View in EDS HTML Full Text PDF Full Text

Does Blended Instruction Enhance English Language Learning in Developing Countries? Evidence from Mexico

Saved in:
Bibliographic Details
Title: Does Blended Instruction Enhance English Language Learning in Developing Countries? Evidence from Mexico
Language: English
Authors: Xu, Di, Glick, Danny, Rodriguez, Fernando, Cung, Bianca, Li, Qiujie, Warschauer, Mark
Source: British Journal of Educational Technology. Jan 2020 51(1):211-227.
Availability: Wiley-Blackwell. 350 Main Street, Malden, MA 02148. Tel: 800-835-6770; Tel: 781-388-8598; Fax: 781-388-8232; e-mail: cs-journals@wiley.com; Web site: http://www.wiley.com/WileyCDA
Peer Reviewed: Y
Page Count: 17
Publication Date: 2020
Document Type: Journal Articles
Reports - Research
Education Level: Higher Education
Postsecondary Education
Descriptors: English (Second Language), Second Language Learning, Second Language Instruction, Online Courses, Conventional Instruction, Blended Learning, Teaching Methods, Language Teachers, Teacher Effectiveness, State Universities, Grades (Scholastic), Teacher Student Ratio, School Policy, Mexicans, Computer Assisted Instruction, Educational Quality, Cost Effectiveness, Outcomes of Education, College Students, Foreign Countries
Geographic Terms: Mexico
DOI: 10.1111/bjet.12797
ISSN: 0007-1013
Abstract: Despite steady investment in English language education made by developing countries over the past few decades, results continue to be constrained by lack of high-quality instructors and language learning resources. Thus, using technology in language instruction has increasingly been recognized as a potential approach for addressing these constraints. This study uses administrative data from a large public university in Mexico to examine the impact of a technology-enhanced blended program on students' English course grades and course completion rates. Specifically, we focus on a campus-wide policy change in all compulsory English language courses that replaces half of the traditional face-to-face class time with an interactive online learning environment developed by a leading technology-mediated English language learning and assessment provider. Our results suggest that, compared to traditional face-to-face instruction, blended learning had a significant, positive impact on students' course grades and course completion rates. In addition, the enrollment-teacher ratio increased after replacing half of the face-to-face instructional time with online learning, suggesting that blended learning environments hold promise for providing high-quality and cost-effective language instruction.
Abstractor: As Provided
Entry Date: 2020
Accession Number: EJ1240961
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwEMyamYbiR_kJt0G4eX37m1AAAA4jCB3wYJKoZIhvcNAQcGoIHRMIHOAgEAMIHIBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDGwrJ9tCt_bRY9YYIwIBEICBmjooYHOJlpVL-vgvcptloK3oH_7ilNSiONWQj9405ycRXHO7ZoiDGatX7b99UbiU09GjBFMRB4EGApwKQcaTnAeORKRa6vgaxpxC00C6q0cWI9yhH_lw8XiaHGjk4cA1V9RvKj98K4G5Dr4A_9dFXaklEA1aRhSGC85ldWvMgFOOJY_9hAVmf9yBWKPVsAPXOCwLTW2b9IBsjUQ=
Text:
  Availability: 1
  Value: <anid>AN0141383096;58i01jan.20;2020Jan27.04:16;v2.2.500</anid> <title id="AN0141383096-1">Does blended instruction enhance English language learning in developing countries? Evidence from Mexico </title> <sbt id="AN0141383096-2">Practitioner Notes</sbt> <p>Despite steady investment in English language education made by developing countries over the past few decades, results continue to be constrained by lack of high‐quality instructors and language learning resources. Thus, using technology in language instruction has increasingly been recognized as a potential approach for addressing these constraints. This study uses administrative data from a large public university in Mexico to examine the impact of a technology‐enhanced blended program on students' English course grades and course completion rates. Specifically, we focus on a campus‐wide policy change in all compulsory English language courses that replaces half of the traditional face‐to‐face class time with an interactive online learning environment developed by a leading technology‐mediated English language learning and assessment provider. Our results suggest that, compared to traditional face‐to‐face instruction, blended learning had a significant, positive impact on students' course grades and course completion rates. In addition, the enrollment‐teacher ratio increased after replacing half of the face‐to‐face instructional time with online learning, suggesting that blended learning environments hold promise for providing high‐quality and cost‐effective language instruction.</p> <p>What we already know about this topic</p> <p></p> <ulist> <item> National efforts to develop students' English language proficiency in developing countries often remain ineffective.</item> <p></p> <item> The use of technology has been viewed as a possible way to address the barriers associated with traditional English language learning (ELL).</item> <p></p> <item> The evidence to date on the possible impact of technology use on English learning in developing countries remains limited, particularly in relationship to large‐scale online and blended learning initiatives.</item> </ulist> <p>What this paper adds</p> <p></p> <ulist> <item> This study is one of the first large‐scale studies that examines a campus‐wide policy change from traditional face‐to‐face instructional to blended learning in a large public university in Mexico.</item> <p></p> <item> Blended learning has a significant, positive impact on students' course grades.</item> <p></p> <item> This paper examines the change in enrollment‐teacher ratio before and after the policy change to compare the per‐student instructional personnel costs between blended and traditional face‐to‐face instruction.</item> </ulist> <p>Implications for practice and/or policy</p> <p></p> <ulist> <item> Blended learning is a promising strategy to leapfrog resource constraints and disrupt the status quo in ELL.</item> <p></p> <item> This paper informs policy makers that transferring from traditional courses to blended English language programs may potentially reduce per‐student instructional personnel costs.</item> </ulist> <hd id="AN0141383096-3">Introduction</hd> <p>English language learning around the world is increasingly being seen as a vital skill for personal as well as national development. Yet, often constrained by both economic and human capital resources, the effectiveness of national efforts to develop students' English language proficiency remains unsatisfactory in developing countries due to low teacher quality, large class sizes, low parent human capital and limited access to high‐quality language learning resources (Cronquist & Fiszbein, [<reflink idref="bib15" id="ref1">15</reflink>]; Davies, [<reflink idref="bib17" id="ref2">17</reflink>]; Johnson, [<reflink idref="bib25" id="ref3">25</reflink>]; OECD, [<reflink idref="bib34" id="ref4">34</reflink>]; OECD, [<reflink idref="bib35" id="ref5">35</reflink>]). Additionally, previous studies conducted in developing countries point out that students have substantial variation in their levels of English preparation, and this heterogeneity severely impedes any effort by the English instructors to ensure all students are making meaningful learning gains (Banerjee & Duflo, [<reflink idref="bib4" id="ref6">4</reflink>]; Glewwe, Kremer, & Moulin, [<reflink idref="bib22" id="ref7">22</reflink>]; Pritchett & Beatty, [<reflink idref="bib37" id="ref8">37</reflink>]). In view of these challenges, incorporating digital learning tools has been increasingly seen as a way to remove these barriers by providing low‐cost and high‐quality personalized education (eg, Sife, Lwoga, & Sanga, [<reflink idref="bib40" id="ref9">40</reflink>]). Yet, the evidence to date on the possible impact of incorporating digital tools in language learning in developing countries remains limited.</p> <p>In this paper, we present evidence on the impact of substituting a portion of in‐person language instruction time with well‐designed online instruction, typically referred to as blended language learning, on university students' English learning outcomes in a large public 4‐year university in Mexico. Specifically, we examine a campus‐wide policy change in all compulsory English language courses in the Fall of 2011 from traditional face‐to‐face instruction to blended instruction, which replaced half of classroom time with self‐paced learning using a customized online learning environment. Since every student at this university is required to take four levels of English language courses to fulfill the baccalaureate degree requirements, this creates a natural experiment that allows us to use a two‐way fixed effects model. Specifically, we include student fixed effects to control for both observable and unobservable between‐individual differences, such as differences in demographic characteristics and academic capacity. We further use course fixed effects to take into account possible variations between the four levels of English courses in course content, grading criteria and average student performance. We further examine the changes in enrollment‐teacher ratio before and after the policy change. This analysis could inform policy makers about the potential of reducing average per‐student instructional personnel costs by transferring some instructional activities to online learning environments.</p> <hd id="AN0141383096-4">Research background</hd> <p></p> <hd id="AN0141383096-5">English proficiency and economic opportunity in developing countries</hd> <p>In a globally integrated marketplace, English has become the lingua franca for engaging in business and trade. Existing reports consistently suggest that, in developing non‐native English‐speaking countries, English language proficiency has a positive association with high‐paying jobs for individuals (British Council, [<reflink idref="bib14" id="ref10">14</reflink>]; Davies, [<reflink idref="bib17" id="ref11">17</reflink>]; OECD, [<reflink idref="bib34" id="ref12">34</reflink>]). In their analysis of over a 100 countries spanning over 30 years, Ku and Zussman ([<reflink idref="bib27" id="ref13">27</reflink>]) further found that a country's English proficiency score was positively correlated with its economic growth.</p> <p>The importance of promoting English language learning for increasing economic opportunity is especially relevant for Mexico, which is one of the largest trading partners with the United States (Burfisher, Robinson, & Thierfelder, [<reflink idref="bib9" id="ref14">9</reflink>]; Krueger, [<reflink idref="bib26" id="ref15">26</reflink>]). For example, based on an online survey of 1000 people from the general Mexican population, a recent British Council report ([<reflink idref="bib14" id="ref16">14</reflink>]) found positive associations between English language proficiency and occupation, level of education attained, and household income. The importance of English proficiency also seem to be widely recognized, where 78% of the survey respondents viewed English as a major route to improve their economic prospects.</p> <p>While younger Mexicans on average have attained higher levels of education than previous generations (OECD, [<reflink idref="bib34" id="ref17">34</reflink>]), Mexico remains an English‐limited country, and national efforts to increase the number of English speakers have not adequately improved English instruction or acquisition (Sayer, [<reflink idref="bib39" id="ref18">39</reflink>]). In the sections below, we review how a confluence of factors among Mexico's public education system contribute to Mexico's low English proficiency rates, and discuss how blended language learning may hold the key to addressing these issues.</p> <hd id="AN0141383096-6">Mexico's public education system and challenges to English language instruction</hd> <p>The quality of Mexico's public education, defined as the system's impact on students' academic, economic and social capabilities, remains low despite showing significant improvements over time (Cronquist & Fiszbein, [<reflink idref="bib15" id="ref19">15</reflink>]; OECD, [<reflink idref="bib34" id="ref20">34</reflink>]). For instance, the 2015 and 2016 Global Competitiveness Report (The Global Competitiveness Report series is the world's most comprehensive assessment of national competitiveness. It assesses the competitiveness landscape of 140 economies, providing insight into the drivers of their productivity and prosperity. The report is based on 12 pillars. Pillar 5—Education—captures the overall quality of the education system and the extent it teaches the relevant skills) (World Economic Forum, [<reflink idref="bib50" id="ref21">50</reflink>]) showed that Mexico's quality of education was ranked 117 out of 140 countries. Similarly, in the 2015 PISA report, Mexico was ranked last in educational attainment among OECD countries (Sayer, [<reflink idref="bib39" id="ref22">39</reflink>]). Mexico also ranks low in English language attainment. The English First English Proficiency Index (EF EPI), which ranks 88 countries based on online English proficiency tests taken by more than 910 000 test takers, ranked Mexico in the bottom half, ranking 57th out of 88 countries (Education First, [<reflink idref="bib21" id="ref23">21</reflink>]). In a study that assessed the level of English proficiency of undergraduate students in three Mexican states, the authors found that students had a very low proficiency level in English (Lemus, Durán, & Martinez, [<reflink idref="bib29" id="ref24">29</reflink>]). Likewise, Davies ([<reflink idref="bib17" id="ref25">17</reflink>]) found that very few Mexican high school graduates have a good command of English, with a very small percentage of Mexicans able to use English effectively.</p> <p>Over the last decade, the Mexican government has made considerable efforts and investments in public education. In recent years, Mexico has spent 6.2% of its GDP on educational expenditures, which is above the OECD average of 6.1% (OECD, [<reflink idref="bib34" id="ref26">34</reflink>]). Additionally, Mexico has worked in partnership with non‐state actors (eg, EDUCANDO) to deliver trainings and ongoing support to teachers from underserved schools (Cronquist & Fiszbein, [<reflink idref="bib15" id="ref27">15</reflink>]; Educando, [<reflink idref="bib20" id="ref28">20</reflink>]).</p> <p>Despite continued national efforts and investment in public education, teacher quality remains low to modest (British Council, [<reflink idref="bib14" id="ref29">14</reflink>]; Davies, [<reflink idref="bib17" id="ref30">17</reflink>]; OECD, [<reflink idref="bib34" id="ref31">34</reflink>]; Vazques, Guzmán, & Roux, [<reflink idref="bib47" id="ref32">47</reflink>]). In 2005, eg, only about 58% of academic staff had a 5‐year undergraduate degree as their highest credential (OECD, [<reflink idref="bib34" id="ref33">34</reflink>]). While shortage of qualified teachers is a persisting problem in the Mexico public education system, there is a particularly serious shortage of qualified English teachers. Sayer ([<reflink idref="bib38" id="ref34">38</reflink>]) found that only one‐third of Mexican English teachers have an English proficiency level of B2 (Upper Intermediate) or better on the CEFR (Common European Framework of Reference for Language) scale, which is an international standard for determining language ability. Most of the state school English teachers use Spanish extensively in class (Dietrich, [<reflink idref="bib19" id="ref35">19</reflink>]). Researchers also found that students typically felt that their English teachers' English language ability is weak in Mexico (Davies, [<reflink idref="bib17" id="ref36">17</reflink>]). In order to successfully implement the national English language program, it is estimated that over 80 000 additional English teachers will need to be recruited and trained (British Council, [<reflink idref="bib14" id="ref37">14</reflink>]).</p> <p>These persistent barriers have substantially limited the extent of improvement in students' English language skills from nationwide endeavors in English instruction. Some researchers point out that the strategic management of Mexican public English language programs has involved little more than a "rush to more and more English," increasing quantity without a thorough consideration of the appropriateness, quality and real results of existing programs (Cummins & Davidson as cited in Davies, [<reflink idref="bib17" id="ref38">17</reflink>]).</p> <hd id="AN0141383096-7">Existing literature on blended language learning</hd> <p></p> <hd id="AN0141383096-8">Blended language learning</hd> <p>One promising approach to improving Second‐Language learning is to provide students with a mix of face‐to‐face instruction time and instruction through an online platform, which is often referred to as blended learning (Porter, Graham, Bodily, & Sandberg, [<reflink idref="bib36" id="ref39">36</reflink>]). Blended learning environments are unique in that instructors preserve the traditional face‐to‐face interaction experience while also providing students with learning resources and more individualized learning opportunities through technology‐enhanced online learning environments.</p> <p>It should be noted that the specific technology‐enhanced tools vary from study to study, and therefore may have different impacts on student learning processes and outcomes. For instance, blending learning environments can incorporate relatively straightforward tools, such as having students use discussion electronic workbooks and complete listening comprehension or rehearsal exercises on CD‐ROMs (Stracke, [<reflink idref="bib41" id="ref40">41</reflink>]). However, these tools are not necessarily designed to provide students with a personalized learning experience, but instead offer individual rote‐level practice. More recent blended learning environments, however, take advantage of computer‐assisted language learning technologies (CALL). These new technologies are delivered online through a learning management system and provide a more adaptive and comprehensive learning experience. For instance, in addition to providing essential language learning instruction (eg, listening, reading, writing), CALL technologies can utilize speech recognition technology to evaluate and improve students' speaking and general pronunciation (Glick & Davidson, [<reflink idref="bib23" id="ref41">23</reflink>]; Talebi & Teimoury, [<reflink idref="bib42" id="ref42">42</reflink>]). CALL technologies also offer a range of assessments, such as tests and short quizzes, which can then be used to build an adaptive lesson plan that is based on students' skill levels. Finally, CALL technologies can also help students understand their own progress by providing real‐time feedback, monitoring and performance reports (Glick & Davidson, [<reflink idref="bib23" id="ref43">23</reflink>]).</p> <hd id="AN0141383096-9">Evidence from developing countries</hd> <p>Blended learning environments with CALL components have the potential to improve student learning outcomes in developing countries, especially when students have limited access to rich and high‐quality language resources (A developing country is defined as a country with middle‐ and low‐income base, a lower‐living standard, underdeveloped industrial base and low Human Development Index (HDI) relative to other countries (Al‐Nasrawi & Zoughbi, [<reflink idref="bib1" id="ref44">1</reflink>]). The UN Human Development Socio‐economic Sustainability Index, which ranks 189 countries based on a range of socioeconomic factors, contains a selection of 11 economic and social sustainability indicators. The UN Human Development Index puts Latin America and sub‐Saharan Africa at the middle and bottom thirds, medium‐ and low‐human development, respectively (UNDP Human Development Index, [<reflink idref="bib45" id="ref45">45</reflink>])), such as in a context where providing strong oral language instruction is a significant hurdle for English language instructors who are not native English speakers (Weinberg & Knoerr, [<reflink idref="bib48" id="ref46">48</reflink>]). For instance, in a study conducted at Iranian junior high, half of the 60 students in an English foreign language class were randomly assigned into two pronunciation session conditions. In the experimental condition, students used CALL to practice their English pronunciation, whereas the control condition students practiced rote pronunciation. Both the experimental and control conditions did these independent learning sessions eight times over the course of a month, and each session lasted 90 minutes. The study found that students in the treatment condition received significantly higher scores on a pronunciation test than students in the control condition (Talebi & Teimoury, [<reflink idref="bib42" id="ref47">42</reflink>]).</p> <p>Similarly, a more recent study in a Malaysian English foreign language class found that 20 students enrolled in the course made significant gains in pronunciation from pre to posttest following the use of an instructional pronunciation video (Chang, Gregory, & Shak, [<reflink idref="bib10" id="ref48">10</reflink>]). In addition to helping students with oral language skills, CALL has also been found to be beneficial for listening comprehension (Bañados, [<reflink idref="bib3" id="ref49">3</reflink>]; Barani, [<reflink idref="bib5" id="ref50">5</reflink>]; Chen & Zhang, [<reflink idref="bib11" id="ref51">11</reflink>]; Vahdat & Eidipour, [<reflink idref="bib46" id="ref52">46</reflink>]), vocabulary acquisition (Tosun, [<reflink idref="bib43" id="ref53">43</reflink>]) and reading comprehension (Liu, [<reflink idref="bib31" id="ref54">31</reflink>]; Marzban, [<reflink idref="bib32" id="ref55">32</reflink>]), based on experimental and quasi‐experimental evidence from classroom studies conducted in developing countries that compared between CALL and traditional language instruction.</p> <hd id="AN0141383096-10">The current study</hd> <p>While the existing studies provide promising evidence on the possible benefits of CALL in second language learning in developing countries, they are all small in scale, typically including students in a single classroom, and therefore it is difficult to determine the extent to which CALL can be scaled to have positive impacts on student language learning at an institution. The current study aims to determine whether a large‐scale blended EFL course implemented in a public Mexican university could lead to improved learning outcomes compared to face‐to‐face courses. Furthermore, this study examines whether implementing the blended English language program changed the average student‐teacher ratio to explore whether a blended delivery model of English language courses may lead less expenditure of personnel costs.</p> <p>As such, this study makes several important contributions to the literature on blended learning environments in the context of computer‐assisted language learning. First, the current study attempts to expand on previous research by analyzing the learning outcomes of 22 023 students taking online and blended English language courses over a period of 4 years. The large sample size coupled with data collected from English courses taught over a period of 4 years can yield more precise estimates that are also more generalizable. Second, the current study investigates the academic success for a population that so far has been given insufficient attention to date in the computer‐assisted language learning literature: EFL students in developing countries. Finally, while government spending on Mexico's public education is significant and has been on a sharp upward trend since the 1980s (Brunner, Santiago, Guadilla, Gerlach, & Velho, [<reflink idref="bib8" id="ref56">8</reflink>]), the results of English language programs in the Mexican public education system are generally poor (Davies, [<reflink idref="bib17" id="ref57">17</reflink>]). This study could guide policy makers, instructional designers and EFL practitioners mainly in developing countries on how to use cost‐effective technologies to improve language instruction.</p> <hd id="AN0141383096-11">Data and context</hd> <p></p> <hd id="AN0141383096-12">Research context</hd> <p>This study was conducted at a second‐tier public university in Northwestern Mexico. The university has several campuses across the state of Chihuahua and enrolls roughly 32 000 students annually in 54 degree programs. About two‐thirds of the enrollees come from low or middle socioeconomic backgrounds. Each student is required to take four English language courses at four corresponding levels to fulfill their degree requirement: Beginner (Level 1), Upper Beginner (Level 2), Intermediate (Level 3) and Upper Intermediate (Level 4). Most students finish the course requirement within 2 years.</p> <p>Prior to Fall 2011, English language courses at the university were delivered through a traditional face‐to‐face format that included 7 hours of in‐person instruction each week using a popular EFL textbook series. Starting in Fall 2011, however, the university underwent a campus‐wide policy change in English instruction. English language courses were no longer taught fully face‐to‐face; instead, they were offered through a blended delivery format that combined 4 hours of weekly face‐to‐face instruction with 3 hours of weekly activities in an online learning environment, still totaling 7 hours of instruction. The blended courses used the same textbooks as in the face‐to‐face courses prior to the policy change, and the online learning activities have been developed to align with each individual lesson of the textbook to ensure that the computer assignments reflect the textbook objectives in both the topic and the proficiency level.</p> <p>The online learning environment—<emph>English Discoveries</emph>—was developed by Edusoft, technology‐mediated language learning and assessment provider. <emph>English Discoveries</emph> provides students with access to language learning resources, including interactive learning tools to aid them in their self‐exploration of learning activities. The online environment employs the following learning components: listening, reading, speaking, writing, grammar and vocabulary. Additionally, the environment has a range of online assessment tools such as lesson tests, midterm and final tests, interactive dialogs evaluated by the program's speech recognition technology, and writing assignments assessed by an automated writing evaluation tool. Tests are automatically scored by the system upon submission. Students can view their test scores after completion and compare their responses with the correct answers, thus enabling immediate feedback on their grasp of the language learning content (During the period of this study, in addition to these blended courses—which combined face‐to‐face and online instruction—a handful of fully online courses were offered to students at each level. A very small number of students took these fully online courses and they have thus been excluded from the analysis).</p> <hd id="AN0141383096-13">Course assessment criteria and comparability over time</hd> <p>Final course grades in both the face‐to‐face and online sections were assigned on a 10‐point scale, with a grade of 6.5 required to pass and continue on to the next course. The university used similar criteria to grade students and similar cut‐off scores to decide whether a student passed a course before and after the implementation of blended courses. Specifically, the final grade of the face‐to‐face and blended courses was broken down into the following four components: three exams, based on questions randomly drawn from a large test bank tied to international language learning standards (60%), assessment of speaking using a standardized rubric (10%), assessment of writing using a standardized rubric (10%), completion of homework (10%) and class participation (10%). To ensure the reliability and consistency of these component scores, the instructors were required to complete a face‐to‐face training course on how to assess writing and speaking using standardized scoring rubrics; and to pass a certification test to demonstrate their ability to use the rubrics to score students in an objective way.</p> <hd id="AN0141383096-14">Sample description</hd> <p>The data set used in this study contains detailed student‐ and course‐level information on English course enrollments between Fall 2007 and Fall 2016. Student‐level variables in the data set include whether the student passed the course, their final grade and their academic major at the time. Course‐level variables include the level, instructor, term and instructional mode of the course. After excluding courses taught in a fully online format, the data for this study include 22 023 students who took at least one English language course in this university before the policy change in Fall 2011 (ie, students who took at least one course between Fall 2007 and Summer 2011).</p> <p>Among the 22 023 students in our sample, 8814 took at least one blended course and 21 865 took at least one face‐to‐face course. A total of 8656 students took a mixture of face‐to‐face and blended courses, which provide sufficient number of observations to implement the individual fixed effects approach. Table presents the average course grade by course level as well as by the delivery format. On average, students scored highest in level 1 courses and lowest in level 2 courses, which highlights the importance to control for between‐level differences in course difficulty and grading criteria.</p> <p>Descriptive statistics of students' course performance by course format and course level</p> <p> <ephtml> <table><thead valign="top"><tr><th align="left" /><th align="center">All formats</th><th align="center">Face‐to‐face</th><th align="center">Blended</th></tr></thead><tbody><tr><td align="left">All levels</td></tr><tr><td align="left">Course grade</td><td align="char" char=".">7.728</td><td align="char" char=".">7.746</td><td align="char" char=".">7.667</td></tr><tr><td align="char" char=".">(1.334)</td><td align="char" char=".">(1.318)</td><td align="char" char=".">(1.384)</td></tr><tr><td align="left">Passing rate</td><td align="char" char=".">0.949</td><td align="char" char=".">0.951</td><td align="char" char=".">0.945</td></tr><tr><td align="left">Level 1</td></tr><tr><td align="left">Course grade</td><td align="char" char=".">8.026</td><td align="char" char=".">8.030</td><td align="char" char=".">7.250</td></tr><tr><td align="char" char=".">(1.348)</td><td align="char" char=".">(1.344)</td><td align="char" char=".">(1.836)</td></tr><tr><td align="left">Passing rate</td><td align="char" char=".">0.952</td><td align="char" char=".">0.953</td><td align="char" char=".">0.827</td></tr><tr><td align="left">Level 2</td></tr><tr><td align="left">Course grade</td><td align="char" char=".">7.511</td><td align="char" char=".">7.482</td><td align="char" char=".">7.652</td></tr><tr><td align="char" char=".">(1.321)</td><td align="char" char=".">(1.297)</td><td align="char" char=".">(1.425)</td></tr><tr><td align="left">Passing rate</td><td align="char" char=".">0.936</td><td align="char" char=".">0.936</td><td align="char" char=".">0.931</td></tr><tr><td align="left">Level 3</td></tr><tr><td align="left">Course grade</td><td align="char" char=".">7.622</td><td align="char" char=".">7.630</td><td align="char" char=".">7.605</td></tr><tr><td align="char" char=".">(1.327)</td><td align="char" char=".">(1.287)</td><td align="char" char=".">(1.399)</td></tr><tr><td align="left">Passing rate</td><td align="char" char=".">0.949</td><td align="char" char=".">0.953</td><td align="char" char=".">0.940</td></tr><tr><td align="left">Level 4</td></tr><tr><td align="left">Course grade</td><td align="char" char=".">7.715</td><td align="char" char=".">7.705</td><td align="char" char=".">7.728</td></tr><tr><td align="char" char=".">(1.272)</td><td align="char" char=".">(1.206)</td><td align="char" char=".">(1.344)</td></tr><tr><td align="left">Passing rate</td><td align="char" char=".">0.963</td><td align="char" char=".">0.969</td><td align="char" char=".">0.955</td></tr></tbody></table> </ephtml> </p> <ulist> <item>2 Note.</item> <item>3 The course grade is measured on a 10‐point scale. Students who got a score below 6.5 failed the course and had to retake the exam.</item> </ulist> <hd id="AN0141383096-15">Method</hd> <p></p> <hd id="AN0141383096-16">Basic empirical model</hd> <p>To explore whether blended delivery format positively or negatively influences students' English learning outcomes, we began with a basic ordinary least squares (OLS) model that relates student course outcomes to course delivery format. Letting <emph>i</emph> denote the individual student and <emph>c</emph> denote each course, the basic model is written as:</p> <olist> <item> <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0001" display="block" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><msub><mi>Y</mi><mi mathvariant="italic">ic</mi></msub><mo>=</mo><mi>α</mi><mo>+</mo><mi>β</mi><mi>b</mi><mi>l</mi><mi>e</mi><mi>n</mi><mi>d</mi><mi>e</mi><msub><mi>d</mi><mi mathvariant="italic">ic</mi></msub><mo>+</mo><msub><mi>ε</mi><mi mathvariant="italic">ic</mi></msub></mrow></math> </ephtml> </item> </olist> <p>where <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0002" xmlns="http://www.w3.org/1998/Math/MathML"><msub><mi>Y</mi><mi mathvariant="italic">ic</mi></msub></math> </ephtml> represents student <emph>i</emph>'s outcome in course <emph>c</emph>. We explored two course outcomes, course grade on a 10‐point scale and whether a student passed a course or not. Students who dropped out of the course were assigned a score of 0. <emph>Face‐to‐face course format</emph> is used as the reference category; "<emph>blended"</emph> is the key explanatory variable and is equal to 1 if the course is taken with the blended format. Therefore, the coefficient <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0003" xmlns="http://www.w3.org/1998/Math/MathML"><mi>β</mi></math> </ephtml> indicates the average difference in outcomes between the blended format and the face‐to‐face format. <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0004" xmlns="http://www.w3.org/1998/Math/MathML"><msub><mi>ε</mi><mi mathvariant="italic">ic</mi></msub></math> </ephtml> is the error term, which represents the unobserved student characteristics affecting the dependent variable.</p> <hd id="AN0141383096-17">Addressing student selection bias using student fixed effects approach</hd> <p>One possible challenge to drawing a causal inference using Equation 1 to predict the impacts of course delivery format on student outcomes is that there might be between‐cohort differences in student predetermined characteristics, such as English language proficiency, motivation and academic capacity. For example, if students admitted after 2011 had stronger or weaker academic preparation on average, directly comparing course outcomes between students in the face‐to‐face format (before the policy change) and the blended format (after the policy change) would result in biased estimates.</p> <p>To deal with possible bias due to between‐student variations in language skills and academic capacity, we take advantage of the panel data structure and employ an individual fixed effects approach. This approach has been widely used in the existing literature to identify the causal impacts of course delivery format on student outcomes (eg, Hart, Friedmann, & Hill, [<reflink idref="bib24" id="ref58">24</reflink>]; Xu & Jaggars, [<reflink idref="bib51" id="ref59">51</reflink>]) and we adapt this model in our current exploration. Specifically, since each student is required to take four English language courses, which typically takes students 1 to 4 years, students who were enrolled in English courses at the university between Winter 2007 and Summer 2011 are likely to take a combination of face‐to‐face and blended courses due to the policy change. The individual fixed effects model is written as:</p> <p>2 <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0005" display="block" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><msub><mi>Y</mi><mi mathvariant="italic">ic</mi></msub><mo>=</mo><mspace width="0.166667em" /><mi>a</mi><mo>+</mo><mspace width="0.166667em" /><mi>β</mi><mi>b</mi><mi>l</mi><mi>e</mi><mi>n</mi><mi>d</mi><mi>e</mi><msub><mi>d</mi><mi mathvariant="italic">ic</mi></msub><mo>+</mo><mspace width="0.166667em" /><msub><mi>σ</mi><mi>i</mi></msub><mo>+</mo><mspace width="0.166667em" /><msub><mi>υ</mi><mi mathvariant="italic">ic</mi></msub></mrow></math> </ephtml></p> <p>With the individual fixed effects model, the unobserved student characteristics affecting the dependent variable (in Equation 1) are further decomposed into two parts in Equation 2: those that are constant, such as fixed personality characteristics (<emph>σ</emph><subs><emph>i</emph></subs>), and those that vary across courses, such as time‐varying interest in ELL (<emph>υ</emph><subs><emph>ic</emph></subs>). Averaging this equation over courses for each individual <emph>i</emph> yields:</p> <p>3 <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0006" display="block" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><msub><mover accent="true"><mi>Y</mi><mo>¯</mo></mover><mi>i</mi></msub><mo>=</mo><mspace width="0.166667em" /><mover accent="true"><mi>a</mi><mo>¯</mo></mover><mo>+</mo><mspace width="0.166667em" /><mi>β</mi><msub><mover><mi mathvariant="italic">blended</mi><mo>¯</mo></mover><mi>i</mi></msub><mo>+</mo><mspace width="0.166667em" /><msub><mi>σ</mi><mi>i</mi></msub><mo>+</mo><mspace width="0.166667em" /><msub><mover accent="true"><mi>υ</mi><mo>¯</mo></mover><mi>i</mi></msub></mrow></math> </ephtml></p> <p>where <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0007" xmlns="http://www.w3.org/1998/Math/MathML"><msub><mover accent="true"><mi>Y</mi><mo>¯</mo></mover><mi>i</mi></msub></math> </ephtml>  = T<sups>‐1</sups><ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0008" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mo>∑</mo><msub><mi>Y</mi><mi mathvariant="italic">ic</mi></msub></mrow></math> </ephtml> , and so on. Because <emph>σ</emph><subs><emph>i</emph></subs> is fixed across courses, it appears in both Equations 2 and 3. Subtracting (<reflink idref="bib3" id="ref60">3</reflink>) from (<reflink idref="bib2" id="ref61">2</reflink>) for each course yields:</p> <p>4 <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0009" display="block" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><msub><mover accent="true"><mi>Y</mi><mo>¨</mo></mover><mi mathvariant="italic">ic</mi></msub><mo>=</mo><mover accent="true"><mi>a</mi><mo>¨</mo></mover><mspace width="0.166667em" /><mo>+</mo><mspace width="0.166667em" /><mi>β</mi><msub><mover accent="true"><mi mathvariant="italic">blended</mi><mo>¨</mo></mover><mi mathvariant="italic">ic</mi></msub><mo>+</mo><mspace width="0.166667em" /><msub><mover accent="true"><mi>υ</mi><mo>¨</mo></mover><mi mathvariant="italic">ic</mi></msub></mrow></math> </ephtml></p> <p>where <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0010" xmlns="http://www.w3.org/1998/Math/MathML"><msub><mover accent="true"><mi>Y</mi><mo>¨</mo></mover><mi mathvariant="italic">ic</mi></msub></math> </ephtml>  = Y<subs><emph>ic</emph></subs> ‐ <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0011" xmlns="http://www.w3.org/1998/Math/MathML"><msub><mover accent="true"><mi>Y</mi><mo>¯</mo></mover><mi>i</mi></msub></math> </ephtml> is the course‐demeaned data on course outcome Y, and so on. The important thing about Equation 4 is that through the within‐individual transformation, the individual‐level unobserved effect has disappeared. In other words, any potential unobserved bias is eliminated through the individual fixed effects model if such bias is constant across courses (Since students completed their language courses at a different pace, not all students attempted both blended and face‐to‐face courses. Across all courses in the sample (N = 73,749), about 44.6% (N = 32,875) were taken by students who took a mixture of blended and face‐to‐face courses; about 54.9% (N = 40,520) were taken by students who took an entirely face‐to‐face curriculum (ie, these students completed their language requirement before the policy change). Including students who have no variation in course delivery format does not bias the fixed effects estimator, as long as selection bias is constant within an individual. A smaller degree of within‐individual variation would be problematic if it yielded an imprecise estimator; with large sample sizes, however, this is typically less of a concern. Indeed, as shown in Table and Table , all the estimates have small standard errors and large t‐statistics, indicating that the fixed effects model is delivering precise estimates. In a robustness check, we also limited the sample to students who took a mixture of blended and face‐to‐face courses, and the resulting effect sizes and significance levels were almost identical to those reported here. For a detailed discussion of the properties of the fixed effects estimator and key assumptions underlying fixed effects models using panel data, see Wooldridge ([<reflink idref="bib49" id="ref62">49</reflink>])). Importantly, the model is now effectively comparing between online and face‐to‐face courses <emph>taken by the same student</emph>. Accordingly, the online coefficient β now explicitly represents the within‐student performance gap between face‐to‐face and blended courses.</p> <hd id="AN0141383096-18">Addressing course‐level differences using course fixed effects approach</hd> <p>Although Equation 2 controls for student characteristics that were consistent over time, it cannot account for another potential problem: Since the blended courses were offered after the policy change in 2011 when many students in our analysis sample had already completed one or multiple low‐level courses, a particular student would be more likely to take entry‐level courses with the face‐to‐face format before the policy change and take higher‐level English courses with the blended course format after the policy change. Thus, Equation 2 cannot control for differences across different levels of English courses. For example, a grade of 8.5 in the entry‐level course may have a quite different meaning compared to a similar grade in the advanced level. Indeed, descriptive statistics indicate that, on average, students scored significantly higher at level 1 (the entry level course) than they scored at the other three levels of more advanced coursework. Thus, a direct comparison within one student who had taken a combination of face‐to‐face and blended courses without taking into account between‐course variations in requirements, instructional quality and difficulty levels would then lead to biased estimates.</p> <p>To address between‐course differences, we further add course‐level fixed effects into the model, thus effectively controlling for any potential differences across course levels. Letting <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0012" xmlns="http://www.w3.org/1998/Math/MathML"><msub><mi>ρ</mi><mi>c</mi></msub></math> </ephtml> denote the unobserved course characteristics due to different course levels, the model with both individual fixed effects and course‐level fixed effects is written as:</p> <p>5 <ephtml> <math altimg="urn:x-wiley:00071013:media:bjet12797:bjet12797-math-0013" display="block" xmlns="http://www.w3.org/1998/Math/MathML"><mrow><msub><mi>Y</mi><mi mathvariant="italic">ic</mi></msub><mo>=</mo><mi>a</mi><mo>+</mo><mspace width="0.166667em" /><mi>β</mi><msub><mi mathvariant="italic">blended</mi><mi mathvariant="italic">ic</mi></msub><mo>+</mo><mspace width="0.166667em" /><msub><mi>σ</mi><mi>i</mi></msub><mo>+</mo><mspace width="0.166667em" /><msub><mi>ρ</mi><mi>c</mi></msub><mo>+</mo><mspace width="0.166667em" /><msub><mi>υ</mi><mi mathvariant="italic">ic</mi></msub></mrow></math> </ephtml></p> <hd id="AN0141383096-19">Change in student‐teacher ratio before and after the policy change</hd> <p>In addition to understanding the effectiveness of an educational program, usually measured by student performance, the cost of the program is another important factor that would influence policy makers' management decisions (Cukier, [<reflink idref="bib16" id="ref63">16</reflink>]; Levin & McEwan, [<reflink idref="bib30" id="ref64">30</reflink>]). It should be noted that the upfront costs of developing the online component for a blended course may depend on various factors, such as the existing campus technology infrastructure, and therefore may vary substantially across institutions.</p> <p>Although we are not able to provide a precise estimate of the start‐up expenditures in developing the blended course, we do have information regarding the student‐teacher ratio before and after the policy change in all the targeted English courses. One reason for the support behind the expansion of online learning is that it has the potential to help address funding insufficiencies in higher education by increasing student to teacher ratio and reducing the average cost per student for instruction (Cukier, [<reflink idref="bib16" id="ref65">16</reflink>]; Deming, Goldin, Katz, & Yuchtman, [<reflink idref="bib18" id="ref66">18</reflink>]). Since online learning does not have physical space limitations on enrollment, colleges can increase class sizes in online or blended courses as a response to changes in demand relatively easily compared to brick‐and‐mortar classrooms. Moreover, the consequence associated with increased class size on student learning may also differ substantially by course delivery format: While larger class sizes can negatively influence student learning outcomes through increased classroom disruptions in the traditional face‐to‐face setting (Lazear, [<reflink idref="bib28" id="ref67">28</reflink>]), these mechanisms would be largely muted if an online course has limited synchronous student‐instructor interactions and peer interactions (Bettinger <emph>et al.</emph>, [<reflink idref="bib7" id="ref68">7</reflink>]). Drawing on previous studies, we conducted a brief analysis of personnel costs by comparing student‐teacher ratio—calculated as the total enrollment in English proficiency courses in a year divided by the number of full‐time teachers who taught English courses in that year—before and after the policy change. This analysis could shed light on the potential of blended learning to help reduce the recurring cost of personnel by increasing class size without negatively affecting student outcomes (Bates, [<reflink idref="bib6" id="ref69">6</reflink>]; Twigg, [<reflink idref="bib44" id="ref70">44</reflink>]).</p> <hd id="AN0141383096-20">Results</hd> <p></p> <hd id="AN0141383096-21">Main results</hd> <p>We report the main results on course grades and course passing rates in Tables and. In both tables, Model 1 presents results of basic OLS while Model 2 and Model 3 successively add student fixed effects and course‐level fixed effects. Model 1 simply compares course outcomes between course formats. We found that students in the blended courses scored significantly lower than students in face‐to‐face courses, <emph>b</emph> = −0.079, <emph>p</emph> < 0.001, and were less likely to pass the course, <emph>b</emph> = −0.006, <emph>p</emph> < 0.001.</p> <p>Regression of student course grade on course format</p> <p> <ephtml> <table><thead valign="top"><tr><th align="left" /><th align="center">Model 1</th><th align="center">Model 2</th><th align="center">Model 3</th><th align="center">Model 4</th></tr></thead><tbody><tr><td align="left">Course format</td></tr><tr><td align="left">Blended</td><td align="char" char=".">−0.079<xref ref-type="fn" rid="tfn9" /></td><td align="char" char=".">−0.100<xref ref-type="fn" rid="tfn9" /></td><td align="char" char=".">0.409<xref ref-type="fn" rid="tfn9" /></td><td align="char" char=".">0.222</td></tr><tr><td align="char" char=".">(0.012)</td><td align="char" char=".">(0.012)</td><td align="char" char=".">(0.013)</td><td align="char" char=".">(0.138)</td></tr><tr><td align="left">Course level</td></tr><tr><td align="left">Level 2</td><td align="left" /><td align="char" char="." /><td align="char" char=".">−0.746<xref ref-type="fn" rid="tfn9" /></td><td align="char" char=".">−0.773<xref ref-type="fn" rid="tfn9" /></td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.010)</td><td align="char" char=".">(0.010)</td></tr><tr><td align="left">Level 3</td><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">−0.803<xref ref-type="fn" rid="tfn9" /></td><td align="char" char=".">−0.780<xref ref-type="fn" rid="tfn9" /></td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.011)</td><td align="char" char=".">(0.012)</td></tr><tr><td align="left">Level 4</td><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">−0.807<xref ref-type="fn" rid="tfn9" /></td><td align="char" char=".">−0.797<xref ref-type="fn" rid="tfn9" /></td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.012)</td><td align="char" char=".">(0.013)</td></tr><tr><td align="left">Interaction between course format and course level</td></tr><tr><td align="left">Blended × Level 2</td><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">0.345<xref ref-type="fn" rid="tfn8" /></td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.138)</td></tr><tr><td align="left">Blended × Level 3</td><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">0.117</td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.138)</td></tr><tr><td align="left">Blended × Level 4</td><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">0.164</td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.138)</td></tr><tr><td align="left">Constant</td><td align="char" char=".">7.746<xref ref-type="fn" rid="tfn9" /></td><td align="char" char=".">7.751<xref ref-type="fn" rid="tfn9" /></td><td align="char" char=".">8.197<xref ref-type="fn" rid="tfn9" /></td><td align="char" char=".">8.198<xref ref-type="fn" rid="tfn9" /></td></tr><tr><td align="char" char=".">(0.006)</td><td align="char" char=".">(0.004)</td><td align="char" char=".">(0.007)</td><td align="char" char=".">(0.007)</td></tr><tr><td align="left">N</td><td align="char" char=".">73749</td><td align="char" char=".">73749</td><td align="char" char=".">73749</td><td align="char" char=".">73749</td></tr><tr><td align="left">R<sup>2</sup></td><td align="char" char=".">0.001</td><td align="char" char=".">0.001</td><td align="char" char=".">0.130</td><td align="char" char=".">0.131</td></tr></tbody></table> </ephtml> </p> <ulist> <item>6 Note.</item> <item>7 Standard errors in parentheses. Face‐to‐face format is the baseline category. Model 2 used individual fixed effects. Model 3 and Model 4 used both individual fixed effects and course‐level fixed effects.</item> <item>8 * <emph>p</emph> < 0.05;</item> <item>9 *** <emph>p</emph> < 0.001.</item> </ulist> <p>Regression of passing the course on course format</p> <p> <ephtml> <table><thead valign="top"><tr><th align="left" /><th align="center">Model 1</th><th align="center">Model 2</th><th align="center">Model 3</th><th align="center">Model 4</th></tr></thead><tbody><tr><td align="left">Course format</td></tr><tr><td align="left">Blended</td><td align="char" char=".">−0.006<xref ref-type="fn" rid="tfn12" /></td><td align="char" char=".">−0.002</td><td align="char" char=".">0.032<xref ref-type="fn" rid="tfn12" /></td><td align="char" char=".">0.009</td></tr><tr><td align="char" char=".">(0.002)</td><td align="char" char=".">(0.002)</td><td align="char" char=".">(0.003)</td><td align="char" char=".">(0.029)</td></tr><tr><td align="left">Course level</td></tr><tr><td align="left">Level 2</td><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">−0.049<xref ref-type="fn" rid="tfn12" /></td><td align="char" char=".">−0.050<xref ref-type="fn" rid="tfn12" /></td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.002)</td><td align="char" char=".">(0.002)</td></tr><tr><td align="left">Level 3</td><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">−0.055<xref ref-type="fn" rid="tfn12" /></td><td align="char" char=".">−0.054<xref ref-type="fn" rid="tfn12" /></td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.002)</td><td align="char" char=".">(0.002)</td></tr><tr><td align="left">Level 4</td><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">−0.053<xref ref-type="fn" rid="tfn12" /></td><td align="char" char=".">−0.052<xref ref-type="fn" rid="tfn12" /></td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.003)</td><td align="char" char=".">(0.003)</td></tr><tr><td align="left">Interaction between course format and course level</td></tr><tr><td align="left">Blended × Level 2</td><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">0.032</td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.029)</td></tr><tr><td align="left">Blended × Level 3</td><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">0.021</td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.029)</td></tr><tr><td align="left">Blended × Level 4</td><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">0.021</td></tr><tr><td align="char" char="." /><td align="char" char="." /><td align="char" char="." /><td align="char" char=".">(0.029)</td></tr><tr><td align="left">Constant</td><td align="char" char=".">0.951<xref ref-type="fn" rid="tfn12" /></td><td align="char" char=".">0.950<xref ref-type="fn" rid="tfn12" /></td><td align="char" char=".">0.979<xref ref-type="fn" rid="tfn12" /></td><td align="char" char=".">0.979<xref ref-type="fn" rid="tfn12" /></td></tr><tr><td align="char" char=".">(0.001)</td><td align="char" char=".">(0.001)</td><td align="char" char=".">(0.001)</td><td align="char" char=".">(0.001)</td></tr><tr><td align="left">N</td><td align="char" char=".">73749</td><td align="char" char=".">73749</td><td align="char" char=".">73749</td><td align="char" char=".">73749</td></tr><tr><td align="left">R<sup>2</sup></td><td align="char" char=".">0.000</td><td align="char" char=".">0.000</td><td align="char" char=".">0.014</td><td align="char" char=".">0.014</td></tr></tbody></table> </ephtml> </p> <ulist> <item>10 Note.</item> <item>11 Standard errors in parentheses. Face‐to‐face format is the baseline category. Model 2 used individual fixed effects. Model 3 and Model 4 used both individual fixed effects and course‐level fixed effects.</item> <item>12 *** <emph>p</emph> < 0.001.</item> </ulist> <p>Model 1 does not control for any student characteristics, while Model 2 compares course outcomes between blended and face‐to‐face courses taken by the same student. The relationship remains significantly negative for course grades. On average, students scored 0.100 points lower in blended courses than in face‐to‐face courses. In terms of course passing rate, there is a small and insignificant difference between blended and face‐to‐face courses after controlling for student characteristics that were consistent over time, <emph>b</emph> = −0.002. However, as discussed previously in the method section, this is partly due to the fact that different levels of courses were taken through the face‐to‐face versus blended format: Since blended courses were initially offered in 2011 when some of the earlier cohorts have already taken lower‐level courses through the face‐to‐face format, the face‐to‐face courses consists a higher proportion of lower‐level courses. Model (<reflink idref="bib2" id="ref71">2</reflink>) that does not control for the level of the course would yield biased estimates in favor of face‐to‐face courses because lower‐level courses, which are easier and have higher average grade, are more likely to be taken through the face‐to‐face format than the blended format.</p> <p>Model 3 addresses this problem by further controlling for course‐level fixed effects. When controlling for fixed effects at both the student and course levels, the estimated impacts of blended formation relative to traditional face‐to‐face delivery format become positive for both course grades and course passing rate. Specifically, compared to face‐to‐face course format, the blended format is on average associated with a higher course completion rate by 3.2 percentage points. The blended format is also associated with higher average course grades by 0.409 points, suggesting that students' course grades in blended courses were, on average, 0.409 points higher than their performance in the face‐to‐face courses. With a pooled standard deviation of students' grades in face‐to‐face and blended courses being 1.335, the mean difference between the two conditions would then translate into a standardized difference of 0.306 standard deviations, which is considered to be a moderate effect according to Cohen ([<reflink idref="bib12" id="ref72">12</reflink>]).</p> <hd id="AN0141383096-22">The effect of blended learning by course levels</hd> <p>We further analyzed if the effect of blended course format varied across course levels by adding an interaction term between blended course format and course level (see Model 4 in Tables and). The results show that, taking a course in a blended course format significantly increases a student's course grade and likelihood to pass at any of the course levels except for level 1. The significant interaction term suggests that the effect of course format changes significantly as the course level changes. Results presented in Table show that blended course format increased course grades by significantly more points at course Level 2 than at course Level 1. On the contrary, for course passing rate, the results in Table show that there is no significant difference in the impact of blended course format between course levels. These results indicate that the advantage of taking a blended course varies by course levels and course outcomes.</p> <hd id="AN0141383096-23">Changes in student‐teacher ratio</hd> <p>Although we found that blended learning resulted in better course outcomes as compared to face‐to‐face instruction, the question remains whether these better course outcomes were achieved with equal or even lower instructional personnel costs. Therefore, we further examined the cost of the two course formats by measuring the enrollment‐teacher ratio before and after the policy change. The results in Table show that, before the policy change, the annual enrollment‐teacher ratio ranged from 101.98 to 145.55. After the policy change, the annual enrollment‐teacher ratio ranged from 135.28 to 148.99. Although the annual enrollment‐teacher ratio varied by year, we still see some increase in enrollee to teacher ratio after the policy change. On average, one teacher was assigned with 126.95 enrollees before the policy change; however, the number increased approximately 13% to an average of 142.96 after the policy change. These results suggest that, after using the online learning system, the instructional personnel spend per students decreased while the course outcomes improved.</p> <p>Enrollment‐teacher ratio at the university from 2007 to 2014</p> <p> <ephtml> <table><thead valign="top"><tr><th align="left" /><th align="center">Course Enrollments</th><th align="center">Number of teachers</th><th align="center">Enrollment‐Teacher ratio</th></tr></thead><tbody><tr><td align="left">Before 2011</td><td align="left" /><td align="char" char="." /><td align="char" char=".">126.95</td></tr><tr><td align="left">2007</td><td align="left">11 524</td><td align="char" char=".">113</td><td align="char" char=".">101.98</td></tr><tr><td align="left">2008</td><td align="left">15 464</td><td align="char" char=".">122</td><td align="char" char=".">126.75</td></tr><tr><td align="left">2009</td><td align="left">17 625</td><td align="char" char=".">132</td><td align="char" char=".">133.52</td></tr><tr><td align="left">2010</td><td align="left">20 377</td><td align="char" char=".">140</td><td align="char" char=".">145.55</td></tr><tr><td align="left">After 2011</td><td align="left" /><td align="char" char="." /><td align="char" char=".">142.96</td></tr><tr><td align="left">2011</td><td align="left">21 413</td><td align="char" char=".">146</td><td align="char" char=".">146.66</td></tr><tr><td align="left">2012</td><td align="left">22 051</td><td align="char" char=".">148</td><td align="char" char=".">148.99</td></tr><tr><td align="left">2013</td><td align="left">21 137</td><td align="char" char=".">150</td><td align="char" char=".">140.91</td></tr><tr><td align="left">2014</td><td align="left">20 562</td><td align="char" char=".">152</td><td align="char" char=".">135.28</td></tr></tbody></table> </ephtml> </p> <hd id="AN0141383096-24">Discussion and conclusions</hd> <p>The primary aim of this study was to examine the impact of blended instruction on university students' language learning outcomes in a context where language learning resources are constrained (Davies, [<reflink idref="bib17" id="ref73">17</reflink>]; Lemus <emph>et al.</emph>, [<reflink idref="bib29" id="ref74">29</reflink>]; World Economic Forum, [<reflink idref="bib50" id="ref75">50</reflink>]). Taking advantage of a policy change in a large university in Mexico and using a model that controls for both course‐ and student‐level fixed effects, we found that blended learning is associated with higher probability of passing a course by more than 3 percentage points, as well as better course grades by an average of 0.409 points on a 10‐point scale. The impact on course grade—corresponding to a moderate effect a 0.306 standard deviations—is particularly promising.</p> <p>The positive language learning outcomes from blended instruction are consistent with a number of other quantitative studies conducted in other settings, especially in developing countries where language learning resources are limited (eg, Bai, Mo, Zhang, Boswell, & Rozelle, [<reflink idref="bib2" id="ref76">2</reflink>]). These results, together with the findings from our study, therefore support using blended learning as a promising strategy to leapfrog resource constraints and disrupt the status quo in language learning.</p> <p>To provide insight into the instructional personnel cost of applying blended instruction compared to traditional face‐to‐face instruction, we examined the change in student‐teacher ratio before and after the implementation of blended courses in the targeted English courses. The results lend further support to using blended instruction. Specifically, in additional to the better course outcomes associated with blended instruction, the average enrollment‐teacher ratio increased by nearly 13% after the university switched to the blended delivery format. The possibility that blended learning might have the potential to boost students' language learning outcomes with lower recurring personnel costs echoes a number of existing studies that compared the cost effectiveness of technology‐mediated learning to traditional face‐to‐face instruction (eg, Cohen & Nachmias, [<reflink idref="bib13" id="ref77">13</reflink>]; Muralidharan, Niehaus, & Sukhtankar, [<reflink idref="bib33" id="ref78">33</reflink>]).</p> <p>These findings provide evidence that blended learning can help improve ELL while lessening costs in a higher education context of a developing country. Specifically, this study contributes to the literature on online language learning in several important ways. First, based on a rigorous quasi‐experimental design, our study is one of the first large‐scale studies that provide evidence on the impacts of online language learning on student outcomes at the postsecondary education level in Mexico—a previously understudied but critically important setting. By answering a highly policy‐relevant research question—whether replacing part of a traditional face‐to‐face course with a well‐developed online program may positively impact learning outcomes of college English language learners—the results of our study provide important implications for policy options to overcome resource constraints and address the needs of students at various proficiency levels in developing countries. More specifically, colleges in developing countries that have insufficient high‐quality English instructors may consider adopting well‐designed online language learning technological tools and integrating these tools in English instruction as important language learning resources to students. Although integrating these technological platforms may require substantial upfront costs, the blended learning delivery format may serve as a cost‐saving strategy in the long run compared to traditional face‐to‐face instruction.</p> <p>More importantly, our results may speak even more broadly to the potential for using new technologies to enable developing countries to leapfrog constraints to development. A handful of studies have provided examples of such technology‐enabled leapfrogging in other sectors, such as the use of biometric authentication to circumvent literacy constraints to financial inclusion (Muralidharan <emph>et al.</emph>, [<reflink idref="bib33" id="ref79">33</reflink>]). Our study adds to this growing literature by providing evidence on using technology to address resource constraints in language instruction at the postsecondary education level.</p> <p>Finally, the current study performs a brief cost analysis to examine whether a technology‐enhanced blended delivery model of English language courses implemented over a period of 4 years could potentially reduce instructional personnel costs compared to traditional face‐to‐face courses. The examination of the student‐teacher ratio provides important information to policy makers regarding the potential of blended courses to reduce recurring costs through larger class size while in the meantime not to negatively impact student learning outcomes.</p> <p>Our study is also subject to several limitations. The current study examined student outcomes in one university in Mexico based on quasi‐experimental methods. While we controlled for unobserved characteristics that are fixed either at the individual level or at the course level, the estimates are still subject to potential bias if there are unobserved factors that vary over time. It is also unclear the extent to which the results identified at this particular university may be generalized to other institutions in Mexico and other developing countries. Further studies of online learning, in more diverse settings and with random assignment of students, will be required to confirm the potential benefits of blended learning. Moreover, while this study seeks to understand the cost‐effectiveness of blended instruction as compared to face‐to‐face instruction, we were only able to examine the change in student‐teacher ratio due to data limitation. It will be helpful for future research to conduct more comprehensive cost‐effectiveness analysis that takes into account both start‐up expenditures in course development and in recurring expenditures in delivering the course.</p> <hd id="AN0141383096-25">Acknowledgements</hd> <p>We would like to thank Beatriz Acosta of the Universidad de Autonoma de Chihuahua for helping make available the data for this study as well as providing detailed information on the study context.</p> <hd id="AN0141383096-26">Statements on open data, ethics and conflict of interest</hd> <p>A de‐identified data set is available by request from the corresponding author. The authors have received approval from the relevant institutional ethics committee. One of the authors is employed by the company that produces the platform discussed in this paper. To mitigate any potential conflict of interest, this author was involved only in writing the introduction, literature review and research context sections. The other authors were solely responsible for data collection, statistical analyses and data interpretation.</p> <ref id="AN0141383096-27"> <title> References </title> <blist> <bibl id="bib1" idref="ref44" type="bt">1</bibl> <bibtext> Al‐Nasrawi, S., & Zoughbi, S. (2015). Information society, digital divide, and e‐governance in developing Countries. In Encyclopedia of information science and technology, (3rd edition, pp. 6827 – 6835). IGI Global. https://doi.org/10.4018/978-1-4666-5888-2.ch672</bibtext> </blist> <blist> <bibl id="bib2" idref="ref61" type="bt">2</bibl> <bibtext> Bai, Y., Mo, D., Zhang, L., Boswell, M., & Rozelle, S. (2016). The impact of integrating ICT with teaching: Evidence from a randomized controlled trial in rural schools in China. Computers & Education, 96, 1 – 14. https://doi.org/10.1016/j.compedu.2016.02.005</bibtext> </blist> <blist> <bibl id="bib3" idref="ref49" type="bt">3</bibl> <bibtext> Bañados, E. (2006). A blended‐learning pedagogical model for teaching and learning EFL successfully through an online interactive multimedia environment. CALICO Journal, 23 (3), 533 – 550. https://doi.org/10.1558/cj.v23i3.533-550</bibtext> </blist> <blist> <bibl id="bib4" idref="ref6" type="bt">4</bibl> <bibtext> Banerjee, A. V., & Duflo, E. (2011). Poor economics: A radical rethinking of the way to fight global poverty. New York, NY : Public Affairs.</bibtext> </blist> <blist> <bibl id="bib5" idref="ref50" type="bt">5</bibl> <bibtext> Barani, G. (2011). The relationship between computer assisted language learning (CALL) and listening skill of Iranian EFL learners. Procedia‐Social and Behavioral Sciences, 15, 4059 – 4063. https://doi.org/10.1016/j.sbspro.2011.04.414</bibtext> </blist> <blist> <bibl id="bib6" idref="ref69" type="bt">6</bibl> <bibtext> Bates, A. (1995). Technology, open learning and distance education. London, UK : Routledge.</bibtext> </blist> <blist> <bibl id="bib7" idref="ref68" type="bt">7</bibl> <bibtext> Bettinger, E. P., Fox, L., Loeb, S., & Taylor, E. S. (2017). Virtual classrooms: How online college courses affect student success. American Economic Review, 107 (9), 2855 – 2875.</bibtext> </blist> <blist> <bibl id="bib8" idref="ref56" type="bt">8</bibl> <bibtext> Brunner, J., Santiago, P., Guadilla, C., Gerlach, J., & Velho, L. (2008). OECD reviews of tertiary education‐Mexico. Retrieved from OECD website, <ulink href="http://www.oecd.org/dataoecd/22/49/37746196.pdf">http://www.oecd.org/dataoecd/22/49/37746196.pdf</ulink></bibtext> </blist> <blist> <bibl id="bib9" idref="ref14" type="bt">9</bibl> <bibtext> Burfisher, M. E., Robinson, S., & Thierfelder, K. (2001). The impact of NAFTA on the United States. Journal of Economic perspectives, 15 (1), 125 – 144. https://doi.org/10.1257/jep.15.1.125</bibtext> </blist> <blist> <bibtext> Chang, S. L., Gregory, N. A., & Shak, P. (2018). An experimental study on using instructional pronunciation video to improve students' pronunciation. GSTF Journal on Education, 4 (2). https://doi.org/10.5176/2345-7163_4.2.106</bibtext> </blist> <blist> <bibtext> Chen, L. M., & Zhang, R. (2010). Web‐based CALL to listening comprehension. Current Issues in Education, 13 (4), Retrieved from <ulink href="http://cie.asu.edu/">http://cie.asu.edu/</ulink></bibtext> </blist> <blist> <bibtext> Cohen, J. (2013). Statistical power analysis for the behavioral sciences (2nd ed). Hoboken: Taylor and Francis.</bibtext> </blist> <blist> <bibtext> Cohen, A., & Nachmias, R. (2012). The implementation of a cost effectiveness analyzer for web‐supported academic instruction: An example from Life Science. International Journal on E‐Learning, 11 (2), 107 – 124. Retrieved from https://<ulink href="http://www.learntechlib.org/primary/p/34113/">www.learntechlib.org/primary/p/34113/</ulink></bibtext> </blist> <blist> <bibtext> British Council. (2015). English in Mexico: An examination of policy, perceptions and influencing factors. Mexico City : Author. Retrieved from https://ei.britishcouncil.org/sites/default/files/latin-americaresearch/English%20in%20Mexico.pdf</bibtext> </blist> <blist> <bibtext> Cronquist, K., & Fiszbein, A. (2017). English language learning in Latin America. Inter‐American dialogue: Retrieved from https://<ulink href="http://www.thedialogue.org/wp-content/uploads/2017/09/English-Language-Learning-in-Latin-America-Final-1.pdf">www.thedialogue.org/wp-content/uploads/2017/09/English-Language-Learning-in-Latin-America-Final-1.pdf</ulink></bibtext> </blist> <blist> <bibtext> Cukier, J. (1997). Cost‐benefit analysis of telelearning: Developing a methodology framework. Distance education, 18 (1), 137 – 152. https://doi.org/10.1080/0158791970180110</bibtext> </blist> <blist> <bibtext> Davies, P. (2009). Strategic management of ELT in public educational systems: Trying to reduce failure, increase success. The Electronic Journal for English as a Second Language, 13 (3), n3. Retrieved from https://files.eric.ed.gov/fulltext/EJ898201.pdf</bibtext> </blist> <blist> <bibtext> Deming, D. J., Goldin, C., Katz, L. F., & Yuchtman, N. (2015). Can online learning bend the higher education cost curve? American Economic Review, 105 (5), 496 – 501. https://doi.org/10.1257/aer.p20151024</bibtext> </blist> <blist> <bibtext> Dietrich, S. E. (2007). Professional development and English language teaching in Tamaulipas: Describing the training and challenges of two groups of teachers. MEXTESOL Journal, 31 (2), 31 – 36. Retrieved from <ulink href="http://mextesol.net/journal/public/files/fff70285c37848ffa782bd6855a2dd43.pdf">http://mextesol.net/journal/public/files/fff70285c37848ffa782bd6855a2dd43.pdf</ulink></bibtext> </blist> <blist> <bibtext> Educando. (2018). IAPE: Inter‐American partnership for education. Retrieved from https://worldfund.org/site/iape/</bibtext> </blist> <blist> <bibtext> First, Education (2018). Education first English proficiency index. Retrieved from https://<ulink href="http://www.ef.edu/%5f%5f/~/media/centralefcom/epi/downloads/full-reports/v8/ef-epi-2018-english.pdf">www.ef.edu/%5f%5f/~/media/centralefcom/epi/downloads/full-reports/v8/ef-epi-2018-english.pdf</ulink></bibtext> </blist> <blist> <bibtext> Glewwe, P., Kremer, M., & Moulin, S. (2009). Many children left behind? Textbooks and test scores in Kenya. American Economic Journal: Applied Economics, 1 (1), 112 – 135. https://doi.org/10.1257/app.1.1.112</bibtext> </blist> <blist> <bibtext> Glick, D., & Davidson, T. (2012). Four types of interaction to overcome challenges of e‐learning solutions. In J. Jia (Ed.), Educational, stages and interactive learning: From Kindergarten to Workplace Training (pp. 407 – 426). Hershey, PA : IGI Publishing. https://doi.org/10.4018/978-1-4666-0137-6.ch022</bibtext> </blist> <blist> <bibtext> Hart, C., Friedmann, E., & Hill, M. (2017). Online course‐taking and student outcomes in California community colleges. Education Finance and Policy, 13 (1), 42 – 71. https://doi.org/10.1162/edfp_a_00218</bibtext> </blist> <blist> <bibtext> Johnson, C. R. (2006). El papel de la motivación, la ansiedad y la auto eficacia del estudiante básico en la adquisición del inglés como lengua extranjera de estudiantes mexicanos universitarios (The role of motivation, anxiety and self‐efficacy of the basic student in the acquisition of English as a foreign language of Mexican university students). Unpublished Ph. D. dissertation. Puebla, Mexico : Benemérita Universidad Autónoma de Puebla.</bibtext> </blist> <blist> <bibtext> Krueger, A. O. (1999). Trade creation and trade diversion under NAFTA (No. w7429). National Bureau of Economic Research, https://doi.org/10.3386/w7429</bibtext> </blist> <blist> <bibtext> Ku, H., & Zussman, A. (2010). Lingua franca: The role of English in international trade. Journal of Economic Behavior & Organization, 75 (2), 250 – 260. https://doi.org/10.1016/j.jebo.2010.03.013</bibtext> </blist> <blist> <bibtext> Lazear, E. P. (2001). Educational production. The Quarterly Journal of Economics, 116 (3), 777 – 803. https://doi.org/10.1162/00335530152466232</bibtext> </blist> <blist> <bibtext> Lemus, M. E., Durán, K., & Martinez, M. (2008, October). El nivel de inglés y su problemática en tres estados geográficamente distantes. Paper presented at the Foro Nacional de Estudios en Lenguas (FONAEL), Quintana Roo, México.</bibtext> </blist> <blist> <bibtext> Levin, H. M., & McEwan, P. J. (2000). Cost‐effectiveness analysis: Methods and applications, 2nd ed. Thousand Oaks, CA : Sage Publications.</bibtext> </blist> <blist> <bibtext> Liu, X. (2015). The effectiveness of using CALL environment on reading skills of English learners. Paper presented at the International Conference on Management Science, Education Technology, Arts, Social Science and Economics (MSETASSE), Qingdao, China. https://doi.org/10.2991/msetasse-15.2015.217</bibtext> </blist> <blist> <bibtext> Marzban, A. (2011). Improvement of reading comprehension through computer‐assisted language learning in Iranian intermediate EFL students. Procedia Computer Science, 3, 3 – 10. https://doi.org/10.1016/j.procs.2010.12.003</bibtext> </blist> <blist> <bibtext> Muralidharan, K., Niehaus, P., & Sukhtankar, S. (2016). Building state capacity: Evidence from biometric smartcards in India. American Economic Review, 106 (10), 2895 – 2929.</bibtext> </blist> <blist> <bibtext> OECD. (2015). Education at a glance 2015: OECD indicators. Paris : Author. https://doi.org/10.1787/eag-2015-en</bibtext> </blist> <blist> <bibtext> OECD. (2017). OECD skills strategy diagnostic report: Mexico 2017. OECD Skills Studies : Author.</bibtext> </blist> <blist> <bibtext> Porter, W. W., Graham, C. R., Bodily, R. G., & Sandberg, D. S. (2016). A qualitative analysis of institutional drivers and barriers to blended learning adoption in higher education. The Internet and Higher Education, 28, 17 – 27. https://doi.org/10.1016/j.iheduc.2015.08.003</bibtext> </blist> <blist> <bibtext> Pritchett, L., & Beatty, A. (2015). Slow down, you're going too fast: Matching curricula to student skill levels. International Journal of Educational Development, 40, 276 – 288. https://doi.org/10.1016/j.ijedudev.2014.11.013</bibtext> </blist> <blist> <bibtext> Sayer, P. (2012). Ambiguities and tensions in English language teaching: Portraits of EFL teachers as legitimate speakers. London, UK : Routledge.</bibtext> </blist> <blist> <bibtext> Sayer, P. (2015). Expanding global language education in public primary schools: The national English programme in Mexico. Language, Culture and Curriculum, 28 (3), 257 – 275. https://doi.org/10.1080/07908318.2015.1102926</bibtext> </blist> <blist> <bibtext> Sife, A., Lwoga, E., & Sanga, C. (2007). New technologies for teaching and learning: Challenges for higher learning institutions in developing countries. International Journal of Education and Development Using ICT, 3 (2), 57 – 67. Retrieved from https://<ulink href="http://www.learntechlib.org/p/42360/">www.learntechlib.org/p/42360/</ulink></bibtext> </blist> <blist> <bibtext> Stracke, E. (2007). A road to understanding: A qualitative study into why learners drop out of a blended language learning (BLL) environment. ReCALL, 19 (1), 57 – 78. https://doi.org/10.1017/S0958344007000511</bibtext> </blist> <blist> <bibtext> Talebi, F., & Teimoury, N. (2013). The effect of computer‐assisted language learning on improving EFL learners' pronunciation ability. World Journal of English Language, 3 (2), 52. https://doi.org/10.5430/wjel.v3n2p52</bibtext> </blist> <blist> <bibtext> Tosun, S. (2015). The effects of blended learning on EFL students' vocabulary enhancement. Procedia‐Social and Behavioral Sciences, 199, 641 – 647. https://doi.org/10.1016/j.sbspro.2015.07.592</bibtext> </blist> <blist> <bibtext> Twigg, C. A. (2000). Course readiness criteria: Identifying targets of opportunity for large‐scale redesign. Educause Review, 35 (3), 40 – 44.</bibtext> </blist> <blist> <bibtext> UNDP Human Development Index. (2018). Human development indices and indicators: 2018 statistical update. Retrieved from <ulink href="http://hdr.undp.org/sites/default/files/2018%5fhuman%5fdevelopment%5fstatistical%5fupdate.pdf">http://hdr.undp.org/sites/default/files/2018%5fhuman%5fdevelopment%5fstatistical%5fupdate.pdf</ulink></bibtext> </blist> <blist> <bibtext> Vahdat, S., & Eidipour, M. (2016). Adopting CALL to improve listening comprehension of Iranian junior high school students. Theory and Practice in Language Studies, 6 (8), 1609 – 1617. https://doi.org/10.17507/tpls.0608.13</bibtext> </blist> <blist> <bibtext> Vazques, A., Guzmán, N., & Roux, R. (2013). Can ELT in higher education be successful? The current status of ELT in Mexico. The Electronic Journal for English as a Second Language, 17 (1), 1 – 26. Retrieved from <ulink href="http://www.tesl-ej.org/pdf/ej65/a2.pdf">http://www.tesl-ej.org/pdf/ej65/a2.pdf</ulink></bibtext> </blist> <blist> <bibtext> Weinberg, A., & Knoerr, H. (2003). Learning French pronunciation: Audiocassettes or multimedia? CALICO Journal, 19 (2), 315 – 336. https://doi.org/10.1558/cj.v20i2.315-336</bibtext> </blist> <blist> <bibtext> Wooldridge, J. (2002). Econometric analysis of cross section and panel data. Cambridge, MA : MIT Press.</bibtext> </blist> <blist> <bibtext> World Economic Forum. (2015). The global competitiveness report 2015–2016. Geneva, Switzerland : World Economic Forum.</bibtext> </blist> <blist> <bibtext> Xu, D., & Jaggars, S. S. (2014). Performance gaps between online and face‐to‐face courses: Differences across types of students and academic subject areas. Journal of Higher Education, 5 (5), 633 – 659. https://doi.org/10.1080/00221546.2014.11777343</bibtext> </blist> </ref> <aug> <p>By Di Xu; Danny Glick; Fernando Rodriguez; Bianca Cung; Qiujie Li and Mark Warschauer</p> <p>Reported by Author; Author; Author; Author; Author; Author</p> <p></p> <p>Di Xu is an assistant professor of Educational Policy and Social Context at the University of California, Irvine. She holds a PhD in economics and education from Columbia University. Her research examines the impacts of educational programs and policies on student academic performance, persistence and degree completion at the postsecondary education level, with a particular focus on students from disadvantaged backgrounds.</p> <p>Danny Glick is a research affiliate at the University of California, Irvine's Digital Learning Lab, and Director of Pedagogy and Research at Edusoft, a subsidiary of Educational Testing Service (ETS). He holds a PhD in Learning Technologies from Ben‐Gurion University. He specializes in developing early warning systems to identify at‐risk students in online courses using learning analytics.</p> <p>Fernando Rodriguez is a postdoctoral scholar in the School of Education at the University of California, Irvine. He holds a PhD in Educational Psychology from the University of Michigan. He examines how online learning environments shape the learning experience in college STEM courses.</p> <p>Bianca Cung is a PhD student specializing in Language, Literacy and Technology. She received her undergraduate degree at UCLA. Bianca's research interests include but are not limited to media and technology for education, data mining and STEM education.</p> <p>Qiujie Li is a PhD candidate in the Department of Education at the University of California, Irvine. She completed her BA in Educational Technology in 2011 at Beijing Normal University, and received an MA in Distance Education from the same university in 2014. Her research focuses on learning analytics and online learning environment design.</p> <p>Mark Warschauer is a professor of Education and Informatics at the University of California, Irvine, and director of the Digital Learning Lab (DLL) at UC Irvine where he works on a range of research projects related to digital media in education. He is founding editor of Language Learning & Technology journal and has been appointed inaugural editor of AERA Open.</p> </aug> <nolink nlid="nl1" bibid="bib15" firstref="ref1"></nolink> <nolink nlid="nl2" bibid="bib17" firstref="ref2"></nolink> <nolink nlid="nl3" bibid="bib25" firstref="ref3"></nolink> <nolink nlid="nl4" bibid="bib34" firstref="ref4"></nolink> <nolink nlid="nl5" bibid="bib35" firstref="ref5"></nolink> <nolink nlid="nl6" bibid="bib22" firstref="ref7"></nolink> <nolink nlid="nl7" bibid="bib37" firstref="ref8"></nolink> <nolink nlid="nl8" bibid="bib40" firstref="ref9"></nolink> <nolink nlid="nl9" bibid="bib14" firstref="ref10"></nolink> <nolink nlid="nl10" bibid="bib27" firstref="ref13"></nolink> <nolink nlid="nl11" bibid="bib26" firstref="ref15"></nolink> <nolink nlid="nl12" bibid="bib39" firstref="ref18"></nolink> <nolink nlid="nl13" bibid="bib50" firstref="ref21"></nolink> <nolink nlid="nl14" bibid="bib21" firstref="ref23"></nolink> <nolink nlid="nl15" bibid="bib29" firstref="ref24"></nolink> <nolink nlid="nl16" bibid="bib20" firstref="ref28"></nolink> <nolink nlid="nl17" bibid="bib47" firstref="ref32"></nolink> <nolink nlid="nl18" bibid="bib38" firstref="ref34"></nolink> <nolink nlid="nl19" bibid="bib19" firstref="ref35"></nolink> <nolink nlid="nl20" bibid="bib36" firstref="ref39"></nolink> <nolink nlid="nl21" bibid="bib41" firstref="ref40"></nolink> <nolink nlid="nl22" bibid="bib23" firstref="ref41"></nolink> <nolink nlid="nl23" bibid="bib42" firstref="ref42"></nolink> <nolink nlid="nl24" bibid="bib45" firstref="ref45"></nolink> <nolink nlid="nl25" bibid="bib48" firstref="ref46"></nolink> <nolink nlid="nl26" bibid="bib10" firstref="ref48"></nolink> <nolink nlid="nl27" bibid="bib11" firstref="ref51"></nolink> <nolink nlid="nl28" bibid="bib46" firstref="ref52"></nolink> <nolink nlid="nl29" bibid="bib43" firstref="ref53"></nolink> <nolink nlid="nl30" bibid="bib31" firstref="ref54"></nolink> <nolink nlid="nl31" bibid="bib32" firstref="ref55"></nolink> <nolink nlid="nl32" bibid="bib24" firstref="ref58"></nolink> <nolink nlid="nl33" bibid="bib51" firstref="ref59"></nolink> <nolink nlid="nl34" bibid="bib49" firstref="ref62"></nolink> <nolink nlid="nl35" bibid="bib16" firstref="ref63"></nolink> <nolink nlid="nl36" bibid="bib30" firstref="ref64"></nolink> <nolink nlid="nl37" bibid="bib18" firstref="ref66"></nolink> <nolink nlid="nl38" bibid="bib28" firstref="ref67"></nolink> <nolink nlid="nl39" bibid="bib44" firstref="ref70"></nolink> <nolink nlid="nl40" bibid="bib12" firstref="ref72"></nolink> <nolink nlid="nl41" bibid="bib13" firstref="ref77"></nolink> <nolink nlid="nl42" bibid="bib33" firstref="ref78"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1240961
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Does Blended Instruction Enhance English Language Learning in Developing Countries? Evidence from Mexico
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Xu%2C+Di%22">Xu, Di</searchLink><br /><searchLink fieldCode="AR" term="%22Glick%2C+Danny%22">Glick, Danny</searchLink><br /><searchLink fieldCode="AR" term="%22Rodriguez%2C+Fernando%22">Rodriguez, Fernando</searchLink><br /><searchLink fieldCode="AR" term="%22Cung%2C+Bianca%22">Cung, Bianca</searchLink><br /><searchLink fieldCode="AR" term="%22Li%2C+Qiujie%22">Li, Qiujie</searchLink><br /><searchLink fieldCode="AR" term="%22Warschauer%2C+Mark%22">Warschauer, Mark</searchLink>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22British+Journal+of+Educational+Technology%22"><i>British Journal of Educational Technology</i></searchLink>. Jan 2020 51(1):211-227.
– Name: Avail
  Label: Availability
  Group: Avail
  Data: Wiley-Blackwell. 350 Main Street, Malden, MA 02148. Tel: 800-835-6770; Tel: 781-388-8598; Fax: 781-388-8232; e-mail: cs-journals@wiley.com; Web site: http://www.wiley.com/WileyCDA
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 17
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2020
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Reports - Research
– Name: Audience
  Label: Education Level
  Group: Audnce
  Data: <searchLink fieldCode="EL" term="%22Higher+Education%22">Higher Education</searchLink><br /><searchLink fieldCode="EL" term="%22Postsecondary+Education%22">Postsecondary Education</searchLink>
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22English+%28Second+Language%29%22">English (Second Language)</searchLink><br /><searchLink fieldCode="DE" term="%22Second+Language+Learning%22">Second Language Learning</searchLink><br /><searchLink fieldCode="DE" term="%22Second+Language+Instruction%22">Second Language Instruction</searchLink><br /><searchLink fieldCode="DE" term="%22Online+Courses%22">Online Courses</searchLink><br /><searchLink fieldCode="DE" term="%22Conventional+Instruction%22">Conventional Instruction</searchLink><br /><searchLink fieldCode="DE" term="%22Blended+Learning%22">Blended Learning</searchLink><br /><searchLink fieldCode="DE" term="%22Teaching+Methods%22">Teaching Methods</searchLink><br /><searchLink fieldCode="DE" term="%22Language+Teachers%22">Language Teachers</searchLink><br /><searchLink fieldCode="DE" term="%22Teacher+Effectiveness%22">Teacher Effectiveness</searchLink><br /><searchLink fieldCode="DE" term="%22State+Universities%22">State Universities</searchLink><br /><searchLink fieldCode="DE" term="%22Grades+%28Scholastic%29%22">Grades (Scholastic)</searchLink><br /><searchLink fieldCode="DE" term="%22Teacher+Student+Ratio%22">Teacher Student Ratio</searchLink><br /><searchLink fieldCode="DE" term="%22School+Policy%22">School Policy</searchLink><br /><searchLink fieldCode="DE" term="%22Mexicans%22">Mexicans</searchLink><br /><searchLink fieldCode="DE" term="%22Computer+Assisted+Instruction%22">Computer Assisted Instruction</searchLink><br /><searchLink fieldCode="DE" term="%22Educational+Quality%22">Educational Quality</searchLink><br /><searchLink fieldCode="DE" term="%22Cost+Effectiveness%22">Cost Effectiveness</searchLink><br /><searchLink fieldCode="DE" term="%22Outcomes+of+Education%22">Outcomes of Education</searchLink><br /><searchLink fieldCode="DE" term="%22College+Students%22">College Students</searchLink><br /><searchLink fieldCode="DE" term="%22Foreign+Countries%22">Foreign Countries</searchLink>
– Name: Subject
  Label: Geographic Terms
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Mexico%22">Mexico</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1111/bjet.12797
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 0007-1013
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Despite steady investment in English language education made by developing countries over the past few decades, results continue to be constrained by lack of high-quality instructors and language learning resources. Thus, using technology in language instruction has increasingly been recognized as a potential approach for addressing these constraints. This study uses administrative data from a large public university in Mexico to examine the impact of a technology-enhanced blended program on students' English course grades and course completion rates. Specifically, we focus on a campus-wide policy change in all compulsory English language courses that replaces half of the traditional face-to-face class time with an interactive online learning environment developed by a leading technology-mediated English language learning and assessment provider. Our results suggest that, compared to traditional face-to-face instruction, blended learning had a significant, positive impact on students' course grades and course completion rates. In addition, the enrollment-teacher ratio increased after replacing half of the face-to-face instructional time with online learning, suggesting that blended learning environments hold promise for providing high-quality and cost-effective language instruction.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2020
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1240961
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1240961
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1111/bjet.12797
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 17
        StartPage: 211
    Subjects:
      – SubjectFull: English (Second Language)
        Type: general
      – SubjectFull: Second Language Learning
        Type: general
      – SubjectFull: Second Language Instruction
        Type: general
      – SubjectFull: Online Courses
        Type: general
      – SubjectFull: Conventional Instruction
        Type: general
      – SubjectFull: Blended Learning
        Type: general
      – SubjectFull: Teaching Methods
        Type: general
      – SubjectFull: Language Teachers
        Type: general
      – SubjectFull: Teacher Effectiveness
        Type: general
      – SubjectFull: State Universities
        Type: general
      – SubjectFull: Grades (Scholastic)
        Type: general
      – SubjectFull: Teacher Student Ratio
        Type: general
      – SubjectFull: School Policy
        Type: general
      – SubjectFull: Mexicans
        Type: general
      – SubjectFull: Computer Assisted Instruction
        Type: general
      – SubjectFull: Educational Quality
        Type: general
      – SubjectFull: Cost Effectiveness
        Type: general
      – SubjectFull: Outcomes of Education
        Type: general
      – SubjectFull: College Students
        Type: general
      – SubjectFull: Foreign Countries
        Type: general
      – SubjectFull: Mexico
        Type: general
    Titles:
      – TitleFull: Does Blended Instruction Enhance English Language Learning in Developing Countries? Evidence from Mexico
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Xu, Di
      – PersonEntity:
          Name:
            NameFull: Glick, Danny
      – PersonEntity:
          Name:
            NameFull: Rodriguez, Fernando
      – PersonEntity:
          Name:
            NameFull: Cung, Bianca
      – PersonEntity:
          Name:
            NameFull: Li, Qiujie
      – PersonEntity:
          Name:
            NameFull: Warschauer, Mark
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2020
          Identifiers:
            – Type: issn-print
              Value: 0007-1013
          Numbering:
            – Type: volume
              Value: 51
            – Type: issue
              Value: 1
          Titles:
            – TitleFull: British Journal of Educational Technology
              Type: main
ResultId 1