Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 4 views

    L5 BRM 010327

    Uploaded by

    Sarvesh Deodhar

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    L5 BRM 010327

    Uploaded by

    Sarvesh Deodhar
    4 views34 pages

    Original Title

    L5_BRM_010327

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    4 views34 pages

    L5 BRM 010327

    Uploaded by

    Sarvesh Deodhar

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 34

    Business Research Methods

    L5: HYPOTHESIS TESTING


    Null and Alternate Hypothesis

    Null hypothesis (Ho): Assumes no significant difference or no effect between sample and
    population parameter or between multiple samples
    Alternative hypothesis (Ha) suggests a difference or an effect.

    Considerations in Choosing Null Hypothesis:


     Null hypothesis is what researchers aim to reject
     Alternative hypothesis represents all other possibilities
     Null hypothesis should be specific and not approximate a certain value
     Chosen before drawing the sample to avoid bias.
    Level of Significance

     The significance level (α) is usually set at a certain


    percentage (e.g., 5%)
     It indicates the rejection region for the null hypothesis
    Tails in Hypothesis Testing

     Convert sample data into test


    statistic such as t-values, F-values
    etc.
     Test statistics follow a sampling
    distribution.
     Probability distribution plot gives
    the area under the curve as
    probability of data falling in the
    range
    Two tail Distribution

    Critical Region:
    • Significance level for rejecting the
    null hypothesis
    • When significance level is 5%, the
    critical region in t- distribution with
    two tail will have 2.5% rejection
    region in each tail
    • In this case if p value is less than 5%,
    t value will fall in the critical region
    so reject null hypothesis

    Tests both positive and negative effects


    One Tail Distribution
     They are one-directional
     Determine whether critical region is in left or right tail

    Null: The effect is less than or equal to zero. Null: The effect is greater than or equal to zero.
    Alternative: The effect is greater than zero. Alternative: The effect is less than zero.
    Errors in Hypothesis Testing

    Type I and II error:


    • Type I error represents the probability of rejecting Ho when it's true.
    • Type II error represents the probability of accepting Ho when it’s false
    Example 1 :

     An automotive manufacturer claims that its car provides an


    average mileage of 22 kmpl. Set up a hypothesis test to
    check this claim?
    Example 1 Solution:

     Ho: Average car mileage is not equal to 22 kmpl or x = 22


    kmpl
     Ha: Average car mileage is equal to 22 kmpl or x ≠ 22 kmpl
    Example 2:

     The automotive manufacturer now decides that it is only


    interested whether the average mileage is less than 22 Kmpl.
    What changes would you make from Example 1?
    Example 2 Solution:

     Ho: Average car mileage is equal to or greater than 22 kmpl


    or x ≥ 22 kmpl
     Ha: Average car mileage less than 22 kmpl or x < 22 kmpl
    Example 3:

     The automotive manufacturer now decides that it is only interested


    whether the average mileage is greater than 22 Kmpl. What changes
    would you make from Example 1?
    Example 3 Solution:

     Ho: Average car mileage is less than or equal to 22 kmpl or x ≤ 22 kmpl


     Ha: Average car mileage greater than 22 kmpl or x > 22 kmpl
    Steps in Hypothesis Testing

     Specify null and alternative hypothesis


     Select significance level
     Select appropriate test
     Find critical value
     Result and conclusion
    Business Research
    Methods
    ONE SAMPLE Z AND T-TEST
    One sample t-test

     Two tailed:  One tailed:


     Ho: No effect or difference  Right tailed:
     H1: There is significant effect  Ho:The effect is less than or equal
    to zero.
     H1: The effect is greater than zero.
     Left tailed:
     Ho:
    The effect is greater than or
    equal to zero
     H1: The effect is less than zero
    Assumptions:

     Our variable of interest is continuous


     The sample units are randomly extracted from the population
     Our variable is normally distributed
     There are no significant outliers in our data
    t-value calculation using mean

     X is sample mean
     µ is population mean
     s is sample standard deviation
     n is number of sample
     df or degree of freedom is n-1
     Find tcrit (critical t value) from t-distribution table
     If t value is greater than tcrit then reject null hypothesis or else fail to reject
    Example:

    Sample Sugar Content


     Imagine you have collected a (grams)
    random sample of 10 honey bottles of 1 20
    100 ml qty from a number of different 2 22
    stores and measured their sugar 3 26
    content in laboratory. The labels on 4 18
    the honey bottle claim that bottle 5 24
    contains 20 grams of sugar. At 95% 6 14
    confidence level, is the claim valid? 7 12
    8 14
    9 10
    10 30
    Steps:

     Develop Hypothesis
     Find t value
     Calculations for t-test:
     Identify tcrit using df and
     Calculate mean: x = significance level from t-distribution
    (x1+x2+x3+….xn)/n table

     Calculate variance :  Compare tcrit with t value

     Calculate sample standard  Reject or fail to reject null hypothesis


    deviation (S) as square root of
    variance
    Solution:

     Hypothesis:
     Ho: Average sugar content is equal to 20 gm (x=20)
     H1: Average sugar content is not equal to 20 gm (x≠20)

     T-test to measure the claim


    Solution:

    Calculations for t-value


    Sample Xi Xi-X (Xi-X)^2
     X= (20+22+28+18+….+30)/10 = 190/10=19
     Variance (S^2) = 386/(10-1)= 42.88 1 20 =20-19=1 1
     Standard deviation (S) = √42.88=6.54 2 22 =22-19=3 9
    3 26 =26-19=7 49
     t = (X-µ)/(s/ √n) 4 18 =18-19=-1 1
     t = (19-20)/(6.54/ √10) 5 24 =24-19=5 25
     t = -0.48 6 14 =14-19=-5 25
    7 12 =12-19=-7 49
    df = 10-1=9 and alpha=0.05
    8 14 =14-19=-5 25
    tcrit = 2.262
    9 10 =10-19=-9 81
    10 30 =30-19=11 121
    As t < tcrit, retain the null hypothesis
    Sum 190 386
    t-table
    Using SPSS
    Z-value calculation using mean

     When sample size is more than 30 go for z test


     X is sample mean
     µho is population mean
     σ is population standard deviation
     n is number of sample
     df or degree of freedom is n-1
     Find zcrit (critical z value) from z-distribution table
     If z value is greater than zcrit then reject null hypothesis or else fail to reject
    Example:

     A sample of 400 male students is found to have a mean height


    67.47 inches. Can it be reasonably regarded as a sample from
    a large population with mean height 67.39 inches and
    standard deviation 1.30 inches? Test at 5% level of
    significance. (Zcrit for 5% significance level is 1.96)
    Steps:

     Develop Hypothesis

     Calculations for z-test:


     Calculate sample mean (if not given): x = (x1+x2+x3+….xn)/n
     Find z value
     Compare zcrit with z value

     Reject or fail to reject null hypothesis


    Solution:

     Hypothesis:
     Ho: Average height of sample is equal to 67.47 (x=67.47)
     H1: Average height of sample is not equal to 67.47 (x≠67.47)

     z-test to measure the claim


    Solution:

    Calculations for z-value  z= (X-µ)/(σ / √n)


     n =400  z = (67.39-67.47)/(1.3/ √400)
     X = 67.47  z = - 0.08/0.065 = -1.231
     µ = 67.39
     σ = 1.3 Zcrit for 95% confidence or 5%
    significance level is 1.96

    As z< zcrit, retain the null


    hypothesis
    z value using population proportion

     is the sample proportion


     p is population proportion
     q is 1-p
     n is sample size
    Example:

     A sample survey indicates that out of 3232 births, 1705 were


    boys and the rest were girls. Do these figures confirm the
    hypothesis that the gender ratio is 50 : 50? Test at 5 per cent
    level of significance.
    Steps:

     Develop Hypothesis

     Calculations for z-test:


     Calculate sample proportion
     Find z value
     Compare zcrit with z values

     Reject or fail to reject null hypothesis


    Solution:

     Hypothesis:
     Ho: Sample gender ratio is equal to 0.5 (p̂ = 0.5)
     H1: Sample gender ratio is not equal to 0.5 (p̂ ≠ 0.5)

     z-test to measure the claim


    Solution:

    Calculations for z-value  Z= (0.0275)/ √(0.000077)


     n = 3232  z = 0.0275/0.0087 = 3.12
     p̂ = 1705/3232 = 0.5275
     p = 0.5 Zcrit for 95% confidence or 5%
     q = 1-p = 1-0.5=0.5 significance level is 1.96

     z= (p̂ - p)/( √(p.q/n)) As z>zcrit, reject the null


    hypothesis
     z = (0.5275-0.5)/(√(0.5*0.5/3232)

    You might also like