Baa8a354 3962 4b14 9085 f579151c21b6

Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

QM353 Class Lecture Notes #7

Testing Hypothesis on two Population


Proportions and Tt2
To test hypotheses on two
population proportions and of some attributes of interest a hypothesis
testing process, consisting of five
steps is explained a.sfollows.
Step#l. (Formulation of Hypotheses
) . The null hypothesis (Ho) and the alternative hypothesis
two population.proportion and (HD on
maybe formulated as one of the following three possible
scenarios.
Scenario #1 n tail Scenario #2 0 tails Scenario #3 ht tail
:% Versa H: % < H :c = Verses H: R* H : ltS Verses H : It >
Step#2. (Test Statistic) : Z = {( PI —pa) —(1tl
- 1%)}/ o p
Where for the sample proportions: p, —x. In, and pa = /nz:
= ( (nt PI + n2P2)/(m+n2) I = [ (x: +x2) is a pooled proportion (weighted average of
proportions)

0 -= (IF + 1/nD I is a standarderror ofthe pooledproportion.

This test statistic is applicable when both the samples are large.
i.e. { not! 25 & n, (I-TCI) 25 } and
{n2Tt225 & na (I-lt2 ) 25 }. Generally, (It! - lt2)=0 unless
the difference is specified.
Step#3. The Acceptance and the Rejection Regions of Ho
To determine whether Ho, ata certain level of significance a, should be
acceptedor rejected the
Acceptance Region (AR) and the Rejection Region (RR) for the three possible
scenarios are msfollows.
Scenario ft tail Scenario#2 0 tails Scenario#3 t tail

RR:z<-zan AR:-Zanszszan b zan AR:zsza


Step #4 (Decision Making): Rule. Accept (or do not reject) Hoif the value of the test statistic Z lies
in the acceptance region (AR); otherwise reject Ho and accept HI.

Step#5 Identification of the type of possible decision making error


Type one error is committed when an attempt is made to reject a true null hypothesis Ho with
probability a and type two error is committed when an attempt is made to accept a false null
hypothesis Ho with probability ß.

Example #1. The managementof Nolege believes that its women employees are more satisfied with
Vthepresent work environment than the men employees. To
test its claim a researcher randomly selected
40 women employees and 60 men employees of Nolege and asked a question on satisfaction of work
environment. Out ofthese 100 sampled employees 32 women and 42 men employees expressed their
0.15
let us see whether
satisfaction with the present work environment. On basis of these
management's claim at 15% level of significance is true or not.
and p, =
Here , the proportionof satisfied sampled women and mep are : p, —xt/nt and the
—42/60=0.7repectively, Pooled proportion = ( (Xl+x2) = (32+42)/100=0.74
Standard error of pooled proportion o —4[p• (I-p- 4(0.74

thern.
Based on these samplai information let us formulate appropriate hypotheses and test
Step #1 (Hypotheses): He: verses Hi: (as pa—O.7)
Step Crest Statisdc): Z— (pi —pa) lop- — (0.8-0.7)/0.09= -1.11
—1.04
Step#3. Acceptance and the Rejection)Regions of Ho . At 15%a, the level of significance ZA
which gives us: AR: Z' 1.04( relevant to Ho) and RR: Z> 1.04 (relevant to HI).
It
Step (Decision).Hois rejected (as Z 1.11> m—1.04; i.e. Z lies in the RR) and HI is accepted.
with 85%
leads to conclusion at that the management's claim (HI: Tti> ) is due. "Iherefore,
confidence I agree with the management claim.
Step Type of possible decision making error. The researcher is vulnerableto commit type one
error with probabilitya—O.1Sif Ho tums out true rather than false.

Point Estimate of - ConfidenceInterval of -


A
I 'CI-xal - (Pi-pa) I - p,-P21 ± Qa

Example . Based on above data, where the sample proportions pu=0.8 and pr0.7.
i) Point Estimate of (Iti - Tt2)= (x, - (Pi-pa) = (0.8-0.7)=0.1
ii) 79.6% Confidence Interval to estimate is A= | 0.8-0.7 ± (0.8-0.7)±
(0.7(o.3Y601 1.274(0.0075) 0.1±0.11 (0.01, 0.21)

Note that the confidence interval of I - I may computed ± p-


-+0.1± 1.27(0.09) (0.0143, 0.2143) (0.01,0.21)
Exercise #7
uestions Answers
#1. Based on the above exampled data: #1.a) is rejected
a)Test against at 40 % level of and 7t2is accepted
siB1ificanceand identifr the type oferror that may b)Type one error with probability a=0.40
possibly be committed. c) (PI-PD
b)State type of possible error.
c)What is point estimate of (ct- lt2)? d) = (0.8-0.7)± 1.51 (0.09)
87% interval for (Iti- CD. 0.0359S (m-
#2. Consider the following responses (Y for yes, N #2. a) Here and
for no) to a question: " Do you have health insurance 15/30=0.5; = (24+15)/70-0.557
?" asked to 40 men and 30 women selected at p- = Mcp-
random from a certain localityof a town. = 4[0.557 =o.12

= 0.833
Women:
Ho:ltMS1twis rejected (as Z=O.833
Based on this sampled information: and He. ltM> is accepted.
a)Can we conclude at 20.9% level of significance
whether the proportion of men health insurers is more b)Type one error with probability a=0.209 if Ho
than the proportion ofwomen insurers? tums out true rather than false.
b)State type of possible decision making error along
with its probability.

You might also like