DATA ANALYSIS AND RESULTS
First, we import the data set into Eviews, and then we put factors affecting the level of spending to get the multiple regression model:
REPUTATION 4.433333
_Jarque-Bera | Jarque-Bera 4.030691 16.96804 1842651 1652367 1667382 1844544 1733520 Probability Probability 0.133274 0000207 0000100 0000288 0000240 0000099 0000172 -_ Sum Sum 410.9500 9270000 9110000 6670000 6650000 9180000 9370000 Sum $q.Dev SumSq Dev 2701744 1774140 1860193 173441073 1570833 1877840 1939873
Correlation AMOUNT INCOME |TIME_SPE | EVENTS /REPUTATION) DISTANCE | SECURITY | AMOUNT 1.000000 0895260 0.889336 -0.812786 -0.784652 0884435 0.861637
INCOM 0.895260 1000000 0.926438 -0.860532 -0.815685 0.929054 0.897142 TIME_SPE | 0.889336 0.926438 1000000 -0841425 -0.820034 0.939371 0.922443 EVENTS -0812796 -0860532 -0841425 1.000000 0918714 -0870256 -0851037 REPUTATION -0.784652 -0815685 -0820034 09187144 1000000 -0854619 -0.854726 DISTANCE | 0.884435 0.929054 0.939371 -0.870256 -0854619 1.000000 0.947582 SECURITY | 0.861637 0.897142 0922443 -0851037 -0854726 0.947582 1.000000
Covanance Analysis: Ordinary Date: 04/10/23 Time: 09:51 Sample: 1 150 Included observations: 150
SPE EVENTS REPUTATIL DISTANCE _ SECURITY
There are two main patterns in the graphs, a positive relationship between expenditure - income, time spend on spot, distance, and security, negative relationship between events - reputation That means that students who are higher in group one will have higher levels of expenditures In group one, income has the most influence with the highest slope In inverse patterns, events and reputation have a negative relationship, the more trending of tourist attractions, FMT students still choose not to participate
Based on the EViews’ result, we have the equation:
Amount of money willing to spend= B1 + B2*Income + B3*Time spend on spot + B4*Events + B5*Reputation + B6*Security + B7*Distance
Amount of money willing to spend= B1 + B2*Income + B3*Time spend on spot + B4*Events + B5*Reputation + B6*Security + B7*Distance + uj
Dependent Variable: AMOUNT_OF_MONEY_WILLING_TO_SPEND Method: Least Squares
Variable Coefficient Std Error t-Statistic Prob
INCOME 0.155064 0.041567 3.730445 0.0003 TIME_SPEND_ON_SPOT 0.110334 0.043556 2533167 00124
Adjusted R-squared 0.825096 S.D dependent var 1.346569 S.E of regression 0.563156 Akaike info criterion 1.735023 Sum squared resid 45.35170 Schwarz criterion 1.875519 Log likelihood -123.1267 Hannan-Quinn criter 1.792102
Figure 4 Estimation of the best model
Thus, we have the Sample regression function is:
AMOUNT_OF_MONEY_WILLING_TO_SPEND = 0.778313638001 + 0.15506370882*INCOME + 0.110333930083*TIME_SPEND_ON_SPOT - 0.0177369531702*EVENTS -0.00753047005522*REPUTATION + 0.0202402937306*SECURITY + 0.0520891514634*DISTANCE
= 0.778314: Regardless of other variables, the amount of money willing to spend is expected to increase 778,314 million VND per year
= 0.155064: There was a positive relationship between the amount of money willing to spend and Income If income increases by | million VND and other variables unchanged, the amount of money willing to spend is expected to increase by 0.155064 million VND
= 0.110334: There was a positive relationship between the amount of money willing to spend and time spend on spot If time spend on spot increases by | hour and other variables unchanged, the amount of money willing to spend is expected to increase by 0.110334 million VND
= -0.017737: There was a negative relationship between the amount of money willing to spend and events If events increases by 1 unit and other variables unchanged, the amount of money willing to spend is expected to decrease by 0.017737 million VND
= -0.007530: There was a negative relationship between the amount of money willing to spend and reputation If reputation increases by 1 unit and other variables unchanged, the amount of money willing to spend is expected to decrease by 0.007530 million VND
= 0.020240: There was a positive relationship between the amount of money willing to spend and security If security increases by | unit and other variables unchanged, the amount of money willing to spend is expected to increase by 0.020240 million VND
= 0.052089: There was a positive relationship between the amount of money willing to spend and distance If distance increases by | kilometer and other variables unchanged, the amount of money willing to spend is expected to increase by 0.052089 million VND
R-squared = 0.832139 is measure of “Goodness of fit’, which means that approximately 83.21% of total variation of amount of money willing to spend can be explained by the variation of six factors: Income, Time spend on spot, Events, Reputation, Security, Distance
2.3 The practical implication of this model
According to the purpose of the report, this survey will discuss factors directly influencing the payment for travel of 3rd and 4th-year FMT students at Ha Noi University After doing an analysis, which is shown in the regression below:
AMOUNT_OF MONEY _WILLING_TO_SPEND= 0.778314 +0.155064*INCOME + 0.110334*TIME SPEND _ON_SPOT - 0.017737*EVENTS -
When students pay more attention to income factors than entertainment events taking place in that tourist destination, when they care more about time on spot, distance and security factors than reputation, it clearly leads to an increase in travel spending The degree of change in the amount of money willing to pay for every 1 unit of change in the income variables is the highest
3.1 Testing the overall significance of all coefficient
Method: Using the F-test to test the overall significant test The purpose of the test is to check the general effect of all independent variables
Dependent Variable: AMOUNT_OF_MONEY_WILLING_TO_SPEND Method: Least Squares
Date: 04/08/23 Time: 19:55 Sample: 1 150 Included observations: 150
Variable Coefficient Std Error t-Statistic Prob c 0.778314 0317566 2450870 0.0155
TIME_SPEND_ON_SPOT 0.110334 0.043556 2533167 00124 EVENTS -0.017737 0.039402 -0.450151 0.6533 REPUTATION -0.007530 0.038833 -0.193919 0.8465 DISTANCE 0.052089 0.052572 0.990821 0.3234
R-squared 0.832139 Mean dependent var 2.739667 Adjusted R-squared 0.825096 S.D dependent var 1.346569 S.E of regression 0.563156 Akaike info criterion 1.735023 Sum squared resid 45.35170 Schwarz criterion 1.875519 Log likelihood -123.1267 Hannan-Quinn criter 1.792102 F-statistic 118.1492 Durbin-Watson stat 1.807402 Prob(F-statistic) 0.000000
Figure 5 Model of test overall significance
We follow the function below to conduct the test:
AMOUNT_OF_MONEY_WILLING_TO_SPEND= B+ B2*INCOME+
Bs*TIME_SPEND_ON_SPOT +B; *EVENTS + Bs *REPUTATION + B.*DISTANCE + By*SECURITY + u;
Ha: at least one variable is not equal to 0 Step 2: Test statistic:
=> F-stat= 118.1492 Step 3: Critical value: oô = 5% = 0.05
Reject Ho !1f F-statistie > Fô,k—1,n—k Step 5: Compare:
There is enough evidence to conclude that at least one variable is statistically significantly different from zero with 95% confidence level
3.2 Testing the individual partial coefficients
In the test, we follow the function below:
AMOUNT_OF_MONEY_WILLING_TO_SPEND= 8, Bằ*INCOME + Bs*TIME_SPEND_ON_SPOT +B; *EVENTS + Bs *REPUTATION + B.*DISTANCE + By*SECURITY + ui
Ho: Bi = 0 G = 2,3,4,5,6,7), Xi has no effect on the amount of money willing to spend Ha: Bi # 0, Xihas effect on the amount of money willing to spend
Step 2: Test statistic: t-statistic: tStep 3: critical value: tw2,n-k = t0.05/2,150-6 = t0.025,144 = 1.96
Step 4: Decision rule: if Test statistic > Critical value -> Reject Ho Step 5: Compare: Test statistic with critical value
Based on EViews above, we have the results in this table:
Hypothesis: | Test statistics: t= ==3.73 Ho: 8;=0 Critical value: There is enough evidence INCOME Ha: B;# 0 2mkEE0025.144= 1.96 to conclude that the
=> Reject Ho level TIME SPEND | Hypothesis: | Test statistics: There is enough evidence
ON MONEY t== to conclude that the time
Ho: B;=0 = 2.53 spend on spot variable is
Ha: BH 0 Critical value: c c t a/2n-k=t 0.025,144= 1.96 statistically significant
=> Reject Ho Hypothesis: | Test statistics: t= There is not enough
Ho: 8,=0 = -0.45 evidence to conclude that
Ha: B;# 0 Critical value: events variable is EVENTS to/2n-k e = t 0.025,144 = 1.96 e statistically significant ơ co
=> Do not reject Ho Hypothesis: | Test statistics: t= = There is not enough
Ho: Bs=0 = -0.1939 evidence to conclude that Ha: Bs# 0 Critical value: reputation variable is REPUTATION ° ° - " to/2n-k = t 0.025,144 = 1.96 statistically significant Conclusion: with a 95% of confidence
=> Do not reject Ho Hypothesis: | Test statistics: ơ There is not enough
Ho: Be=0 = 1.006 evidence to conclude that
Ha: Bz# 0 Critical value: distance variable is
Ua2nk=t0.025,144 1.96 statistically significant Conclusion: with a 95% of confidence
=> do not reject Ho SECURITY | Hypothesis: | Test statistics: There is not enough
Ho: B0 = 0.4666 evidence to conclude that
Ha: B;# 0 Critical value: security variable is tỄw/2n-k= 0025,144= 1.96 statistically significant Conclusion: with a 95% of confidence
F%.05.7-3,150-7 Step 5: Compare: F-statistics = 1.1285 < F°o05,7.3,150.7= 2.435
=> Do not reject Ho Step 6: Conclusion: There is not enough evidence to keep the EVENTS, REPUTATION, DISTANCE, SECURITY variables in the model
3.3.2 Comparison between the original model and the model after dropping From the results that we have tested from the two models, we witness that the adjusted R- squared before and after removing the EVENTS, REPUTATION, DISTANCE, SECURITY variables is different compared to the original model, which are respectively 83.21% and 82.68% That means, R-squared has decreased 0.53% when we drop those four variables
Nevertheless, the change in adjusted R-squared is not enough to define which model is fitted For this reason, we would look into the F-statistics and t-statistics of the two models
The F-statistics of the dropping model are approximately twice as much as larger than the old one Hence, there was a considerable increase in the t-statistics from the model after we decided to drop the insignificant variables, all of the t-statistics are bigger than 2 As the results, we could conclude that the dropping variable model is the best model
Now, we would use this model for the following tests:
AMOUNT_OF_MONEY_WILLING_TO_SPEND= Đ, ô.B:*INCOME +
B›*TIME_SPEND_ON_SPOT
For our research, we use the dummy variable to see if sanitary is a factor affecting the willingness of FMT students to pay for travel After adding a dummy variable, we have the following model below:
AMOUNT_OF_MONEY_WILLING_TO_SPEND= 8+ Bi°*D1 + B:*INCOME + B›*
TIME SPEND _ON_SPOT
CHECKING ERRORS IN THE MODEL
Multicollinearity is a statistical concept where several independent variables in a model are correlated Multicollinearity happens when there are high intercorrelations between two or more
21 independent variables in a multiple regression model In general, multicollinearity can cause broader confidence intervals, which might result in probabilities that are less trustworthy when predicting the impact of independent variables in a model There are four popular techniques in identifying multicollinearity errors among independent variables are the R-squared, the correlation matrix, the variance inflation factor (VIF), and auxiliary regressions
From the previous outcome from the best fit model that we have given above The R- squared is quite high, which is around 82.68% while the independent variables are bigger than 0.05 This indicates that there is probably no multicollinearity in the test that we are running
When rj 1s bigger than 0.8, the model could be enormously correlated which means there is a high chance of multicollinearity From the table above, as we could see, there are rjj in the table that are greater than 0.8 So the chance of having multicollinearity is possible in our model
A variance inflation factor (VIF) provides a measure of multicollinearity among the independent variables in a multiple regression model A large VIF on an independent variable denotes a highly collinear link to the other variables, which should be taken into account or accounted for in the model's structure and independent variable selection If VIF > 10 then there is definitely multicollinearity
Variance Inflation Factors Date: 05/01/23 Time: 12:01 Sample: 1 150 Included observations: 150
Variable Variance VIF VIF c 0009027 4.254362 NA
TIME_SPEND _
SUMMARY, CONCLUSION AND RECOMMENDATIONS
The Covid-19 pandemic wreaks havoc on all sectors of the world, and travel is one of the most detrimental fields and the trend for travel is changed To specify all the components that have a link to travel decisions for Hanoi University students, namely: (1) Income, (2) Time spend on the spot, (3) events, (4) reputation, (5) distance, and (6) securities
We find that all the independent variables are significant to the expenditure on travel
However, the changes and trends in each variable are different, each variable shows a different influence on the dependent variable Income, time spend on the spot, distance, and security have a positive relationship with money spent; and income has the most influential factors (highest coefficient) Events and reputation have a negative impact; events have the lowest coefficient
The paper shed light on how various information links to student travel expenditure Our results give out the trend in travel of the young after Covid, this can be used in travel companies to have the most optimal price for the young
From the results, we suggest the travel agents who aim at the students should care more about income, distance and securities in the tourist attraction and arrange the events that can satisfy the demand of the young For any future research, we suggest to use more independent variables such as customer services, online payments, wifi service, and cuisine Other indexes can be used are: travel and tourism development index, readiness for information and communication technology index, price competition index, and environmental sustainability index
" —- - ˆ - ỉ œ >ơ1 — 8G hũ C hở Íỉ 6 Ccn anew ` aan m kh! đc mo E tà cà c 9 CN hộ G Bộ Nể 6 02 CĐ n9 cú Bộ PỢ CƠ ở GẦN cỏ CƠ CC 0 AM Cl 4D 0U GD N GD ÚU (2 4< TC cỏ ƠI GUỐN ca (2 BÍ ĐC 6 BÍ SỐ SẠC Bộ Cỏ ki G ca Bộ G9 — GG C9 ca Mở G0 CO cỏ C08 Bộ Cú c Đở ƠI hộ G08 sử = mở Rở Gà ở AI bà sỉ AI Cễ NI G0 Bỡ PỔ cà Nể C Gà PỔ (0 0D cị cụ Nể CÀ N07 (ễ ch TẢ m 0 Ca hà bờ G7 R3 ƠI ^I @0 Bỡ Aở G2 nở CC Œ TỔ nở ca G8 Bộ cỏ Q0 CC G G2 Tỏ 2 CO CO C9 NO Cễ Cể C Q6 GP Sa P2 CƠ N ở Ơ? BỘ Ơ) é
APPENDIX n G G N ơn Œ@ Có Bi € @ — d6 (6 tỳ ( hò (Õ bẻ Gò TH Gò Ơn sở GÒ Cò dò G0 Co @ cò ss woneod epudima Evens Time ependon scet Cistence
‘Repuistion Everts Time spond on spot Distance Soourity
C3 ca I Œ CC 00c 02 C2 G0 S4 02 TỔ Tỉ l Gở ƠI Si TỔ cá ƠI SỈ G0 G2 DĐ #3 0 02 G2 ~N ƠI 2 ƠI Gò Sĩ O
Swernowwscvo Si ni eowin ReweuwoSnrwwun
Sanitary W Yes Amount of money wil Durmy Sanitary
3 ee ee ee — —lel~|~ el.|_|_ '~ elel_|_ _ le le l_ e ~ |_ le |_ — le |_ |_ ~ ơˆˆˆˆ ˆ ˆ ,Qa ẽốÊố ố 7 76 6 ŠốốÊố 7 ` ` 7
NON RK (ễ Œ Cú Gũ Cú Œ SỔ Gv Độ mm +4 bũ Rồi cỏ cỏ Cễ HC ể2 (ễ Nể ở (O ca 4 NO N02 cỏ Nể SỈ G2 G2 Q0 ở GJ cỏ CŨ NG c c GJ Cễ 4Œ CC (0 T G0 G ca Œ Cễ U — 0ề ca Ss Ẻ DNSOOONWON - Ca Œ PỞ 6 (0.00 hị PẢ Nỡ bà cụ Đ ca Cú Gỡ G2 (Đ 0D RJ 07 hà SỈ bà Nể ca C2 con Ê Cễ BƠI G2 0S Sa 0 00 009 6 h G63 dđN 3œ (co t6) (3 d k2 c6 0 Go Œ C ệ — Go hà nộ Go Cú G0 0 0 ĐỔ hũ c0 0G (2 6) si 0 hũ cú Ổ 0 @
HOn On COV seen Onn nnn wenn nnn sO inosine ho in CÔ