Calculation of Learning GainThe use

Calculation of Learning Gain
The use of pre-test/post-test design studies is a particularly common methodology to assess learning gain, with two measurements recorded from a single subject either side of an intervention. In this case the intervention is a specific learning tool (screencast or textbook), with the overall test scores compared to assess the impact on changes to both the acquired (post-test), and retained (post-tests 1-week and 4- week) knowledge. To calculate the absolute gain from the pre-test and post-test scores a simple bivariate to univariate data transformation is performed according to Equation (Eq.) 1, revealing the difference in retention at two points in time. However, due to the maximum score of a test instrument being 100% a strong negative correlation is observed between students’ absolute gain (Eq. (1)) and their pre-test scores (i.e., a higher pre-test score results in reduced absolute gains). In order to reduce the influence of pre-test scores the normalized learning gain is calculated by dividing the abso- lute gain by the maximum possible gain (Eq. (2)). Normaliza- tion, therefore, allows for the actual realized change in learning gain to be recorded independent of pre-test scores and permits comparisons between groups to be made. With a diverse range of students yielding a wide range of pre-test scores the normalized gain values should be equal, and with all other conditions being controlled, any observed changes in gain can be attributed to the intervention. This is sup- ported from other studies which reveal that the mean normal- ized gain values are uncorrelated with the mean pre-test scores (Hake, 1998, 2002). Absolute learner gain (gabs)
gabs 5Ppost2test2Ppre2test (1)
P5mean score (%) Normalized learning gain (gi)
gi 5
Ppost2test 2 Ppre2test 100% 2 Ppre2test
(2)
P5mean score (%) In order to calculate the group mean normalized gain either Eq. (3) or Eq. (4) can be employed. The advantages of calculating the normalized gain of averages (Eq. 3) have been
4 Pickering
discussed elsewhere (Hake, 1998). However, for this study the average of the individual normalized learning gain (Eq. 4) has been used as it generates standard deviations which allow for the respective effect sizes to be calculated. Normalized gain of average (g)
g5
Ppost2test 2 Ppre2test  100% 2 Ppre2test (3)
P5mean score (%) Average of normalized learning gain (gave)
gave 5
P 1!n giðÞ  n
(4)
gi 5normalized learner gain

Results (Indonesian) 1:

Perhitungan belajar GainThe use of pre-test/post-test design studies is a particularly common methodology to assess learning gain, with two measurements recorded from a single subject either side of an intervention. In this case the intervention is a specific learning tool (screencast or textbook), with the overall test scores compared to assess the impact on changes to both the acquired (post-test), and retained (post-tests 1-week and 4- week) knowledge. To calculate the absolute gain from the pre-test and post-test scores a simple bivariate to univariate data transformation is performed according to Equation (Eq.) 1, revealing the difference in retention at two points in time. However, due to the maximum score of a test instrument being 100% a strong negative correlation is observed between students’ absolute gain (Eq. (1)) and their pre-test scores (i.e., a higher pre-test score results in reduced absolute gains). In order to reduce the influence of pre-test scores the normalized learning gain is calculated by dividing the abso- lute gain by the maximum possible gain (Eq. (2)). Normaliza- tion, therefore, allows for the actual realized change in learning gain to be recorded independent of pre-test scores and permits comparisons between groups to be made. With a diverse range of students yielding a wide range of pre-test scores the normalized gain values should be equal, and with all other conditions being controlled, any observed changes in gain can be attributed to the intervention. This is sup- ported from other studies which reveal that the mean normal- ized gain values are uncorrelated with the mean pre-test scores (Hake, 1998, 2002). Absolute learner gain (gabs)gabs 5Ppost2test2Ppre2test (1)P5mean Skor (%) Dinormalisasi belajar keuntungan (gi)GI 5100% Ppost2test 2 Ppre2test 2 Ppre2test(2)P5mean Skor (%) Untuk menghitung kelompok berarti menormalkan keuntungan baik EQ (3) atau EQ (4) dapat digunakan. Keuntungan dari menghitung keuntungan menormalkan atau rata-rata (EQ 3) telah4 Pickeringdibahas di tempat lain (semacam ikan, 1998). Namun, untuk studi ini rata-rata pembelajaran individual menormalkan gain (EQ 4) telah digunakan sebagai dihasilkannya deviasi standar yang memungkinkan untuk ukuran masing-masing efek harus dihitung. Menormalkan keuntungan atau rata-rata (g)G5100% Ppost2test 2 Ppre2test 2 Ppre2test (3)P5mean Skor (%) Rata-rata atau menormalkan belajar keuntungan (view)Lihat 51 P! n n giðÞ(4)Mendapatkan GI 5normalized pelajar