GB/T 36953.3-2018 in English

This standard constructs an evaluation system with five major criteria: safety, reliability, convenience, comfort, and economy, with 32 secondary indicators. A dual-track verification mechanism of service performance model (SP) and structural equation model (SEM) is adopted, in which:

1. Data collection specifications

The sampling design must follow the stratified random sampling principle, and the sample size of super-large stations should be ≥102/station (at a confidence level of 95%). Appendix C provides the sample size calculation formula:

n=(μ_1-α/2×σ/d)²

Where the allowable deviation d is recommended to be 0.5-1.0, and the population standard deviation σ is usually 6-8.

2. Key points of questionnaire design

Use a 5-level Likert scale (1-5 points correspond to dissatisfied-satisfied), which must pass:

• Reliability test: α coefficient> 0.7 (the total scale must reach 0.891)
• Validity test: KMO> 0.9 and Bartlett test p<0.05

Application of importance-performance matrix

According to the quadrant analysis method in Appendix F, the typical indicators are distributed as follows:

Empirical data show that the total effect of passenger expectations on satisfaction is 0.36 (direct 0.05 + indirect 0.31), indicating that basic service experience is more influential than marketing promises.

This standard introduces the structural equation model into the field of rail transit evaluation for the first time, breaking through the linear assumption of traditional satisfaction evaluation. Compared with GB/T 19038-2009, the following are added:

• Dynamic weight adjustment mechanism (Appendix A)
• Multi-level sampling plan (Clause 9.1.4)
• Strategic matrix analysis tool (Appendix F)

The practice of 32 participating units across the country shows that this standard can increase the correlation between satisfaction evaluation results and passenger complaints to 0.82.