With 6069 households in Linh Nam ward, the sample size reaching the 95% significance level is 375 households. It is calculated according to the formula of Yamane as follows [14]:

n = N 1 + N × e 2

where: N is the sample population.

n is the number of samples needed to be investigated to ensure representativeness.

e is the level of statistical significance (this study usede = 0.05).

Randomly select 375 households in 3 locations: Linh Nam Street, Nam Du Street, and Thuy Linh Street. Three adjacent streets all belong to Linh Nam ward.

The Binary Logistic regression function used in this study is as follows [15] [16]:

Y = ln ( P i 1 − P i ) = β 0 + β 1 b i d + β i X i

where: X is a vector of socioeconomic factors such as age, gender, education, household size, and income.

β ( i = 0 , 1 , 2 ) are the estimated parameters.

The Odds Ratio: P 1 − P .

P is the probability that the respondent agrees to pay to reduce plastic bags in daily life. On the contrary, (1−P) is the probability that respondents do not agree to pay to reduce the use of plastic bags in daily life.

The Binary Logistic regression equation is rewritten as follows:

Y = ln ( O d d s ) = β 0 + β 1 b i d + β i X i

The Mean WTP is estimated by using binary logistic model as follows [17]:

Y = ln ( O d d s ) = β 0 + β 1 b i d + β i X i

We have:

MeanWTP = β 0 + ∑ 1 n β i X i − β b i d

Based on DC1 model, 77.87% of the respondents are willing to pay to reduce plastic bags in daily life (Table 1). Moreover, the percentage of “Yes”-“No” responses indicates that the acceptance rate falls when the suggested bid gets higher. About 85.29% of people agree to pay at the bid of 50,000 VND. This ratio decreased to 78.49% at the bid of 60,000 VND. And the lowest rate is 69.57% at the proposal of 90,000 VND. This proves that the higher the bid, the more households tend to refuse to pay to reduce the use of plastic bags in daily life.

The Binary Logistic regression model was used to estimate the influence of the main parameters on the explanatory variable. The Omnibus fit test (Table 2) was performed before model estimation. The test result is Sig. < 0.05. It indicates the model’s suitability.

Table 3 describes the variables in the model. It consists of 7 independent variables and one dependent variable.

Table 4 presents the output of the logistic regression model. This model estimates the proportion of respondents from individual categories who would be

Note: total percent of “yes” was calculated by dividing the number of households agreed to pay by the total number of households surveyed.

willing to pay a fee to reduce household use of plastic bags, i.e., the odds that a variable “willing” will take on the value “Yes”.

Household size and gender of the household head were not statistically significant due to the Sig. coefficient. > 0.05. We will remove these two variables from the model.

When we remove the two variables “household size” and “gender of the household head” from the model, the β coefficients of the remaining variables do change, but not much.

We used a parametric approach to estimate the interviewed households’ mean WTP level. Specifically, this method will use the coefficients obtained from the Binary Logistic regression model in Table 5. The strength of this method is that it allows to combine and analyze the characteristics of the interviewees to calculate mean willingness to pay. The mean willingness to pay for reducing the use of plastic bags in daily life in Linh Nam ward is VND 67,221.89/month. It shows that respondents have a rather positive response to reducing plastic bags in daily life. In addition, it can be very effective to establish fundraising policies that fluctuate around willingness to pay. We need to determine how factors influence willingness to pay to develop the most effective policies. It is necessary to quantify the influencing factors.

In Table 5, the variable P has a coefficient of −0.577, which means that when the proposed price increases, the number of households that agree to reduce plastic bags will decrease. This conforms to the conclusions of Jin et al. [18] and Nasreen et al. [19]. The variable Inc has a coefficient of 3399, which proves that income is a very important variable affecting the “Yes” or “No” decision to pay for reducing the use of plastic bags. Construction policies that help reduce the use of plastic bags should be associated with policies to increase income. It will bring great benefits in ensuring social security. The Age variable has a coefficient of −1.178, showing that people in the older age group are more likely to refuse to pay for the reduction in plastic bag use. It may be because living habits and customs heavily influence their perception. We need to have appropriate policies to raise awareness of the elderly group, thereby stimulating them to reduce plastic bags in daily life. Education level is also an important variable affecting the “Yes” or “No” decision to pay for reducing the use of plastic bags in daily life. In this study, the variable Edu has a coefficient of 0.915, which means that as the level of education increases, the level of consent to pay for reducing the use of plastic bags in daily life also increases.

When giving the opinion “people should sacrifice part of their income to protect the environment”, most respondents agreed with this view. Moreover, the percentage of households who agree with the given point of view also agree to pay the proposed price to reduce the use of plastic bags in daily life is 81.36%. It is consistent with the coefficient of the Cog variable when the coefficient 1.767 has a positive sign.

To better quantify the influence of the independent variables on the dependent variable, we assume that the probability that the household agrees to pay for reducing the use of plastic bags in their primary activities is 10%. Or P₀ = 10%. P_i is the probability of willingness to pay to reduce the use of plastic bags in daily life when the variable X_i fluctuates while other variables remain constant. We have:

P i = P 0 e B i 1 − P 0 ( 1 − e B i ) = 0.1 × e B i 1 − 0.1 × ( 1 − e B i )

Through Table 6, we see that income has the greatest influence on the probability of answering “Yes” to reducing the use of plastic bags in daily life. The proposed price has the lowest impact. The ranking of influencing factors is important to help develop suitable policies for the study area.

One strong application of Binary Logistic regression is prediction. Predicting the probability of agreeing or disagreeing with reducing the use of plastic bags in daily life is necessary to estimate the effectiveness of the policy. The correct prediction for the whole model is 88%—calculated based on the Binary Logistic regression analysis results. To predict the probability of agreeing or disagreeing to pay for reducing the use of plastic bags in daily life, we use an equation as follows:

E ( Y X ) = e ln ( o d d s ) 1 + e ( o d d s )

where, E ( Y X ) is the probability that the dependent factor reaches the value 1

when the independent factors have a specific value.

ln ( o d d s ) = β 0 + β 1 G e n + β 2 A g e + β 3 E d u + β 4 I n c + β 5 N f + β 6 C o g + β 7 P

We get:

ln ( o d d s ) = − 5.933 − 0.577 P + 3.399 I n c + 0.915 E d u − 1.178 A g e + 1.767 C o g

To make this clear, we can take the following example:

Mr. A lives in Linh Nam ward. This year he is 27 years old. He graduated from high school. Currently, he has an income of 10 million VND a month. He agrees with the view that “people should sacrifice part of their income to protect the environment”. When he withdraws money for 60,000 VND/month, then the probability of answering “yes” is:

E ( Y X ) = e − 5.933 − 0.577 × 2 + 3.399 × 2 + 0.915 × 2 − 1.178 × 1 + 1.767 × 2 1 + e − 5.933 − 0.577 × 2 + 3.399 × 2 + 0.915 × 2 − 1.178 × 1 + 1.767 × 2 = 0.993638 = 99.3638 %

The probability that Mr. A agrees to pay is 99.3638%, with a correct prediction of 88%.