Sie sind auf Seite 1von 24

NATIONAL UNIVERSITY OF RWANDA FACULTY OF ECONOMICS AND MANAGEMENT DEPARTMENT OF APPLIED STATISTICS ACADEMIC YEAR: 2011-2012, BACC: IV

ASSIGNMENT OF MULTIVARIATE ANALYSIS BASIC CONCEPTS OF EXPERIMENTAL DESIGN AND ONE WAY ANALYSIS OF VARIANCE

Presented by:
NDIBANJE Gilbert UG10105473

Lecturer: Mr. NIRAGIRE Francois Done at Huye, The14th , May2012

Table of content
PART I: BASIC CONCEPTS OF EXPERIMENTAL DESIGN 1. Definition 2. Description of experimental design 3. Assumtions of experimental design 5. Illustration 5. Conlusion PART II: ONE-WAY ANALYSIS OF VARIANCE 2.1. Definition 2.2. Description 2.3. Assumptions 2.4. Example 2.5. Conclusion

PART I: BASIC CONCEPTS OF EXPERIMENTAL DESIGN 1. DEFINITION

Experimental design is the process of planning a study to meet specified objectives (SAS, 2005). It includes both Strategies for

organizing data collection and data analysis procedures matched


to those data collection strategies .

2. DESCRIPTION OF EXPERIMENTAL DESIGN 2.1. Designing an Experiment

When design the experiment the following steps will be


perfomed:
Define the problem and the questions to be addressed. Define the population of interest. Determine the need for sampling. Define the experimental design.

2.1.1. Define the problem and the questions to be addressed

This first step of experimental design is consisting in identifying clearly the specific questions that the researcher is lanning to examine .The researcher should identify the sources of valiability in

experimental conditions. The objectives of design experiment:


To partition the effects of the sources of variability into distinct

components in order to evaluate the specific questions on which the researcher is interested.
To improve the precision of the results in order to examine or to test

the research hypothesis.

2.1.2. Define the population of interest


A population is a collective or a set of all people, animals,

plants, or other items that researchers collect data from.


The designed experiment should designate the population for which the problem will be examined. The whole population on which the researcher will based in making inferences will be the focus of the experimental design.

2.1.3. Determine the need for sampling


A sample is finite part of a statistical population whose

properties are studied to gain information about the whole


population.The results from a sample are then used to draw valid inferences or conclusions about the population.

2.1.4. Define the experimental design


A clear definition of the details of the experiment makes the

desired statistical analyses possible, and almost always


improves the usefulness of the results. It is consisting of the following steps:
Identify the experimental unit.
Identify the types of variables.

Define the treatment structure.


Define the design structure.

3. ASSUMPTIONS OF EXPERIMENTAL DESIGN Data from experiments are analyzed using linear regression and analysis of variance . The standard assumptions for data analysis that apply to linear regression and the analysis of variance are now summarized as follows: 1. No model specification error The response Y is the dependent variable. The independent variables, x1,..,xp, influence Y. The form of the relationship between Y and (x1,..,xp) is linear (not nonlinear) in the parameters. 2. No measurement error The dependent variable(s) are interval or ratio data (not ordinal or nominal). The independent variables are measured accurately.

Assumptions
3.No collinearity (a high correlation between any two independent . variables is not allowed). 4.The error term,residuals,are well-behaved when the following conditions hold: A zero mean and Homoscedasticity No autocorrelation (usually of most concern with time series or spatial data) No large correlations between any of the independent variables and Normal distribution

4. ILLUSTRATION The example is for a study that can be perfomed in evaluating the contribution of NGOs in economic growth of their beneficiaries. Grobal fund in Nyamagabe and Huye Districts is taken as case study. 4.1. Define the problem and the question to be addressed The problem was to know how Grobal fund as NGOs is

contributing on the economic growth,as the number of NGOs in


Rwanda is increasing and the poverty remains the serious problem. The following questions: Is the income generated by the beneficairies depending on the kind of support they are given?

4.2. Defining the population of interest


Within this study the whole population on which the research will be concerned is the set of 60000 beneficiaries of Grobal fund from Nyamagabe and Huye Districts. 4.3. Determine the need for sampling Using techniques such as random selection after stratification or blocking is often preferred. The sample of 100 beneficiaries was randomly selected from all two Districts. Sample size was been calculated as follows :
n N 1 N (e)
2

60000 60000 99 .83 1 60000 (0.1) 2 1 600

4.4. Defining the experimental design


The experimental units in this study are the beneficiaries of

Grobal fund.
The variable that can be identified at this level are so many;

Primary variables (income, kind of suport given), background variables ( age,sex, number of family members,),etc
Treatment structure: the study will treat the income of

beneficiaries according to the kind of suport they are given.


The design structure that will be used is Completely

Randomized Design (CRD) where Subjects are assigned to


treatments completely at random.

Completely Randomized Design (CRD) Suppose the study is going to be conducted on the evaluation of the contribution of Grobal Fund on the economic growth of their beneficiaies in 2 Districts (Nyamagabe and Huye). Their Beneficiaries from 2 Districts of sourthern provinces for Completely Randomized Design (CRD) will be randomly assigned to one of two treatment groups (income and kind of support provided) .The total number of beneficiaries in two districts, on which to conduct a

research is 100. Randomly assign 1/2 of them, or 50 beneficiaries, to


each of the 2types of evaluations.

PART II: ONE-WAY ANALYSIS OF VARIANCE

2.1. Definition
Analysis of Variance is a linear model that relates nominal

predictor variables to a continuous outcome variable.


According to Silicon Genetics (2003), One-way analysis of variance (ANOVA) tests is allowing determining if one given factor, such as drug treatment, has a significant effect on gene expression behavior across any of the groups under study.

Description
The analysis of variance model examines the association between nominal predictor variables or factors and a continuous outcome variable or dependent. The mathematical model that describes the relationship between the response and treatment for the one-way ANOVA is given by :
yij i ij

Where Yij represents the j-th observation (j = 1, 2, ...ni) on the i-th treatment (i = 1, 2, ..., k levels. ij is the common effect for the

whole experiment, i represents the i-th treatment effect and ij


represents the random error present in the j-th observation on the i-th treatment.

2.3.

Assumptions

Zero mean value of disturbances: E(u)=0

No serial correlation, or cov(ui,uj)=0


Homoscedasticity, or var(ui)=2 Error term are normaly distributed with (0, 2 )

2.4. EXAMPLE
In the example of completely randomized design of evaluating the
contribution of Grobla Fund in economic growth, we evaluate the dependance between the income of 100 beneficiaries of Grobla Fund and their kind of suport (Health insurance, Livestock, Construction of house and schooll fees)the data cfr: word doc.

H0: 1 = 2 = 3 = 4, i.e. the income of beneficiaries of gaining


different suports is the same or the income of beneficiaries is equally affected by the kind of suport they are provided. H1: The means are not all equal.

Summary of data
TA1=315102 TA2=266278 TA3=213978 TA4=216837
( y ij ) 2 N (1012196 ) 2 1024538718 0 100

T 2 Aj
Aj

3151022 2662782 2139782 2168372 10259122949 32 25 22 21

SSTotal yij
2

( yij ) 2 N

758573123

T 2 Aj ( y Aj ) SSBetween 13735769 N Aj N
2

yij 11003960303
2

T 2 Aj SSWithin yij 744837354 N AJ


2

The ANOVA Table:


Source A(or Treatment,or explained Error(Residual) SS SS Between =13735769 SS Within =744837354 SS Total =758573123 DF J-1 4-1=3 Mean squre SS Between/(J-1)
13735769 4578589.576 3

MSBetween MSwithin

N-J SS Within/(N-J) 744837354 100-4=96 7758722.438


96

Total

N-1 100-1=99

SS Total/(N-1)

4578589.576 0.590122 7758722.438

Using SPSS
Using SPSS, the following is the output provided:

As conclusion, we fail to reject the null hypothesis because the p-value of 0.623 is greater than the level of signicance of 5%.

Conclusion
To sum up the experimental designs can be analyzed by using a
specific ANOVA that is designed for that experimental design. The statisticians have to recognize the various experimental designs and help clients or users create a design and analyze the experiments by using appropriate methods and software.

References

http://stat.stanford.edu/~jtaylo/courses/stats203 http://stattrek.com/experiments/ http://support.sas.com/ SAS (2005), Concepts of Experimental Design, USA www.sas.com Zar, J. (1999) Biostatistical Analysis (4th ed.) Upper Saddle River, NJ, Prentice Hall. http://www.themeasurementgroup.com http://www.smartersolutions.com Israel, Glenn D. 1992. Sampling the Evidence Of Extension Program Impact. Program Evaluation and Organizational Development, IFAS, University of Florida. PEOD-5. October.

THANK YOU FOR YOUR KIND ATTENTION!!!!

Questions, ideas and comments are welcome!!!!

Das könnte Ihnen auch gefallen