Abstraction
Analysis of variance is part of the statistics was first introduced by Sir Ronald & Fisher. This analysis has been utilized in all areas of research that uses quantitative data. In the implementation, analysis of variance using ANOVA (Analysis of Variance). In the education sector analysis of variance was used to test whether there are similarities or differences between the learning methods and sebagianya.
Preliminary
In a study of quantitative data obtained must be processed and analyzed to become an information that we want. The process of acquisition, processing, and data analysis are usually referred to as the statistical method. According to Hadi Sutrisno (1996: 3) statistic is knowledge related to the ways of data collection, processing or penganalisisannya and conclusion based on the collection and analyzing data that are made. While the analysis of the data itself, according to Hasan, Iqbal (2004: 29) can be interpreted as follows: (1) to compare two things or two variable value / more to know the difference / ratio is then drawn a conclusion, (2) regulate the process of sequence data, mengorganisaikannya into a pattern, categories, and the basic outline of the unit, (3) detailing the process of formal efforts to find a theme and to formulate hypotheses as suggested by the data and in an effort to provide assistance on themes and hypotheses. And according Setyowati, Eni (2008: 1) Statistics is the science of collecting, organizing, presenting, analyzing, and interpreting the data into information for effective decision making emmbantu. The excess of data with statistical analysis are: (1) allows an exact description of something. Verbal symbols is more efficient than verbal language, (2) allows a person to work exactly in the process and ways of thinking. Although not absolutely true, but it can be set up to which level of the conclusion is correct, (3) researchers provide summaries of research results in a more meaningful and more compact because it provides certain rules, (4) can draw general conclusions, and (5) possible to conduct forecasts (Hasan, Iqbal, 2004: 30). One of the quantitative data with statistical analysis is the analysis of variance.
Before we perform analysis of variance we need to first understand the concept of experimental design, because the analysis of variance was an experiment or experimental research.
Experimental design
Experiments are generally carried out to find something. Therefore, theoretically, be interpreted as a test or experiment planned inquiry to obtain new facts (Steel and Torrie, in Herawati Nita, 2007: Online). And experimental design can be interpreted as a test or series of tests in which significant changes made to the variables of a process or system so that we can observe and identify the reasons for the change in output response. While according to the Herawati Miliken and Johnson, Nita (2007: Online), the experimental design is very related to the planning of research to obtain the maximum information from materials that are available. And can also be interpreted as a set of rules / way / procedure to apply the treatment to the experimental units.
From the various definitions of the above, it is clear that the purpose of the experiment is similar to that answer one or more questions to get the maximum information by: (1) Determine which variables most affect the response (response), (2) determine how sets of independent variables that influence The dependent variable thus approaching the desired nominal value, (3) Determine how the set of independent variables that influence the dependent variable range of small, (4) Determine how to set up an independent variable so that the uncontrollable variables as small as possible.
In designing a study, researchers often control the specific influences such as treatment, population, or a combination of treatments. Therefore, before the research goes it raises some questions that must be answered: (1) How many treatments that must be applied, (2) How many times each treatment must be observed, (3) What are the units of parole, (4) How to apply to the treatment unit experiment and observing the response, (5) Can the design was analyzed and compared?
To be able to answer these questions directly and do not have can not be answered in general. Here, the experimental design is used so that it can play an important role in the development process and the process of finding and solving problems in order to enhance research.
Three main principles in experimental design are: (1) Deuteronomy. Deuteronomy is applied one treatment to more than one experimental unit. Apling Deuteronomy is of importance in an investigation and has the function to provide experimental error, increase precision by reducing the standard deviation, improve generalization, (2) randomization. Randomization is the underlying statistical methods in experimental design. Randomization is the application of treatments to experimental units so that all / any single experiment has an equal opportunity to emnerima a treatment. The concept of randomization is valid also for decision or determination of a sub sample of observation units. Randomization function to avoid bias, ensure the existence of freedom between observations, and overcome the sources of variability that are known but unpredictable effects; (3) Grouping. Grouping is a technique used to improve the accuracy of the experiment. Grouping done if there is a source of diversity that can be known and its influence can be estimated. Experimental material arranged in groups of experimental units are relatively uniform (Herawati, Nita, 2007: Online)
Variety Analysis
During this time, analysis of variance was used in the biometric field (agriculture, biology, medicine, engineering, etc.). In this paper we will focus on the use of analysis of variance in education. Analysis of variance was first introduced by Sir Ronald & Fisher and is basically an arithmetic process for dividing the total sum of squares into components associated with a known source of variability (Steel and Torrie, 1991: 168). In Hadi, Sutrisno (1988: 367), analysis of variance was also known as variance analysis is a tool to test hypotheses about differences in the mean zero value of more than two samples simultaneously. According Suharyadi (2004: 440), analysis of variance (ANOVA) was used to distinguish three or more with the middle value assumptions underlying ANOVA are (a) sample obtained from a normal population, (b) each population has the same standard deviation, and ( c) all populations are independent of each other. From bebarapa definition above we can conclude that the analysis of variance is a method of describing the total diversity of the data into components that measure various sources of diversity with the aim of testing the similarity some central values in one go.
Analysis of variance is the separation of the total number of kuadarat into components associated with a known source of variability. In the analysis of variance involving real test basic assumptions are: (1) Effect of additive treatment and environment, (2) are random experimental error, to spread freedom and normal dis ekitar middle value range of zero and common (same). These basic assumptions is an absolute thing that must be filled in using analysis of variance. Non-fulfillment of one or more assumptions can affect both the real level (level of significance) and the sensitivity of F or t the deviation from the null hypothesis is true.
In the case of abnormalities, the actual real rate is usually larger than those stated. This resulted in rejection of the null hypothesis opportunity hypothesis is true when larger, in other words too often be said when in fact not significantly different. Researchers may think that he used the real level 5 percent when in fact 7a tau 8 percent. If the assumption is not fulfilled aditifitas heterogeneity will cause an error. Various error components contributed by various observations did not expect the same variety. This resulted in a general range of assumptions automatically become not met as well. While the assumption of freedom of error will didapats ecara done directly if the randomization.
Because of these basic assumptions is absolute, after the data obtained from the research, then the first step is to test the crew if the data meet the assumptions there. Recently many statistical software that can help researchers to test these assumptions more easily. And if these assumptions are not met after testing proved one way around that is to perform data transformation.
Basically the analysis of variance was divided into five categories, namely (1) Analysis of Variety I (Classification of One Way / One Way Anova), (2) Variety Analysis II (Classification Many Directions), (3) Analysis of Variety III (Factorial Experiments), ( 4) Variety Analysis IV (Split plot design), and (5) V Variety Analysis (Number of Child-Class is not the same) (Steel and Torry, 1991: 168). In this paper only be described analysis of variance was I, II, and III.
Variety Analysis I (One-Way ANOVA)
According to Santoso, Singgih (2007: xxvi), Analysis of variance I used to completely randomized design. This design is the simplest and only examine the contents of the data column. Statistical procedures used are one-way anova. Now this with the help of existing software such as SPSS, statistical and sebagainyakita will be easier to do anova test without doing the calculation manually. In this discussion we will try to use the aid of statistical software SPSS 16.0 application.
Guide to do a one-way ANOVA test were as berikuit:
1. Determine and Hi Ho
2. Determine the confidence level or a significant level; in general level of confidence (confidence level) is 95% so that the significant level (significant level) is 100% - 95% = 5%
3. Determine the statistical procedures that will be used:
SSB formula above can be calculated by:
SSW formula above can be calculated by:
Where SST (total sum of square) is a quadratic toatal all values of data, which can be expressed by the formula:
4. Take the conclusion to accept or reject H0
If statistics count
Example application: A school melakukan research to determine whether a variety of teaching methods baru give different results, or not with the academic ability of students to examine the effectiveness of both methods, in this study included standard methods that have been used in the schools as a comparison, while two The new method is the method of X and Y. methods Each method is applied to 5 students with different students. After three months of training, academic ability is measured through a test score. Data from the study are as follows:
Student Name X Y Standard
A 49 71 83
B 60 60 87
C 57 65 89
D 59 59 92
E 55 69 95
(Kurniawan, 2007: Online)
Analysis: In the case of so-called factor (independent variable) is a teaching method while the level / treatment are three methods of teaching and the score was the dependent variable.
Testing procedures are:
1. Create a hypothesis:
Ho: score values using three methods of teaching the same relative to each other.
(Μ1 = μ2 = μ3)
Hi: at least one score value of a different teaching methods with other methods.
2. 95% confidence level and 5% significance level.
3. With SPSS 16.0 statistical procedures
STUDENT'S NAME SCORE METHOD
1 1 49
2 1 60
3 1 57
4 1 59
5 1 55
6 2 71
7 2 60
8 2 65
9 2 59
10 2 69
11 3 83
12 3 87
13 3 89
14 3 92
15 3 95
SPSS output is as follows:
Obtained:
F count = 64.877
F table (0:05; 2.12) = 3.88
Conclusion:
Therefore, calculated F> F table, so Ho rejected
It was concluded that teaching methods are applied to produce at least one score in three different teaching methods or in other words, these three methods of teaching have a different impact on students' academic abilities.
Variety Analsisi II (Classification Many Directions)
In the analysis of variance II we not only use a single classification, but many classifications. Entered into this dual classification is not only possibly be done in many investigations but is also very useful to get more information and more accurate. According to Santoso, Singgih (2007: xxviii) II analysis of variance was used for the randomized block design. This design to test the contents of columns and rows of data. Same statistical procedures used with a one-way anova.
Example applications: the same problem but variable analysis of variance was first included in the testing of students to determine whether the student 1 to 5 resulted in a score of academic ability. A school conducted a study to determine whether the various new teaching methods give different results, or not with the academic ability of students to examine the effectiveness of both methods, in this study included standard methods that have been used in schools as a comparison, while the two new methods The method is a method of X and Y. Each method is applied to five students. After three months of training, academic ability is measured through a test score. Data from the study are as follows:
Student Name X Y Standard
A 49 71 83
B 60 60 87
C 57 65 89
D 59 59 92
E 55 69 95
(Kurniawan, 2007: Online)
In this model there is a block variable, namely STUDENT. Now there will be two tests, namely the influence of the method and influence of students, in the language of the test statistics are columns and rows.
To test the column that contains the variable method of teaching procedures performed:
1. Create a hypothesis:
Ho: score values using three methods of teaching the same relative to each other.
(Μ1 = μ2 = μ3)
Hi: at least one score value of a different teaching methods with other methods.
2. 95% confidence level and 5% significance level.
3. With SPSS 16.0 statistical procedures
Obtained:
F count = 64.877
F table (0:05; 2.12) = 3.88
Conclusion:
Therefore, calculated F> F table, so Ho rejected
It was concluded that teaching methods are applied to produce at least one score in three different teaching methods or in other words, these three methods of teaching have a different impact on students' academic abilities.
The procedure to test the line that contains the variable students is as follows:
Ho: no significant differences between the academic ability of students in existing schools
Hi: at least one school with different academic scores than other schools
In SPSS, use the menu GENERAL LINEAR MODEL
STUDENT'S NAME SCORE METHOD
1 1 49
2 1 60
3 1 57
4 1 59
5 1 55
1 2 71
2 2 60
3 2 65
4 2 59
5 2 69
1 3 83
2 3 87
3 3 89
4 3 92
5 3 95
P value (sig) for variable p-value method = 0.000 (sig) for the variable name = 0.796. The value of having a probability value above 5% so thank Ho.
It can be concluded that the average score of academic ability was not significantly different for students who are there, the average score of academic ability relative the same four methods for both students A, B, C, D, and E. While the variables of teaching methods are analyzed with the conclusion remains that there are clear differences in scores of academic abilities seen from the method of teaching provided.
Variety Analysis III (Factorial Experiments)
In the analysis of variance III there are a number of treatment in each of several categories that form a grid treatment. Selection of decomposition of the design of treatment leads to the treatment sum of squares into components of the additive-konponen following hypothesis test (Steel and Torrie, 1993: 403).
Factor is a kind of treatment, and in the factorial experiment, each factor has several treatments. Factorial experiment is an experiment so that treatment consists of all possible combinations of levels of several factors. This experiment gives a huge advantage in explanatory research, which to our knowledge about the optimum level of each factor is still very minimal. This experiment is testing the contents of columns and rows of data, as well as whether there is any interaction between the contents of columns and rows. Satatistik procedure used is TWO-WAY ANOVA.
Example applications:
School Name X Y Standard
A 49 71 83
A 60 60 87
B 57 65 89
B 59 59 92
In this model the interaction between variables was tested columns and rows.
1. Create a hypothesis
Ho: There is no interaction between the methods with these schools
Hi: There is an interaction between the methods with these schools
2. Procedures with SPSS 16.0
REFERENCES
Steel and Torrie, Principles and Procedures of Statistics A Biometrics Approach, Jakarta, Gramedia Pustaka Utama, 1993.
Hasan, Iqbal, Research Analysis With Statistics, Jakarta, Earth Literacy, 2006.
Hadi, Sutrisno, Statistics 3, Yogyakarta, Andi Offset, 1993.
Suharyadi, Statistics for Modern Economy and Finance, Jakarta, Salemba Four, 2004.
Setyowati, Eni, Diktat Maetode Statistics, Tulungagung, STAIN, 2008.
Sudjana. Statistical Methods. Bandung, Tarsito, 1996.
Setyowati, Eni, Application Module Statistics with SPSS 16.0. Tulungagung, STAIN, 2008.
Hamang, Abdul, Statistical Methods, Jakarta, Graha Science, 2005.
Santoso, Singgih, Questions answered Statistics with SPSS and EXCEL, Jakarta, Gramedia, 2007.
Herawati, Nita, 2007, Experimental Design (Online) http://lemlit.unila.ac.id/file/makalah% 20pdf/BAHANMETODOL.DOSEN.pdf, accessed on January 28, 2008
Kurniawan, Deni, 2007. Analysis of variance of a Direction (One Way ANOVA) ineddeni.files.wordpress.com/2007/11/oneway.pdf - Similar pages, accessed on 28 January 2008
Posts
0 comments:
Post a Comment