Anda di halaman 1dari 6

1

Kruskal - Wallis Test and Friedman Test (DR SEE KIN HAI)
(NON-PARAMETRIC TESTS) 1. If you have violated the assumptions to carry out the One-Way Between Groups ANOVA, then you can use Kruskal-Wallis Test which is not so stringent on the assumptions as stated below: 1. Data is randomly collected from the population. (Normality Test of samples see BASIC 3) 2. Variance for the data distribution is equal. (Homogeneity of Variance Levenes Test see L48) 3. Independence of samples. Example A Geography teacher wanted to investigate the effectiveness of three different teaching methods on students performance in Geography. He divided the students into 3 groups (Group 1 using Method 1, Group 2 using Method 2 and Group 3 using Method 3). After 3 weeks, all the students sat for a Geography test. (Assuming the data has violated the assumptions to carry out the 1-way ANOVA e.g. there is a significant difference in the Levenes test of equal Variance where p < 0.05).

1=Group 1, 2=Group 2, 3= Group3

How to run Kruskal-Wallis Test 1. Select [Analyze] [Nonparametric Tests] [Legacy Dialogs] [K Independent Samples..] to open the dialogue box below.

2. Move [Geography Marks] into [Test Variable List] and [Groups] into [Grouping Variable] box then click on [Define Range] to open the sub-dialogue box. 3. Enter [1] into the [Minimum] box and [3] into the [Maximum] box, then [Continue] and [OK].

The means for the 3 groups are shown with Group 1 higher than Group 2 and Group 2 higher than Group 3.

2 =12.50 with DF=2 and p =0.002 indicating that there is a significant difference in the mean scores between the 3 groups.

3 Reporting the output The Kruskal-Wallis Test with 2 =12.50, DF =2 and p = 0.002 indicating that there is a significant difference in the mean scores between the 3 Groups. Thus, the different teaching methods are effective in causing the differences between the 3 Groups. COURSEWORK A Biology teacher wanted to investigate the effectiveness of 3 different teaching methods (1,2 and 3) on Form 6 students performance in Biology. A quasi-experimental research design has been chosen for this study. By using the intact classes, the students are divided into 3 Groups using 3 different teaching strategies (Group 1 using strategy 1 , Group 2 strategy 2 and Group 3 strategy 3). After 3 weeks, the students were given a Biology test and the scores are shown below. (Assuming that the data has violated the assumptions to use a 1-way between groups ANOVA test)

Note: Quasi experiments resemble quantitative and qualitative experiments, but lack random allocation of groups or proper controls, so firm statistical analysis can be very difficult. Quasi-experimental design involves selecting groups, upon which a variable is tested, without any random pre-selection processes. For example, to perform an educational experiment, a class might be arbitrarily divided by alphabetical selection or by seating arrangement. The division is often convenient and, especially in an educational situation, causes as little disruption as possible. After this selection, the experiment proceeds in a very similar way to any other experiment, with a variable being compared between different groups, or over a period of time.
ADVANTAGES

Especially in social sciences, where pre-selection and randomization of groups is often difficult, they can be very useful in generating results for general trends.

Friedman Test
If the data has violated the assumptions to use One-Way Repeated measures/ Related ANOVA then use Friedman Test. Assumptions 1. Data is randomly collected from the population. (Normality Test of samples see BASIC 3) 2. Variance for the data distribution is equal. (Homogeneity of Variance Levenes Test see L48) 3. Independence of samples. A Malay Language teacher wanted to investigate the effectiveness of 3 different teaching methods on the same Year 7 students performance in Bahasa Melayu. The students were first taught using strategy 1 then after 3 weeks were given a test. The same sample was then taught using strategy 2 and given a similar test after 3 weeks. The sample was again taught using strategy 3 and given a similar test after another 3 weeks. It was hypothesized that the different strategies will affect significantly students performance in Bahasa Melayu. (Assuming the data violated the assumptions to carry out the 1-way Related Measures ANOVA e.g. there is a significant difference in the Levenes test of equal Variance where p < 0.05).

5 How to run Friedman Test 1. Select [Analyze] [Nonparametric Test] [Legacy Dialogs] [K Related Samples..] to open the dialogue box below.

2. Move [Method1], [Method2] and [Method3] into the [Test Variables] box and select [Statistics].

3. Click [Descriptive] [Continue] [OK].

Means, SD, Min and Max for the 3 methods are shown in this table
Method 2 is the best strategy and Method 3 is the weakest strategy

2 = 10.00 with DF =2 and p = 0.007 Results of the Friedman test indicate significant differences did exist across the 3 different methods of teaching.

6 COURSEWORK A Biology teacher wanted to investigate the effectiveness of 3 different teaching methods (1,2 and 3) on the same Form 6 students performance in Biology. A quasi-experimental research design has been chosen for this study. By using the intact classes, the students are taught with Strategy 1 and after 2 weeks were given a test. The same group of students were then taught using strategy 2 and after 2 weeks were given a similar test. The students were again taught with strategy 3 and a similar test was given after 2 weeks. The scores are shown below. (Assuming the data has violated the assumptions to use a 1way repeated measures ANOVA test)