Investigating college students’ choice of train trips for homecoming during the Spring Festival travel rush in China: results from a stated preference approach

Figures & data

Figure 1. Number of trips in the Spring Festival travel rush for the latest 4 years (unit: 100 million)

Table 1. Attributes and corresponding levels

Attributes	Description	Levels
Travel time (CT)	Travel time from origin station to destination station for a conventional train (hour).	15, 20, 25, 30
Travel time (HS)	Travel time from origin station to destination station for a high-speed train (hour).	5, 6, 7, 8
Ticket fare (CT)	Price of a ticket for the whole trip for a conventional train (RMB)	¥150, ¥200, ¥250, ¥300
Ticket fare (HS)	Price of a ticket for the whole trip for a high-speed train (RMB)	¥400, ¥500, ¥600, ¥700
Delay (CT)	Average delay for a conventional train (hour).	0 ~ 1, 1 ~ 2, 2 ~ 3, 3 ~ 4
Delay (HS)	Average delay for a high-speed train (min).	0 ~ 15, 15 ~ 30
Arrival time	The moment arriving at the destination station.	2:00 AM, 8:00 AM, 2:00 PM, 8:00 PM
Egress time	Travel time for the next trip, from destination station to home (hour).	1 ~ 3, 3 ~ 5, 5 ~ 7, 7 ~ 9
Ticket type (CT)	Ticket type for a conventional train.	Middle/bottom bunk, hard seat, upper bunk, standing room only
Ticket type (HS)	Ticket type for a high-speed train.	Second class seat, standing room only

Table 2. Descriptive statistics of respondents’ socio-demographic characteristics

Socio-demographic characteristic	Level	Percentage
Gender	Male	42.77
	Female	57.23
Age	18 ~ 25	86.13
	26 ~ 30	13.87
Expenditure (monthly)	≤1000	19.08
	1001 ~ 2000	63.29
	2001 ~ 3000	14.16
	3001 ~ 4000	2.31
	4000 ~ 5000	1.16
Family-college locations	Same	44.22
	Different	55.78
Family location	Rural	43.64
	Urban	56.36
Studying years	1	16.18
	2	22.83
	3	27.75
	4	18.21
	5	4.62
	6	5.49
	7	3.47
	8	0.29
	9	0.58
	10	0.58

Figure 2. Model structure

Table 3. Goodness of fit for the latent class nested logit models

Number of classes	Number of parameters	Convergent log-likelihood	Adj. Rho-squared	AIC	BIC
1	30	−3037.676	0.2006	6135.55	6313.13
2	56	−2821.965	0.2500	5755.93	6087.78
3	82	−2741.776	0.2641	5647.55	6113.47
4	108	−2679.226	0.2736	5574.45	6214.45

Table 4. Estimation results for latent class nested logit model with two classes

	Class 1			Class 2
	Estimate	Robust SE	Robust p	Estimate	Robust SE	Robust p
Alternative-specific attributes
Travel time (CT)	−0.0477	0.0103	0.0000 ***	−0.1065	0.0143	0.0000***
Travel time (HS)	−0.2497	0.0725	0.0006***	−0.2238	0.0372	0.0000***
Ticket fare (CT)	−0.5301	0.1379	0.0001***	−0.3505	0.1123	0.0018***
Ticket fare (HS)	−0.7089	0.1170	0.0000***	−0.3719	0.0496	0.0000***
Arrival time
2:00 AM	−0.2705	0.1035	0.0090***	−0.3063	0.0722	0.0000***
8:00 AM	0.1439	0.0781	0.0655*	0.1222	0.0605	0.0436**
2:00 PM	0.1305	0.1074	0.2244	0.1909	0.0686	0.0054***
8:00 PM	−0.0039			−0.0068
Egress time
1 ~ 3 h	0.3459	0.1015	0.0007***	0.4334	0.0727	0.0000***
3 ~ 5 h	0.1872	0.0804	0.0199**	0.2856	0.0609	0.0000***
5 ~ 7 h	−0.2034	0.0944	0.0311**	−0.1489	0.0570	0.0090***
7 ~ 9 h	−0.3297			−0.5701
Ticket type (CT)
Middle/bottom bunk	0.8240	0.1539	0.0000***	1.0617	0.1202	0.0000***
Upper bunk	0.6691	0.1701	0.0001***	0.7108	0.1359	0.0000***
Hard seat	−0.2053	0.1246	0.0996*	−0.0482	0.1281	0.7065
Standing room only	−1.2878			−1.7243
Ticket type (HS)
Second class seat	0.4763	0.1194	0.0001***	0.8440	0.0992	0.0000***
Standing room only	−0.4763			−0.8440
Socio-demographics ^a
Family location
Urban	0.3831	0.2286	0.0938*	0.3238	0.1135	0.0043***
Rural	−0.3831			−0.3238
Family-college locations
Same	−0.2698	0.1502	0.0725*	−0.2987	0.1038	0.0040***
Different	0.2698			0.2987
Studying years	0.1861	0.1416	0.1888	0.1399	0.0696	0.0445**
Alternative-specific constants
CT_1	0.0000
CT_2	0.0267	0.0650	0.6813
HS_1	1.2976	0.5071	0.0105**
HS_2	1.1210	0.5011	0.0253**
IV parameters ^b
IV_CT	0.9665	0.1145	0.2926
IV_HS	0.8319	0.0911	0.0650*
Membership constants
Class 1	−0.5841	0.1933	0.0025***
Class 2	0.0000
Initial log-likelihood	−3837.263
Convergent log-likelihood	−2830.782
Rho-squared	0.2623
Adj. Rho-squared	0.2519

***p-value < 0.01; **p-value < 0.05; *p-value < 0.1.

All categorical variables are effects coded.

^aSocio-demographic characteristics only contribute to the utilities of high-speed train trips.

^bThe null hypothesis is the estimated value equals 1.

Table 5. Willingness-to-pay for ticket type changing in conventional trains (class 1)

	Middle/bottom bunk	Upper bunk	Hard seat	Standing room only
Middle/bottom bunk	––––	−0.2922	−1.9417	−3.9838
Upper bunk	0.2922	––––	−1.6495	−3.6916
Hard seat	1.9417	1.6495	––––	−2.0421
Standing room only	3.9838	3.6916	2.0421	––––

Table 6. Willingness-to-pay for ticket type changing in conventional trains (class 2)

	Middle/bottom bunk	Upper bunk	Hard seat	Standing room only
Middle/bottom bunk	––––	−1.0011	−3.1666	−7.9486
Upper bunk	1.0011	––––	−2.1655	−6.9475
Hard seat	3.1666	2.1655	––––	−4.7820
Standing room only	7.9486	6.9475	4.7820	––––

Table 7. Willingness-to-pay for arrival time changing in conventional trains (class 1)

	2:00 AM	8:00 AM	2:00 PM	8:00 PM
2:00 AM	––––	0.7817	0.7565	0.5029
8:00 AM	−0.7817	––––	−0.0253	−0.2788
2:00 PM	−0.7565	0.0253	––––	−0.2535
8:00 PM	−0.5029	0.2788	0.2535	––––

Table 8. Willingness-to-pay for arrival time changing in high-speed trains (class 1)

	2:00 AM	8:00 AM	2:00 PM	8:00 PM
2:00 AM	––––	0.5846	0.5657	0.3761
8:00 AM	−0.5846	––––	−0.0189	−0.2085
2:00 PM	−0.5657	0.0189	––––	−0.1896
8:00 PM	−0.3761	0.2085	0.1896	––––

Table 9. Willingness-to-pay for arrival time changing in conventional trains (class 2)

	2:00 AM	8:00 AM	2:00 PM	8:00 PM
2:00 AM	––––	1.2225	1.4185	0.8545
8:00 AM	−1.2225	––––	0.1960	−0.3680
2:00 PM	−1.4185	−0.1960	––––	−0.5641
8:00 PM	−0.8545	0.3680	0.5641	––––

Table 10. Willingness-to-pay for arrival time changing in high-speed trains (class 2)

	2:00 AM	8:00 AM	2:00 PM	8:00 PM
2:00 PM	––––	1.1522	1.3369	0.8053
8:00 AM	−1.1522	––––	0.1847	−0.3469
8:00 PM	−1.3369	−0.1847	––––	−0.5316
2:00 AM	−0.8053	0.3469	0.5316	––––