Influence of media exposure to school success--decision tree analysis.

Simicevic, V. ; Hribar, J. ; Pejic-Bach, M. 等

Abstract: The paper is focused on connection between children's school success, free time variable and the level of exposure to media contents. The analysis was conducted with 'decision tree' method. Assumption that was started with was that the larger exposure to media contents has more unsuitable influence on school success. The paper put special emphasis on negative influence of watching television, as the most significant media of today's world on children in school age period.

Key words: statistical analysis, media, decision tree, school Success

1. INTRODUCTION

From the very beginning of the world, human community was occupied with questions of what are the best ways of bringing children up and who needs and can participate in such mission. Among existing socialization factors in last century, mass communication media imposed themselves as the part of our everyday life which has enormous influence on socialization process of every child. In contemporary society every new generation grows up in environment that is richer with new media and its contents. According to that, the potential influence of media on children or more accurately, the perception of that influence on children is changed. Just for that reason the research of potential influence of mass media on children is as necessary as always actual.

The aim of the paper is to interrogate the connection between the children's characteristics and the level of exposure to media contents using the 'decision tree' method. Assumption that was started with was that the larger exposure to media contents has more unsuitable influence on school success. In paper the special emphasis will be put on watching television.

After the introduction, second section of the paper will analyze the media influence on children. Third section describes the characteristics of 'decision tree' method, and fourth is analyzing the connection between children's characteristics and children's exposure to the media. The last section of the paper consists of conclusion considerations.

2. MEDIA INFLUENCE ON CHILDREN

As one of the strongest video media television with its beginning ensured itself more important place in the average family home. The television influence is in dependence to contextual, social and personal factors. According to that the TV influence is bigger that the TV program is more in accordance to attitudes of the viewer. We often ask ourselves, what is so interesting on television, which occupies such large part of our free time?! The most representing population in front of the TV tube is the over 50 years old population. Children watch television simultaneously with all other media or with some other activity.

It is interesting that the length of the day TV watching almost doesn't don't depend on observed social characteristics of the examined group. Ilisin, Bobinac and Radin (2001) examined the connection between the amount of free time and school success with television watching. Authors checked if the amount of day free time was connected to television watching and it was shown that the connection exists.

Mass media shows violence in real, fictional and imagined lives. What is especially disturbing to parents and school experts in media-communication developed world is not the fact that such amount of violence exists in news and documentary movies, which represent serious stories, but the fact that it exists in the fun part of the program. The most considering consequences of violence in mass media (Kosir et al.) are mimicking and dumbness. For the consequences of use of mass media the content of television shows is not so important, but in what social and communication environment they do so.

Second most certain consequence of watching television and violent scenes is dumbness, which as well as mimicking depends on great number of factors. The children who are exposed to television on every day basis and have they minds focused on numerous violent scenes, know that blood and death they see on TV is not truthful and with time they watch those scenes very easily. Children's feelings due to overexposure to media violence along with problematic life circumstances go dumb to the point that cannot be 'unfrozen' any more.

It can be concluded that this media can be considered potentially dangerous, especially for those people who its TV consuming cannot control. It should be pointed out again that also in this case the most threatened is the youngest population of viewers. It is hard to turn of the head from the danger of what the experts point to. Even if it still isn't with most certainty established the connection between the real violence and watching TV violence, we cannot claim that the certain connection doesn't exist.

In this research, the groups of the pupils will be made considering the amount of TV watching, free time and school success, which are not defined in advance. So, we will try to quantify more accurately the connection between those variables, because in the research we just have explained the methods that detect the variable connections were used, but they don't quantify it.

3. THE METHODOLOGY OF RESEARCH

'Decision tree' method can be used for classification and regression problems, during which the model in form of the rule is generated and can explain the relation between incoming and out coming variables. Rules generated in this way can be expressed in the form of SQL orders and simply be incorporated in program solution.

If some kind of the problem should be solved using 'decision tree' it has to have following characteristics (Mitchell, 1997): (1) the data have to be described in the form of final number of attributes; for example for every bank there are attributes; (2) the number of those attributes is known in advance and it is represented by the final number; for example it is precisely known what number of attributes one bank can have; (3) if the case is classification tree, every data should belong to only one class; (4) great number of data is needed - at least few hundred. But the data can be incomplete and content mistakes.

Decision tree is classification algorithm (McLachlan, 1992). There are two types of knots that are connected with branches: terminal node (leaf node) and the node which is final for certain branch on the tree (decision node), which defines certain condition in the form of certain attribute. It creates itself through algorithm which finds regularities between data, and the most famous algorithms are CHAID, EXHAUSTIVE CHAID, C&RT. (Breiman et al., 1984) and Quest (Loh et al.1997.)

Algorithm consists of selection of attributes for making decision nods, during what every data is on the beginning are in one group. The data is then divided towards all possible criteria in branches, and one criterion that divides the data in more homogeneous groups than starting group is chosen. When it is no more possible to divide data in groups that are more homogeneous than starting group, the decision tree is finished. As homogeneous measure of the group of entities the entropy is used. (Han et al., 2000.)

In sample was chosen 500 children from lower grades of elementary school, and for the needs of this paper only 3 variables were analyzed: school success, the number of daily TV watching hours, and the number of the free time hours.

The questionnaire was filled out by the parents of the children, and this research can be advised as preliminary and there is the conduct of real research in plan, on the same sample, which should collect daily data on children's behavior throughout shorter time period. The CHAID algorithm was applied in decision tree and the Statistica Data Miner software was used.

4. CONNECTION BETWEEN CHILDREN'S CHARACTERISTICS AND MEDIA CONTENTS EXPOSURE

Picture 1. consists of decision tree generated on the basis of sample data. The data showed clear connection between TV watching and school success, whilst the free time of the pupils hadn't bigger influence on the model. The pupils can be divided in two sections: those who watch TV less then 2, 5 hours a day and those who watch TV more then 2, 5 hours a day.

[FIGURE 1 OMITTED]

The pupils who watch TV more then 2, 5 hours a day have average success of 3, 9-. However those who watch TV more then 4,5 hours a day have very low success average of 2,8. On the other side, watching TV in the time interval from 2, 5 hours a day to 4, 5 hours a day had as consequence the average of school success 3, 9 in all pupil cases.

Those pupils who watch TV less then 2,5 hours have considerably higher average school success of 4,42. The best results have shown those pupils who watch more than 1,5 hours TV per day, but their free time is probably occupied with some more extra curriculum activities, and they have average success of 5,0.

However, there are only 3% of such pupils in the sample. The pupils that watched TV in interval of 1,5 to 2,5 hours per day have achieved the average school success of 4,25. Those who watched TV less then 1, 5 hours per day achieved the average school success of 4, 62.

It is important to point out the fact that the children watch the most of the time programs for adults and especially those that include the elements of excitement. (Ilisin, et al., 2001.) On the other side, programs that are created for children the examined group watches only temporary, so the interesting question is why those programs represent more interest to them than the others.

5. CONCLUSIONS

The paper was written with the aim of examining the connection between the characteristics of children and the level of the exposure to the media contents with the use of the decision tree. Assumption that was started with was that the larger exposure to media contents has more unsuitable influence on school success.

The research was conducted on sample of 500 children in lower grades of elementary school. The data that was analyzed were on school success, free time and number of TV watching hours. The 'decision tree' method was used in examination. The results showed that there is negative kind of connection between the number of TV watching hours and school success of the pupils.

The violence influence of the media on children could be the object of future research work.

6. REFERENCES

Breiman, L., Friedman, J. H., Olshen, R. A. and Stone, C. J., 1984. Classification and Regression Trees. Belmont: Wadsworth.

Han, J., and Kamber, K., 2000. Data Mining: Concepts and Techniques. San Francisco: Morgan Kaufman.

Loh W. Y. and Shih Y. S. (1997) Split Selection Methods for Classification trees. Statistica Sinica, Vol. 7, No.4, 815-840

McLachlan, G. J., 1992. Discriminant Analysis and Statistical Pattern Recognition. New York: Wiley Interscience.

Mitchell, T. "Decision Tree Learning", in T. Mitchell, Machine Learning, The McGraw-Hill Companies, Inc., 1997, pp. 52-78.