> <; \pJava Excel API v2.6 Ba==h\:#8X@"1Arial1Arial1Arial1Arial + ) , * `PDC, 4$titletitle[alternative]title[translated]author[google(nobreakline)] contributor[author(nobreakline)]"contributor[scopusid(nobreakline)]date[issued]relation[journaltitle]identifier[issn] citationsidentifier[major]subject(nobreakline) publisheridentifier[thesisdegree]contributor[advisor]relation[ispartofseries]relation[index(nobreakline)]typeabstractidentifier[govdoc]identifier[uri]identifier[doi(doi)]identifier[isbn]identifier[ismn]
identifierTЄD \ սݲX ?Verification of the Effect of Flipped Learning in Meta Analysis2018 Y P!YtTŐYP YDoctortDoctoral Thesis#The purpose of this study is to present the direction of future research by examining the learning effect of flipped learning and exploring the effects of various moderator variables. The effectiveness of flipped learning was analyzed in meta analysis. The meta analysis examined the effectiveness of flipped learning and verified the effectiveness of flipped learning by integrating individual studies.
I selected a total of 126 publications including journal articles, theses, and dissertation which employed a pretest-posttest design and standardized measurement tools from 2007 to 2017 in Korea and abroad. A total of 433 effect size values were calculated by using the CMA(Comprehensive Meta Analysis) program.
Meta analysis can suggest not only the conclusions about the direction between variables but also the direction in which research needs to be done in the field and what direction should be studied in the future(Borenstein, Hedges, Higgins, & Rothestein, 2009). Therefore, I reviewed the flow of flip learning research based on the previous research, investigated whether there is a difference in the effect of flipped learning according to the moderator variable using meta analysis, and search for future research direction. Below are the research questions of this study:
1. What is the overall effect size of flipped learning on learning effectiveness?
2. What is the mean effect size and the mean effect size of each sub-element on the effects of learning by flipped learning (cognitive domain, positive domain, interpersonal domain)?
3. Are there any differences in the mean effect sizes of the flipped learning according to the moderator variables?
3-1. What is the mean effect size according to the general characteristics of flipped learning?
3-2. What is the mean effect of flipped learning design characteristics (before, during, and after class) of flipped learning?
Below are the results of this study:
First, the total program effect size value based on the random effect model was medium level(effect size= .41). The effect size of flipped learning is significant, and bigger than instructor-led courses. These results suggest that flipped learning has achieved plenty of educational outcomes.
Second, the effect size of flipped learning education showed statistically significant effect on cognitive, affective, and interpersonal relations domain. It showes that all domains were medium level such as cognitive(.65), affective(.51), and interpersonal relations domain(.68).
Specifically, the cognitive domain, the effects are critical thinking(.71), academic performance(.62), problem solving ability(.49), and self-directed learning ability(.45) from the highest to lowest. The affective domain, the effects were class participation(.80), learning motivation(.69), learning attitude(.59), learning satisfaction(.55), self efficacy(.44) from the highest to lowest. And Interpersonal relationship domains showed communication skills (.74), interaction (.59), and cooperation (.56) from the highest to lowest. In addition, all variables showed median effect. Thus, we need to actively use flipped learning when increase learning effects of cognitive, affective and interpersonal relations domain.
Third, as a result of the moderator effect analysis on the general characteristic variable, the statistically significant variables are school type and core subject. In case of flipped learning design characteristic variables are video make< r, video time, homework, check up on pre study, number of team member, and time of discussion.
Especially, flipped learning is most effective for students in high schools. In addition, the effect size of flipped learning is increased when gave lessons at high school, took the music and korean language classes, used single subject research design, made 10-15mins video times, checked up on pre study, assigned homework, was composed of 4 team member, gave 20mins for discussion time. In addition, we made design implication to gave guideline for instructional designer based on meta analysis results.
Specifically, when it is done in high school, when elementary school is used for English, music, and computer curriculum, when middle school for mathematics and home economics, when high school for English and computer curriculum, when university for science and sociology, when the experimental treatment period is 7 weeks, when the video is not produced by teacher, when the video is given for 10 minutes to 15 minutes, when homework is given, When the group member is four, discussion time is most effective when given 20 minutes.
Other moderator variables are not statistically significant such as research design, implementation frequency, publication type, publishing area, class type, disability, number of participants, presenting alternatives without prior learning, orientation, presence of self-directed learning activities in advance learning, Whether they are presenting their learning according to the learner level, and presence of supplement and enrichment after class.
In this study, design implications are provided based on these results. Design implications are classified into pre learning activity design, face to face class composition, and video production. Based on these results, it can be concluded that flipped learning has a positive effect on learning. However, there is a lack of flipped learning meta-analysis, which makes it difficult to present in-depth discussions. In addition, due to the characteristics of traditional meta analysis methods, there are limitations in calculating the effect size by simultaneously taking into account variables that can affect learning effects. This study also has some limitations in that the proportion of dissertations and domestic research is high.;սݲt 1<\ P!֥ ȩ ɍ<\ 0 t ƥD Ȕ 䲑\ x ƌD X| \. Lɔ սݲ
x/x ƥD Ȕ $Ĭ ƌ 4x \ l| >0Ŕ $t . lX Š\ D ȩt| X Š\ D 0t| XՔ t\ $Ĭ ƌ \ lx t DՔX(, M\, 2016).
lX @ ֬L ɉ mx սݲ \ lD iXՔ TЄD \ սݲX $Ĭ tܴ|xD XՔ p . t| t սݲX x ĳ 1x ȩD \ pt ^<\ D )D X.
սݲ Yյ¨ \ TЄt DՔ\ t ǔ L (i1ٳ, 2015). , սݲ \ `t |0 0 t \ \ `D DՔ . ٳ|\ ٳ|\ P ȩXLĳ lX \ И <\ t \ U\ `D DՔ . X, Ĭt Uŕ t DՔX0 L8t. ٳ|\ | Ǭ\ )<\ \ l (<\h |0 lX D(1 8 , 0 x t` \ սݲD 0 ƥ%X l0| ִp ι@ \t . , սݲt ȵ P DPX YյX x, X, xĬ ȹ̹| ǔ ix | DՔ . K, սݲX 1 ƥD ` ǔ px(| 1, սݲ $Ĭ1)X ƥ% \ t X0 L8t. px 0x l0| Ĭ` , t\ | X T x P! \ D t .
TЄ@ X ) \ `п̹ DȲ| | Ť l T DՔX ^<\ Ť )<\ l t踴8| XՔ \ )D H` (Borenstein, Hedges, Higgins, & Rothestein, 2009). \ l| <\ սݲ lX PׄD X, TЄD \֩X px 0x սݲX | , lX )D Xଐ\. t| \ lX l8Ȕ L .
1. սݲt Yյ XΔ ȴ l0(effect size) Š\?
2. սݲt Yյ ļ (x , X , xĬ ) XΔ l0 Xǔƌļ l0 Š\?
3. px 0x սݲX l0 (t ǔ?
3-1. սݲX | 1 0x l0 Š\?
3-2. սݲX $Ĭ1(, , ) 0x l0 Š\?
l Ȭ $Ĭ@ \T !ĳl| \ l l $\ 0 iXՔ 126X Y Y|8D <\ X. 433X l0 Ь@
!"#$%&'()*,-./012456789:Root EntryWorkbookVSummaryInformation(+DocumentSummaryInformation83