Mind the Gap, wave 1

01

Six things worth knowing before you design the session

1. The gap is real on 4 of 7 traits, and it leans one way every timeManagers rate Gen Z less favourably whatever the question asks. On the traits where a high score is a compliment they score Gen Z lower; on the traits where a high score is a complaint they score them higher. A genuine disagreement would point different ways on different items.
2. The deference gap is about Thai obligation, not about hierarchySplit the deference items into the ones about เกรงใจ, บุญคุณ and saving face, and the ones about who gets to decide. The obligation items are 2.20 apart (g = -2.16). The hierarchy items are 0.97 apart (g = -1.04). The headline gap is mostly the first kind, and that is a different workshop from a hierarchy workshop.
3. Gen Z cannot see the gap, and they guess wrong in one directionAsked to guess what employers believe about them, Gen Z were kinder to themselves than the managers were on every scale that shows a real gap. They are not ignoring this or arguing with it. They do not know.
4. Nothing about the manager explains itSplit the 35 managers by role, seniority, years managing Gen Z, organisation size, share of Gen Z staff, and their own generation. The largest spread any split produces on the focal scale is 0.47, against a headline gap of 1.74. And 29 of the 35 are Millennials, so this is not older people judging younger people.
5. Both sides are arguing about different thingsThe managers' top written worry is other-orientation (57%). What Gen Z expect to be accused of is laziness (56%), and only 11% of them saw other-orientation coming. Both sides then used their one open letter to ask the other side to be the one who changes.
6. Two measurement problems to fix before wave 2Both reverse-worded items behave as if they belong to a different scale (negative item-rest correlations on every referent), and three scales have reliability below the usual 0.70 threshold on at least one referent. Neither invalidates the headline, and both need fixing.

Everything below is folded. Each section opens with the finding in a sentence; click any heading marked with a triangle to get the full table underneath it. Nothing on this page is typed by hand: it is generated from the response export every time the analysis is re-run, so it cannot drift out of step with the numbers.

02

Who answered

Read every employer number as "these managers", not as "Thai employers". 20 of the 35 are one hospital group, both arms volunteered rather than being sampled, and the Gen Z respondents do not work for the managers who answered. Section 07 tests whether the hospital cohort behaves like the rest.

03

All seven scales, with the full statistics

Four scales show a gap the data can stand behind: seniority deference, loyalty / low mobility, learning adaptability, workplace entitlement. Three cannot be told from zero and should not be presented as findings: work-life segmentation, motivation to lead, power distance.

04

The lean runs one way

All 4 gaps that survive point in the less flattering direction, on both the scales where a high score is a compliment and the scales where a high score is a complaint.

This is the argument that the finding is a bias rather than a set of opinions. If managers simply disagreed with Gen Z about young workers, the direction would flip somewhere across seven different questions. It does not flip once. Use this in the room right after the overview chart, because it is the point participants most often push back on.

05

Item by item, all 36 statements

The five widest single-item gaps in the whole study are all deference items, and all five are about obligation and face rather than about rank. That is what section 05 tests.

Items are shown in the Thai they were fielded in, because that is what the respondent read and a translation would put a second author between the team and the data. Every scale is below, folded.

06

What the deference gap actually is

The obligation items are 2.20 points apart (g = -2.16). The hierarchy items are 0.97 apart (g = -1.04). The headline gap is mostly about Thai social obligation, not about accepting rank.

The deference scale was built in two layers on purpose: standard power-distance items that exist in the international literature, and Thai obligation items that do not. This split is the payoff. Gen Z report the obligation feelings strongly (5.00, well into agreement) and managers do not believe it (2.81, disagreement). On plain hierarchy both sides sit much closer, and the standalone power-distance scale shows no gap at all.

For the workshop this changes the script. A session about "young people do not respect authority" is aimed at the layer where the gap is smallest. The thing managers are missing is that these employees report feeling เกรงใจ and บุญคุณ toward them, and are read as not feeling it. Caveat: the split was defined by reading the item wording, it was not pre-registered, and it needs a factor analysis on a bigger sample before it goes in a paper.

07

Meta-accuracy: what Gen Z think managers think

On every scale that shows a real gap, the Gen Z guess is kinder to Gen Z than what managers actually said. The biggest miss is deference, where they expected 3.98 and got 2.72.

This is the study's lead contribution and the part that is hardest to present without sounding accusatory. It is not that Gen Z are wrong about their managers in a general way. It is that the error runs one direction: they assume they are seen more generously than they are. A person who believes they are already read correctly has no reason to change anything, which is why "they know and do not care" is the wrong story to tell in the room.

08

What managers say the contact is actually like

Positive contact averages 3.81 of 5 and negative contact 1.73, so most managers report mostly good day-to-day experience with Gen Z. The belief that travels with worse reported contact is entitlement: the more a manager believes Gen Z expect to be accommodated, the less positive they say the contact was (r = -0.46).

How to read this: These two blocks were in the instrument from the start and no analysis had touched them until today. They matter because every other measure in this study is a BELIEF and these are the closest wave 1 gets to behaviour: how the manager says the contact actually went. Positive and negative contact are two scales rather than two ends of one, which is deliberate. A manager can report plenty of both, and that person is exactly who gets hidden by averaging them together.

Does what a manager believes go with how the contact goes?

How to read this: Correlations between a manager's belief score and their reported contact, across 35 managers. At this sample size an r of about 0.33 clears p < .05, and 5 of the 14 pairings tested do. Teal means the two rise together, amber means one rises as the other falls. These are associations on unplanned pairings and the direction of cause is genuinely open: a manager who has had poor contact may form the belief rather than the belief shaping the contact.

The intention block, and why it is not scored

These five items were meant to form one willingness-to-invest scale and they do not. Reverse-keyed as the construct requires, Cronbach alpha is -0.29, which is worse than useless. The raw correlation matrix says why: all 10 inter-item pairs are POSITIVE, including between statements that contradict each other, and 5 of 35 managers agreed with BOTH "developing Gen Z is the better investment" and "I would rather not work closely with Gen Z". That is acquiescence, agreeing with whatever is put in front of you, so a total here would measure willingness to agree and not willingness to invest. The items are reported one by one and never summed. Wave 2 needs balanced wording and a forced-choice or behavioural format for this block.

One more instrument note: QID231 is in the survey and collected nothing from anybody, because its rows were piped fields that were never populated. Recorded here so nobody rediscovers it as a bug.

09

A third of the Gen Z arm does not think they are Gen Z

9 of 27 people in the Gen Z sample say that, setting age aside, they are a different generation. Eight say Millennial and one says Baby Boomer, on birth years between 1997 and 2002.

How to read this: Everyone was shown the generation their birth year puts them in, then asked: setting age aside, which generation do you believe you are? This is the one question in the instrument that tests the assumption the whole study rests on, that a birth-year cohort is the identity carrying the belief. Nothing had read it until today.

The meta-stereotype question asks Gen Z what employers believe about "Gen Z employees". 9 of the 27 people answering it do not consider themselves Gen Z. That is not fatal, and it is not nothing either: for a third of the arm the referent of the question is a group they place themselves outside. It belongs in the limitations of anything written up, and in wave 2 it should be a planned moderator rather than a discovery.

10

Does who the manager is change what they believe?

No. Six different splits of the 35 managers, and the widest spread any of them produces on the focal scale is 0.47 against a headline gap of 1.74. The belief is uniform across role, seniority, experience, org size and the manager's own generation.

How to read this: Everyone was shown the generation their birth year puts them in, then asked: setting age aside, which generation do you believe you are? This is the one question in the instrument that tests the assumption the whole study rests on, that a birth-year cohort is the identity carrying the belief. Nothing had read it until today.

The meta-stereotype question asks Gen Z what employers believe about "Gen Z employees". 9 of the 27 people answering it do not consider themselves Gen Z. That is not fatal, and it is not nothing either: for a third of the arm the referent of the question is a group they place themselves outside. It belongs in the limitations of anything written up, and in wave 2 it should be a planned moderator rather than a discovery.

11

Measurement problems, stated plainly

Both reverse-worded items misbehave on all three referents, and three scales fall below the 0.70 reliability threshold somewhere. The four headline gaps survive all of it.

Alpha asks whether the items in a scale move together. Below 0.70 the scale is not measuring one thing cleanly. Item-rest r asks whether one item belongs: near zero or negative means it is measuring something else.

The two reverse-worded items are the clearest problem. SD_5 ("current ability matters more than age or seniority") and MTL_4 ("not a natural leader") both come out negative or near zero on every referent, which is the signature of respondents not registering the flip rather than of a bad idea. The standard fix is to drop them and re-word, and dropping SD_5 would raise the deference alpha from 0.66 to 0.70. Nothing here changes the headline: the deference gap is 1.74 points, and no single item decision moves it enough to matter.

Also worth carrying into wave 2: motivation to lead is weak on every referent and is flagged supplementary anyway, and power distance is weak on the manager side. On sample size, only one of the three flat scales is worth chasing. Work-life segmentation is at g = 0.47 and would need about 71 per group to settle, against the 27 Gen Z and 35 managers we have. Power distance (g = -0.009) and motivation to lead (g = 0.020) are so close to zero that asking how many people would settle them is the wrong question: the formula returns tens of thousands per group, which is the arithmetic saying there is probably nothing there rather than a target to aim at.

12

The written answers, in full

Confidence and speaking up is the only theme heavily present on both sides of the manager ledger: 54% name it a strength and 29% a worry. That is a double bind rather than a criticism.

Percentages are of the people who answered that box, and each person is counted once per theme however many times they said it. "Both boxes" counts people who put the same theme in their strengths AND their worries. The coding matches vocabulary, not meaning: it cannot see negation or sarcasm and Thai compounds hide inside longer strings, so every count is a floor on its theme and never the whole of it. The word lists are in code and are meant to be argued with.

Every written answer

All 429 of them, in the Thai they were written in, grouped by the box they came from and nothing removed. This is the check on every count above: a number nobody can read behind is a number nobody can argue with.

13

Reading the written answers properly

Two analysts read all 429 written answers in full. The strongest thing they found is a mirror the scales cannot see: Gen Z claim listening and consideration as their leading strength, and across 105 things-managers-value answers, the number crediting Gen Z with any relational quality is zero.

1. Gen Z claim the exact virtue managers never once credit11 of 69 Gen Z strength answers claim listening, empathy, compromise or preserving others' feelings. On the manager side, across 105 things-we-value answers, the number crediting Gen Z with any relational or listening quality is ZERO. The three manager answers containing เข้าใจ all mean understanding technology, not understanding people. Meanwhile the manager worry list carries ไม่ค่อยมี empathy and ไม่รับฟังความคิดเห็น.The single trait Gen Z most want credit for is the one the manager side never credits and sometimes names as the deficit. This is a sharper mirror than any scale produced, and it is invisible to the instrument.count re-derived by script
2. Gen Z hear the loud accusation and not the quiet oneAsked what employers believe about them, 16 of 27 name an effort or endurance accusation. The obligation family, respect for elders, manners, loyalty, sacrifice, appears in 3 of 27 answers, in three unrelated words, never once in the reciprocity vocabulary the construct is built from. Across the WHOLE Gen Z corpus of 185 answers the words เกรงใจ, บุญคุณ and กตัญญู appear ZERO times. On the manager side the same family appears in the worry box and in 5 of 33 letters.This is the qualitative confirmation of the quantitative meta-accuracy result, and it locates the blind spot precisely. They can hear the endurance complaint. They cannot hear the obligation complaint.count re-derived by script
3. CORRECTION: they are not blind to the trait, only to the attributionTheir own challenge box contains five answers admitting exactly the manners and deference failure: มารยาทอาจไม่ค่อยมี, ทัศนคติต่อหัวหน้าที่อายุเยอะกว่า, บางทีชอบตัดสินคนอายุมากว่าเขาล้าหลัง, ชอบชักสีหน้า, พูดจาไม่ค่อยอ้อมค้อม. So they know it about themselves. Only 3 of 27 predict a manager would say it.This corrects the framing on the participant page. The failure is in predicting the other side's judgment, not in self-knowledge, and that is both more accurate and more testable. It also changes the intervention: they do not need telling what they do, they need telling that it is what you are grading.count re-derived by script
4. The meta-stereotype answers are bimodal, so the average hides half the roomOf 27 answers, 16 expect an accusation, 6 expect a wholly favourable stereotype (adaptable, smart, fast learner, tech), and 5 are mixed or neither. A mean meta-accuracy score averages a group that expects to be attacked with a group that expects to be admired.Do not present meta-accuracy as one number to the room. Roughly a fifth of these young employees think their employer already thinks well of them.count re-derived by script
5. บุญคุณ appears once in 244 manager answers, and the gap on it is the biggest in the studyThe construct with the sharpest quantitative divergence is almost silent in free text. Its single appearance is in a letter written at length: อย่าลืมคนที่ เลี้ยงดูฟูมฟักพวกเรามา อย่าละเลย อย่าดูถูก.Reading: obligation is felt but is not sayable in a survey box. That is itself a finding about the construct and an argument for the interview arm.count re-derived by script
6. Some managers teach deference as tactics without endorsing itTwo letters transmit the ritual while openly distancing from it. One: ให้ปรับตัวเข้าหาผู้ใหญ่แบบนุ่มนวล เพราะบางองค์กร ผู้ใหญ่ก็ตัดสินคนจากที่ว่าคนรอบตัวเข้าหาเขายังไง เข้าใจว่ามันเป็นระบบที่ฟังแล้วแอบเอิ่ม แต่หลายองค์กรเป็นแบบนี้จริงๆ. Another frames the wai as points scored: ไหว้ผู้ใหญ่ คือการกระทำที่น่ารัก และคุณจะได้จิตพิสัยเพิ่มแน่ ๆ.A Likert item cannot separate a manager enforcing a norm they believe from one enforcing a norm they think is faintly absurd. Directly relevant to the obligation finding, and a strong interview question for wave 2.a reading, not a count
7. Both sides used their one free message to ask the other side to change23 of 26 Gen Z letters ask the manager to change something the manager does. Of 33 manager letters, roughly 2 ask the manager side to change anything (เราเจอกันตรงกลางบ้าง and ขอแค่อย่าตัดสินกันล่วงหน้าไปก่อนทั้งสอง gen). The rest instruct Gen Z.The symmetry is the workshop. Neither side wrote about itself.a reading, not a count

What turns a praised quality into a complaint

Nine qualities appear in both the value box and the worry box: being oneself, arguing, staying in role, rights, fairness, directness, confidence, speed, and the work-life boundary. What separates the two framings is not the trait and not its intensity. In every complaint version the sentence names somebody who absorbs the cost, usually with a จน clause: มีความเป็นตัวเองสูงจนไม่สนใจวัฒนธรรมองค์กร, เคารพสิทธิตัวเองจนเห็นแก่ตัว. Every praise version stops at the individual.

This is the most directly usable thing in the whole qualitative read. It says the manager's objection is almost never to the behaviour itself, it is to an uncosted externality. That reframes the feedback conversation from 'be less confident' to 'here is who picks up what you put down', which is a sentence a manager can actually say and a young employee can actually act on.

A READING of all 429 written answers, not a computation. Every other JSON in this folder is regenerated from the response export by a script; this one is not. It was produced on 2026-08-21 by two analysts reading the employer corpus (244 answers) and the Gen Z corpus (185 answers) in full, and the counts marked 'verified' were then re-derived by script against the same data. It is stored here rather than pasted into the page so the page stays generated and the reading stays arguable. If the sample grows, this file goes stale and must be re-read: it does NOT update itself, and that is the one way it differs from everything beside it.

14

What this means for running the session

1. Gen Z claim the exact virtue managers never once credit11 of 69 Gen Z strength answers claim listening, empathy, compromise or preserving others' feelings. On the manager side, across 105 things-we-value answers, the number crediting Gen Z with any relational or listening quality is ZERO. The three manager answers containing เข้าใจ all mean understanding technology, not understanding people. Meanwhile the manager worry list carries ไม่ค่อยมี empathy and ไม่รับฟังความคิดเห็น.The single trait Gen Z most want credit for is the one the manager side never credits and sometimes names as the deficit. This is a sharper mirror than any scale produced, and it is invisible to the instrument.count re-derived by script
2. Gen Z hear the loud accusation and not the quiet oneAsked what employers believe about them, 16 of 27 name an effort or endurance accusation. The obligation family, respect for elders, manners, loyalty, sacrifice, appears in 3 of 27 answers, in three unrelated words, never once in the reciprocity vocabulary the construct is built from. Across the WHOLE Gen Z corpus of 185 answers the words เกรงใจ, บุญคุณ and กตัญญู appear ZERO times. On the manager side the same family appears in the worry box and in 5 of 33 letters.This is the qualitative confirmation of the quantitative meta-accuracy result, and it locates the blind spot precisely. They can hear the endurance complaint. They cannot hear the obligation complaint.count re-derived by script
3. CORRECTION: they are not blind to the trait, only to the attributionTheir own challenge box contains five answers admitting exactly the manners and deference failure: มารยาทอาจไม่ค่อยมี, ทัศนคติต่อหัวหน้าที่อายุเยอะกว่า, บางทีชอบตัดสินคนอายุมากว่าเขาล้าหลัง, ชอบชักสีหน้า, พูดจาไม่ค่อยอ้อมค้อม. So they know it about themselves. Only 3 of 27 predict a manager would say it.This corrects the framing on the participant page. The failure is in predicting the other side's judgment, not in self-knowledge, and that is both more accurate and more testable. It also changes the intervention: they do not need telling what they do, they need telling that it is what you are grading.count re-derived by script
4. The meta-stereotype answers are bimodal, so the average hides half the roomOf 27 answers, 16 expect an accusation, 6 expect a wholly favourable stereotype (adaptable, smart, fast learner, tech), and 5 are mixed or neither. A mean meta-accuracy score averages a group that expects to be attacked with a group that expects to be admired.Do not present meta-accuracy as one number to the room. Roughly a fifth of these young employees think their employer already thinks well of them.count re-derived by script
5. บุญคุณ appears once in 244 manager answers, and the gap on it is the biggest in the studyThe construct with the sharpest quantitative divergence is almost silent in free text. Its single appearance is in a letter written at length: อย่าลืมคนที่ เลี้ยงดูฟูมฟักพวกเรามา อย่าละเลย อย่าดูถูก.Reading: obligation is felt but is not sayable in a survey box. That is itself a finding about the construct and an argument for the interview arm.count re-derived by script
6. Some managers teach deference as tactics without endorsing itTwo letters transmit the ritual while openly distancing from it. One: ให้ปรับตัวเข้าหาผู้ใหญ่แบบนุ่มนวล เพราะบางองค์กร ผู้ใหญ่ก็ตัดสินคนจากที่ว่าคนรอบตัวเข้าหาเขายังไง เข้าใจว่ามันเป็นระบบที่ฟังแล้วแอบเอิ่ม แต่หลายองค์กรเป็นแบบนี้จริงๆ. Another frames the wai as points scored: ไหว้ผู้ใหญ่ คือการกระทำที่น่ารัก และคุณจะได้จิตพิสัยเพิ่มแน่ ๆ.A Likert item cannot separate a manager enforcing a norm they believe from one enforcing a norm they think is faintly absurd. Directly relevant to the obligation finding, and a strong interview question for wave 2.a reading, not a count
7. Both sides used their one free message to ask the other side to change23 of 26 Gen Z letters ask the manager to change something the manager does. Of 33 manager letters, roughly 2 ask the manager side to change anything (เราเจอกันตรงกลางบ้าง and ขอแค่อย่าตัดสินกันล่วงหน้าไปก่อนทั้งสอง gen). The rest instruct Gen Z.The symmetry is the workshop. Neither side wrote about itself.a reading, not a count

What turns a praised quality into a complaint

Nine qualities appear in both the value box and the worry box: being oneself, arguing, staying in role, rights, fairness, directness, confidence, speed, and the work-life boundary. What separates the two framings is not the trait and not its intensity. In every complaint version the sentence names somebody who absorbs the cost, usually with a จน clause: มีความเป็นตัวเองสูงจนไม่สนใจวัฒนธรรมองค์กร, เคารพสิทธิตัวเองจนเห็นแก่ตัว. Every praise version stops at the individual.

This is the most directly usable thing in the whole qualitative read. It says the manager's objection is almost never to the behaviour itself, it is to an uncosted externality. That reframes the feedback conversation from 'be less confident' to 'here is who picks up what you put down', which is a sentence a manager can actually say and a young employee can actually act on.

A READING of all 429 written answers, not a computation. Every other JSON in this folder is regenerated from the response export by a script; this one is not. It was produced on 2026-08-21 by two analysts reading the employer corpus (244 answers) and the Gen Z corpus (185 answers) in full, and the counts marked 'verified' were then re-derived by script against the same data. It is stored here rather than pasted into the page so the page stays generated and the reading stays arguable. If the sample grows, this file goes stale and must be re-read: it does NOT update itself, and that is the one way it differs from everything beside it.

15

What this cannot tell you

• Nobody has been interviewedBoth headline readings, the obligation finding and the wrong-accusation finding, are readings of written answers. The interview arm exists to test them and has not run.
• Neither sample is randomBoth arms volunteered. People who answer a survey about a generation gap are not a random draw from either group.
• The two arms are unmatchedThese Gen Z do not work for these managers, so nothing here is a within-organisation comparison.
• The employer arm is mostly one hospital group20 of 35. Section 07 shows the headline gaps hold across the split, which helps, but does not make the sample representative.
• Every subgroup and layer split here is exploratoryNone was specified in advance. They are a wave-2 agenda, not results.
• The qualitative coding matches words, not meaningCounts are floors on their themes. Read the verbatims in section 09 before quoting any of them.
• One flat scale needs a bigger sample, two probably need droppingWork-life segmentation would need about 71 per group to settle, against 27 Gen Z and 35 managers today. Power distance and motivation to lead sit at effect sizes of -0.009 and 0.020, which is not "not yet significant", it is "there is very likely nothing here".
• Two items need re-wording before wave 2Both reverse-scored items, on the evidence in section 08.