A synthetic two-level dataset with pupils nested within schools, generated to mimic the structure and parameters of the popular dataset from Hox (2010). It can be used to demonstrate multilevel bootstrap methods without depending on external data sources.
Format
A data frame with 2000 rows and 5 variables:
- pupil
Pupil identification number within school (integer).
- school
School identification number, 1–100 (integer).
- popular
Pupil popularity score on a 0–10 scale (numeric).
- sex
Pupil sex: 0 = boy, 1 = girl (integer).
- texp
Teacher experience in years (numeric).
Source
Simulated from parameters estimated from the popular dataset in: Hox, J. J. (2010). Multilevel analysis: Techniques and applications (2nd ed.). Routledge. Data available at https://stats.oarc.ucla.edu/stat/stata/examples/mlm_ma_hox/popular.dta.
Details
The dataset was generated by fitting a two-level random intercept model to the original popular data and simulating new data from the estimated parameters:
Fixed effects: intercept = 3.56, sex = 0.84, texp = 0.093
School-level random intercept SD: 0.69
Residual SD: 0.68
Continuous predictions were rounded to the nearest integer and clamped to the [0, 10] range to match the original Likert-type scale.