People Push and Pull Factors in Adopting a Crowdsourced Delivery System Transportation Research Record 2019, Vol. 2673(7) 529–540 Ó National Academy of Sciences: Transportation Research Board 2019 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/0361198119842127 journals.sagepub.com/home/trr Aymeric Punel1, Alireza Ermagun2, and Amanda Stathopoulos1 Abstract This study explores push and pull factors affecting the adoption of crowdsourced delivery or crowd-shipping systems using a structural equation method. The core data used in this study are obtained from an online survey distributed among 800 individuals in June 2016 in the United States (U.S.) Analyzing the answers of 533 respondents, the direct and indirect effects of the personal attitudes, socio-demographic, and built-environment variables on the likelihood to be a crowd-shipper are looked at. The model suggests that crowd-shipping is more likely for men, full time employed, younger respondents, and for areas of higher population density yet lower density of employment opportunities. Concerning attitudinal motivations, the results suggest that financially motivated users are less likely to become crowd-shipping users, while community orientation and viewing the platform as a platform for helping relationships has only indirect effects on use. This leads to the observation that the early adopters appear to hold quite distinct, more critical, views of the ability of the new shipping platforms to deliver affordable and community-building service. The adoption of sharing economic and collaborative systems is growing, fueled by the development of mobile connected technology and public interest in service affordability (1–6). Sharing economic and collaborative systems have attracted attention in the transportation sector with ridesharing and on-demand mobility services as common examples (7). The rise of the sharing economy has affected both the passenger and freight industries. Crowdsourced delivery or crowd-shipping is one of the on-demand mobility services that is currently growing in the freight industry. Crowd-shipping operates as online platforms where senders are matched with enrolled couriers from the crowd, sharing the empty space in their vehicle for monetary compensation. Crowdsourced delivery services are relatively new but their penetration in the parcel delivery market is growing (8). Advantages of the system, such as affordability and flexibility, attract a growing customer base (9, 10). Weaknesses, such as privacy and trust challenges, limit its diffusion. In line with the theory of diffusion of innovation, crowd-shipping is in its early stages, and current users can be considered as ‘‘early adopters’’ (11). Knowing their characteristics helps foster the growth of the system as it is maturing over time. Crowd-shipping has attracted a sizeable body of research, which, until now, has focused on the logistics and business aspects (9, 12–14). To date, few studies have analyzed acceptance and motivations for using crowd-shipping. This paper contributes to the existing body of literature by defining latent constructs and studying their direct and indirect effects on the use of crowd-shipping. The goal of the research is to identify the main factors, including personal motivations, socio-demographic and built-environment variables, which play a role in the adoption of crowdsourced delivery systems. Of particular interest is whether latent factors related to saving money and sense of community can foster the adoption of crowdshipping, in line with motivations in other sharing economy systems. Using online survey data, a confirmatory factor analysis is run to define attitudes held by the respondents to the survey. Then, a structural equation model is developed where the explanatory power of the defined factors for adopting crowd-shipping is assessed. The research continues by examining the direct versus the indirect effects generated by the broader context represented by socio-demographic information and builtenvironment variables. The results contribute to the understanding of push and pull factors in crowd-shipping 1 Department of Civil and Environmental Engineering, Northwestern University, Evanston, IL 2 Department of Civil and Environmental Engineering, Mississippi State University, Mississippi State, MS Corresponding Author: Address correspondence to Aymeric Punel: AymericPunel@u.northwestern.edu 530 use, and especially highlight the categories of people who are more likely to be early adopters. The findings can benefit the developing crowd-shipping industry by helping businesses to expand their user base. On the one hand, current users’ socio-demographic information allows companies to identify the categories of people who are more likely to enroll in the system. On the other hand, early adopters’ preferences and attitudes provide insights regarding the aspects of the service that led to adoption. Jointly, these insights would allow crowdshipping firms to adjust their marketing strategy to attract new customers and build critical mass. The rest of the paper is organized as follows. First, there is a summary of the literature on consumers’ motivations for using sharing economic systems. Second, the data is presented by introducing the sample and describing the variables used in the analysis. Then, the method employed to design the structural equation model is explained, followed by the discussion of the results. Finally, the paper is concluded by summarizing the main findings, and setting the limits of the current work that need to be addressed for future studies. Background As part of the sharing economy, crowd-shipping is likely to attract customers who have specific motivations and expectations toward the service. Most of the existing research on crowd-shipping concerns the logistics (9, 12–14). But some research also examines the behavioral aspects (15, 16). However, none of the previous work has focused on individual attitudinal motivations for using a crowdsourced delivery system. The remainder of this section reviews the literature on consumer motivations for using sharing economic systems more broadly by distinguishing studies using structural equation modeling and others. Finally, the existing literature involving survey related to crowdshipping is looked at. Several studies use structural equation modeling to understand individuals’ behavioral motivations for using sharing economy systems (5, 17, 18). Hamari et al. (5) define four motivational dimensions that they believe explain attitudes toward sharing economy and behavioral intention to use these systems: (a) sustainability, (b) enjoyment, (c) reputation, and (d) economic benefits. They test their hypotheses in a structural equation model using data from 168 users of a collaborative consumption platform in 2013. Their results show that intrinsic motivations (perceived sustainability and perceived enjoyment) do have a significant positive role in forming attitudes toward sharing economic services, while extrinsic motivations (expected gains in reputation and economic benefits) do not. Likewise, two of the four Transportation Research Record 2673(7) constructs significantly predict behavioral intention of using collaborative consumptions service: perceived enjoyment and economic benefits, which both have a positive effect. Möhlmann focuses on two specific examples, Business-to-Consumer car sharing services and the Consumer-to-Consumer accommodation marketplace, by collecting answers from 236 car2go and 187 Airbnb users (17). To understand the determinants of choosing a sharing service, he uses partial least squares path modeling analysis. The framework controls for a range of factors, either directly leading to choosing a sharing service again, or indirectly, as mediated through the choice satisfaction. His global findings are that satisfaction and likelihood for choosing a sharing economic service are essentially explained by utility, trust, cost savings, and familiarity, which appear to be determinants serving users’ self-benefit. He also finds evidence that service quality and sense of community have a positive role in the satisfaction and likelihood of choosing a ridesharing service. Airbnb is also studied in So et al.’s research looking at consumer motivations (18). They build a partial least squares path model. This model looks at the impact of motivations and constraints on the overall attitude towards Airbnb. Analysis of 519 respondents from an online survey reveals that price value, enjoyment, social interactions, and home benefits significantly contribute to positively explain the overall attitude towards Airbnb. Other research explores individual motivations for using collaborative platforms in the context of ridesharing (19–21). Amirkiaee and Evangelopoulos establish a list of 11 hypotheses based on the literature, which are categorized into two categories: (i) personal attributes such as altruism or sustainability concern, and (ii) contextual attributes such as economic benefits or trust (in rideshare) (20). They test these hypotheses in a structural equation model and assess direct effects on the intention to participate in rideshare, and indirect effects mediated through attitudes toward ridesharing. To evaluate the model, they used the answers of 300 undergraduate students from a public university via an online survey. Results suggest that personal attributes do not play a significant role in ridesharing participation intention. Indeed, a high sense of community, altruism, or environmental concerns are not associated with interest in rideshare programs. The authors do find that people who believe in mutual assistance are more likely to use ridesharing. A different story emerges for their defined contextual attributes. Trust, especially, is one of the strongest factors to explain intention to participate in ridesharing directly, and mediated through attitudes toward ridesharing. This feature is the focus of Mittendorf’s work (19). He assesses the impact of trust and familiarity with a ridesharing company and drivers, and evaluates how this trust influences the likelihood to Punel et al 531 Table 1. Summary of Studies on Crowd-Shipping Involving Survey Analysis Study Devari et al. (23) Marcucci et al. (24) Miller et al. (25) Le and Ukkusuri (26) Serafini et al. (27) Ta et al. (28) Goal of the study Study the potential benefits of people’s social network for supporting crowd-shipping activities Investigate people’s willingness to send or deliver packages via crowd-shipping Measure driver’s willingness to work as crowd-shipper Investigate crowd-shipping sender’s behavior and potential driver’s willingness to work Model people’s willingness to act as crowd-shipper using public transport Analyze effects of driver disclosure and ethnicity on crowd-shipping customers’ attitudes toward drivers and retailers Model developed in the paper Sample size Logistic Regression Model 104 Exploratory Analysis 190 Multinomial Logit Model 236 Exploratory Analysis 1,058 Multinomial Logit Model 225 Measurement Model with Confirmatory Factor Analysis 761 use the ridesharing service. To estimate the structural equation model, Mittendorf developed a survey that was disseminated in 2016 and completed by 221 participants, where he asked the respondents to answer a list of 30 statements about Uber using a 5-point Likert scale. He adds nuance to Amirkiaee and Evangelopoulos’s research as he distinguishes the trust in the company from that in its drivers (20). He finds that the degree of trust in a driver does not have an impact on the intention to use the system but the degree of trust in the company does. Wang et al. build a Technology Acceptance Model (TAM), which consists in a Confirmatory Factor Analysis and Path Analysis, to evaluate the behavioral intention to use ridesharing systems (21). They employ Davis’ work as a base model that they enhance with three innovative constructs, which are (a) personal innovativeness, (b) environmental awareness, and (c) perceived risk (22). They develop an online questionnaire distributed in Spring 2017 to ca. 600 college students from Chinese universities. Using the responses of 426 respondents, they show that personal innovativeness, environmental awareness, and perceived usefulness have a positive role in the intention to use ridesharing, while perceived risk significantly decreases the willingness to use rideshare services. Finally, several studies investigate the crowd-shipping phenomenon by designing and disseminating surveys and experiments (23–28). A summary is presented in Table 1. Devari et al. evaluate the potential benefit of customer’s social network for supporting crowd-shipping activities Population of the survey U.S. population paid to complete the survey online on Survey Monkey Students at Roma Tre University (1) Students at Northwestern University completing the survey online, and (2) online panel from Illinois from Amazon’s Mechanical Turk paid to complete the survey online Respondents from U.S. and Vietnam recruited via different platforms to complete an online survey Population from Roma reached by social media, email, and face to face interviews in metro stations Online panel from Amazon Mechanical Turk and Qualtrics, both coming from the U.S. and compensated (23). Using a binary logistic regression model, they show that participants are more willing to perform a delivery when the service requester is a friend. Marcucci et al. investigate the key elements to promote a successful development of a crowd-shipping system by focusing on the early adopters (24). They find from their exploratory analysis that 93% of the respondents are willing to use crowd-shipping service to send their packages, and that 87% are willing to act as crowd-shippers against a correct monetary compensation, although no one would be willing to stop more than 5 times along their regular journey. Miller et al. also look at people’s willingness to use crowd-shipping service, but with a focus on the driver side (25). They design a mixed logit model using results from a stated preference experiment. They estimate the median value of the willingness to work as a crowdshipper to be $19 per hour. Their findings also highlight that low and high income earners are less likely to work as crowd-shippers. Two similar studies investigating people’s willingness to act as crowd-shipper were carried out by Le and Ukkusuri, and Serafini et al. (26, 27). On the one hand, Le and Ukkusuri examine both senders’ and drivers’ behavior by developing a survey including revealed preference (RP) and stated preference (SP) questions tailored to cover both sides of a crowd-shipping system (26). Their main findings reveal a high interest in participating in a crowd-shipping system as 80% of the respondents are willing to work as crowd-shippers, essentially motivated by economic factors. On the other hand, Serafini et al. innovate by focusing on the use of public 532 transport (27). They also develop a multinomial logit model which identifies the most important features for a crowd-shipper to be remuneration, bank crediting modes, delivery booking, and location of delivery points. Finally, Ta et al. develop a measurement model to investigate the effects of driver disclosure and ethnicity on customers’ attitude towards drivers and retailers in a crowd-shipping system (28). Their results show that companies which disclose drivers’ information increase the trust and satisfaction of their customers, who are more likely to use the system again. Reviewing the literature on sharing economy and crowd-shipping helped the identification of research opportunities that are addressed in this paper. Structural equation modeling (SEM) has seen several applications in the field of sharing economy, but none of them actually treated specifically crowd-shipping. Developing an SEM would be a novel contribution in this industry and would allow the derivation of new findings and results regarding customers’ attitude and behavior. Besides, the majority of studies which investigated crowd-shipping using survey data had a small sample size, with a median of less than 200 respondents. This study benefits from a larger sample which will allow more significant results to be obtained and more insights to be drawn. To investigate individual motivations for using a ridesharing system, the literature studies a limited number of determinants, which most frequently are trust, environmental concern, economic benefits, or sense of community. The latter two are discussed in Bellotti’s paper that highlights the motivation to make social connections and self-interest motivations in sharing economic systems (29). Taking inspiration from them, this study looks at how motivation for saving money and for belonging to a community impact the likelihood of being a crowdshipping user. Further, both direct and indirect effects of various socio-demographic information and builtenvironment variables on the likelihood of being a crowd-shipping user will be tested for. Finally, in comparison with the reviewed literature, this study has the advantage of having one of the largest samples and will be able to provide relatively more accurate results. Data The core data used in this study are obtained from an online survey distributed among 800 individuals in June 2016 using Qualtrics survey software (www.qualtrics.com) and disseminated on Amazon Mechanical Turks (www.mturk.com). Participants were chosen from California, Florida, Georgia, and Illinois as these four states represent a high level of crowd-shipping use at the national level. Out of 800 requests, 587 individuals completed the survey, which indicates a 67% return rate. The Transportation Research Record 2673(7) quality of the responses was controlled using two methods: (a) adding two attention-test questions in the questionnaire and (b) checking the speed of completion and repeated logins. Screening the responses, it is found that 54 individuals failed to meet the two criteria. This resulted in 533 valid observations. The questionnaire consists of five sections preceded by an introduction of the crowd-shipping concept to participants, as the survey targets both users and non-users of the crowd-shipping system. The exact wording is quoted below: ‘‘Crowd-shipping involves using the general public to deliver packages. The idea is to match people who need to ship a package from an origin to a destination, with a driver planning to commute with the same origin and destination. This system is similar to ridesourcing and carpooling promoted by companies such as Uber and Blablacar. With some conditions, anyone can become a crowd-shipping courier in order to earn additional income by delivering packages for the participants in the crowd-shipping community. The system is generally managed by a third party (a crowd-shipping company) which provides the platform of communication (website or mobile app). Crowd-shipping is frequently less expensive than traditional shipping methods.’’ The first section asks participants about their experience with parcel and crowd-shipping industries with questions such as ‘‘Before taking this survey, were you familiar with the concept of crowd-shipping?’’ or ‘‘On average, how many packages are you sending through a crowdsourced delivery service each year?’’ The second section presents to respondents a list of nine stated preference scenarios randomly selected from a full factorial design. Each scenario presents a delivery context and asks survey participants to make a choice between three crowd-shipping alternatives, one traditional delivery service alternative, and the alternative of not sending the package at all. Further details regarding this section are available in Punel and Stathopoulos (30). The current study only accounts for the number of times each participant selects a crowd-shipping option. The third section examines respondents’ perceptions of shipping attributes, as well as opinions of crowdshipping advantages and disadvantages, respectively. The fourth section is comprised of attitude and preference statements toward crowd-shipping. This section provides two categories of statements used in this study which are highlighted as two motivations for using sharing economic system in Bellotti et al.’s work (29): Sense of Community (SC): Statements about the specific potential of crowd-shipping to build a community as well as personal belief and Punel et al 533 Table 2. Demographic Characteristics of Respondents Demographic characteristics Gender Female Male Other Age Below 34 35 to 44 45 to 64 65 and older Household annual incomeb Less than $39,999 $40,000 to $79,999 More than $80,000 Education (.25 years old) High school or less Some college or Associate’s Degree Bachelor’s Degree or higher Survey (count = 533) U.S. Census Bureau (2012)a 59.10% 40.34% 0.56% 49.13% 50.87% – 53.84% 22.33% 20.83% 3.00% 46.44% 14.19% 28.21% 11.16% 36.21% 37.53% 21.20% – – – 10.13% 35.46% 54.41% 34.15% 28.36% 37.49% Note: – = not applicable. a Data issued from the U.S. Census Bureau, Reported Internet Usage for Individuals 3 Years and Older, by Selected Characteristics: 2012. b Twenty-seven respondents did not share their household annual income. motivation for belonging to a community in general. Sense of community is defined in this paper as the membership to a group where its members interact and help each other to improve their common well-being. Monetary Saving (MS): Statements about the economic advantages of crowd-shipping and personal motivations for saving money. Each category is divided into two sub-categories: (1) a list of items about individual motivations in the context of crowd-shipping, and (2) a list of items about individuals’ motivations in general. This allows comparison of respondents’ general attitude with how they apply in the specific context of using crowd-shipping. 15 items were included: 5 items related to the sense of community in crowd-shipping systems (SCCS); 3 items related to the sense of community in general (SCGE); 3 items related to monetary saving in crowdshipping systems (MSCS); 4 items related to monetary saving in general (MSGE). These items were designed using 7-point Likert scales, where respondents were asked to rate their level of agreement from ‘‘Strongly Disagree’’ to ‘‘Strongly Agree’’. The fifth section includes socioeconomic and demographic information such as age, gender, employment status, level of education, and annual household income. Table 2 summarizes socioeconomic and demographic information of the 533 valid respondents. The results show that the sample is younger and more educated than the overall United States (U.S.) population, and females are more predominant than males by almost 20 points. Looking at the crowd-shipping experience statistics, it is found that 7.88% of the respondents have already used a crowd-shipping service, which is twice as high as Ghajargar et al. obtained from a similar study in Italy (31). Just under half, 47.84%, are familiar with the crowdshipping concept. This latter data reveals a general lack of awareness of the notion of crowd-shipping, also supported by Marcucci et al.’s findings (24). Consequently, because of their low proportion, current crowd-shipping users can be considered as ‘‘early adopters’’ of the system. To build the measurement model, the items of the fourth section were used to identify the latent factors. The structure was then enhanced by controlling for the effects of respondents’ socioeconomic information and the built-environment variables. The latter are extracted from the EPA Smart Location Database and matched with block-level indicators for survey responses. Table 3 depicts variables used in the analysis along with their summary statistics. Methodology: A Structural Equation Model To explain push and pull factors in the adoption of crowdshipping, a structural equation model is developed. The conceptual framework of the model is depicted in Figure 1. This framework aims to understand how attitudes related to sense of community and saving money, positive 534 Transportation Research Record 2673(7) Table 3. Description of the Variables Used in the Structural Equation Model Type Name Items of Factor 1 Support Platform SCCS1 SCCS2 SCCS3 Items of Factor 2 Community Items of Factor 3 Finance Respondent information and stated choice SCCS5 SCGE1 SCGE2 MSCS2 MSCS3 Gender—Male Age \ 44 HHIncome \ 19k Employment—FT Shipping Experience SP—Crowd Built-environment variables Age Median Working Pop. Density Job Access Endogenous Variable Crowd-shipping User Description Mean Std. dev. I believe crowd-shipping is a platform where its users, both senders and drivers, can help each other. I think crowd-shipping is a great way of creating a community of drivers and people looking to send packages. Crowd-shipping provides advantages for both drivers and senders. I would easily create relationships with other crowd-shipping users if I were a member of such a community. I like being part of a community. I like being active in contributing to a community. Crowd-shipping is a way of saving money. The main advantage of crowd-shipping over traditional delivery services is its attractive cost. 1: when the respondent indicates male as gender information | 0: otherwise 1: when the respondent indicates an age lower than 44 | 0: otherwise 1: when the respondent indicates an annual household income of less than $19,000 | 0: otherwise 1: when the respondent indicates he/she is full-time employed | 0: otherwise 1: when the respondent indicates he/she sends more than 1 package a month | 0: otherwise Number of times the respondent selected a crowd-shipping option in the Stated Preference scenario (from 0 to 9) Median age of the population Proportion of population in working age Gross residential density (HU/acre) on unprotected land Jobs within 45 minutes auto travel time, time-decay (network travel time) weighted (divided by 100,000) 1: when the respondent indicates he/she has already used crowdshipping systems | 0: otherwise 5.86 0.87 5.72 0.99 5.91 0.90 4.78 1.39 5.16 5.13 5.71 5.80 1.22 1.24 1.01 1.05 0.40 - 0.76 - 0.10 - 0.49 - 0.31 - 5.64 1.94 37.55 0.76 3.14 1.51 10.44 0.16 6.92 1.97 0.08 - Note: SCCS = sense of community in crowd-shipping systems; SCGE = sense of community in general; MSCS = monetary saving in crowd-shipping systems; Std. dev. = standard deviation. attitudes towards crowd-shipping, socio-demographic information, and built-environment variables impact adoption of crowd-shipping. As shown in Figure 1, Crowd-shipping User is the endogenous variable, which represents whether an individual has already used a crowd-shipping service in reality at the time of the survey. The framework tests the following hypotheses: Hypothesis H1: Viewing crowd-shipping as a support platform where users help each other increases the likelihood of being a crowd-shipper; Hypothesis H2: Having a high sense of community increases the likelihood of being a crowd-shipper; Hypothesis H3: Considering crowd-shipping’s main advantage is to save money increases the likelihood of being a crowd-shipper; Hypothesis H4: Socio-demographic and builtenvironment variables affect both directly and indirectly the likelihood of being a crowd-shipper. The remainder of this section discusses (a) the factor analysis used to create latent variables and (b) testing the conceptual framework. R version 3.4.1. (32) is used along with its packages lavaan (33) and psych (34) for the calculation and statistical analysis. Factor Analysis The factor analysis defines the factors of the measurement model, which add on to the items related to the attitudinal statements from the survey. In a preliminary step, an item analysis is performed to test each item versus the rest and remove the ones which perform poorly. A visual technique is used by plotting the Locally Weighted Scatterplot Smoother (35) plots of each item against the rest, and it is checked that the average of each item becomes higher as the remaining item scores becomes higher. This translates by a monotonically increasing line. This step leads to removing two items. To determine the appropriate number of latent factors, an Exploratory Factor Analysis is used. Several criteria are employed in tandem with critical interpretation Punel et al 535 SUPPORT PLATFORM Socio-demographic Information & Builtenvironment Variables Socio-demographic Information & Builtenvironment Variables SCCS1 Socio-demographic Information & Builtenvironment Variables SCCS2 SCCS3 COMMUNITY SCCS5 Socio-demographic Information & Builtenvironment Variables SCGE1 Crowd-shipping User SCGE2 FINANCE MSCS2 MSCS3 Figure 1. The conceptual framework of the model. for extracting and determining the content of the factors. Cattell’s scree test is applied to have a preliminary visual indication of the number of factors by looking for the natural break point in the data (36, 37). The proportion and cumulative variance of the factors is also looked at and it is checked that the proportion of variance of each factor is above 10%. Finally, it is ensured that there are at least two items with significant loadings (greater than 0.5) on each factor, and that these variables, for each factor, share a conceptual meaning which allows the factor to be defined. All the criteria suggest adopting a three-factor structure, the contents of which are presented in Table 4. The three factors are defined as follows: (a) Support Platform, which refers to an individual’s interest in the help aspect provided in crowd-shipping service, (b) Community, which refers to an individual’s sense of community and willingness of creating bonds with other members, and (3) Finance, which refers to an individual’s willingness to save money when using a crowdsourced delivery system. A Confirmatory Factor Analysis is then run to validate the structure of the model and assess its goodnessof-fit. To estimate the model, the maximum likelihood estimator is used with robust (Huber-White) standard errors. Results of the latent constructs are presented in Table 4, where the robust values are reported. Regarding the overall fit of the model, the values for CFI and TLI are greater than 0.95, however, the SRMR and RMSEA are not below 0.08 suggesting a non-optimal fit. Structural Equation Modeling The structural equation model then regresses the crowdshipping user status against various combinations of the latent factors SUPPORT PLATFORM, COMMUNITY, and FINANCE. Both direct and indirect relationships are tested and the model with the best fit and interpretability has a direct relationship between FINANCE and user status, while the other factors have an indirect association. The model is then completed by testing for the different causal effects of the sociodemographic and built-environment variables on the endogenous variable and the latent variables, and selecting the ones which are significant at the 90% confidence interval and have an intuitive explanatory power. Results of the final model are presented in Table 5, while Figure 2 presents the overall final structure of the model. To avoid cluttering, the figure items are omitted but clearly listed in Tables 3 and 4. While the SRMR value is still above 0.08, the CFI, TLI and RMSEA values outperform the cut-off values indicating a good model fit. Results and Discussion The results demonstrate that out of the three factors, only FINANCE has a significant direct effect on the crowd-shipping user likelihood. Its negative sign indicates that individuals who consider the main advantage of crowd-shipping as saving money are less likely to use 536 Transportation Research Record 2673(7) Table 4. CFA Results Estimate Std. err. SUPPORT PLATFORM SCCS1 1.000 – SCCS2 0.953 0.050 SCCS3 0.938 0.057 FINANCE MSCS2 1.091 0.055 MSCS3 0.819 0.067 Comparative Fit Index (CFI) Tucker-Lewis Index (TLI) Standardized Root Mean Square Residual (SRMR) Standardized Estimate Std. err. 0.927 0.847 0.900 COMMUNITY SCCS5 SCGE1 SCGE2 0.994 0.746 0.956 0.935 0.221 Root Mean Square Error of Approximation (RMSEA) Lower Bound 90% Confidence Interval Upper Bound 90% Confidence Interval 1.000 1.187 1.220 – 0.052 0.052 Standardized 0.673 0.918 0.925 0.103 0.084 0.124 Table 5. SEM Results Latent variables Estimate SUPPORT PLATFORM SCCS1 1.000 SCCS2 0.921 SCCS3 0.909 FINANCE MSCS2 0.750 MSCS3 0.587 Regression Estimate Factor FINANCE against SUPPORT 0.897 PLATFORM Working Pop. –0.626 Job Access. 0.051 Std. err. Standardized – 0.048 0.054 0.936 0.854 0.906 0.049 0.043 0.976 0.762 Std. Err. Standardized 0.098 0.672 0.216 0.022 –0.070 0.070 SP—Crowd 0.088 0.029 Factor SUPPORT PLATFORM against COMMUNITY 0.278 0.041 Density –0.012 0.005 SP—Crowd 0.126 0.020 Factor COMMUNITY against Age Median –0.009 0.004 Shipping 0.361 0.090 Experience Overall Fit of the Model Comparative Fit Index (CFI) Tucket-Lewis Index (TLI) Standardized Root Mean Square Residual (SRMR) 0.120 0.264 –0.077 0.228 Estimate SCCS5 SCGE1 SCGE2 Standardized COMMUNITY 1.00 – 1.151 0.050 1.166 0.050 0.677 0.922 0.917 Estimate Std. Err. Standardized 0.011 –0.109 0.048 0.102 0.024 0.029 0.087 0.175 0.041 -0.064 0.040 0.010 –0.012 0.022 0.017 0.023 0.002 0.006 0.065 -0.070 0.075 0.247 –0.086 Crowd-shipping user against FINANCE –0.021 Gender—Make Shipping Experience Age \ 44 HHIncome \ 19k Employment—FT Density Job Access. Std. err. –0.092 0.164 0.922 0.905 0.118 Root Mean Square Error of Approximation (RMSEA) Lower Bound 90% Confidence Interval Upper Bound 90% Confidence Interval the service. Regarding the result, hypothesis H3 can be rejected as considering crowd-shipping’s main advantage is to save money does not increase the likelihood of being a crowd-shipper. Instead, SUPPORT PLATFORM has an indirect effect on the use of crowd-shipping as it 0.061 0.053 0.069 positively impacts FINANCE. This means individuals who believe that crowd-shipping systems represent a support platform with mutual help are more likely to maintain the view that crowd-shipping makes shipping more affordable, thereby presenting only an indirect Punel et al 537 COMM COMMUNITY SP - Crowd Age Median Survey Results SUPPORT PLATFORM Working Pop. Density Crowd-shipping User FINANCE Job Access Built-environment Variables Shipping Experience Employment - FT HHIncome < $19k Age < 44 Gender Male Socio-demographic Variables Figure 2. Final structural equation model. Note: The items are not represented here (see Figure 1). relationship to use. That means that belief in the platform’s support aspects is mediated by motivations to derive economic benefits, when deciding to become a crowd-shipping user. The positive sign suggests a rejection of hypothesis H1 as individuals who view crowdshipping as a support platform are indirectly less likely to use such a system. Finally, people who have an elevated sense of community are more likely to accept the mutual help aspect of the crowd-shipping system, but no evidence was found that such an attitude has any direct effect on the crowd-shipping use probability. Consequently, hypothesis H2 cannot be confirmed or rejected. As far as the socio-demographic information is concerned, evidence is found of direct effects related to gender, age, employment, experience, and household income. Men are more likely to be crowd-shippers, which aligns with results from other studies related to the sharing economy (25, 38). People below the age of 44 are also more likely to use a crowd-shipping system, as well as full-time employed individuals who might prefer the convenience of the platform and do not have to adapt their work schedule for pickup and delivery timings. The positive sign for the shipping experience variable shows that people who regularly send packages using traditional channels are more likely to try a crowd-shipping system. This variable also impacts the COMMUNITY factor, revealing that experienced senders tend to have a higher sense of community. Finally, low household income has a negative effect on crowd-shipping usage. Individuals with low income are less likely to use crowd-shipping systems. The model does not reveal any indirect effects of the socio-demographic information on the crowdshipping experience. As far as the built-environment variables are concerned, density impacts crowd-shipping both positively and directly. Crowd-shipping is more likely to be adopted by people living in densely populated areas. Such environments are predisposed to the development of crowdsourced activities likely because of network effects. Further, job accessibility has both direct and indirect negative effects for the use of crowd-shipping, suggesting that areas with high job accessibility are not the ideal locations to develop crowd-shipping activities. These results support the idea that crowd-shipping preferences vary across locations and that the spatial distribution of actors and characteristics of the network impacts the performance of crowd-shipping matching system (39, 14). Furthermore, the working-age population has a negative role on the factor FINANCE, meaning that working people tend to less be attracted by the money-saving benefits of crowd-shipping systems. Consequently, this variable indirectly affects positively the use of crowdshipping. Furthermore, the density variable has a negative effect on the factor SUPPORT PLATFORM. This reveals that people living in densely populated areas are less likely to believe in the mutual help aspect of crowdshipping platforms, which may not constitute their main motivation for using a crowdsourced delivery system. Both the findings from the socio-demographic information and built-environment variables confirm hypothesis H4 which supports that being a crowd-shipper is directly and indirectly affected by the socio-demographic and built-environment variables. 538 Conclusion Crowd-shipping combines sharing economy and ondemand features to offer a new delivery solution relying on internet platforms and crowd-carriers. The system distinguishes itself from traditional delivery with its specific strengths and weaknesses, such as affordable delivery cost, convenience, and social interaction, versus privacy and liability concerns, and reliability issues, respectively. While crowd-shipping is not likely to replace established traditional logistics companies, it holds promise as a complementary shipping model owing to its ability to attract new demand. Little is currently known about the profile and motivation of users. Identifying the customer segments would help crowdshipping companies to better tailor their service to address demand more effectively. Further, results of such studies would also benefit research on the logistics and operations side of crowd-shipping, which would be able to build more realistic models informed by improved behavioral assumptions. This study attempted to gain an understanding of crowd-shipping users’ motivations by analyzing their socio-demographic characteristics and personal attitudes. Their effects on crowd-shipping acceptance were tested for through a structural equation model. This model was developed using data from 533 respondents from an online panel survey. Using the responses to attitudinal statements, three factors were defined for the measurement model: (i) interest in reciprocal helping relationships in the crowd-shipping platform, (ii) sense of community, and (iii) interest in crowd-shipping’s economic benefits. The direct and indirect effects of these factors were then tested, along with the sociodemographic and built-environment variables to assess the relationship to using crowd-shipping. The model results suggest some unexpected findings. Crowd-shipping use is more likely among respondents who do not claim to be motivated by the service affordability, and simultaneously, the affordable price is not viewed as the main advantage of crowd-shipping. Similarly, through an indirect effect, it is shown that people who view crowd-shipping as a platform where members help each other are less likely to use the system. While financial and community aspects do not seem to be the main motivators for using crowd-shipping, users do appear to be animated by other goals and preferences, such as the convenience of the system. In line with expectations, looking at the socio-demographic profile, evidence is shown of a direct relationship for the following features: men, full-time employed, individuals above 44 years of age, and experienced senders are more likely to use crowd-shipping systems, in contrast to low-income respondents. The built-environment variables of population and employment density also impact the Transportation Research Record 2673(7) development of crowd-shipping. Notably, higher population density and fewer accessible job opportunities lead to more likely crowd-shipping adoption. From a business angle, findings encourage crowd-shipping companies not to emphasize the monetary advantage of the system in their marketing campaigns but rather to emphasize its convenience for controlling pick-up and delivery conditions. Further, insights indicate that crowd-shipping businesses should focus their operations into dense residential areas to increase their activity. On the whole, the structural equation model revealed that it is challenging to find the causal structure of the motivations to use crowd-shipping. In many cases the actual users hold different, more critical, beliefs about the service benefits than non-users. This experiential change in values suggests the need to explore the adoption dynamics carefully as the user base expands. Some caveats need to be noted so as to be addressed in future studies. First, the low share of actual users in the sample, as well as the general lack of awareness of crowd-shipping, limit the generalizability of the results. Second, because the sample is small (although it remains larger than most similar studies), it is challenging to highlight all significant socio-demographic variables and motivations. Third, while several types of crowd-shipping businesses exist, the current study especially focused on the deliveries for and made by private customers (9, 40). The survey did not identify respondents who were business owners or reliant on logistics activities. Future research should explore this feature to further help tailor business strategies. Importantly, companies can have a collective value orientation with social goals, similar to the individuals in this research. The connection between company needs and supply chain, customer base, and value orientation merits further research. Finally, the present research considered only two main motivations: economic benefits and sense of community. Future studies should also consider other motivations and attitudes highlighted in the literature, such as trust, openness to innovation, and convenience. Acknowledgments This material is based upon work supported by the U.S. National Science Foundation Partnerships for Innovation: Building Innovation Capacity (PFI:BIC) Grant No. 1534138 Smart CROwdsourced Urban Delivery (CROUD) System, and by the Northwestern University Transportation Center (NUTC). Author Contributions The authors confirm contribution to the paper as follows— study, conception, and design: AE, AP, AS; data collection: AP, AS; literature review: AP; structural equation model: AE, AP; interpretation of the results: AP; draft manuscript preparation: AE, AP, AS. Punel et al 539 References 1. Cherry, C. E., and N. F. Pidgeon. Is Sharing The Solution? Exploring Public Acceptability of the Sharing Economy. Journal of Cleaner Production, Vol. 195, 2018, pp. 939–948 2. Zervas, G., D. Prosperio, and J. W. Byers. The Rise of the Sharing Economy: Estimating the Impact of Airbnb on the Hotel Industry. Journal of Marketing Research, Vol. 54, No. 5, 2017, pp. 687–705. 3. Sundararajan, A. The Sharing Economy: The End of Employment and the Rise of Crowd-Based Capitalism. MIT Press, Cambridge, MA, 2016. 4. Belk, R. You Are What You Can Access: Sharing and Collaborative Consumption Online. Journal of Business Research, Vol. 67, No. 8, 2014, pp. 1595–1600. 5. Hamari, J., M. Sjoklint, and A. Ukkonen. The Sharing Economy: Why People Participate in Collaborative Consumption. Journal of the Association for Information Science and Technology, Vol. 67, No. 9, 2016, pp. 2047–2059. 6. Buczynski, B. Sharing is Good: How to Save Money Time and Resources through Collaborative Consumption. New Society Publishers. Gabriola Island, British Columbia, Canada, 2013. 7. Cohen, B. D., and J. Kietzmann. Ride On! Mobility Business Models for the Sharing Economy. Organization & Environment, Vol. 27, No. 3, 2014, pp. 279–296. 8. Rai, H. B., S. Verlinde, J. Merckx, and C. Macharis. Crowd Logistics: An Opportunity for More Sustainable Urban Freight Transport? European Transportation Research Review, Vol. 9, No. 3, 2017, p. 39. 9. Rouges, J., and B. Montreuil. Crowdsourcing Delivery: New Interconnected Business Models to Reinvent Delivery. Proc., 1st International Physical Internet Conference, Quebec City, Canada, May 28–30, 2014, pp. 1–19. 10. Goetting, E., and W. N. Handover. Crowd-Shipping: Is Crowd-Sourced the Secret Recipe for Delivery in the Future? German Industry and Commerce Ltd/GCC, 2016. 11. Rogers, E. M. Diffusion of Innovations. The Free Press of Glencoe, New York, 1962. 12. Archetti, C., M. Savelsbergh, and M. G. Speranza. The Vehicle Routing Problem with Occasional Drivers. European Journal of Operation Research, Vol. 254, 2015, pp. 472–480. 13. Arslan, A., N. Agatz, L. G. Kroon, and R. A. Zuidwijk. Crowdsoruced Delivery – A Pickup and Delivery Problem with Ad-Hoc Drivers. ERIM Report Series Reference. Erasmus Research Institute of Management, 2016. 14. Chen, W., M. Mes., and M. Schutten. Multi-Hop DriverParcel Matching Problem with Time Windows. Flexible Services and Manufacturing Journal, Vol. 30, No. 3, 2017, pp. 1–37. 15. Ermagun, A., and A. Stathopoulos. To Bid or Not to Bid: An Empirical Study of the Supply Determinants of CrowdShipping. Transportation Research Part A: Policy and Practice, Vol. 116, 2018, pp. 468–483. 16. Punel, A., A. Ermagun, and A. Stathopoulos. Studying Determinants of Crowd-Shipping Use. Travel Behaviour and Society, Vol. 12, 2018, pp. 30–40. 17. Mohlmann, M. Collaborative Consumption: Determinants of Satisfaction and the Likelihood of using a Sharing 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. Economy Option Again. Journal of Consumer Behavior, Vol. 14, No. 3, 2015, pp. 193–207. So, K. K. F., H. Oh, and S. Min. Motivations and Constraints of Airbnb Consumers: Findings from a MixedMethods Approach. Tourism Management, Vol. 67, 2018, pp. 224–236. Mittendorf, C. The Implications of Trust in the Sharing Economy – An Empirical Analysis of Uber. Proc., 50th Hawaii International Conference on System Sciences, University of Hawaii at Manoa, Honolulu, HI, 2017. Amirkiaee, S. Y., and N. Evangelopoulos. Why Do People Rideshare? An Experimental Study. Transportation Research Part F: Traffic Psychology and Behaviour, Vol. 55, 2018, pp. 9–24. Wang, Y., S. Wang, J. Wang, J. Wei, and C. Wang. An Empirical Study of Consumer’s Intention to Use Ride-Sharing Services: Using an Extended Technology Acceptance Model. Transportation, 2018, pp. 1–19. https://doi.org/10.1007/s11116018-9893-4. Davis, F. D. Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology. MIS Quarterly, Vol. 13, 1989, pp. 319–340. Devari, A., A. G. Nikolaev, and Q. He. Crowdsourcing the Last Mile Delivery of Online Orders by Exploiting the Social Networks of Retail Store Customers. Transportaion Research Part E: Logistics and Transportation Review, Vol. 105, 2017, pp. 105–122. Marcucci, E., V. Gatta, M. Le Pira, C. S. Carrocci, and E. Pieralice. Connected Shared Mobility for Passengers and Freight: Investigating the Potential of Crowdshipping in Urban Areas. Proc., 2017 5th IEEE International Conference on Models and Technologies for Intelligent Transportation Systems (MT-ITS), Naples, Italy, IEEE, New York, 2017. Miller, J., Y. Nie, and A. Stathopoulos. Crowdsourced Urban Package Delivery: Modeling Traveler Willingness to Work as Crowdshippers. Transportation Research Record: Journal of the Transportation Research Board, 2017. 2610: 67–75. Le, T. V., and S. V. Ukkusuri. Crowd-Shipping for Last Mile Delivery: Analysis from Survey Data in Two Countries. Presented at 97th Annual Meeting of the Transportation Research Board, Washington, D.C., 2018. Serafini, S., M. Nigro, V. Gatta, and E. Marcucci. Sustainable Crowdshipping using Public Transport: A Case Study Evaluation in Rome. Transportation Research Procedia, Vol. 30, 2018, pp. 101–110. Ta, H., T. L. Esper, and A. R. Hofer. Designing Crowdsoureced Delivery Systems: The Effect of Driver Disclosure and Ethnic Similarity. Journal of Operations Management, Vol. 60, 2018, pp. 19–33. Bellotti, V., A. Ambard, D. Turner, C. Gossman, K. Demkova, and J. M. Carroll. A Muddle of Models of Motivation for using Peer-to-Peer Economy Systems. Proc., 33rd Annual ACM Conference on Human Factors in Computing Systems, Seoul, Republic of Korea, April 18–23, 2015, pp. 1085–1094. Punel, A., and A. Stathopoulos. Modeling the Acceptability of Crowdsourced Goods Deliveries: Role of Context 540 31. 32. 33. 34. 35. 36. Transportation Research Record 2673(7) and Experience Effects. Transportation Research Part E: Logistics and Transportation Review, Vol. 105, 2017, pp. 18–38. Ghajargar, M., G. Zenezini, and T. Montanaro. Home Delivery Services: Innovations and Emerging Needs. IFAC-PapersOnline, Vol. 49, 2016, pp. 1371–1376. R Core Team. R: A Language and Environment For Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, 2014. http://Rproject.org/. Accessed July 18, 2018. Rosseel, Y. The Lavaan Tutorial. Department of Data Analysis, Ghent University, 2014. Revelle, W., and M. W. Revelle. Package ‘Psych’. The Comprehensive R Archive Network, 2015. Cleveland, W. S. Robust Locally Weighted Regression and Smoothing Scatterplots. Journal of the American Statistical Association, Vol. 4, 1979, pp. 829–836. Cattell, R. B. The Scree Test for the Number of Factors. Multivariate Behavioral Research, Vol. 1, No. 2, 1966, pp. 245–276. 37. Costello, A. B., and J. W. Osborne. Best Practices in Exploratory Factor Analysis: Four Recommendations for Getting the Most from Your Analysis. Practical Assessment, Research & Evaluation, Vol. 10, 2005, pp. 1–9. 38. Anderson, D. N. Not Just a Taxi? For-Profit Ridesharing, Driver Strategies, and VMT. Transportation, Vol. 41, 2014, pp. 1099–1117. 39. Mladenow, A., C. Bauer, and C. Strauss. ‘‘Crowd Logistics’’: The Contribution of Social Crowds in Logistics Activities. International Journal of Web Information Systems, Vol. 12, No. 3, 2016, pp. 379–396. 40. McKinnon, A. Crowdshipping: A Communal Approach to Reducing Urban Traffic Levels? Working Paper, 2016. DOI: 10.13140/RG.2.2.20271.53925. The Standing Committee on Freight Transportation Planning and Logistics (AT015) peer-reviewed this paper (19-04917).