Synopsis....

 Normal Distribution
 Type of Errors
 Hypothesis & Types
Test of Hypothesis
 Test of Significance
Parametric Tests Nonparametric Tests
 A random variable X is said to have a normal distribution
with parameter (,2) if its density function is given by,
 f(x) = (x, ,) = 1 / 2 e – (x-)2/22
 - < x <  - <  <  ,  >0
 Reality True Reality false
Researcher True Correct X  II error (β)
Researcher False X  I error (α) Correct  Power (1 – β)

Type I Error (α): - the probability that difference shown occurred due to
chance
It is known as level of significance / critical region / region of rejection
P [reject H0 when it is true]
Type II Error (β): - if a true difference of a stated magnitude existed but
the study would not have picked up as statistically significant.
P [accept HA when it is false]
Power of Study (1 – β): - the probability that if a true difference of stated
magnitude existed then the study would have picked up as statistically
significant.
 A hypothesis is a statement of predicted
relationship between 2 or more variables.
 A definite statement about the population
parameter.

 Hypothesis are never proved / disproved in

an ultimate sense – they are accepted /
rejected supported / not supported by the
data.
 It provides bridge between theory and reality and in
this sense unifying the two domains.
 It provides direction for any research endeavor by
tentatively identifying the anticipated outcome.
 It is a guide to the thinking process and the
process of discovery.
 It serves as a frame work for drawing conclusions.
Advantages:
 Hypothesis builds researches confidence in his
results
 Sound hypothesis gives direction to the inquiry
 A good hypothesis enriches theory.

Sources: General culture, scientific theory, personal

experience, analogies.
 Simple hypothesis: expected relationship b/w one independent
and one dependent variable.
 Complex hypothesis: prediction of relationship b/w 2 or more
Independent Variable & Dependent Variable
 Directional hypothesis: Specifies not only the existence but the
expected direction of the relationship b/w variables.

 Non directional hypothesis: predicts the existence of

relationship not their direction.
 Null or statistical hypothesis: There is no relationship b/w the
IV & DV
The hypothesis of no difference or no effect in the
population is called as Null hypothesis.
 Fix Null Hypothesis: there is no difference / relation between
the groups
 Fix Alternative Hypothesis: there is significant difference /
relation between the groups
 Level of Significance (α)
 Test of Significance (ex) Z =  x1 – x2  / SE (x1 – x2)
 If Z> Zα  if the computed value of test statistic is greater than
the critical or significance value then we say that it is significant
and the null hypothesis is rejected at level of significance α.
Parametric Tests : -
 Mean and SD as Parameters
 Large Sample Size
 Followed Normal Distribution
Nonparametric Tests: -
 Followed not normal distribution
 Small sample size
 Use median / ranks
 Calculated value > Table Value
 P<0.05
 95% Confidence Interval = {Χ ±2SE(Χ)}
should not include null value

 Reject H0 => Accept HA

 Similar to normal distribution with wide tails
 Assumes normality assumption

Steps: – t test: -
 Calculate the difference between 2 means.
 Calculate SE of diff of mean. This will give us a variation which
can occur purely by chance between the mean of random
samples from the same population
 Find out ‘t’ by t = x1 – x2 / SE (x1 – x2)
SE (x1 – x2):-
 Final out sum of sq for both the groups
 SE (x1 – x2) =  SD2 / n1 + SD2 / n2
 Calculate pooled d.f from formula n1+n2-2
Test statistic for an experimental comp 2 samples of
equal size:-
 t = x1 – x2 / S2(1/n1 + 1/n2)
 S2 = ∑ (x-x1)2 + (x-x2)2 / (n1-1)+(n2-1)
Two sample of unequal size:-
 t = x1 – x2 / S2(n1+n2 / n1n2) where S2 = ∑ (x-x1)2 + (x-x2)2
/ (n1-1)+(n2-1)
 The study was conducted to find out whether there is a significant
difference in the serum cholesterol of sedentary workers and hard
laborers. A representative sample of 150 sedentary workers had an
average mean serum cholesterol of 230 mg% with SD of 15
mg% another representative sample of 100 hard workers had
mean serum cholesterol level of 210 mg% with SD of 20
mg%. On the basis of the above study can you say that serum
cholesterol in both the groups is significantly different?
 Group 1  n 150, Mean 230 , SD 15
 Group 2  n 100, Mean 210 , SD 20

SE (X1 – X2):-

 (15*15/150)+(20*20/100)
 =5.5 = 2.34
 t = (230-210) / 2.34
 = 20 / 2.34 = 8.54
 tcal = 8.54 > ttab = 1.96

 Significant.
 Calculate the difference in each set of paired observation.
 Find out mean difference (d)
 Find out S.E (d)
 t = d / SE(d)
 Calculate df = (n-1)
 Refer to a table ‘t’ & see the probability of calculation ‘t’
corresponding to the degrees of freedom t = d / ( S2 / n)
Is there any significant difference exits?
Pre test: 6 8 8 6 5 9 6 7 6 6 4 8
Post test: 8 8 10 7 6 10 9 8 5 7 4 6
Solution: -
 Diff (di): 2 0 2 1 1 1 3 1 -1 1 0 -2
 d = ∑di / n = 0.75 s2 = 1.84
 tcal = 0.75/ (1.84 / 12)
 tcal= 1.91< ttab = 2.2
 Not significant
 One way ANOVA (Analysis of variance)
 Two way ANOVA
 Multivariate Analysis
 Discriminant Analysis
 Principle Component Analysis
 Factor Analysis
 Cluster Analysis
 To find out if there is any correlation between the two
variables under study.
 Karl pearson a British biometrician developed a formula
for correlation.
 Correlation coefficient between two random variables X
and Y usually denoted by r(X,Y) = Cov (X,Y) / xy
 X – Independent Variable, Y – Dependent Variable
Assumptions: -
 The variables x and y under study are linearly related.
 Follow normal distribution
 Not independent of each other but are related in a casual
fashion.
 Range -1 to 1 usually represented as -100% (perfect negative
correlation) to 100% (perfect positive correlation)
 The linear Regression Model: Y = a + b X ,
 Where Y is Dependent Variable, X is Independent
variable
 b is Slope, b = cov (x,y) / var(x)
 a= Y–b*X
 a = intercept for which X = 0
 the slope b will explain that for each unit change in x, y
increase.
 U test : - test difference in ranks score of two different groups
 Wilcoxon Signed rank test: - within difference in ranks of two
related groups
 Median Test:- To test difference b/w medians of two different
groups
 Kruscal Wallis Test: - Difference in ranks of three or more
independent groups
 Chi square test:- find difference b/w 2 categorical data
 McNemar chi sq: - within difference in proportion for paired
2x2
Pearson Chi square test:-
 Qualitative variable with 2 or more categories
 The no of observation in each cell of the table must be known
 Frequencies in the different categories should be mutually
exclusive and exhaustive
 2 = ∑(Oi – Ei)2 / Ei
 Where Oi = observed value
 E i = R i * Cj / N
 Considering a data where we want to find association
between prenatal care received by mother and survival
status of infants at one month. Dead Alive

Less Care 20 373

 Calculation: More
 O1 = a = 20; O2 = b = 373; 6 316
Care
 O3 = c = 6; O4 = d = 316
 R1 = 393 R2 = 322C1 = 26 C2 = 689N = 715
 E1 = 14.3 E2 = 378.7 E3 = 11.7 E4 = 310.2
 2 calculated value = 5.24 > 2 table value = 3.84
 It is used to compare the sample proportion with population
or two different proportions. Usually, denoted as Z test when
N is large.
 Single Proportion test:-
 Z = p-P / (PQ/N)  Z (n-1)df, 5%

 Where Q = 1 – P; P = Prevalence or Proportion

 Ho: P1 = P2
 HA: P1 = P2
 Z = p1 – p2/ (p1q1 / n1 + p2q2/n2)  Z (n1+n2-1) df,

5%

 Where q1 = 1 – p1; q2 = 1-p2,

 n1 & n2 are sample size of group 1 & group 2
 A study was under taken to compare the proportions of
malnourished school children in urban and rural areas. Out of
150 urban school children 30 were found malnourished,
whereas 50 children in rural areas in schools were found to
have malnutrition out of total 200. On the basis of above
data, can you say that proportion of malnutrition among
school children of urban and rural areas is same?
 n1 = 150
 n2 = 200
 p1 = 30/150 = 0.2 q1 = 0.8
 p2 = 50/200 = 0.25 q2 = 0.75
 SE (p1 – p2) =  (0.2*0.8/150+0.25*0.75/200)
 = 0.04
 p1 – p2 = 0.05
 Z = 0.05/0.04 = 1.12

Z cal = 1.12 < Z tab = 1.96

Not significant
 Variable Research Parametric Nonparametric
Question
A vs B t test U test
Quantitative
Variables
Pre vs Post Paired t test Signed rank test
A vs B vs C ANOVA with Kruscal wallis test
multiple comparison

Relationship Pearson Spearman/Kendals

b/w X & Y Correlation correlation

Counts of Event Chi square test

Qualitative
Variables Proportion Z test