1 : ONLINE COURSE TAKING BEHAVIOR AMONG COMMUNITY COLLEGE STUDENTS The Need for High Speed: Online Course Taking Behavior Among Community College Students Jillian Gross Molly Kleinman University of Michigan
2 2 Online Course Taking Behavior Among Community College Students Colleges and universities of all types and sizes offer online courses to meet student demand, reduce teaching costs, and increase enrollments (Allen & Seaman, 2011). By 2011, thirty-one percent of all postsecondary students were enrolled in at least one online course (Allen & Seaman, 2011). Though growth in the number of online offerings has been dramatic in recent years, empirical evidence does not consistently support many of its intended benefits. The target audience for these courses is frequently students who are non-traditional in some way, either due to age, employment or parental status, military service, or other factors (Bacow, Bowen, Guthrie, Lack, & Long, 2012), the same groups of students that community colleges serve (Bragg, 2011). Because community colleges are rapidly adding new online offerings to meet perceived demand (Mullin, 2010) many questions about the impact of online education on community college students arise. The present study examines for-credit, online course taking behavior among community college students. The data come from the Pearson Foundation Community College Student Survey Most previous studies focus on retention and success rates among online course takers and highlight differences based on distal characteristics such as race, gender, age, and academic ability (Jaggars, 2011). Because online course taking is an increasingly common element of the higher education experience, it is important to develop a holistic understanding of students who take courses online. Using the emerging conceptual framework of digital choice and exclusion as a guide (Eynon & Helsper, 2010), we aim to move beyond basic demographics and paint a more nuanced picture of who online students are, and why they have chosen to take classes online. To achieve this end, we examine additional factors that influence the probability of enrollment in online courses, including socioeconomic status, Internet access, employment
3 3 status, ownership of computers and portable electronic devices, and reasons for enrolling in community college. The role of community colleges Literature Review As the pressure to pursue postsecondary education has increased, both as an economic necessity and a national policy, community colleges are playing a crucial role in the higher education landscape (Bragg, 2011). In 2011, over 12 million students were attending a community college, comprising forty-five percent of all students enrolled in higher education in the United States (American Association of Community Colleges, 2012; National Student Clearinghouse, 2012). Demand for seats in community college courses and programs rises every year (Rhoades, 2012). Because the majority of community colleges maintain an open admissions policy, they serve as the primary mode of access to higher education for underserved groups including racial and ethnic minorities, immigrants, low-income students, and students of nontraditional age (Bragg & Durham, 2012). Rhetoric around community colleges often treats them as a panacea for solving a wide range of problems, from job training and workforce development to improving equity and access to higher education for underrepresented students of all kinds (Bragg, 2011). However, whether success is defined as transfer to a four-year institution, or the completion of a degree or certificate, unique challenges conspire to make that success more difficult for community college students. These challenges include a higher likelihood of employment and family responsibilities, weaker high school preparation, and less social support, on average (Hagedorn, 2010). As a result, researchers, policy makers, and administrators are expending a great deal of energy and resources to improve retention, graduation, and transfer rates for community college students (Mullin, 2012). One popular solution to the challenges
4 4 facing community colleges and higher education more broadly is online education. Online education The definition of online education varies widely. For instance, Lorenzo (2010) defines an online course as one in which instruction, assignments, and evaluation occur entirely online (Lorenzo, 2010). Much of online education is therefore distance education, that is, the student enrolls in an institution that is not physically near her home, and she has no face-to-face relationship with the school. However, many colleges and universities are offering online courses to students who may live just down the street or blend face to face with online instruction. Certain trends in online education, such as Massive Open Online Courses (MOOCs) and free online lectures like Khan Academy, receive a disproportionate amount of press coverage and attention, but many schools are simply adapting the courses they already offer in person to an online format (Allen & Seaman, 2011; Jaggars, 2011). The number of colleges and universities offering online courses, and the number of students enrolled in those courses, has been rising steadily over the past decade (Allen & Seaman, 2011). In the fall term of 2010, 6.1 million higher education students were taking at least one course online, an increase of 560,000 students over the number reported the previous year (Allen & Seaman, 2011). Like community colleges themselves, online education is often touted as a cure for much of what ails higher education. It can provide scheduling flexibility for campuses that are stretched to capacity by increased enrollment and budget cuts and for students who have jobs or families making demands on their time (Allen & Seaman, 2011). As economies of scale are reached, online learning may allow colleges to spend less money per student (Bacow, et. al, 2012), and to educate students more efficiently than traditional classroom models (Singh, Rylander, & Mims, 2012). Online education s asynchronous nature can improve access for non-traditional students by permitting them to take their courses from home and do schoolwork at convenient times
5 5 (Kramarae, 2001). Among academic leaders, perceived flexibility and potential for revenue generation are among the most cited rationales for pursuing online education, yet supporting evidence remains primarily anecdotal (Bacow et. al, 2012). Despite lofty goals, high attrition rates and concerns about both quality and legitimacy continue to plague many online learning opportunities (Jaggars, 2011; Johnson & Berge, 2012; Lorenzo, 2010). Furthermore, only limited evidence exists linking online education to improved learning outcomes generally, and for lowincome, first generation, and underrepresented minority students in particular (Jaggars, 2011). It is also not clear why some students choose online courses and others do not a question that has largely been ignored by research (Jaggars & Bailey, 2010; Roblyer, 1999). Researchers have only just begun to identify possible factors that influence online course taking behavior. Because there are very few fully online programs, Jaggars (2011) points out that most studies have focused on students enrolled in traditional or blended programs who take one or more of their courses online. Research suggests that minority students are taking online courses at lower rates than white students (Angiello, 2010; Xu & Jaggars, 2011). Several studies indicate that female students are taking online courses at higher rates than male students (Kramarae, 2001; Roy, & Schumm, 2011; Xu & Jaggars, 2011), but this effect has not been separated from a higher prevalence of women in higher education overall. Older students with full time jobs may also be more likely to choose online course options (Halsne & Gatta, 2002). A small body of research includes measures of learning style, locus of control, aptitude and other student-centered constructs to understand how students choose to take courses online. One constant refrain throughout the literature on online education is convenience. Almost universally, students indicate that the flexibility of online education is desirable (Jaggars, 2011). Students who prefer traditional face to face learning environments may do so because of a perceived match with their learning styles and a desire for in-class engagement relative to
6 6 students who prefer online or hybrid offerings (Clayton, Blumberg, & Auld, 2010). A study of community college and virtual high school students found that for students who chose online options control over pace and timing was the primary consideration, while students who chose face to face courses were more interested in interaction with the instructor (Roblyer, 1999). In a recent working paper, community college students reported a preference for online courses in easy subjects, but feel that face-to-face classes are more appropriate for hard or important classes (Jaggars, 2013). Other factors that may influence online course choice, such as reasons for enrolling in college, access to technology, and preparation for college, have not been thoroughly examined. Online education and community colleges Despite the lack of clarity regarding the effectiveness and efficiency of online education, community colleges appear to be riding the swell toward online education along with the rest of higher education. Indeed, community colleges have been under pressure to adopt and expand their online course offerings, and many have responded (Cox, 2005). Online offerings at community colleges range from technically sophisticated and comprehensive programs that have been refined over many years to newer programs that use rudimentary course management software and provide little support for instructors or students (Petrides, Karaglani, & Nguyen, 2005). While some research suggests that there is little difference in outcomes between face-toface and online education, that assumption does not appear to hold true for community college students (Jaggars & Xu, 2011, Xu & Jaggars, 2010, Xu & Jaggars 2013). A few clear trends in community college online course taking include higher attrition rates among low-income and underrepresented students that result in overall barriers to college persistence (Bailey, Jeong, &
7 7 Cho, 2010), a dearth of fully online degree program options, and little reduction in cost for online tuition (Jaggars, 2011). Furthermore, whereas increasing availability of online courses is often described as a means to increase access to community colleges, online courses are not increasing enrollment among new or low-income students (Jaggars, 2011; Jaggars & Xu, 2011). On average, community college students taking (and succeeding in) online courses are white, higher income women who are academically well prepared for college coursework (Halsne & Gatta, 2003; Jaggars & Xu, 2011; Jenkins & Downs, 2003; Xu & Jaggars, 2011; Xu & Jaggars, forthcoming). In other words, the intended benefits of increased access, retention, and completion that online course offerings aim to furnish, may not be realized. Students often cite technical difficulties as a reason for withdrawing from or not taking online courses (Bambara, Harbour, Davies, & Athey, 2009; El Mansour & Mupinga, 2007). Lowincome households disproportionately lack the necessary infrastructure, e.g. high-speed Internet and home computer, to take full advantage of online courses (Jaggars, 2011). This digital divide also creates an obstacle for historically underrepresented minority students, because in 2012 only fifty-one and forty-seven percent of African Americans and Hispanics respectively had high-speed Internet access at home (National Telecommunications & Information Administration, 2013). As Lennett and Kehl (2013) suggest, in the 21st century, ensuring equal access to education may also depend upon equal access to broadband (para. 13). Therefore, the convenience of online course taking may not extend to students who must go out of their way to access the Internet. Lack of access is likely one among many factors that explain low completion rates among this population of students. Additional factors, borne out in the research, include inadequate guidance regarding the requirements of online course taking and low interpersonal interaction between students and instructors (Jaggars, 2011; Jaggars, 2013). However, in a comprehensive review of current literature concerning online course taking in community
8 8 colleges (Jaggars, 2011), no studies controlled for factors related to technological access and only a handful included measures for academic motivation or preparation. With this study, we aim to shed light on important but often-overlooked factors associated with a student s choice to engage in online learning at a community college. Because online students constitute a growing proportion of the community college population, we view this study as an essential step in building theory about course choice in a technologically infused educational environment. As Jaggars (2011) concludes in her review of literature pertaining to online education among community college students, the majority of studies exploring this issue employ naive models, highly specific case study data, and generally small sample sizes. Only a handful of studies (Jaggars & Xu, 2013; Jaggars & Xu, 2011; Jaggars, 2010; Xu & Jaggars, 2010; Xu & Jaggars, 2013) diverge from this trend by using statewide, longitudinal data and employing more sophisticated statistical techniques including propensity score matching, instrumental variables, and multi-level regression. While our analysis is limited to survey data, our sample size is sufficiently large and representative to begin to make generalizations about the broader community college population. Furthermore, the technology focus of the survey instrument provides a unique insight into the characteristics of online course takers that generally has not been captured in large-scale datasets. Conceptual Framework Grounded in an emerging framework of digital exclusion and choice that has sprung from research on the digital divide (Eynon & Helsper, 2010), our research explores the decision of community college students to take online courses. Digital exclusion refers to the often involuntary process of technological disengagement due to limited resources (Eynon & Helsper, 2010; Livingstone & Helsper, 2007). Certain individuals or groups may be excluded from information consumption or digital participation because they do not have access to the tools,
9 9 connectivity, or knowledge necessary to join in. Digital choice explains how individuals select and use the technological resources available to them. Choice, in this context, is defined as select[ing] among media choices based on how well each option helps them meet specific needs or goals (Cho, Gil de Zuniga, Rojas, & Shaw, 2003, p. 48). Students who overcome digital exclusion may still choose not to participate in online education. Both digital choice and digital exclusion are tied to socioeconomic status, race, sex, and other demographic characteristics related to the digital divide (Eynon & Helsper, 2010). Digital exclusion and choice offer a more nuanced continuum than early examinations of the digital divide that focused on material access to technology and the internet (Livingstone & Helsper, 2007). Recent research suggests that in developed countries where internet access is increasingly ubiquitous across classes, socioeconomic status and other demographic factors affect how people use the internet (Mancanelli, 2007). For example, using the internet for informational purposes rather than solely for entertainment is associated with higher socioeconomic status (Wei & Hindman, 2011). The interaction between choice and exclusion in this context can be complex; individuals may choose not to use the internet as a result of insecurity or ignorance that result from a lifetime of educational or social exclusion (Eynon & Helsper, 2010). Digital inclusion or exclusion may constrain students choice sets well before they enroll in a particular class. In the present study, digital exclusion and choice may influence online course taking behavior because the community college sector enrolls a large portion of low-income and historically underrepresented minority students in higher education, a population that would be prone to digital exclusion. Simultaneously, because American community colleges exist in an environment where the Internet is widespread, certain groups of students may choose to participate in online learning opportunities, or not, based on factors other than access to the web (Wei & Hindman, 2011; Livingstone & Helsper, 2007). In a recent qualitative study, participants
10 10 cited flexibility, convenience, and efficiency among the most prevalent explanations for the decision to enroll in an online course (Jaggars, 2013). Less frequently, students reported a fit between learning style and preferred format for interaction as reasons for taking online courses. Conversely, students reported the desire to stay connected both physically and personally with campus and instructors as a reason for continuing to enroll in a blended course schedule (Jaggars, 2013). While this study hints toward factors associated with digital choice, additional research is needed to explore how factors associated with exclusion may have mediated or moderated student decision-making and perceptions of online learning. Furthermore, a recent report by the Federal Communications Commission states that 28.2% of Americans living in rural areas do not have access to high speed internet in their communities (2011). With roughly 60% of community colleges located in rural communities (Hardy & Katsinas, 2006), geography may play a critical, but overlooked role in digital exclusion and choice among students. Even for students who do have home internet access, increasingly stringent data caps on broadband access with costly penalties for exceeding limits may exacerbate factors related to digital exclusion as online courses quickly devour expensive bandwidth (Hussain, 2012; Lennett & Kehl, 2013; Young, 2013). Furthermore, personal network exposure among economically advantaged groups may increase choice, while lack of exposure and confidence in using technology may result in exclusion among less economically advantaged individuals (Hsieh, 2008). Building on this emerging body of research, our study applies a digital exclusion/choice framework to online course taking in community colleges to explore the following research questions: 1 What student characteristics are associated with a student s decision to enroll in an online course at a community college?
11 11 2 What student characteristics are associated with the decision to enroll in a program of study that is fully online, partially online, or exclusively face-to-face? Data The data for the study come from the Pearson Foundation Community College Student Survey 2011, which collected responses from a sample of 1,205 students in the United States between the ages of 18 and Participants were enrolled in a U.S. community college and pursuing at least one course for college credit at any point between August 1, 2011 and September 26, Questions on the survey instrument span from student demographics to educational aspirations and motivations, and from interaction with technology to course taking behavior. See Tables 2 through 4 in Appendix A for complete descriptive statistics of the variables used in this study. Although the structure of the data do not allow us to make causal claims, the results of this analysis offer clear implications for both policy and practice. Variables Building on emerging theories of digital exclusion and choice as a mediating force in a student s decision to engage in online learning opportunities, we explore the enrollment behavior of community college students in online courses as a function of three groups of variables: demographic and socio-economic characteristics, educational motivation and goals, and access to technology. Outcome variables. The outcome variables of interest are 1) if a student has ever taken an online course, and 2) if a student is participating in a fully online program, a hybrid (blended) program including face-to-face and online courses, or a program that includes only face-to-face 1 After dropping several cases for missing data the final sample size for this study was 1,086 students. See Appendix A for a missing data analysis and other data management information.
12 12 instruction. See Table 1 in Appendix A for descriptive statistics on each outcome variable and information about the manipulation of raw data. 2 Whereas there was specificity in terms of program type for the second outcome variable, the survey question related to our first outcome provided no definition of online course. Hence, it is likely that students who answered this question affirmatively have taken a combination of fully online courses and hybrid courses (i.e., courses that have both online and face to face components). Demographic and socioeconomic variables. Because most prior research in online course taking among community colleges focuses on differences based on gender, age, and ethnicity, we include these measures as covariates in our models. Additionally, we incorporate measures of socio-economic status (education level, income, parents education 3, and employment status) that are not commonly included. See Table 2 in Appendix A for complete descriptive statistics. Prior research suggests that factors related to socioeconomic status and demographic characteristics such as age and gender have been linked to the phenomena of digital exclusion and digital choice respectively (Eynon & Helsper, 2010). We also include a control for status as a first semester student because prior work has concluded that first semester students are less likely to enroll in online courses (Jaggars, 2011), which could indicate digital exclusion or choice among such students depending on the context of the decision. Motivation and goal variables. A complementary extension of digital choice/exclusion theories includes student motivation and goals as a measure of choice utility (Fuller, Manski, & Wise, 1982). For example, students motivated to take courses for career related reasons or who have educational goals tied to upward mobility (e.g., skills upgrade or changing careers), may choose online courses because they perceive those courses to be the most flexible and efficient 2 We performed exploratory factor analysis on the technology variables included in these models but the KMO statistics, which were miserable/middling (McCall, 2012), did not warrant extracting factor scores for each participant. Additionally, we tested numerous interaction terms but none proved statistically significant in any model. 3 Measured as a dichotomous indicator variable; 1 = some college or more, 0 = high school diploma or less.
13 13 option for achieving their goal and find that online courses best fit their schedules and budgets (Jaggars, 2011). The Pearson survey included several questions pertaining to students motivations for enrolling in community college, such as a desire to change careers or upgrade skills, as well as their goals for taking community college courses, such as receiving a degree or transferring to another institution. Our analysis includes this series of covariates as a predictor of online course taking behavior; see Table 3 in Appendix A for descriptive statistics. Technology access variables. The last vector of characteristics we include reflects a student s relationship with technology. Digital choice/exclusion theory suggests that as familiarity with technology and ease of use increase, the probability of engaging in onlinelearning opportunities also increases. The vast majority of students primarily access the Internet at home, therefore, in order to account for ease of use, we include an indicator variable for students who do not access the web at home most often, and another variable regarding the type of Internet connection a student has at home. As a proxy for familiarity we include a series of technology ownership variables, such as whether the student owns a laptop, desktop computer, e- reader, smartphone, or other similar devices. Table 4 in Appendix A provides descriptive statistics for each covariate in this block. Methodology We employed a logistic regression technique to examine the factors that influence the dichotomous outcome variable of ever enrolling in online courses, using three nested models to determine the effectiveness of the covariates in explaining the outcome. Our most restricted model contains control variables including the set of demographic and socioeconomic characteristics described above. The second model adds the set of variables encompassing a student s motivation and goals for attending a community college, and the third model includes the vector concerning technology access and use. After entering each block of regressors into the
14 14 model, we tested the relationship between the explanatory variables and the probability of the outcome variable using chi square tests; all were significant (p<.001). In order to determine the effectiveness of the regressors in explaining students online course taking behavior, we further assessed each model using a number of goodness of fit tests; results for all goodness of fit tests are reported in Appendix B. Given the categorical nature of the second dependent variable of interest, program type, we employed a multinomial logistic regression (MNLM) technique to estimate the effects of the three vectors of student characteristics on his or her choice in program. MNLM allows us to estimate and compare results for each pair of possible outcomes. We, therefore, are able to report differences between students enrolled in programs that are face to face relative to hybrid, face to face relative to fully online, and hybrid relative to fully online. Using nested models, we tested the goodness of fit using the same techniques described for the logistic regression models. We also conducted a likelihood ratio and a Wald test for all regressors in the model. Both tests have a null hypothesis that coefficients for all explanatory variables are equal to zero. 4 Additionally, we conducted likelihood ratio and Wald tests to combine outcome categories. Statistically, these tests compute whether or not there are differences in coefficients between categories with a null hypothesis that categories can be collapsed (Long, 1997). Both tests for combining outcomes are significant (p<.001) for all three outcomes, indicating that student responses about type of educational program are distinctly different and cannot be combined. For a full presentation of the equations estimated in this analysis, refer to Appendix C. Limitations 4 Both Wald and Likelihood ratio test statistics were similar. We report Likelihood ratio results in Appendix B as this is considered a superior test (DesJardins, 2012)
15 15 The data have several limitations that we believe require brief discussion here. First, it should be noted that the Pearson Foundation did not design this survey for academic purposes; rather it was designed as market research. 5 These origins resulted in a question structure that was not always conducive to rigorous statistical analysis. Survey questions suffered from three main problems: 1) Chronological ambiguity which made it impossible to determine whether certain opinions led to certain actions. For example, the questions regarding perceptions of online courses such as whether or not they were difficult did not ask whether these opinions were formed before or after taking an online course. 2) Undefined terminology, such that important terms like online program and online course are undefined or poorly defined in the survey. Is an online course one that is entirely online, or a course that has some online components? 3) Unranked responses in the questions regarding motivations and goals for attending community college meant that participants could select all that apply, without ordering them from most important to least important. In addition, while summary materials describe the survey sample as nationally representative, it relies heavily on weights, and on some dimensions the sample population deviates considerably from the national population. It over represents Asian and Hispanic students while underrepresenting white and African American students compared with community college students nationally (American Association of Community College Students, 2012). Not surprisingly for a web-based survey that recruited participants online, 79 percent of respondents own a laptop and 73 percent have high speed internet access at home; however, according to the Pew Research Center, in 2011 only 57 percent of American adults owned laptops (2012a) and 62 percent had high speed broadband at home (2012b). 5 Information obtained via personal communication.
16 16 Furthermore, because all the information was acquired via an electronic survey, the data are subject to self-reporting bias and analyses are subject to omitted variable bias. For instance, there were no measures of student ability available which did not create issues of reverse causation when included in the model. Additionally, the data are confined to a point-in-time snapshot of students experience with online course taking whereas longitudinal data (e.g. Jaggars, 2011; Jaggars & Xu, 2010) would allow for more rigorous analysis of the decisions prompting online course taking behavior. Imprecise measurement, indicated by large standard errors, is another significant limitation regarding the nature of the data. Hence, our analysis is limited in the claims we can make regarding the factors that are related to students decisions to choose online courses. Despite these limitations, the Pearson survey is unique in studying technology and motivation variables in relation to community college student behavior, and we believe that the results remain relevant for identifying policy implications and avenues for future research, which we address in the discussion. Results Our most restricted model (1) proves to be the best fit for both outcomes according the goodness of fit tests described above. Nesting the models did not result in changes of significance among variables related to basic demographics or socioeconomic status. However, the strength of the results related to motivation and goals as well as technology access and ownership bring to light important policy and practice issues that are under-theorized in extant research. Hence, for theoretical reasons related to the influence of digital choice and exclusion, we have confidence that Model 3 can provide important insight on our research questions. Results for Model 3 are reported below and included in Appendix D. The reference group for our analysis is respondents not in their first semester of college, who report having some
17 17 college, and who self identify their employment status as a student making less than $15,000 a year. Their parents both have at least some college and they access the Internet primarily at home through a high speed connection. Overall, within each vector, the significance, magnitude, and direction of the relationship between individual explanatory variables and the probability of the possible outcomes vary widely. Question 1: What factors influence a student s decision to enroll in an online course at a community college? Demographic and socioeconomic variables. The basic demographic variables of age, sex, and race are rarely significant across the estimated models. Holding all other variables constant, 6 Black students have odds that are 43 percent lower than white students of ever having taken a class online (p<0.10), while the odds of female students ever having taken an online course are 1.57 times those of male students (p<0.05). There were no significant differences among age groups. Although what we can say about employment is limited by the problematic construction of the variable, 7 results suggest that students who identify as self-employed are highly engaged in online learning. They have 507 percent higher odds of ever enrolling in an online course (p<.01), but only account for three percent of the sample. Like employment, income is sporadically significant as a predictor of online course taking; many estimates are significant but nothing was consistent across the regressions. In general, the results suggest that higher income is associated with a greater likelihood of being involved in online learning opportunities relative to the lowest income bracket (less than $15,000/year). One specific contrast includes: students earning 6 From this point forward, the interpretation will assume that results are reported while holding all other variables in the model constant, unless otherwise specified. 6 7 Student was an option for best descriptor of employment status, but every participant in the study was a student. Given that many community college students hold jobs, it seems odd that over fifty percent of respondents would self report their employment status as student and yet only 58 report earning less than $15,000 per year.
18 18 $35,000-$49,999 versus those earning less than $15,000 are almost two times as likely to have ever enrolled in an online course (OR=1.90, p<.05) and are almost four times as likely to be enrolled in a fully online program relative to a face-to-face program (RRR=3.84, p<.05). Motivation and goal variables. Only one variable in this block was a significant predictor of ever having taken an online course. In keeping with the literature suggesting that online remediation courses present serious challenges for students (Bambara, Harbour, Davies, & Athey, 2009; Xu & Jaggars, 2011), students who report being motivated to attend a community college for remediation have odds that are 58 percent lower than students not reporting remediation as a motivation to have ever taken an online course, (p<.01). Although a growing phenomenon, few remedial courses are offered online currently, which may contribute to the decreased likelihood of students motivated by remediation to engage in online learning (Xu & Jaggars, forthcoming). Technology access variables. Many of the technology ownership and access variables are not significantly associated with online course taking; overall the majority of respondents own laptops, access the Internet at home, and have a high-speed connection. However, lack of access to high speed Internet is negatively associated with ever having taken an online course and may be a barrier for digital participation. Students with a slower Internet connection at home are almost twice as likely to have never enrolled in an online course relative to their peers with a high-speed connection (p<.10). Question 2: What student characteristics are associated with the decision to enroll in a program of study that is fully online, partially online, or exclusively face-to-face? Demographic and socioeconomic variables. Neither age, sex, nor race are significant predictors of enrollment in online programs. Students who identify as self-employed are enrolled in online programs at higher rates; their odds of being enrolled in a fully online program relative to a face-to-face program are higher by a factor of 8.37 (p<.01) when compared to those who
19 19 identify as students. Similarly, relative to participants who identify their employment status as student, individuals employed full-time are 3.60 times as likely to be enrolled in a fully online program relative to a face-to-face program (p<.01). In combination with results from our first research question, it appears that the overall probability of self employed and full-time workers participating in online education is greater than self-identified students, even controlling for factors like age and income. As in the model for online experience, income is only sporadically significant as a predictor of program type, but the results suggest that higher income is associated with a greater likelihood of being enrolled in a fully or partially online program relative to the lowest income bracket (less than $15,000/year). Motivation and goal variables. Keeping in mind that respondents were able to choose multiple motivations and goals without ordering them, results suggest that certain motivations and goals are significantly related to online course taking. Students who report being interested in getting or upgrading skills are more likely to be enrolled in either a fully online or blended program rather than face to face, relative to students who do not report such a motivation. Though it seems related to upgrading job skills, there are no significant differences in online course taking behavior between people who report attending a community college in order to change careers relative to those who do not. Meanwhile, students who view community college attendance as the next step after graduating from high school, and students who reported returning to school after having been in the workforce to earn a credential are both more likely to be enrolled in a blended program over a face-to-face program relative to students who did not have those motivations. Specifically, students who view community college attendance as the next step after graduating from high school are 1.67 times as likely to be enrolled in a blended program over a face-to-face program (p<.05).
20 20 The results regarding students reporting a degree, transfer, or certificate as their goal are mixed. Students reporting a degree or transfer as their goal for attending a community college are more likely to be involved in either a face-to-face program or a blended program over a fully online programs. Being enrolled in a face-to-face program relative to a fully online program is associated with odds for students hoping to earn a degree (transfer) that are 2.46 (2.87) times that of students not reporting a degree (transfer) as their goal. Degree (transfer) seeking students are also 2.41 (3.08) times as likely to be enrolled in a blended program over a fully online program relative to their peers not reporting this as a goal. 8 Students reporting a certificate as a goal for attending community college have generally lower likelihood of engaging in online learning relative to their peers with alternate goals. Technology access variables. As in the online experience model, many of the technology ownership and access variables are not significantly associated with program type. However, owning a laptop is associated with a significantly higher likelihood of being enrolled in both fully online and blended course schedules. For students who own a laptop relative to their peers who do not, the odds of being enrolled in a fully online program (blended) relative to a face-to-face program are 2.99 (1.93). 9 While owning many other types of technology, like smartphones and tablets, appears unrelated, it is not surprising that a laptop - which students can use both to access and complete coursework, from anywhere - is associated with participation in online education. Finally, students who access the internet primarily away from home are significantly more likely to be enrolled in a blended program relative to a face-to face program (p<.001) while speed of internet connection at home is not a significant predictor of student s choice of program type. Discussion 8 All results for the goal of degree and transfer are significant at an alpha level of Results are significant at an alpha level of.01 and.05 respectively.