publisher colophon
abstract

This paper uses student survey data from a regional university in the southeastern U.S. to gauge the impact of a variety of socio-demographic and personal preference variables to determine which factors are most significant on the demand side of the market, and specifically on the willingness to pay for online classes. While gender was not a significant factor in willingness to pay, students in the bachelor program were willing to pay more for online classes than the students in the associate program. We also found that willingness to pay for online classes increased with work hours, yet decreased with commuting time. Given the student responses on willingness to pay, we derived the demand curve and the marginal revenue curve for online classes to find that a revenue-maximizing price premium for an online course is approximately $150 per course, which is lower than the average amount ($226.3) that a student is willing to pay. We also find that a student expectation of a better grade is the most distinctive factor in the choice of an online course.

introduction

Online learning has shown a substantial increase in postsecondary institutions and has become an indispensable element of college education. According to the National Center for Education Statistics, in the fall of 2015, almost 6 million students, which represents 29.8% of total college enrollment, were enrolled in some form of distance education courses at degree-granting institutions. Out of these 6 million students, 3.1 million students were enrolled exclusively in distance learning courses, while the other 2.9 million students were taking at least one course as a distance learning. In particular, college enrollment in distance education courses increased by 3.5% for the period 2014-2015; meanwhile, college [End Page 232] enrollment in non-distance education courses declined by 3% for the same period. Allen and Seaman (2017), using the Babson Survey, showed the same pattern of an increase in online courses; for the past decade, online course offerings have outgrown total course offerings.

The substantial increase in online enrollment is credited to both the demand for and supply of course delivery. On the demand side, online course delivery provides additional flexibility to students, who, in particular, can have a variety of restrictions, e.g., employment, family responsibilities, disabilities, or time commitments associated with extracurricular activities (Jaggars et al., 2013). On the supply side, online course delivery frees up institutions from some restrictions, i.e., capacity constraints, and helps generate additional revenue for institutions. The potential for additional revenue is more relevant given that public funding for higher education has decreased over the past decades (Ehrenberg, 2012; Turner, 2006). Given declining state and federal funding, online course offerings represent one way to overcome any physical space constraints that limit the supply of face-to-face courses (Fink, 2017; Sturgeon, 2007). Another option is to increase revenues through an increase in tuition and fees (Oliff et al., 2013; Ehrenberg, 2012).

While many studies focus on the reason why online course delivery has increased, only a few studies have focused on how much a student is willing to pay for the online course and why a student is willing to pay that amount. Byrd et al. (2015), utilizing data from 2004 to 2009 for public universities in the southeastern region of the U.S., showed that the demand for online classes is price elastic (ranging from -4.43 to -4.47) while demand for face-to-face classes is price inelastic (ranging from -0.33 to -0.35). Han et al. (2018), collecting data from 22 public colleges and universities in Georgia, confirmed that the demand for online classes is price elastic, although the magnitude of elasticity was smaller than the estimates from Byrd et al. (2015). Han et al. (2018) also noted that online credit hours increased 1.0 to 1.3 times more than overall student enrollment by showing that online courses and traditional courses are substitutes, rather than complements.

Despite the studies on the relationship between tuition and credit hours taken online, we lack an understanding of any individual characteristics in the relationship between tuition and enrollment (in terms of credit hours). That is, as the previous studies relied on institutional and aggregate-level socioeconomic data, i.e., tuition rates, enrollments, faculty-student ratios, state funding, average household incomes, etc., we do not know how individual characteristics affect the choice of online classes with the determination of a student's willingness to pay for an online class.

In this study, with financial support from the University of North Georgia, we [End Page 233] survey students using a questionnaire to address the following questions:

  1. 1). How much, on average, is a student willing to pay for an online class?

  2. 2). What factors play a role in a student's willingness to pay for an online class and how much does each factor affect willingness to pay?

  3. 3). What does the demand and the marginal revenue for online classes look like and what are the policy implications in regard to price setting of an online course?

  4. 4). What are the characteristics of online course takers?

  5. 5). What factors affect students' decisions to take online classes and how influential are those factors?

In the sections that follow, previous research related to our current project will be summarized. Then, we will explain how our survey was created, and data was collected. We will show the descriptive characteristics of all respondents, and we will employ various statistical analysis including regression analysis to address the above questions.

literature review

Review of Factors Affecting Student Success and Satisfaction with Online Education

The first line of study related to online education is to ask whether students are satisfied with online delivery methods, or, in other words, whether an online course is worth more to students. Studies of this nature, performed by consulting companies, have yielded mixed results. Noel-Levitz (2014), in its online survey of 122,403 students from 117 institutions from 2011 through 2014, assessed student satisfaction of online classes and found that students showed relatively high satisfaction. Aslanian and Clinefelter (2013) similarly found a high level of student satisfaction. Based upon their survey of 1,500 students recently enrolled, currently enrolled, or planning to enroll in a fully online program, Aslanian and Clinefelter (2013) noted that approximately two-thirds of respondents agreed that their online program was a good financial investment while three-quarters agreed that it was a good investment of their time. In contrast to these online students, community college students and some employers, surveyed by Public Agenda (2013), remained skeptical about the value of online learning. Pubic Agenda (2013) showed that some employers would prefer a job applicant with a traditional degree from an average school over one with an online degree from a top university.

Given the mixed results on the value of an online program, and student satisfaction with online courses, another line of study explored what makes online education successful and reputable. Student commitment, family background, quality of a program, school reputation, self-efficacy, prior online experience, and [End Page 234] interactions are listed as the factors affecting student success in online education. Illustrated below are some studies on these factors.

Regarding success in online education, student commitment is most important, as the literature of marketing has long recognized (Kotler and Fox, 1995; Beck and Milligan, 2014). From their survey of 831 students enrolled in online courses at a southeastern university during the 2007-2008 academic year, Beck and Milligan (2014) found that student commitment would lead to higher satisfaction and higher grades, and that student commitment is affected by factors such as family background, quality of academics and school reputation, and even age. Dehghan et al. (2013) reached similar conclusions for the master's level of online education.

Similar to student commitment, self-efficacy, an ability to produce a desired or intended result, has been identified as a key component in successful online learning. Lee and Choi (2011) observed that the higher drop-out rate in online learning than in face-to-face learning was related in part to a lack of self-efficacy. Shen et al. (2013) further demonstrated the multi-dimensions of self-efficacy that are required to complete a course, to interact socially and academically with classmates and instructors in online classes, and to handle tools such as a computer.

The importance of interactions in online classes has been noted by many studies. Regarding effects on online course success, Mark et al. (2005) found that any course-related interactions, i.e., instructor-student, student-student, and student-content interaction, were more important than outside factors, e.g., work and family flexibility and antecedent personal characteristics such as experience and gender. Mark et al. (2005) also argued that, among all three types of course-related interactions, instructor-student interaction was most important. In contrast, Bernard et al. (2009) in their meta-analysis of the experimental literature of distance education showed that all three types of interactions, student-student, student-instructor, and student-content interactions, were strongly associated with increasing achievement outcomes in distance education.

Prior experience in online classes was also identified as a crucial element to success in an online environment. Volery and Lord (2000) identified prior experience in online classes, together with technology and instructor, as critical factors in online class results. Similarly, Hachey et al. (2015) also found that prior online experience and GPA had significantly positive impacts on the successful completion of online courses. Hachey et al. (2015) particularly showed that students who had withdrawn or earned a grade of D or F in prior online courses had significantly lower rates of successful STEM course completion. Murphy and Stewart (2017), over an eight-year observation period, focused on on-campus students taking online courses and determined that unsuccessful course completion was associated with higher rates of students repeating the class and with [End Page 235] early disengagement.

In addition to the factors of successful online education mentioned above, Jaggers (2013) summarized some factors affecting the choice of online classes and face-to-face classes. From interviews with online faculty, staff, and 47 students taking online classes, Jaggers (2013) found that flexibility and convenience were key reasons to take online courses. As identified by Aslanian and Clinefelter (2013) and Noel-Levitz (2014), some students preferred the online learning environment, e.g., reduced travel time and efficient use of their time, lack of interaction with other students. Jaggers (2013) also found that a general reason to take face-to-face classes was to maintain a connection to the campus and their peers and a stronger student-instructor connection.

Review on Willingness-to-Pay

Numerous approaches to measure willingness-to-pay have been proposed and applied in the literature within the fields of economics and marketing. To illustrate different approaches, Breidert et al. (2006) hierarchically classified them into two categories: data approach or survey approach. Data approach has two distinctive methods: analysis of market data and experiments (e.g., lab experiment, field experiment and auction). Survey approach also has two distinctive methods: direct survey (e.g., expert judgement and consumer surveys) and indirect surveys (e.g., conjoint analysis and discrete choice analysis). Each approach has its own merits and obstacles, and there is no single dominant approach. The merits and obstacles of each approach are compared below.

An analysis of market data provides estimates of price-quantity relationships by using market data on sales, based on the assumption that historic data can be used to predict current and future demand. While this approach provides quantitative estimates on "real" purchases, not "intentions" of the decision makers, the approach is only valid when data are available and past consumer behavior does not change in the future (Nagle and Holden, 2002). Related to their work on online education, Han et al. (2018), using institutional level data on online enrollment and tuition rates, estimated the price elasticity of demand for online education in 22 public universities and colleges to show that a 10% increase in online tuition could reduce online credit hours by 30-36% when traditional tuition price was controlled.

Experiments are used to simulate a consumer's purchase behavior in the artificial setup (lab experiment) or in the real-world shopping environment (field experiment) to determine willingness to pay. One great merit of this approach is to give decision-makers a systematic variation of the prices, while obstacles include any possible bias from the artificial setup (Nessim and Dodge, 1995) of lab experiments or higher (implicit and explicit) expenses for the field experiment. [End Page 236]

Direct survey includes the polling of experts or potential consumers about their willingness to pay. The sales estimates, judged by experts, e.g., sales or marketing managers, are more time and cost efficient than those obtained from consumer surveys, but the estimates can also be a poor measurement with low validity (Balderjahn 2003). Consumer surveys have become popular as the willingness to pay can be elicited by directly asking the individuals who will be making the decision on purchases. However, there are challenges associated with consumer surveys. A major limitation is that consumers do not necessarily reveal their true willingness to pay for a product (Neissim and Dodge, 1995).

Indirect surveys allow consumers to select their choice from among the different combinations of quality (or product) and price. Conjoint analysis, first introduced into marketing by Green and Rao (1971), presents the different combinations that are built upon product-attributes and the worth of the attributes. While such conjoint analysis helps develop a product by combining different worth of different attributes, it is often criticized that the calculation of the worth of the attribute is heuristic. To avoid the heuristic combination of product-attributes, McFadden (1985) developed a method to make consumers choose between alternative product profiles. Based upon a consumer choice, a latent preference at the aggregate level is estimated in accordance with utility theory.

survey and sampling

A survey was administered to students currently enrolled in a regional university in the southeastern U.S. to assess their willingness to pay for online classes and the determinants of their choice of online class. The survey collected information on student sociodemographic characteristics, the burden of educational expenses, perceived factors affecting a student's choice of online class, and a student's willingness to pay for online classes. After a pilot study to assess quality, a structured questionnaire was developed to collect this information.

Surveys were distributed at all five campuses of the university. Two sections of an economics class, principles of macroeconomics (ECON 2105), were selected for each campus except for one smaller campus. One history class and a mathematics class were selected for that campus, as economics classes were not offered. By choosing the same economics class, we attempted to avoid the possibility that the same student could respond twice. The surveys were administered in classrooms from September to October 2018.

A total of 243 respondents were collected after dual-enrollment students (high school students taking classes for both high school and college credit) were excluded. Dual-enrollment students were excluded as they are not only minors but they are also not responsible for tuition and fees, and thus we believe that they were not able to realistically answer questions related to tuition and costs. Survey [End Page 237] respondents were diversely distributed across campuses and academic status when measured by credit hours taken. Across campuses, our sample had 62 respondents from the first campus, 61 from the second, 58 from the third, 34 from the fourth, and 28 from the fifth. For academic status, the sample had 124 sophomores (30-59 hours) and 80 freshmen (0-29 hours), followed by 36 juniors (60-89 hours) and three seniors (90 hours and more).

Male respondents (64.6%) slightly outnumbered female respondents (35.0%). The sample was predominantly single (96.7%), with only six respondents who were married or living in a domestic partnership. On degree level, 66.5% of respondents were enrolled in the bachelor program, while 33.5% were enrolled in the associate program.

socioeconomic characteristics of respondents

Socioeconomic status can be an important factor in many of a student's choices. For this reason, information on commuting time, work hours, and household income was compiled as we a priori think these factors should be important to the choice of online classes. On the commuting time to campus from a current residence, 94.1% of respondents reported spending less than 50 minutes traveling one way. One-half of respondents lived within a 15-minute commute, while only a few respondents (3%) lived one hour or more away from their campus. Regarding employment, 27.8% of respondents said that they do not work, thus roughly three-quarters of respondents worked. Of the respondents, 33.6% of reported working for 15-25 hours a week, while 9.5% of respondents worked for more than 35 hours a week. When asked about income (parents' income was not included), 61.4% of respondents made less than $1,000 a month, and 13.9% of respondents did not have any earned income.

Educational expenses are important to enrollment decisions. For this reason, we asked students what percentage of educational expenses were paid by family, by loans, by aid, etc. In particular, we emphasized to students that the sum of all percentages must be 100%. Based upon student responses, 78.7% of respondents reported that their educational expenses were paid, to some degree, by their families, of which 22.6% of respondents said that all educational expenses were paid by their families. Only 21.3% of respondents did not depend at all on their family for educational expenses. Among all respondents, 63.2% received some form of financial aid, of which 11.7% of respondents said that their educational expenses were paid fully by financial aid, although about one-third of respondents did not have any aid. About three-quarters of respondents did not use student loans at all, while 27.2% of respondents reported that they use student loans to a varying degree, including 3.8% said that they depend fully upon student loans. [End Page 238]

characteristics and perception of online course takers

Descriptive Characteristics

In the survey, we asked students about the number of (fully) online courses that they have taken. The average number of online courses taken by respondents was 0.89, which is less than one course. We dissected this average number by gender, marital status, and degree level. As shown in Table 1, on average females took more online courses than males, as the average number of online courses taken by female respondents was 1.0, while that of online courses taken by male respondents was 0.81. A similar pattern was noted by Noel-Levitz (2006). On average, married students took twice as many online courses as unmarried students. That is, married students, on average, took 1.87 courses, while unmarried students took only 0.86 courses. However, this finding is based on just six married respondents. Students enrolled in the bachelor program took more online courses than students in the associate program. That is, the average number of online classes taken by students in associate programs was 0.76, while the average number by students in bachelor programs was 0.96.

Given these differences by subgroups in each demographic category, we performed an F test on the equal mean and the non-parametric rank sum test (Wilcoxon, 1945; Mann and Whitney, 1947) on the equal median to see if the difference was statistically significant. It turned out that any differences in the number of online courses taken based upon gender, marital status, and degree level were statistically insignificant. In testing the null hypotheses that the means and the medians of the subgroups were the same, all F statistics and z statistics were small enough to fall inside the critical values.

Of the socioeconomic characteristics of online course takers, only household income was statistically significant. We found no consistent pattern of online class taking over commuting time; the number of online courses seems to increase with longer commuting time in the beginning and then fluctuates before decreasing. The same finding held true for work hours, implying that a student's work hours do not influence the number of online courses taken. The number of online courses taken over the household income revealed a negative relationship between household income and the number of online classes. Respondents tended to take fewer online courses as their household income rises.

student perception of online classes

Based upon the literature review, and our own experiences and anecdotes, we selected nine factors that students might think important to their choice of online class: grade expectation, convenience of online class, unavailability of a face-to-face class, interactions with other students and an instructor, technical challenges, burden of expense, concerns on learning materials, etc. Students were asked [End Page 239]

Table 1. Average Number of Online Courses Taken
Click for larger view
View full resolution
Table 1.

Average Number of Online Courses Taken

to rank the importance of each of the factors in their choice of online class. A 5-point Likert scale was employed to measure the importance of each factor in a student's perception, ranging from 'very important' to 'not at all important.' Table 2 below summarizes the results of the survey on the perceived factors.

Four factors might positively affect the choice to take an online course. Most respondents (i.e., 90.9%) reported that an expectation of a better grade is important or somewhat important at the very least, of which one-half of respondents think that a better grade is important (25.6%) or very important (26.9%). Only 9.1% of the respondents reported that grade concerns are not important at all. The second question was if an online class is more convenient for schedule and/or location. Compared to the expectation of a better grade, more respondents (95.5%) considered convenience important or somewhat important. A third of respondents considered convenience important (37.0%) or very important (37.9%). Only 4.5% of respondents indicated convenience as not important at all.

The third question was if students chose an online course when a face-to-face class was not available. The responses to this question showed a similar pattern to the expectation of a better grade and convenience in scheduling. The last question asked if a student chooses an online class because he or she does not like interacting in-person with other students. More than one-half of respondents (i.e., 66.7%) declined this factor. Only a few said that they chose to take an online class for this reason.

Given the responses to the above questions, students perceived that the expectation of a better grade, convenience in scheduling, and the unavailability of a face-to-face class mattered to their choice of online class. Particularly, our survey shows the importance of convenience in scheduling as found in previous studies (e.g., Jaggers, 2013; Clinefelter, 2013). Meanwhile, interaction with students seemed to be less important to their choice.

The next set of questions asked students whether several factors were influential [End Page 240]

Table 2. Student Perception on Online Class
Click for larger view
View full resolution
Table 2.

Student Perception on Online Class

or not in choosing to take an online class. The factors included technical challenges, less interaction with an instructor, a burden of expense for an online class, concerns on learning, and deadline issues. The first question was if technology associated with an online class is too challenging. In response, 43.8% of those surveyed said that technical challenges were not an important factor at all, while 56.2% of respondents admitted that technical challenges were important or somewhat important in relation to avoiding an online class. However, only 8.7% of respondents thought technical challenges were very important to avoiding an online class. The second question was if insufficient interaction between instructor and student deters students from taking an online class. Almost 90% of respondents considered insufficient interaction a deterring factor to taking an online class, of which one-half of respondents thought this deterring factor important (25.2%) or very important (26.0%). Only 10.7% of the respondents thought that insufficient interaction has no importance in their choice of taking an online class.

The third question asked about online classes being more expensive than face-to-face classes. Almost 90% of respondents considered the higher expenses of taking online classes an enrollment deterrent, of which 20.8% and 27.4% of respondents answered that the higher cost is important and very important, respectively, in a decision to not take an online class. The fourth question was whether a student does not learn well in an online class. In a similar response to the cost burden, 84.4% of the respondents considered the concerns related to learning important or somewhat important in any choice to avoiding enrolling in an online class. Only 16.6% of the respondents thought that the concerns related to learning did not factor into their choice of taking an online class. The last question asked if a student has trouble keeping up with deadlines in an online [End Page 241] class. Among all respondents, 25.2% said that the deadline issue does not matter, while 74.8% of the respondents said that the issue matters to some degree, of which only 14.5% of the respondents said that the deadline issue is very important.

Given the results, insufficient interaction with an instructor, higher cost of taking online classes, and concerns related to learning seemed to be important factors that deterred students from taking online classes. These survey results supported the findings of Mark et al. (2005) that instructor-student interaction is the most important of all types of interactions. It also supported the findings of Bernard et al. (2009) that interactions are strongly associated with learning outcomes in online classes. On the contrary, technical challenges and deadline issues seemed to be less important factors.

willingness to pay for an online class

Student Preference for an Online Class

In the survey, we asked students about their preference between online classes and traditional classes. To find student preference regarding delivery format of a class, we asked students to choose whether they were willing to pay more for an online course or more for a face-to-face course. For the former question, students were provided three choices with varying price differences, e.g., an increase by $50; paying more for an online class, paying more for a traditional face-to-face class, and no difference between an online class and a traditional class. Almost 75% of respondents did not want to pay more for an online class, and 35.5% of respondents wanted to have no differential tuition between an online class and a traditional face-to-face class. Only 20.7% of respondents said that they are willing to pay more for an online class, of which one-half of them chose to pay less than $50 more for an online course.

Maximum Willingness to Pay (WTP) for Online Class

Given the student preference of a traditional face-to-face class over an online class, we asked students about the maximum amount of additional money that they are willing to pay for an online course over a face-to-face course. The average additional amount that a student would pay for an online class over a face-to-face class was $226.30 per course. Specifically, about one-half of the respondents would pay less than the average amount ($226.30 per an online course). The modal amount falls in the range of $100 to $200, with only a few respondents willing to pay over $500 more for an online course.

The maximum willingness to pay for an online course was further dissected by gender and degree program. As Table 3 shows, on average, male students reported being more willing to pay more than female students would pay for an online [End Page 242]

Table 3. Willingness to Pay by Gender and Degree Program
Click for larger view
View full resolution
Table 3.

Willingness to Pay by Gender and Degree Program

course, as the average WTP of males is $241.50 while that of females is $181.70. Is this difference in WTP statistically meaningful? To answer this question, we performed an F test on the equality of two means and a non-parametric rank sum test on the equal median. The null hypotheses of the equal mean and the equal median cannot be rejected. Thus, there is no statistically significant difference between males and females on their willingness to pay for an online course.

We then saw the average WTPs over the degree level in which students were enrolled. Those who were enrolled in the bachelor program were, on average, willing to pay $238 more for an online course, while students enrolled in the associate program were, on average, willing to pay $184.70 more for an online course. Hence, for an online class, on average, students in the bachelor program would pay more than students in the associate program would. We also performed an F test and a rank sum test to see if the difference was statistically significant. The null hypotheses of the equal mean and the equal median cannot be rejected at the 5% level. But the hypotheses can be rejected at the 10% level. The test results implied that the difference in the willingness to pay for an online course between the students in the bachelor program and the students in the associate program is marginally different as the difference is statistically significant at the 10% level.

We also looked at the WTP over socioeconomic status, i.e., commuting time, work hour, and household income. We found that WTP declined as commuting time increased, indicating a negative relationship between WTP and commuting time. The average WTP of non-commuting students was high at $329.80. The [End Page 243] average WTP then decreased to $259.90 for students residing within 15 minutes of campus and further to $166.60 for students residing within 30 minutes. This downward trend continued further as commute time increased. This negative relationship is, at first sight, counter-intuitive to our presumption that we thought that students who spend more time commuting would be more willing to pay for an online class to reduce travel time. We now speculate that our presumption may be true for students who take all online classes, but not for students who take only one or two online courses when they have a full-time schedule of four or five courses. If our speculation is correct, one possible reason for the negative relationship is that, under the situation where the benefit of saving commuting time is small, those who reside far away from campus are less willing to pay for an online class as they may opt to choose other schools.

On the contrary, we found a positive relationship between WTP and work hours; those who work longer hours, in general, reported a greater willingness to pay more for online classes. This positive relationship fit our presumption that students working a lot prefer to take online classes because they provide flexibility and convenience.

What about the relationship between household income and WTP? We found that household income did not seem to affect WTP much, as many WTPs were clustered at very low levels of household income. Given the fact that most respondents did not have much income and the survey finding that 63.2% of respondents depended upon financial aid for their educational expenses, it is not surprising to see that household income was not crucial in forming the WTP.

Demand and Marginal Revenue for Online Class

Following Foreit and Foreit (2004)'s manual, we derived the demand for an online class and its marginal revenue curve with student responses on the maximum amount they were willing to pay for an online course. The Foreit-Foreit method, which is simple to construct a derived demand but consistent with economic principles, i.e., law of demand, is that a cumulative frequency of respondents serves as the quantity while WTP is regarded as the price. Once a derived demand curve is constructed, it is easy to derive its counterpart, a marginal revenue curve.

By examining the demand curve and the marginal revenue curve, the revenue-maximizing price setting is inferred. Total revenue would be maximized as the marginal revenue becomes 0 when the quantity measured by the cumulative frequency of respondents is slightly less than 40%. At this point, a revenue-maximizing price premium for an online course is to charge approximately $150 more for an online course than for a traditional face-to-face course, which is less than the average amount of maximum willingness to pay for an online course. [End Page 244]

statistical analysis of the determinants of online course taking and wtp

Determinants of Online Course Taking

We explored the following equation to find what determines a student's choice of an online class,

inline graphic

where Onlinei is a dependent variable indicating that it is 1 if a student i has taken any online class and 0 if a student i has not taken any online class. Xi is a vector of explanatory variables in (1) and δ is a vector of the coefficients to variables Xi. F(∙) is a certain functional form while vi is an error term capturing any statistical noises. Since the dependent variable, Onlinei, is binary, a typical ordinary least square method assuming a linear relationship leads to an inaccurate interpretation of δ as well as heteroskedastic disturbances violating a classical assumption on the ordinary least squares. To avoid these problems, we employed non-linear functional forms, e.g., Probit and Logit. The log-likelihood function, log L, for the estimation is given below;

inline graphic

where P(∙) denotes the probability. The difference between Probit and Logit is the tails in the cumulative distribution functions, of which the density function of Probit is inline graphic and that of Logit is 1 / (1 + ez).

In the absence of previous statistical work on the determinants of online course taking, we conjectured that any perception that a student may have about online classes affects his or her choice of online class. Together with student perception about online classes, we also conjecture that socioeconomic differences among students and the cost of taking classes play some role in the decision to take an online class. We included the variables from the following categories: student perceptions about online classes, student responsibility for educational expenses, and socioeconomic characteristics.

The estimation results, shown in Table 4 below, revealed some factors that affected a student's decision whether he or she would take an online class. Those factors, statistically significant at the 5% and at the 10% level, were student expectation of a better grade in an online class, the portion of educational expenses paid by aid, age, marital status, and household income. These factors are robustly significant even if different sets of explanatory variables are included.

First, of the nine perceptional factors, we surveyed about student perception of online classes, only student expectation of a better grade in an online course significantly influences whether or not to take an online class. The impact of [End Page 245]

Table 4. Estimation Results for the Determinants of Online Class
Click for larger view
View full resolution
Table 4.

Estimation Results for the Determinants of Online Class

student expectation of a better grade was substantial enough to sway students away from an initial choice of a face-to-face course once the student expected to obtain a better grade from an online class. Specifically, the probability of taking an online class almost doubles, precisely jumping by 80% to 126.8%, if a student expects a better grade in the online class. It should be noted that other perceptional factors, e.g., convenience and unavailability of a traditional face-to-face class, did not significantly influence the decision to choose an online class, even if a student perceives those factors as important.

Second, on the cost side, who pays educational expenses is, generally speaking, not crucial to the choice of online class, as it is statistically insignificant or as its impact is too small. Out of three sources of educational expenses, i.e., family, loans, and aid, only aid is statistically significant at the 5% level. However, the impact of aid is very small, although statistically significant; the probability that a student takes an online class decreases by 9% to 16% if educational expenses paid by aid increases by 10%. That is, students take more face-to-face classes as [End Page 246] more of their expenses are covered by aid. We speculate that students may opt to work more by taking more online classes if financial aid decreases because online classes provide flexibility and convenience in scheduling and location.

Lastly, of the socioeconomic factors, only three factors, i.e., age, marital status, and household income, turn out to be statistically significant for the decision to take an online class. The probability of taking an online class increases by 18.7% to 29.8% as a student ages a year. On the contrary, the probability plunges as a student gets married or lives with a domestic partner, and the probability also drops somewhat, i.e., 1.2% to 2.0%, if a student's household income increases by $1,000. However, the impact of the three socioeconomic factors stated on the above must be downplayed for the following reasons. First, it is natural that a student is more likely to have more opportunities to take online classes as a student stays longer at school (meaning that a student gets older). In fact, seniors take more online classes than juniors, and juniors take more than sophomores, and so on. Second, the number of online classes taken by married students may be biased as our data sample has only six married respondents. Third, the impact of household income is not only very small but also is statistically significant only at the 10% level.

Given these results, it can be concluded that the most distinctive determinant in the choice of an online class is a student's expectation of having a better grade from an online course. Other factors such as the burden of educational expenses as well as household income, affect the decision to take an online class, but only minimally. Furthermore, it seems natural that a student takes more online classes as a student takes more classes at school.

Determinants of WTP for Online Classes

We use a censored regression analysis, i.e., Tobit model, to address the determinants of WTP for online class. As we asked students to indicate their maximum amount to pay for an online course, the values of WTP are non-negative. No negative WTP was possibly recorded as the value of zero is the lowest value for WTP. To correctly capture the censored dependent variable, WTP*, at zero, the equation for the Tobit model is given below;

inline graphic

where inline graphic is a latent variable in which we observe inline graphic only if inline graphic and inline graphic is a vector of explanatory variables while β is a vector of the coefficients to the explanatory variables. ui is an error term capturing statistical noise. Ordinary least square method would be biased as only non-negative observations are used. The log-likelihood function, log L, for the equation (3) is given below; [End Page 247]

inline graphicinline graphicinline graphicinline graphic
Table 5. Estimation Results for the Determinants of Willingness to Pay for Online Class
Click for larger view
View full resolution
Table 5.

Estimation Results for the Determinants of Willingness to Pay for Online Class

inline graphic

where di is a dummy which takes 1 if WTP is observed and Φ otherwise, and (·) denotes the cumulative density function of the Probit while inline graphic is inline graphic deflated by the standard error of WTP.

As in the analysis of online course taking behavior, we considered different sets of variables representing socioeconomic status, cost burden, and perceptional factors of online classes as the explanatory variables in the equation (3). Table 5 above shows the estimation of (4).

Throughout the three sets of explanatory variables, denoted by (1), (2) and (3) in the first row of Table 5, only three variables were statistically significant at the 5% level: degree program, work hours, and commuting time. It should be noted that any of cost burden and perceptional factors turned out to be irrelevant to forming a student's willingness to pay for an online class. Following is what we found on the determinants of WTP from the statistical analysis.

First, a student enrolled in the bachelor program was more willing to pay for online classes than a student enrolled in the associate program. How much more? When other factors were controlled, a student in the bachelor program would pay $59.50 to $68.70 more per one online course than a student in the associate program. This result is consistent with our earlier analysis on the data while a simple data comparison (shown in Table 3) yields only a $53.30 difference [End Page 248] between bachelor and associate programs. We speculate that this difference may be related to student valuation of a college class. A student in the bachelor program probably considers a college education more valuable than a student in the associate program and thus a student in the bachelor program would want to invest more into human capital than a student in the associate program.

Second, a student who works longer is willing to pay more for online classes, which is consistent with our earlier data analysis and intuitive to our presumption that online classes provide convenience in scheduling. Precisely, WTP would increase by $2.80 to $3.10 as a student works one more hour per week. This result means that a student who works 20 hours a week would pay $56 to $62 more for an online class than a student who does not work.

Third, students with longer commutes would want to pay less for online classes. Specifically, a student would pay $1.10 to $1.30 per one minute's commute. That is, a student's WTP decreases $66 to $78 per one online class if a student has to spend one more hour (one-way) each time he or she comes to a campus. This result is counter-intuitive. We speculate that a student residing farther away from campus could still need to take a face-to-face class and thus may view the drive a sunk cost while a student residing nearby may not correctly recognize the time value of commuting. In this context, this counter-intuitive result may reflect a self-selection bias. That is, those who place a high value on the time spent commuting could already have transferred to nearby schools and thus these students' time values are not captured in our sample. So, we speculate that those who do not have to take a face-to-face class would still want to pay more for online class.

conclusion

This study intends to broaden our understanding of student choice of taking online classes. In particular, this study attempts to address the following questions regarding the choice to take online classes: how much more a student would pay for an online class, what factors affect students' willingness to pay for online classes and how much those factors affect the students' decisions, what price could maximize revenue from online courses, what characteristics online course takers exhibit, and what determines whether students take online classes over face-to-face classes and how much each factor affects students' decisions.

A questionnaire was designed to answer the above questions, and then the questionnaire was administered to students enrolled at a regional university in the southeastern U.S. A total of 243 observations from the university's five campuses was collected after dual enrollment students were excluded because they are still in high school and do not pay any tuition and fees for their college classes. [End Page 249]

We first analyzed the sample data collected from the survey to understand how educational expenses are paid. Of all respondents, 78.7% reported that their expenses are paid, to some degree, by family, while 63.2% of respondents said expenses were paid by financial aid. Three-quarters of respondents reported that they do not use any student loans.

We then analyzed the data to characterize online course takers and to understand how they perceive online classes. For the characteristics of online course takers, we found that the average number of online classes taken by our respondents was 0.89 courses. We found that females (1.00 course), married students (1.67), and students in the bachelor program (0.96), on average, took more online classes than males (0.81), unmarried students (0.86), and students in the associate program, although these differences by gender, marital status, and degree program were not statistically significant. Regarding socioeconomic characteristics, only household income was negatively associated with the number of online classes taken. Regarding student perceptions, we compiled a list of factors that may be important in selecting online courses. Out of those perceptional factors, more than 90% of respondents thought the specific factors of expectation of a better grade, convenience, and unavailability of a face-to-face class were important, to varying degrees, in their decision to take online class. A substantial number of students said that technical challenges and deadline issues were not important at all in their decision, while more than 80% of respondents indicated that insufficient interaction with an instructor, higher expenses of taking online classes, and concerns about learning were deterring factors in their consideration of an online class.

On the maximum willingness to pay more for an online course (WTP), we found, from the sample, that the average amount was $226.30 per one course. If dissected by gender and degree program, males ($241.50) were willing to pay more for online classes than females ($181.70), although this difference was not statistically supported. On the contrary, the students in the bachelor program ($238) would pay more for online classes than the students in the associate program ($184.70), and this difference was statistically supported at the 10 percent significance level. We also found that WTP increased with work hours, while WTP decreased with commuting time, contrary to our initial presumption. Given the student responses on WTP, we derived the demand curve and the marginal revenue curve for an online class to find a revenue-maximizing price premium for an online course. The premium is approximately $150 per course.

Lastly, we performed statistical analysis to find the determinants of online course taking behavior and WTP. We found that the most influential determinant was a student's expectation of receiving a better grade from an online [End Page 250] course. Other factors, i.e., the burden of educational expense and household income, can also affect the decision to take an online class, but the size of the effect was minimal. Age was important, as it seems natural that a student takes more online classes over time as the student takes more classes at a school.

On the determinants of WTP, we found that three variables, i.e., degree program, work hours, and commuting time, were statistically significant at the 5% level. Students in the bachelor program would pay $59.50 to $68.70 more per course compared to the students in the associate program. A student would pay $2.80 to $3.10 more as work hours increased by one minute, while a student would pay $1.10 to $1.30 less as commuting time increased by one hour. Cost burdens and perceptional factors turned out to be irrelevant in forming a student's willingness to pay for an online class.

Kelly Manley

Kelly Manley is Associate Professor of Economics

Yongseung Han

Yongseung Han is Professor of Economics

Michael Ryan

Michael Ryan is Associate Professor of Economics

Christopher Serkan

Christopher Serkan is Associate Professor of Mathematics, at the Univsity of North Georgia.

references

Allen, I. E., and J. Seaman. 2017. Digital Learning Compass: Distance Education Enrollment Report 2017. Babson Survey Research Group.
Aslanian, C. B., and D.L. Clinefelter. 2013. Online College Students 2013: Comprehensive Data on Demands and Preferences. Louisville, KY: The Learning House, Inc.
Beck, H. P., and M. Milligan. 2014. "Factors Influencing the Institutional Commitment of Online Students." Internet and Higher Education 20: 51-56.
Bernard, R. M., P.C. Abrami, E. Borokhovski, C.A. Wade, R.M. Tamim, M.A. Surkes, and E.C. Bethel. 2009. "A Meta-Analysis of Three Types of Interaction Treatments in Distance Education." Review of Educational Research 79 no. 3: 1243-1289.
Boys, K.A., D.B. Willis, and C.E. Carpio. 2013. "Consumer Willingness to Pay for Organic and Locally Grown Produce on Dominica: Insights into the Potential for an "Organic Island." Environment, Develpment, and Sustainability 16: 595-617.
Breidert, C., M. Hahsler, and T. Reutterer. 2006. "A Review of Methods for Measuring Willingness-to-pay." Innovative Marketing 2 no. 4: 8-32.
Byrd, J., J. Roufagalas, and P.A. Mixon. 2015. "Tuition Sensitivity in Online Education." Journal of Economics and Economic Education Research 16 no. 30: 25-41.
Dehghan, A., J. Dugger, D. Dobrzykowski, and A. Balazs. 2013. "The Antecedents of Student Royalty in Online Programs." International Journal of Education Management 28 no. 1: 15-35.
Dorantes, A.R. and J.R. Low. 2016. "Financial Crisis Management in Higher Education: Responses by 20 Colleges and Universities to the 2007-2009 Financial Crisis." Journal of Education Finance 42 no. 2: 189-219.
Ehrenberg, R.G. 2012. "American Higher Education in Transition." Journal of Economic Perspectives 26 no. 1: 193–216.
Fink, I. 2017. "Classroom Use and Utilization." Facilities Manager Jan/Feb: 28-35.
Foreit, K. G. F. and J.R. Foreit. 2004. "Willingness to Pay Surveys for Setting Prices for Reproductive Health Products and Services: A User's Manual." The Population Council, Available on https://www.popcouncil.org/uploads/pdfs/popbriefs/frontiers/Capacity_Bldg/WTP_Manual.pdf.
Green, P. E. and V.R. Rao. 1971. "Conjoint Measurement for Quantifying Judgmental Data." Journal of Marketing Research 8 no. 3: 355-363.
Hachey, A. C., C. Wladis, and K. Conway. 2015. "Prior Online Course Experience and G.P.A. as Predictors of Subsequent Online STEM Course Outcomes." Internet and Higher Education 25: 11-17.
Han, Y., M. Ryan, and K. Manley. 2018. "Online Course Enrollment and Tuition: Empirical Evidence from Public Colleges in Georgia, U.S.A." International Journal of Education Economics and Development 10 no. 1: 1-21.
Jaggars, S.S. 2014. "Choosing between Online and Face-to-face courses: Community College Student Voices." American Journal of Distance Education 28 no. 1: 27-38.
Jaggars, S.S., N. Edgecombe, and G.W. Stacey. 2013. What We Know about Online Course Outcomes. Community College Research Center, Teachers College, Columbia University.
Klein, M.W. 2015. "Settling a U.S. Senatorial Debate: Understanding Declines in State Higher Education Funding." Journal of Education Finance 41 no. 1: 1-29.
Kotler, P. and K.F. Fox. 1995. Strategic Marketing for Educational Institutions, Prentice-Hall, Englewood Cliffs, NJ.
Lee, Y. and J. Choi. 2011. "A Review of Online Course Dropout Research: Implications for Practice and Future Research." Educational Technology Research and Development 59: 593-618.
Mann, H.B., and D.R. Whitney. 1947. "On a Test of Whether One of Two Random Variables is Stochastically Larger than the Other." The Annals of Mathematical Statistics 18: 50-60.
McFadden, D. 1986. "The Choice Theory Approach to Market Research." Marketing Science 5 no. 4: 275-297.
Murphy, C. A. and J.C. Stewart. 2015. "On-campus Students Taking Online Courses: Factors Associated with Unsuccessful Course Completion." Internet and Higher Education 34: 1-9.
Nagle, T. T. and R.K. Holden. 2002. The Strategy and Tactics of Pricing, Prentice Hall, Upper Saddle River, NJ.
Nessim, H. and R. Dodge. 1995. Pricing-policies and Procedures, MacMillan Press, London.
Noel-Levitz. 2006. National Online Learners Priorities Report. Ruffalo Noel Levitz, LLC. Available on https://www.ruffalonl.com/upload/Papers_and_Research/2006/06ONLINE_report.pdf
Noel-Levitz. 2014. National Online Learners Priorities Report. Ruffalo Noel Levitz, LLC. Available on https://www.ruffalonl.com/upload/Papers_and_Research/2014/2014-15_OnlineLearners.pdf
Oliff, P., V. Palacios, I. Johnson, and M. Leachman. 2013. Recent Deep State Higher Education Cuts May Harm Students and the Economy for Years to Come. Center on Budget and Policy Priorities.
Public Agenda. 2013. What Employers and Community College Students Think about Online Education. Available online at https://www.publicagenda.org/files/NotYetSold_PublicAgenda_2013.pdf
Serna, G.R. and G. Harris. 2014. "Higher Education Expenditures and State Balanced Budget Requirements: Is there a relationship?" Journal of Education Finance 39 no. 3: 175-202.
Shen, D., M. Cho, C. Tsai, and R. Marra. 2013. "Unpacking Online Learning Experiences: Online Learning Self-Efficacy and Learning Satisfaction." Internet and Higher Education 19: 10-17.
Sturgeon, J. 2007. "Lost in SPACE." University Business 10: 58-62.
Tobin, J. 1958. "Estimation of Limited Dependent Variables." Econometrica 26: 24-36.
Turner, S. 2006. "Higher Tuition, Higher Aid, and the Quest to Improve Opportunities for Low-income Students: The Case of Virginia." In Ehrenberg, R.G. (Ed.), What's Happening to Public Higher Education: The Shifting Financial Burden, Johns Hopkins University Press, Baltimore, MD., pp. 251–274.
Volery, T. and Deborah Lord. 2000. "Critical Success Factors in Online Education." International Journal of Educational Management 14 no. 5: 216-223.
Wilcoxon, F. 1945. "Individual Comparisons by Ranking Methods." Biometrics Bulletin 1: 80-83.

Additional Information

ISSN
1944-6470
Print ISSN
0098-9495
Pages
232-252
Launched on MUSE
2020-03-16
Open Access
No
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.