Study Links E-Learning Access Costs to Weaker Internal Efficiency at Nairobi Universities

Between August and October 2025, researchers studying e-learning in Nairobi County, Kenya, sent a structured questionnaire to 201 second-year university students. All of them were enrolled in purely online programs. The question was whether the cost of getting online, and staying online, affects how well those programs actually work.

The case for e-learning is familiar. It lets students reach course material from anywhere at any time, and it can lower the cost of buildings and infrastructure. The paper cites one earlier financial survey of web-based distance learning that found costs lower than in conventional teaching. But the authors point to a counterweight in the literature: the expense of access itself. Students pay for institutional fees, e-books, a laptop or computer, a mobile phone, internet connectivity and writing materials, and the authors argue that this burden can undercut the results online learning is supposed to deliver. They also note that earlier research often lumped costs together, leaving unclear how much any single kind of cost matters.

Nairobi County was chosen as the setting because of its size and its concentration of higher education. The paper cites Kenya National Bureau of Statistics figures from 2023 putting the population at about 5.5 million across roughly 696 square kilometres. The county has 29 private and 9 public universities, and many of them have a main or satellite campus there with established e-learning programs.

The team narrowed the field in stages. Only universities with a pure e-learning program that had already graduated students qualified, and blended programs were excluded. From the 28 universities in the study area that met the framing used in the paper, 10 were selected: five public and five private. Within them the researchers targeted social science schools, specifically education and business departments, and the second-year students there. That gave a population of 1,329.

Using a standard sample-size formula with a 30 percent coefficient of variation and a 2 percent margin of error, they calculated a required sample of 193. They added eight extra questionnaires, about 5 percent, to cover missing or wrongly entered data, bringing the total to 201. The sample was split in proportion to enrollment: 126 from private universities (62.7 percent) and 75 from public ones (37.3 percent). Respondents were picked systematically from student lists supplied with permission from department heads, and the questionnaire went out online by phone number and email.

Several findings emerged. Male students were more represented in e-learning than female students, which the authors describe as a gender disparity. Laptops or computers were the most commonly used access device. Most respondents came from private universities, which the authors read as a sign of student preference and choice.

The central result is more qualified than the conclusion. The OLS regression of accessibility costs on internal efficiency was not statistically significant at the p<0.1 level. The direction of the effect, however, was positive across all five measures: academic performance, distance effect, affordability, student retention and accessibility. The random factors in the model, by contrast, were positive and significant at p<0.001. The authors take this as a sign that unobserved or omitted variables matter a great deal and deserve further research.

Despite the non-significant cost estimates, the authors conclude that accessibility costs impede e-learning's internal efficiency. They call on the government to ease the burden on vulnerable students through tax incentives and budget allocations. The open question for later work is what those omitted variables are.