Modelling factors affecting attitudes of 6, 7 and 8th grade students towards science via artificial neural network method: Mus province sample

The aim of this study is to determine secondary school students' (6th, 7th and 8th grades) attitudes towards science and to examine variables that have an impact on these attitudes via Multi-layer Artificial Neural Networks. According to the date obtained from the Provincial Directorate for National Education in 2015-2016 academic year, the total number of students studying at secondary schools located in Muş center is 6465 (3135 male; 3330 female) whereas the total number of students studying at secondary school located in districts is 7362 (3767 male; 3595 female). The current study was conducted with 617 students (295 male; 322 female) of 9 different secondary schools located in Muş province and its districts in 2015-2016 academic year. Personal information form and Attitude towards Science Scale were applied to these students who were elected randomly. The sample group consists of 291 students (134 male; 157 female) studying at secondary schools located in Muş center and 326 students (161 male; 165 female) studying at secondary schools located in districts of Muş. To determine the secondary school students' attitudes towards science, with 0.79 reliability co-efficient Attitude towards Science Scale developed by Duran (2008) was used and a 19-item Personal Information Form involving various socio-demographic variables was administered in order to determine what kind of variables affect attitude. SPSS and Excel programs were utilised for data analysis. Regarding students' score of attitude towards science the findings revealed that the most effective variable was "The frequency of asking question to teacher that s/he could not solve in science lesson" (Significance level: 100%; significance value = 0.101); secondary variable having impact on the predicted variable was "The frequency of searching science subject on Internet" (significance level: 90.9%; significance value = 0.092); the third effective variable was "Last year Science courses passing grade" (significance level: 78.7%; significance value = 0.079); the fourth effective variable was "Studying at schools located in center or districts" (significance level: 77.0%; significance value = 0.078)"; the fifth effective variable was "Like dealing with difficult problems" (significance level: 74.6%; significance value = 0.075)". The predictors whose severity is moderate are "Reading topics covered in Science from different sources", "Like watching documentary", and "Occupation of your father". Variables which have less effect on attitude towards Science are "Favourite type of computer games", "Have a room of your own", "Reason for using the computer", "In which grade (6., 7., 8.)You are studying", "Ideal future profession", "Mother's education level", "Number of siblings," "Family monthly average income", and "Gender". In addition, the students' scores of attitude towards science was found to be high. Key words: Multilayer Perceptron, Science, Attitude, Artificial Neural Networks, Secondary School.