The effectiveness of blended model of Argumentation Flipped Learning (AFL) in increasing students’ critical thinking skills: A case study of junior high schools during the Covid-19 period

Research design and procedures

A weak experimental design approach was used in this study because there was no control class. This method is often classified as non-experimental or pre-experimental (Campbell & Stanley, 1963). The independent variable studied was the Argumentation Flipped Learning (AFL) model which is a fairly long online education program, and its implementation requires (1) good devices such as internet networks, laptops or computers, and cellphone or smartphones to ensure the online learning run effectively and (2) approximately 3 meetings plus two others for the pretest and posttest. Several obstacles are, however, associated with this model such as the ability of students and parents to provide the equipment, the ability of teachers to supervise online learning, and the lack of even distribution of the network. These conditions make it quite difficult to conduct the research using the control class, therefore, the focus was placed on the conditions of student learning in an observation class where their learning experiences, interactions with teachers, interactions with their colleagues, and difficulties encountered during the learning process were intensively recorded.

The pre-experiment approach used in this study was a one-group pretest-posttest design (Fraenkel et al., 2012) which consists of a group of participants observed before the treatment (pretest) and after the application of the Argumentation Flipped Learning (AFL) treatment on the material "Temperature and its Changes and Heat and its Transfer" (posttest) with the focus on the dependent variable which was the critical thinking skill. The treatment effect was measured by comparing the means of the posttest and pretest (Dimitrov & Rumrill, 2003). It is expected that this research design has the ability to determine the effectiveness of the AFL learning model in improving the critical thinking skills of junior high school students.

Research sample

The population used in this study were junior high school students in Wates Kulon Progo, Yogyakarta Special Region with the 31 students in even semester of class VII in the 2020/2021 school year selected as samples using purposive sampling techniques based on their National Examination data and the results of the initial profile test conducted in relation critical thinking skills.

Research instruments

The data was obtained using critical thinking skill test instrument which was in the form of essay questions arranged based on the critical thinking skill indicators according to Bloom and revised by Rukmini (2008) and Ruwaida (2019) to include analyzing (C4), evaluating (C5), and creating (C6). The questions were validated by certified science education experts, science experts, and science teachers as well as empirical tests through experiments conducted to show the instrument is valid and reliable. It is important to note that the instrument has an assessment interval of 1-5 according to the perspective criteria, and the validity was determined using the percentage of agreement (R) = [(A-B)/(A+B)]x100% with A being the highest score from all validators and B being the lowest score from all validators (Borich, 2016).

The results of the validity test conducted on the AFL model and its learning tools are presented in the following Tables 1 and 2.

Table 1. The results of the validity for the AFL model

N	Aspects of Assessment	Validation score	Validation criteria	Reliability coefficient (%)	Reliability
A	Content validity
1	AFL model requirement	4.00	Very valid	87.68	Reliable
2	State of the art of knowledge	4.33	Very valid	89.75	Reliable
3	AFL model theory support	4.00	Very valid	88.00	Reliable
4	Implementing the AFL model	4.67	Very valid	87,68	Reliable
5	AFL learning environment	5.00	Very valid	87.68	Reliable
6	Use of state-of-the-art evaluation techniques	4.67	Very valid	88.00	Reliable
B	Construct validity		Very valid		Reliable
1	Theoretical and empirical support for the AFL model	4.67	Very valid	90.25	Reliable
2	Model syntax	4.33	Very valid	89.89	Reliable
3	Social system	5.00	Very valid	90.00	Reliable
4	Reaction principle	5.00	Very valid	89.89	Reliable
5	Learning environment	4.67	Very valid	88.89	Reliable
6	Implementation of evaluation	4.33	Very valid	90.25	Reliable

Table 1 shows the average scores for the content and construct validities are in the very valid category and their reliability coefficients are also above 75% which indicates they are valid and reliable.

Table 2. The results of the validity test for the AFL model learning tools

N	Material	Average validator			Amount Average	Criteria	Coefficient %	Reliability
N	Material	1	2	3	Amount Average	Criteria	Coefficient %	Reliability
1	Syllabus	4.67	4.33	4.00	4.33	Valid	87.67	Reliable
2	Lesson plan (LP)	4.33	4.33	4.33	4.33	Valid	86.00	Reliable
3	Teaching materials	4.33	4.00	4.00	4.11	Valid	87.67	Reliable
4	Video-Worksheet	4.00	4.33	4.00	4.11	Valid	86.67	Reliable
5	Critical thinking skills test sheet	4.00	4.00	4.67	4.22	Valid	88.70	Reliable

Table 2 shows that the AFL learning tools have an average score above 4.0 which implies they are valid. They also have a reliability coefficient above 75% which denotes they are also reliable. This shows that the AFL model and its learning tools can be applied in the classroom for learning.

The pretest-posttest data during the implementation of the learning were also scored according to the assessment rubric and later statistically analyzed using SPSS 23.0 for Window. This was used for the descriptive analysis while normality was determined using one-sample Kolmogorov-Smirnov, homogeneity using Levene tests while ANOVA was also applied to determine the difference between the pretest and posttest. Moreover, n-gain was calculated in line with the interpretation of Hake (1999) while the effect size was determined according to Cohen (2007) which was assisted by a calculator.