protestant at 37%, Pentecostal-Charismatic at 25, Roman Catholic at 21, and a cross-traditional ascetic register at 17. The repertoire available to the room is far wider than the part it uses.
Where the room reaches for a cultural script it reaches for a migrant and mercantile one: expatriate and transnational life 44 times, global mission 36, the multi-faith workplace 27. Ancestral obligation barely surfaces, though the coding has a category for it. And where the room takes a position on these scripts, it contests them more often than it affirms them.
For the 28 speakers whose church could be verified independently, the tradition they sounded like matched their actual congregation 9 times. Always naming the room's commonest tradition would have scored 14. Doctrinal accent is not a membership card, and you cannot infer someone's church from how they pray.
The attributes that predict which register a believer uses are the believer's own: sex and daily trade. The formal church, the film shown and the room sat in predict nothing the coding could detect. The theological selection is made by the people present, not by the institutional setting.
Each carries the measurement behind it. Where a result is suggestive rather than established, it says so.
Theology arrives the moment the room is asked something it cannot answer from a CV, and it does not build through the evening. In every one of the 24 recorded sessions the discussion carried more doctrinal markers per contribution than the introduction round before it, 1.77 against 0.66 (paired Wilcoxon p = 1.2 x 10^-7). The rise is a step at the boundary between the two segments, not a slope: once the discussion starts, how late a person speaks tells you nothing about how deeply they speak (median rho = -0.01, p = 0.66).
DoKeep the introduction round and keep it short, then put a question tied to the teaching. A fellowship with no second question has asked only the one whose answer is a business card.
More voices meant fewer turns each, with no gain in depth. Recorded rooms ran from 8 to 27 voices.
DoTables of six to eight, each with a briefed host.
Facilitator share of turns varied fourfold across recorded sessions. Higher share went with fewer participant turns (rho = -0.46, p = 0.027) and no more depth (rho = +0.01). The correlation does not survive multiple-comparison correction.
DoAim under a third of the turns, and treat this as a direction rather than a proven effect.
Across 15 standard and 9 substituted sessions, marker density was indistinguishable (1.89 against 1.58, Mann-Whitney p = 0.20). What a substitute film changes is the subject matter, not the quality of the conversation.
DoPick the film for the conversation you want, then write questions about the film you actually showed.
Three measures appeared to separate sessions held in worship buildings from those held elsewhere, at q = 0.069. All three collapse under an exact permutation test that holds cohort fixed.
DoThe room does not change what gets said. Choose for parking and quiet.
Women's contributions carry Pentecostal-charismatic markers at 30.4% of their weighted marker mass against 21.1 for men, and raise Spirit-guidance and prayer at 30.1 against 18.3. Men raise wealth and stewardship at 24.3 against 13.8. This is the most robust contrast in the corpus, and it is an emphasis rather than a partition: of the 146 people who voice either register, 49 voice both.
DoMove the addressee, not the topic. When the conversation turns to money, put the question to the women first; when it turns to guidance and prayer, put it to the men.
What someone does all week shapes what they bring. Speakers working in finance and investment raise wealth and stewardship in 37.9% of their contributions against 17.7% for everyone else (q = 0.015, and it survives dropping any one cohort). Sector conditions the subject matter, though not the doctrinal register.
DoLook at the roster before choosing the question. A table of financiers will take you to stewardship whether you planned it or not.
Across 24 recorded sessions the mean length was 98 minutes, ranging from 75 to 131. Longer sessions produced more turns of conversation (Spearman rho = 0.64, p = 0.0007) but not deeper ones, so roughly 90 to 110 minutes is the working range.
Yes. In 24 of 24 recorded sessions the discussion that followed was deeper than the introduction round preceding it, with doctrinal markers per contribution rising from 0.66 to 1.77. The two segments also elicit different subjects: calling and the church dominate introductions, while ethics, identity, risk and suffering rise in discussion.
More voices meant fewer turns each (rho = -0.55, q = 0.020) with no gain in depth. Recorded rooms ran from 8 to 27 voices. Above roughly sixteen, split into tables of six to eight.
Less than convenors assume. Across 15 sessions using the standard film and 9 using a substitute, marker density was statistically indistinguishable (1.89 against 1.58, Mann–Whitney p = 0.20). What the film changes is the subject matter, not the depth.
No effect survives testing. Three measures appeared to separate sessions held in worship buildings from those held elsewhere, but all three collapse once cohort is held fixed by an exact permutation test. Meet wherever is convenient.
Less. Facilitator share ranged from 14 to 60% of turns. Higher facilitator share was associated with fewer participant turns (rho = -0.46, p = 0.027), though this does not survive multiple-comparison correction. Treat it as a direction.
Yes, and it is the most robust contrast in the corpus. Women's turns carry Pentecostal-charismatic markers at 30.4% of their weighted marker mass against 21.1% for men (q = 0.013), and raise Spirit-guidance and prayer at 30.1% against 18.3 (q = 0.036). Men raise wealth and stewardship at 24.3% against 13.8 (q = 0.046).
Between four and eight were put in recorded sessions, with a median of six. No recorded session got through more than about three at any one table. Choose three and drop the rest.