Name Jahangir Khan

Father Name Soorat Khan

Roll No. CB643097

Semester Autumn 2021

Tutor Name Din Muhammad

Course Name Educational Statistics

Course Code 8614

Assignment No. 2

Contact No 03138263473

Date Of Submission 01/04/2022

1
 QUESTION NO.1

What is the importance of inferential statistics in educationa

.research? Elaborate with examples

In educational research, it is never possible to sample all of the people that we want to draw a
conclusion about. For example, to the purpose of a research study may be to determine if a new
way of teaching mathematics improves mathematical achievement for all children in Primary 1.
However, it would be impossible to test all children in Primary 1 because of time, resources,
and other logistical factors. Instead, the researcher chooses a sample of the population to
conduct a study. However, the researcher does not want to limit her conclusions to only the few
children who participated in the study. Instead, she wants to say that because her new way of
teaching mathematics improved mathematical achievement for the children in her sample, it
will also improve mathematical achievement for all children in Primary 1. Thus, the researcher
wants to draw conclusions - or inferences - about the entire population based on the results from
the sample.
The purpose of inferential statistics is to determine whether the findings from the sample can generalize - or be
applied - to the entire population. There will always be differences in scores between groups in a research study.
For example, imagine the following scores on WAEC from students who attended either a government or a
private school. (Note that the data was created by a random number generator and thus does not reflect actual
scores.)

Course Government Private

English 24.13 28.18

Mathematics 58.90 27.87

2
 Integrated Science 61.19 47.89

Chemistry 34.38 34.25

Social Studies 99.10 86.97

CRK/IRK 70.90 70.53

In all courses listed in Table, there is a difference in the average scores between the government
and private schools. However, which differences are large enough to be significant? In
mathematics, there is a clear difference in the mean scores. On the other hand, the difference
between government and private schools on CRK/IRK is less than 1 point. This clearly does not
represent a meaningful difference between the two types of schools. How big must the
difference be in order to conclude that there is a meaningful difference in the academic
performance between government and private schools?
Inferential statistics must be used to determine whether the difference between the two groups
in the sample is large enough to be able to say that the findings are significant. If the findings
are indeed significant, then the conclusions can be applied, or generalized, to the entire
population. On the other hand, if the difference between the groups is very small, then the
findings are not significant and therefore were simply the result of chance.
Returning to the WAEC illustration, this research found that there is a difference in CRK/IRK
performance in our sample: government schools had a slightly higher mean score than private
schools. However, after conducting inferential statistics, we will find that the difference is not
significant. This means that the difference between government and private schools in our
sample was too small to conclude that an actual difference exists in the entire population of
government and private schools. This tiny difference in CRK/IRK performance was simply due
to chance. If we collected data from another sample of government and private schools, we will
likely find the opposite of this small difference.
On the other hand, after conducting inferential statistics, imagine that we find that the difference
in mathematics performance is significant. Therefore, the results from our sample - government
schools have higher performance in mathematics than private schools - can also be applied to

3
the entire population. Thus, we conclude that in general, government schools outperform
private schools in mathematics. This means that if we collect data from another sample, we are
almost guaranteed to find the exact same result.
When calculating inferential statistics, the key statistic is the p statistic. This p-value is the
probability that the result is due to chance. The p-value can range from 0.000 to 1.000. The
larger p is, the more likely the results are due to chance. If p is 0.500, the probability that the
result is due to chance is 5 out of 10. This means that there is an equal probability that the result
is significant in the population or that the result is due to chance. A p of 0.850 means that the
probability that the result is due to chance is 85 out of 100, meaning that the finding is most
likely due to chance. A p of 0.050 means the probability that the result is due to chance is 5 out
of 100.
Researchers want to be fairly certain of their conclusions, so they have decided that the
probability that the result is due to chance should be very small - less than 5 out of 100. This is
why the standard in research is that the p must be is less than .05 for the results to be significant.
(This cut-off point for p is oftentimes called the alpha, sometimes stated that alpha is less
than .05.)
Recall from Writing Research Hypotheses that research hypotheses are generally stated as null
hypotheses: There is no significant difference between government and private schools on
mathematics performance. If the calculated p-value is less than 0.050, then the null hypothesis
is rejected. If the statement "There is no significant difference" is rejected, this means that there
is a significant difference. It can therefore be concluded that a difference exists between the two
groups in the population, not just the sample. On the other hand, if p is greater than 0.050, this
means that the null hypothesis is retained. This means the null hypothesis, as stated, is
accurate: in the population, there is no significant difference.
At this point, the data should be coded and total scores calculated for each variable. It is now
time to calculate the inferential statistics. I highly recommend using VassarStats, a free website
that calculates statistics for all but the most complex studies. The specific statistic to calculate
for each research hypothesis should have been already identified in Method of Data
Analysis step. If this has been done, the data analysis should take less than an hour to complete.

4
Each type of inferential statistic is slightly different, so read below for how to calculate each
statistic.
t-test:
The t-test should be used when comparing two groups on a dependent variable. Thus, a t-test
will be used to compare a treatment group to a control group or to compare males and females.
For sake of illustration, refer back to the comparison of government and private schools on
WAEC scores. To calculate the t-test, the data must be sorted according to the independent
variable, in this case type of school. In other words, all of the government school scores are
grouped together, and all of the private school scores are grouped together. In VassarStats, click
on the t-test, then Two-sample t-test for independent or correlated samples. In the new screen,
click Independent Samples. (Correlated samples would be for studies where the two groups
were matched on another relevant variable, such as intelligence, or when both scores came from
the same group of people, such as comparing English and maths scores within the government
schools.) Then enter the mathematics scores for each student in the government school group in
one sample area (Sample A) and the mathematics scores for each student in the private school
group in the other. Click Calculate and the results should pop up.
Scroll down the screen. The first statistic to look at is the p: two-tailed. If this p is greater than
0.050, then the null hypothesis is retained; the result is not significant. If the result is not
significant, analysis and interpretation is finished because there is no significant difference
between groups.
If this p is less than 0.050, then the null hypothesis is rejected and the result is significant. If this
is significant, then the next step is to look at the mean score for each group (scroll back up the
page for this). Based on those mean scores, which group had the higher mathematics scores:
government or private schools? This last step of examining the mean scores is often overlooked
in data analysis, but it is very important to identify which group had the highest mean score. It
is important to record the means and standard deviations for the two groups, the t, df, and two-
tailed p. Click the Reset button and move to the next research hypothesis.
Analysis of Variance (ANOVA):

5
When comparing three or more groups on one dependent variable, an Analysis of Variance is
the statistics to use. There are two basic types of ANOVAs that can be used.
One-way ANOVA: A one-way ANOVA compares multiple groups on the same variable.
For example, perhaps the researcher decides to divide private schools into religious private and
secular private. Now, there are three groups to be compared: government schools, religious
private, and secular private. A one-way ANOVA is now necessary. To calculate the one-way
ANOVA, the data must be sorted according to the independent variable - again, school type. In
Vassar Stats, click on ANOVA, then One-Way ANOVA. Then enter the number of samples
(aka the number of groups; in this example, 3). Then click Independent Samples. Enter the
mathematics scores for each student in the appropriate column. For example, enter government
students' scores in Sample A, religious private students' scores in Sample B, and secular private
students' scores in Sample C. Then click Calculate.
Scroll down the screen. Again, the first statistic to look at is the p in the ANOVA summary
table. Again, if this p is greater than 0.050, then the null hypothesis is retained; the result is not
significant. If the result is not significant, analysis and interpretation is finished because there is
no significant difference between groups.
If this p is less than 0.050, then the result is significant. This only means, however, that there is
a significant difference between groups somewhere, not that there is a significant difference
between all groups. It is possible that government students were significantly higher than
religious private and secular private students, but there are no significant differences between
religious private and secular private students. Down at the bottom of the screen is the result of
Turkey’s HSD (Honestly Significant Difference) test. This test identifies which differences are
really significant. It is important to record the means and standard deviations for all groups, the
ANOVA summary table, and the results of Turkey’s HSD. Click the Reset button and move to
the next research hypothesis.
Factorial ANOVA: The factorial ANOVA compares the effect of multiple independent
variables on one dependent variable. For example, a 2x3 factorial ANOVA could compare the
effects of gender and school type on academic performance. The first independent variable,
gender, has two levels (male and female) and the second independent variable, school type, has

6
three levels (government, religious private, and secular private), hence 2x3 (read "two by
three"). Factorial ANOVAs can also be calculated on Vassar Stats (click on ANOVA then on
Two-way factorial ANOVA for independent samples). However, this interpretation is a bit
more complex so please see an expert statistician to help with interpreting the results.
Analysis of Covariance (ANCOVA);-
When using a pre-post test research design, the Analysis of Covariance allows a comparison of
post-test scores with pre-test scores factored out. For example, if comparing a treatment and
control group on achievement motivation with a pre-post test design, the ANCOVA will
compare the treatment and control groups' post-test scores by statistically adjusting for the pre-
test scores. For an ANCOVA, you must have pre- and post-test scores for every person in the
sample, and these scores must be sorted by the group (aka treatment and control group).
To calculate an ANCOVA with VassarStats, click on ANCOVA. Then VassarStats will ask for
the k. The k is the number of groups. If there is only one treatment and one control group, then
k=2. Click on the correct k for data import. There are two things to bear in mind when doing
ANCOVA with VassarStat. It will ask for the concomitant variable and the dependent variable.
The concomitant variable (CV) is the variable that should be controlled for. In the case of a pre-
post test design, the concomitant variable is the pre-test. The dependent variable (DV) is the
variable that you think has been affected by the independent variable. In the case of a pre-post
test design, the dependent variable is the post-test. To use VassarStats, it is important that the
CV and the DV are side-by-side for each of the two groups. Then enter the CV and DV into the
correct columns and click Calculate.
Scroll down the screen. Just as before, the first statistic to look at is the p in the ANCOVA
summary table. If this p is less than 0.050, then the null hypothesis is rejected and the result is
significant. There are two sets of means that are important to understand in an ANCOVA. First,
the Observed Means are the actual means for the dependent variable (post-test). Then the
Adjusted Means are the means that have been statistically manipulated based on the pre-test
scores. A simple way to imagine this is that the ANCOVA statistically forces the pre-test scores
to be equal between the two groups (meaning that the two groups are now equal at the start of
the study), and then re-calculates the post-test scores based on the adjusted pre-test scores. It is

7
important to record the observed means, adjusted means, and standard deviations for all groups
and the ACNOVA summary table. When creating the tables in the next step, report both the
Observed and Adjusted Means. However, make any figures based with the Adjusted Means.
Add a note to the figure so that readers are clear that these are Adjusted Means. Click the Reset
button and move to the next research hypothesis.
Correlation:-
Correlations should be calculated to examine the relationship between two variables within the
same group of participants. For example, the correlation would quantify the relationship
between academic achievement and achievement motivation. To calculate a correlation, you
must have scores for two variables for every participant in the sample. To calculate a correlation
in VassarStats, click on Correlation & Regression, then Basic Linear Correlation and
Regression, Data-Import Version. Enter the total scores for the two variables and click
Calculate.
Scroll down the screen. Again, the first statistic to look at is the p: two-tailed. The null
hypotheses for correlations state, There is no significant relationship between mathematics and
English achievement. If the p is greater than 0.050, then the null hypothesis is retained; there is
no significant relationship between variables. If the result is not significant, analysis and
interpretation is finished because there is no significant relationship.
If this p is less than 0.050, then the null hypothesis is rejected and the correlation is significant.
If the correlation is significant, then the next step is to look at the correlation itself, symbolized
by r. For more information on how to interpret the correlation, click on Method of Data
Analysis. It is important to record the means and standard deviations for the two variables, the t,
df, two-tailed p, and r. Click the "Reset" button and move to the next research hypothesis.

Question NO:-2
What mean by hypothesis testing? Briefly state the logic
behind hypothesis testing.

8
Hypothesis testing is an act in statistics whereby an analyst tests an assumption regarding a
population parameter. The methodology employed by the analyst depends on the nature of the
data used and the reason for the analysis.
Hypothesis testing is used to assess the plausibility of a hypothesis by using sample data. Such
data may come from a larger population, or from a data-generating process. The word
"population" will be used for both of these cases in the following descriptions.
 Hypothesis testing is used to assess the plausibility of a hypothesis by using sample data.
 The test provides evidence concerning the plausibility of the hypothesis, given the data.
 Statistical analysts test a hypothesis by measuring and examining a random sample of
the population being analyzed.
In hypothesis testing, an analyst tests a statistical sample, with the goal of providing evidence
on the plausibility of the null hypothesis.
Statistical analysts test a hypothesis by measuring and examining a random sample of the
population being analyzed. All analysts use a random population sample to test two different
hypotheses: the null hypothesis and the alternative hypothesis.
The null hypothesis is usually a hypothesis of equality between population parameters; e.g., a
null hypothesis may state that the population mean return is equal to zero. The alternative
hypothesis is effectively the opposite of a null hypothesis (e.g., the population mean return is
not equal to zero). Thus, they are mutually exclusive, and only one can be true. However, one
of the two hypotheses will always be true.
4 Steps of Hypothesis Testing:-
All hypotheses are tested using a four-step: process
1. The first step is for the analyst to state the two hypotheses so that only one can be right.
2. The next step is to formulate an analysis plan, which outlines how the data will be
evaluated.
3. The third step is to carry out the plan and physically analyze the sample data.
4. The fourth and final step is to analyze the results and either reject the null hypothesis, or
state that the null hypothesis is plausible, given the data.

9
If, for example, a person wants to test that a penny has exactly a 50% chance of landing on
heads, the null hypothesis would be that 50% is correct, and the alternative hypothesis would
be that 50% is not correct.
Mathematically, the null hypothesis would be represented as Ho: P = 0.5. The alternative
hypothesis would be denoted as "Ha" and be identical to the null hypothesis, except with the
equal sign struck-through, meaning that it does not equal 50%.
A random sample of 100 coin flips is taken, and the null hypothesis is then tested. If it is found
that the 100 coin flips were distributed as 40 heads and 60 tails, the analyst would assume that
a penny does not have a 50% chance of landing on heads and would reject the null hypothesis
and accept the alternative hypothesis.
If, on the other hand, there were 48 heads and 52 tails, then it is plausible that the coin could be
fair and still produce such a result. In cases such as this where the null hypothesis is
"accepted," the analyst states that the difference between the expected results (50 heads and 50
tails) and the observed results (48 heads and 52 tails) is "explainable by chance alone."

Question NO #3:-
Briefly explain what you understand by “correlation”? Write
down where and why to use correlation? Write down the
consideration that should be kept in mind while
interpreting correlation.
Regression analysis involves identifying the relationship between a dependent variable and one
or more independent variables. A model of the relationship is hypothesized, and estimates of the
parameter values are used to develop an estimated regression equation. Various tests are then
employed to determine if the model is satisfactory. If the model is deemed satisfactory, the
estimated regression equation can be used to predict the value of the dependent variable given
values for the independent variables.
Regression model.
In simple linear regression, the model used to describe the relationship between a single
dependent variable y and a single independent variable x is y = a0 + a1x + k. a0and a1 are

10
referred to as the model parameters, and is a probabilistic error term that accounts for the
variability in y that cannot be explained by the linear relationship with x. If the error term were
not present, the model would be deterministic; in that case, knowledge of the value of x would
be sufficient to determine the value of y.
Least squares method.
Either a simple or multiple regression model is initially posed as a hypothesis concerning the
relationship among the dependent and independent variables. The least squares method is the
most widely used procedure for developing estimates of the model parameters.
As an illustration of regression analysis and the least squares method, suppose a university
medical centre is investigating the relationship between stress and blood pressure. Assume that
both a stress test score and a blood pressure reading have been recorded for a sample of 20
patients. The data are shown graphically in the figure below, called a scatter diagram. Values of
the independent variable, stress test score, are given on the horizontal axis, and values of the
dependent variable, blood pressure, are shown on the vertical axis. The line passing through the
data points is the graph of the estimated regression equation: y = 42.3 + 0.49x. The parameter
estimates, b0 = 42.3 and b1 = 0.49, were obtained using the least squares method.

Correlation.
Correlation and regression analysis are related in the sense that both deal with relationships
among variables. The correlation coefficient is a measure of linear association between two
variables. Values of the correlation coefficient are always between -1 and +1. A correlation

11
coefficient of +1 indicates that two variables are perfectly related in a positive linear sense, a
correlation coefficient of -1 indicates that two variables are perfectly related in a negative linear
sense, and a correlation coefficient of 0 indicates that there is no linear relationship between the
two variables. For simple linear regression, the sample correlation coefficient is the square root
of the coefficient of determination, with the sign of the correlation coefficient being the same as
the sign of b1, the coefficient of x1 in the estimated regression equation.
Neither regression nor correlation analyses can be interpreted as establishing cause-and-effect
relationships. They can indicate only how or to what extent variables are associated with each
other. The correlation coefficient measures only the degree of linear association between two
variables. Any conclusions about a cause-and-effect relationship must be based on the judgment
of the analyst.

Question No#4:-
When will you use ANOVA in your research? Write down
the logic behind using ANOVA.
In some decision-making situations, the sample data may be divided into various groups i.e. the
sample may be supposed to have consisted of k-sub samples. There are interest lies in
examining whether the total sample can be considered as homogenous or there is some
indication that sub-samples have been drawn from different populations. So, in these situations,
we have to compare the mean values of various groups, with respect to one or more criteria. 
The total variation present in a set of data may be partitioned into a number of non-overlapping
components as per the nature of the classification. The systematic procedure to achieve this is
called Analysis of Variance (ANOVA). With the help of such a partitioning, some testing of
hypothesis may be performed. 
Initially, Analysis of Variance (ANOVA) had been employed only for the experimental data
from the Randomized Designs but later they have been used for analyzing survey and secondary
data from the Descriptive Research. 
Analysis of Variance may also be visualized as a technique to examine a dependence
relationship where the response (dependence) variable is metric (measured on interval or ratio

12
scale) and the factors (independent variables) are categorical in nature with a number of
categories more than two. 
Example of ANOVA:-
Ventura is an FMCG company, selling a range of products. Its outlets have been spread over the
entire state. For administrative and planning purpose, Ventura has sub-divided the state into
four geographical-regions (Northern, Eastern, Western and Southern). Random sample data of
sales collected from different outlets spread over the four geographical regions. 
Variation, being a fundamental characteristics of data, would always be present. Here, the total
variation in the sales may be measured by the squared sum of deviation from the mean sales. If
we analyze the sources of variation in the sales, in this case, we may identify two sources:
 Sales within a region would differ and this would be true for all four regions (within-
group variations)
 There might be impact of the regions and mean-sales of the four regions would not be
all the same i.e. there might be variation among regions (between-group variations).
So, total variation present in the sample data may be partitioned into two components: between-
regions and within-regions and their magnitudes may be compared to decide whether there is a
substantial difference in the sales with respect to regions. If the two variations are in close
agreement, then there is no reason to believe that sales are not same in all four regions and if not
then it may be concluded that there exists a substantial difference between some or all the
regions.
Here, it should be kept in mind that ANOVA is the partitioning of variation as per the
assignable causes and random component and by this partitioning ANOVA technique may be
used as a method for testing significance of difference among means (more than two). 
If the values of the response variable have been affected by only one factor (different categories
of single factor), then there will be only one assignable reason by which data is sub-divided,
then the corresponding analysis will be known as One-Way Analysis of Variance.  The
example (Ventura Sales) comes in this category. Other examples may be: examining the
difference in analytical-aptitude among students of various subject-streams (like engineering

13
graduates, management graduates, statistics graduates); impact of different modes of
advertisements on brand-acceptance of consumer durables etc.
On the other hand, if we consider the effect of more than one assignable cause (different
categories of multiple factors) on the response variable then the corresponding analysis is
known as N-Way ANOVA (N>=2). In particular, if the impact of two factors (having multiple
categories) been considered on the dependent (response) variable then that is known as Two-
Way ANOVA. For example: in the Ventura Sales, if along with geographical-regions
(Northern, Eastern, Western and Southern), one more factor ‘type of outlet’ (Rural and Urban)
has been considered then the corresponding analysis will be Two-Way ANOVA. More
examples: examining the difference in analytical-aptitude among students of various subject-
streams and geographical locations; the impact of different modes of advertisements and
occupations on brand-acceptance of consumer durables etc.
Two-Way ANOVA may be further classified into two categories:
 Two-Way ANOVA with one observation per cell: there will be only one observation
in each cell (combination). Suppose, we have two factors A (having m categories) and
B (having n categories), So, there will be N= m*n total observations with one
observation(data-point) in each of (Ai Bj) cell (combination), i=1, 2, ……., m and j= 1,
2, …..n. Here, the effect of the two factors may be examined.
 Two-Way ANOVA with multiple observations per cell: there will be multiple
observations in each cell (combination). Here, along with the effect of two factors,
their interaction effect may also be examined. Interaction effect occurs when the
impact of one factor (assignable cause) depends on the category of other assignable
cause (factor) and so on. For examining interaction-effect it is necessary that each cell
(combination) should have more than one observations so it may not be possible in the
earlier Two-Way ANOVA with one observation per cell.
Conceptual Background:-
The fundamental concept behind the Analysis of Variance is “Linear Model”. 
X1, X2,……….Xn are observable quantities. Here, all the values can be expressed as:
Xi = µi + ei 

14
:-Where µi is the true value which is because of some assignable causes and e i is the error term
which is because of random causes. Here, it has been assumed that all error terms e i are
independent distributed normal variate with mean zero and common variance (σe2). 
Further, true value µi can be assumed to be consist of a linear function of t 1, t2,…….tk, known as
“effects”.
If in a linear model, all effects tj’s are unknown constants (parameters), then that linear model is
known as “fixed-effect model”. Otherwise, if effects tj’s are random variables then that model
is known as “random-effect model”. 
One-Way Analysis of Variance :-
We have n observations (Xij), divided into k groups, A1, A2,…….Ak, with each group having
nj observations.
Here, the proposed fixed-effect linear model is: 
Xij = µi + eij 
Where µi is the mean of the ith group.
General effect (grand mean): µ = Σ (ni. µi)/n 
and additional effect of the ith group over the general effect: αi = µi – µ.
So, the linear model becomes: 
 Xij = µ + αi + eij
with Σi (ni αi) = 0
The least-square estimates of µ and αi may be determined by minimizing error sum of square
(Σi Σj eij2) = Σi Σj (Xij – µ – αi)2 as:
X.. (combined mean of the sample) and Xi. (mean of the ith group in the sample).
So, the estimated linear model becomes:
Xij = X.. + (Xi.- X..) + (Xij– Xi.)
This can be further solved as:
Σi Σj (Xij –X..)2 = Σi ni (Xi.- X..)2 + Σi Σj (Xij –Xi.)2
Total Sum of Square = Sum of square due to group-effect + Sum of square due to error
or
Total Sum of Square= Between Group Sum of square+ Within Group Sum of square 

15
TSS= SSB + SSE
Further, Mean Sum of Square may be given as:
MSB = SSB/(k-1) and MSE = SSE/(n-k), 
where (k-1) is the degree of freedom (df) for SSB and (n-k) is the df for SSE.
Here, it should be noted that SSB and SSE added up to TSS and the corresponding df’s (k-1)
and (n-k) add up to total df (n-1) but MSB and MSE will not be added up to Total MS.
This by partitioning TSS and total df into two components, we may be able to test the
hypothesis:
H0: µ1 = µ2=……….= µk 
H1: Not all µ’s are same i.e. at least one µ is different from others.
or alternatively:
H0: α1 = α2=……….= αk =0
H1: Not all α’s are zero i.e. at least one α is different from zero.
MSE has always been an unbiased estimate of σ e2 and if H0 is true then MSB will also be an
unbiased estimate of σe2. 
Further MSB/ σe2 will follow Chi-square (χ2) distribution with(k-1) df and MSE/ σ e2 will follow
Chi-square (χ2) distribution with(n-k) df. These two χ2 distributions are independent so the ratio
of two Chi-square (χ2) variate F= MSB/MSE will follow variance-ratio distribution (F
distribution) with (k-1), (n-k) df. 
Here, the test-statistic F is a right-tailed test (one-tailed Test). Accordingly, p-value may be
estimated to decide about reject/not able to reject of the null hypothesis H0.
If is H0 rejected i.e. all µ’s are not same then rejecting the null hypothesis does not inform which
group-means are different from others, So, Post-Hoc Analysis is to be performed to identify
which group-means are significantly different from others. Post Hoc Test is in the form of
multiple comparison by testing equality of two group-means (two at a time) i.e. H 0: µp = µq by
using two-group independent samples test or by comparing the difference between sample
means (two at a time) with the least significance difference (LSD)/critical difference (CD) 
= terror-df*MSE/ (1/np+1/nq )1/2  

16
If observed difference between two means is greater than the LSD/CD then the corresponding
Null hypothesis is rejected at alpha level of significance.
Assumptions for ANOVA:-
Though it has been discussed in the conceptual part just to reiterate it should be ensured that the
following assumptions must be fulfilled:
1. The populations from where samples have been drawn should follow a normal distribution.
2. The samples have been selected randomly and independently.  
3. Each group should have common variance i.e. should be homoscedastic i.e. the variability in
the dependent variable values within different groups is equal.
It should be noted that the Linear Model used in ANOVA is not affected by minor deviations in
the assumptions especially if the sample is large.
The Normality Assumption may be checked using Tests of Normality: Shapiro-Wilk Test and
Kolmogorov-Smirnov Test with Lilliefors Significance Correction. Here, Normal Probability
Plots (P-P Plots and Q-Q Plots) may also be used for checking normality assumption. The
assumption of Equality of Variances (homoscedasticity) may be checked using different Tests
of Homogeneity of Variances (Levene Test, Bartlett’s test, Brown–Forsythe test etc.).   
ANOVA vs T-test:-
We employ two-independent sample T-test to examine whether there exists a significant
difference in the means of two categories i.e. the two samples have come from the same or
different populations. The extension to it may be applied to perform multiple T-tests (by taking
two at a time) to examine the significance of the difference in the means of k-samples in place
of ANOVA. If this is attempted, then the errors involved in the testing of hypothesis (type I and
type II error) can’t be estimated correctly and the value of type I error will be much more than
alpha (significance level). So, in this situation, ANOVA is always preferred over multiple
intendent samples T-tests.
Like, in our example we have four categories of the regions Northern (N), Eastern (E), Western
(W) and Southern (S). If we want to compare the population means by using two-independent
sample T-test i.e. by taking two categories (groups) at a time. We have to make 4C2 = 6 number
of comparisons i.e. six independent samples tests have to performed (Tests comparing N with S,

17
N with E, N with W, E with W, E with S and W with S), Suppose we are using 5% level of
significance for the null hypothesis based on six individual T-tests, then type I error will be = 1-
(0.95)6 =1-0.735= 0.265 i.e. 26.5%.

Question No#5:-
What is chi-square (X2) distribution? What are the uses of chi-
square (X2) distribution?
Set up the hypothesis for Chi-Square goodness of fit test:
A. Null hypothesis: In Chi-Square goodness of fit test, the null hypothesis assumes that
there is no significant difference between the observed and the expected value.
B. Alternative hypothesis: In Chi-Square goodness of fit test, the alternative hypothesis
assumes that there is a significant difference between the observed and the expected value.
Compute the value of Chi-Square goodness of fit test using the following formula:

Where,  = Chi-Square goodness of fit test O= observed value E= expected value

Degree of freedom: In Chi-Square goodness of fit test, the degree of freedom depends on
the distribution of the sample.  The following table shows the distribution and an associated
degree of freedom:
Type of distribution No of constraints Degree of freedom
Binominal distribution 1 n-1
Poisson distribution 2 n-2
Normal distribution 3 n-3
Hypothesis testing: Hypothesis testing in Chi-Square goodness of fit test is the same as in
other tests, like t-test, ANOVA, etc.  The calculated value of Chi-Square goodness of fit test is
compared with the table value.  If the calculated value of Chi-Square goodness of fit test is
greater than the table value, we will reject the null hypothesis and conclude that there is a
significant difference between the observed and the expected frequency.  If the calculated value

18
of Chi-Square goodness of fit test is less than the table value, we will accept the null hypothesis
and conclude that there is no significant difference between the observed and expected value.
Chi-Square goodness of fit test is a non-parametric test that is used to find out how the observed
value of a given phenomenon is significantly different from the expected value.  In Chi-Square
goodness of fit test, the term goodness of fit is used to compare the observed sample
distribution with the expected probability distribution.  Chi-Square goodness of fit test
determines how well theoretical distribution (such as normal, binomial, or Poisson) fits the
empirical distribution. In Chi-Square goodness of fit test, sample data is divided into intervals.
Then the numbers of points that fall into the interval are compared, with the expected numbers
of points in each interval.
We use the chi-square test to test the validity of a distribution assumed for a random
phenomenon. The test evaluates the null hypotheses H 0 (that the data are governed by the
assumed distribution) against the alternative (that the data are not drawn from the assumed
distribution).
Let p1, p2, ..., pk denote the probabilities hypothesized for k possible outcomes. In n independent
trials, we let Y1, Y2, ..., Yk denote the observed counts of each outcome which are to be
compared to the expected counts np1, np2, ..., npk. The chi-square test statistic is qk-1 =

= (Y1 - np1)²  + (Y2 - np2)² + ... + (Yk - npk)²

  ----------    ----------          --------
     np1           np2                npk

Reject H0 if this value exceeds the upper   critical value of the  (k-1) distribution, where   
is the desired level of significance.

19