# Codebook — Deidentified Data (HSSC manuscript)

**File:** `06_Data_Deidentified_reviewer.csv`  
**Rows:** 16075 respondents (one row per student, no duplicated records)  
**Provenance:** Cleaned, de-identified survey data underlying the manuscript
"Response styles and value internalisation in entrepreneurship education
evaluation: a survey of 16,075 Chinese students". No direct identifiers
(name, student ID, phone, e-mail, institution name) are present. Institution
type is released only as a 4-level tier code plus an elite/non-elite flag.

## 1. Variable dictionary

| Variable | Type | Description / Coding |
|---|---|---|
| gender | int | 1 = female, 0 = male |
| age | float | Self-reported age in years (missing n = 284; the paper imputes by median for the MNLogit in Table 4) |
| elite | int | 1 = elite institution, 0 = non-elite |
| tier_code | int | 1 = Double First-Class, 2 = Provincial Double First-Class, 3 = Regular undergraduate, 4 = Higher vocational (elite = 1 or 2) |
| region_code | int | 1 = Northeast, 2 = North China, 3 = Central China, 4 = East China, 5 = South China, 6 = Northwest, 7 = Southwest |
| origin | str | rural / urban |
| year_code | int | 1 = Year 1, 2 = Year 2, 3 = Year 3, 4 = Year 4, 5 = Master's or above |
| discipline_code | int | 1 = Humanities & social sciences (incl. econ/management), 2 = Sciences, 3 = Engineering, 4 = Medicine, 5 = Arts & sports, 6 = Other |
| satisfaction | float | Programme satisfaction (1–5) |
| climate | float | Perceived entrepreneurial climate (1–5); missing n = 1,791 (path sample N = 14,284) |
| effectiveness | float | Self-rated programme effectiveness (1–5) |
| courses | int | Number of entrepreneurship courses taken (0–3 or more) |
| value_conflict | str | sacrifice_self / social_primary / self_primary / sacrifice_social |
| dilemma | str | stop_immediately / weigh_options / legal_only / sunk_cost |
| material_code | int | 1 = Material first, 2 = Spiritual focus, 3 = Both, 4 = Unclear |
| activity_count | int | Number of entrepreneurship activities attended |
| profile | int | LPA class assignment (1–6), labels in the manuscript: 1 = full-endorsing, 2 = value-endorsing, 3 = high-value/moderate-instrumental, 4 = moderate, 5 = instrumental-leaning, 6 = disengaged |
| item_1 … item_25 | int | Raw 5-point Likert items (1–5). VR and IR items are interleaved in the questionnaire order; the exact composition is given below |
| VR_mean | float | Mean of the 14 value-orientation items (item_1, 3, 5, 6, 7, 12, 14, 15, 17, 19, 22, 23, 24, 25) |
| IR_mean | float | Mean of the 11 instrumental-orientation items (item_2, 4, 8, 9, 10, 11, 13, 16, 18, 20, 21) |
| VR_IR_diff | float | VR_mean − IR_mean |
| VR_dominant | int | 1 if VR_mean > IR_mean, else 0 |

> **Item ordering note.** The numbering `item_1 … item_25` follows the
> questionnaire order, in which VR and IR items are interleaved. The mapping
> above (VR 14 items / IR 11 items) is the one used to compute the scale
> scores reported in the manuscript; it was verified against the cleaned data
> (16,075/16,075 rows match exactly).

## 2. Key analysis correspondence

| Manuscript element | Data used |
|---|---|
| Table 1 sample characteristics | gender, age, origin, region_code, year_code, discipline_code, tier_code |
| Table 2 measurement model fit | item_1–item_25 |
| Table 3 profile means & covariates | VR_mean, IR_mean + covariates, profile |
| Table 4 multinomial logistic regression | profiles + gender, age, origin, elite, courses |
| Table 5 institution comparisons | satisfaction, climate, VR_mean, IR_mean, elite |
| Table 6 path model | satisfaction, climate, courses, VR_mean, IR_mean (path sample N = 14,284, climate non-missing) |
| Table 7 mediation & robustness | satisfaction, climate, VR_mean, IR_mean (path sample; after-exclusion N = 13,374) |
| Table 8 dose–response | courses × VR_mean / IR_mean |
| Table 9 quadrants & response-style tiers | VR_mean, IR_mean medians (4.50, 3.18); item_1–25 ceiling patterns |

## 3. Reproduction checks (run against this file)

- N = 16075
- elite n = 1407; non-elite n = 14668
- Full-endorsing respondents (all 25 items rated 5) = 1022 (6.4%)
- Respondents with VR = 5 = 2886 (18.0%); IR = 5 = 1057 (6.6%)
- VR mean = 4.35, median = 4.5;
  IR median = 3.18
- Group means (Table 5): elite VR 4.29 / non-elite 4.357;
  elite IR 3.467 / non-elite 3.318
- LPA profile sizes (Table 3): P1–P6 = [1022, 1864, 5196, 4416, 1689, 1888]

## 4. Ethics & availability

- No personally identifiable information is contained in this file; the
  original survey likewise collected no direct identifiers, and institutional
  names are replaced by tier codes.
- The complete original data (including institution-level identifiers) are
  retained by the corresponding author and available to editors/referees upon
  reasonable request, as stated in the manuscript's Data Availability statement.