Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
SlideShare a Scribd company logo

Formulating hypotheses

•Download as PPTX, PDF•
•
Aniket Verma
Aniket Verma

The document discusses key concepts related to formulating and testing hypotheses, including: - Null and alternative hypotheses, which are mutually exclusive statements tested through sample analysis. - Type I and Type II errors that can occur when making decisions to accept or reject the null hypothesis. - The level of significance, critical region, and test statistics used to determine whether to reject the null hypothesis. - The differences between one-tailed and two-tailed tests, parametric vs. non-parametric tests, and one-sample vs. two-sample tests.

Related slideshows

Research Design
Research Design Research Design
Research Design
 
Testing of hypotheses
Testing of hypothesesTesting of hypotheses
Testing of hypotheses
 
Sampling - Stratified vs Cluster
Sampling - Stratified vs ClusterSampling - Stratified vs Cluster
Sampling - Stratified vs Cluster
 
1 of 53
Formulating HypothesesParametric TestsBusiness Research MethodologyMBA : 2nd Semester
Formulating HypothesesBegins with an assumption called Hypothesis.HypothesisClaim (assumption) about a population parameter.An unproven proposition or supposition that tentatively explains certain facts or phenomena.A proposition that is empirically testable.
What is Hypothesis testingA set of logical and statistical guidelines used to make decisions from sample statistics to population characteristics.
Hypotheses TestingThe intent of hypothesis testing is to formally examine two opposing conjectures (hypotheses), H0 and HA.
These two hypotheses are mutually exclusive and exhaustive.
Sample information is collected and analysed.Basic Concepts in Hypotheses Testing Null Hypotheses &Alternate HypothesesLevel of SignificanceCritical RegionDecision Rule(Test of Hypothesis)Type I & Type II ErrorsPower of TestTwo Tailed & One Tailed TestsOne Sample & Two Sample TestsParametric & Non Parametric Tests
Null HypothesisSpecific statement about a population parameter made for the purposes of argument.States the assumption to be tested, is a status quo.Is always about a population parameter, not about a sample statistic.
Alternate HypothesisRepresents all other possible parameter values except that stated in the null hypothesis.Challenges the status quo.Hypothesis that is believed (or needs to be supported) by the researcher –a research hypothesis.
Level of SignificanceThe critical probability in choosing between the null & alternative hypotheses.The probability of making a Type I error.The higher the significance level, the higher the probability of rejecting a null hypothesis when its true.Risk that a researcher is willing to take of rejecting the null hypotheses when it happens to be true.Confidence level: A percentage or decimal value that tells how confident a researcher can be about being correct.
Critical RegionThe rejection region.If the value of mean falls within this region, the null hypothesis is rejected.Critical value The value of a test statistic beyond which the null hypothesis can be rejected.
Decision Rule(Test of Hypothesis)The rule according to which we accept or reject null hypothesis.
Type I Error & Type II Error A Type I error is the mistake of rejecting the null hypothesis when it is true.α- Type I error.A Type II error is the mistake of failing to reject the null hypothesis when it is false.β- Type II error.
Type I Error & Type II Error Probability of Type I error is determined in advance.Level of significance of testing the hypothesis.
Type I Error & Type II Error Accept H0 Reject H0Correct Decision Type I ErrorType II Error Correct Decision Ho (True)Ho (False)
Type I Error & Type II Error Suppose there is a test for a particular disease. If the disease really exists and is diagnosed early, it can be successfully treated
If it is notdiagnosed and treated, the person will become severely disabled
If a person is erroneously diagnosed as having the disease and treated, no physical damage is doneType I Error & Type II Error treated but not harmed by the treatmentirreparable damage would be done
Power of TestThe ability of a test to reject a false null hypothesis.The probability of supporting an alternative hypothesis that is true.Power = 1- βHigh value of 1- β(near 1) means test is working fine, it is rejecting a null hypothesis when it is false.
One Tailed & Two Tailed TestTwo-Tailed TestsIf the null hypothesis is rejected for values of the test statistic falling into either tail of its sampling distribution.A deviation in either direction would reject the null hypothesisNormally α is divided into α/2 on one side and α/2 on the other.
One Tailed & Two Tailed Test
One Tailed & Two Tailed TestOne-Tailed TestsOnly used when the other tail is nonsensical.If null hypothesis is rejected only for values of the test statistic falling into one specified tail of its sampling distribution.
One Tailed & Two Tailed Test
One Tailed & Two Tailed TestA manufacturer of a light bulb wants to produce bulbs with a mean life of 1000 hours. If the lifetime is shorter, he will lose customers to the competitors; if the lifetime is longer, he will have a very high production cost because the filaments will be very thick. Determine the type of test.The wholesaler buys bulbs in large lots & does not want to accept bulbs unless their mean life is at least 1000 hours. Determine the type of test.
One Tailed & Two Tailed TestA highway safety engineer, decides to test the load bearing capacity of a bridge that is 20 yrs old. Considerable data is available from similar tests on the same type of bridge. Which type of test is appropriate? If the maximum load bearing capacity of this bridge must be 10 tons, what are the null & alternative hypotheses?
One Sample & Two Sample TestsOne Sample Test When we want to draw inferences about the population on the basis of given sample.Two Sample Test When we want to compare and draw inferences about 2 populations on the basis of given samples.
Independent & Paired SamplesIndependent SamplesDrawn randomly from different populations.Paired SamplesWhen the data for the two samples relate to the same group of respondents.
Types of HypothesesResearch hypotheses.Logical hypotheses.Null hypothesis (Ho).Alternative hypothesis (Ha).Statistical hypotheses.
Research Hypotheses.Statement in words as to what the investigator expects to find.Example. Students who drink caffeine will be able to memorise information faster than students who do not drink caffeine.
Logical HypothesesStated in terms of null & alternate hypotheses.Null Hypothesis (Ho). Students who drink caffeine will be not be able to memorise information faster than students who do not drink caffeine.Alternative Hypothesis (Ha). Students who drink caffeine will be able to memorise information faster than students who do not drink caffeine.
Statistical HypothesesStatement in statistical terms as to what would be found if the research hypothesis is true.A sales manager has asked her salespeople to observe a limit on travelling expenses. The manager hope to keep expenses to an average of $ 100 per salesman per day. What will be H0 and Ha?
Steps in Hypotheses TestingFormulation of the null and alternate hypothesisDefinition of a test statisticDetermination of the distribution of the test statisticDefinition of critical region of the test statisticTesting whether the calculated value of the test statistic falls within the acceptance region.
1: Formulation of H0The Null hypothesis assumes a certain specific value for the unknown population parameter.Defined as an inequality – greater than or less than.For example, if the mean of a population is considered, thenH0: μ ≤ μ0H0: μ = μ0H0: μ ≥ μ0
2: Formulation of HaThe alternate hypothesis assigns the values to the population parameter that is not contained in the null hypothesis.For example,Ha: μ > μ0Ha: μ ≠ μ0Ha: μ < μ0The null hypothesis is accepted or rejected on the basis of the information provided by the sample.
3: Definition of a Test StatisticA test statistic must be defined to test the validity of the hypothesis.The test statistic is computed from sample information.A number calculated to represent the match between a set of data and the expectation under the null hypothesisA test that uses the z-score as a test statistic is called a z-test.
4: Determination of the distribution of the test statisticThe probability distribution of the test statistic depends on the null hypothesis assumed, the parameter to be tested, and the sample size. Commonly used ones are the Normal, Student’s “t”, Chi-square and F-distributions.
5: Definition of the critical region for the test statisticThe set of values of the test statistic that leads to the rejection of H0 in favour of Ha is called the rejection region or critical region. Depends upon whether the testing is one-sided or two-sided.
6: Decision ruleA decision rule is used to accept or reject the null hypothesis.P- value P < α Reject the null hypothesis Statistically significantTest statistic Test statistic (calculated value) < Table value of α Accept H0 Statistically insignificant
7: OutcomeThe acceptance or rejection of the hypothesis will lead to the following possible outcomes:To accept H0 when H0 is true- Correct decisionTo reject H0 when H0 is false- Correct decisionTo reject H0 when H0 is true- Type I errorTo accept H0 when H0 is false- Type II error
8: Error probabilitiesThe inferences made on the basis of the sample information would always have some degree of error. α = Type I error = Rejecting H0 when H0 is true β = Type II error = Accepting H0 when H0 is falseRejecting a null hypothesis (Type I error) is more serious than accepting it when it is false. Therefore, the error probability α is referred to as the significance level.
Parametric & Non Parametric TestsParametric TestStatistical procedures that use interval or ratio scaled data and assume populations or sampling distributions with normal distributions.Non Parametric TestStatistical procedures that use nominal or ordinal scaled data and make no assumptions about the distribution of the population.
Parametric Testsz-testt-testF-testChi square test
Conditions for using Tests
t-TestUsed when sample size is ≤ 30.Given by W.S.Gosset (pen name Student)Also called Student’s t distribution.Based on t distribution.The relevant test statistic is t.
t-TestConditionsSample should be small.Population standard deviation must be unknown.AssumptionNormal or approximately normal population.
t DistributionCharacteristicsFlatter than normal distribution.Lower at mean, higher at tails than normal distribution.As degrees of freedom increase, t-distribution approaches the standard normal distribution (df=8 or more)Degrees of Freedom: No. of observations minus the no. of constraints or assumptions needed to calculate a statistical term.
t-TestConfidence IntervalDegrees of Freedom: n-1
t-TestThe specimen of copper wires drawn from a large lot to have the following breaking strength (in Kg weight) 578, 572, 570, 568, 572, 578, 570, 572, 596, 544Test whether the mean breaking strength of the lot may be taken to be 578 kg, at 5% significance level.
t-TestGiven a sample mean of 83, a sample standard deviation of 12.5, & a sample size of 22, test the hypothesis that the value of population mean is 70 against the alternative that is more than 70. Use 0.025 significance level.
z -TestBased on the normal distribution.Mostly used for judging the significance level of mean.The relevant test statistic is z. The value of z is calculated & compared with its probable value.If calculated value is less than table value- accept H0
z -TestThe mean of a certain production process is known to be 50 with a standard deviation of 2.5. The production manager may welcome any change in mean value towards higher side but would like to safeguard against decreasing values of mean. He takes a sample of 12 items that gives a mean value of 48.5. What inference should the manager take for the production process on the basis of sample results? Use 5% significance value.z -TestA 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 & standard deviation 1.30 inches? Test at 5% level of significance.
z -TestHinton press hypothesizes that the average life of its largest web press is 14,500 hours. They know that the standard deviation of press life is 2100 hours. From a sample of 36 presses , the company finds a sample mean of 13,000 hours. At a 0.01 significance level, should the company conclude that the average life of the presses is less than the hypothesized 14,500 hours?
F-TestBased on F-DistributionUsed to compare variance of 2 independent samples.Relevant Test statistic is F.Larger the F value, greater the possibility of having statistically significant results.
F-TestCharacteristicsFamily of distributions .Has 2 degrees of freedom.

More Related Content

Formulating hypotheses

  • 1. Formulating HypothesesParametric TestsBusiness Research MethodologyMBA : 2nd Semester
  • 2. Formulating HypothesesBegins with an assumption called Hypothesis.HypothesisClaim (assumption) about a population parameter.An unproven proposition or supposition that tentatively explains certain facts or phenomena.A proposition that is empirically testable.
  • 3. What is Hypothesis testingA set of logical and statistical guidelines used to make decisions from sample statistics to population characteristics.
  • 4. Hypotheses TestingThe intent of hypothesis testing is to formally examine two opposing conjectures (hypotheses), H0 and HA.
  • 5. These two hypotheses are mutually exclusive and exhaustive.
  • 6. Sample information is collected and analysed.Basic Concepts in Hypotheses Testing Null Hypotheses &Alternate HypothesesLevel of SignificanceCritical RegionDecision Rule(Test of Hypothesis)Type I & Type II ErrorsPower of TestTwo Tailed & One Tailed TestsOne Sample & Two Sample TestsParametric & Non Parametric Tests
  • 7. Null HypothesisSpecific statement about a population parameter made for the purposes of argument.States the assumption to be tested, is a status quo.Is always about a population parameter, not about a sample statistic.
  • 8. Alternate HypothesisRepresents all other possible parameter values except that stated in the null hypothesis.Challenges the status quo.Hypothesis that is believed (or needs to be supported) by the researcher –a research hypothesis.
  • 9. Level of SignificanceThe critical probability in choosing between the null & alternative hypotheses.The probability of making a Type I error.The higher the significance level, the higher the probability of rejecting a null hypothesis when its true.Risk that a researcher is willing to take of rejecting the null hypotheses when it happens to be true.Confidence level: A percentage or decimal value that tells how confident a researcher can be about being correct.
  • 10. Critical RegionThe rejection region.If the value of mean falls within this region, the null hypothesis is rejected.Critical value The value of a test statistic beyond which the null hypothesis can be rejected.
  • 11. Decision Rule(Test of Hypothesis)The rule according to which we accept or reject null hypothesis.
  • 12. Type I Error & Type II Error A Type I error is the mistake of rejecting the null hypothesis when it is true.α- Type I error.A Type II error is the mistake of failing to reject the null hypothesis when it is false.β- Type II error.
  • 13. Type I Error & Type II Error Probability of Type I error is determined in advance.Level of significance of testing the hypothesis.
  • 14. Type I Error & Type II Error Accept H0 Reject H0Correct Decision Type I ErrorType II Error Correct Decision Ho (True)Ho (False)
  • 15. Type I Error & Type II Error Suppose there is a test for a particular disease. If the disease really exists and is diagnosed early, it can be successfully treated
  • 16. If it is notdiagnosed and treated, the person will become severely disabled
  • 17. If a person is erroneously diagnosed as having the disease and treated, no physical damage is doneType I Error & Type II Error treated but not harmed by the treatmentirreparable damage would be done
  • 18. Power of TestThe ability of a test to reject a false null hypothesis.The probability of supporting an alternative hypothesis that is true.Power = 1- βHigh value of 1- β(near 1) means test is working fine, it is rejecting a null hypothesis when it is false.
  • 19. One Tailed & Two Tailed TestTwo-Tailed TestsIf the null hypothesis is rejected for values of the test statistic falling into either tail of its sampling distribution.A deviation in either direction would reject the null hypothesisNormally α is divided into α/2 on one side and α/2 on the other.
  • 20. One Tailed & Two Tailed Test
  • 21. One Tailed & Two Tailed TestOne-Tailed TestsOnly used when the other tail is nonsensical.If null hypothesis is rejected only for values of the test statistic falling into one specified tail of its sampling distribution.
  • 22. One Tailed & Two Tailed Test
  • 23. One Tailed & Two Tailed TestA manufacturer of a light bulb wants to produce bulbs with a mean life of 1000 hours. If the lifetime is shorter, he will lose customers to the competitors; if the lifetime is longer, he will have a very high production cost because the filaments will be very thick. Determine the type of test.The wholesaler buys bulbs in large lots & does not want to accept bulbs unless their mean life is at least 1000 hours. Determine the type of test.
  • 24. One Tailed & Two Tailed TestA highway safety engineer, decides to test the load bearing capacity of a bridge that is 20 yrs old. Considerable data is available from similar tests on the same type of bridge. Which type of test is appropriate? If the maximum load bearing capacity of this bridge must be 10 tons, what are the null & alternative hypotheses?
  • 25. One Sample & Two Sample TestsOne Sample Test When we want to draw inferences about the population on the basis of given sample.Two Sample Test When we want to compare and draw inferences about 2 populations on the basis of given samples.
  • 26. Independent & Paired SamplesIndependent SamplesDrawn randomly from different populations.Paired SamplesWhen the data for the two samples relate to the same group of respondents.
  • 27. Types of HypothesesResearch hypotheses.Logical hypotheses.Null hypothesis (Ho).Alternative hypothesis (Ha).Statistical hypotheses.
  • 28. Research Hypotheses.Statement in words as to what the investigator expects to find.Example. Students who drink caffeine will be able to memorise information faster than students who do not drink caffeine.
  • 29. Logical HypothesesStated in terms of null & alternate hypotheses.Null Hypothesis (Ho). Students who drink caffeine will be not be able to memorise information faster than students who do not drink caffeine.Alternative Hypothesis (Ha). Students who drink caffeine will be able to memorise information faster than students who do not drink caffeine.
  • 30. Statistical HypothesesStatement in statistical terms as to what would be found if the research hypothesis is true.A sales manager has asked her salespeople to observe a limit on travelling expenses. The manager hope to keep expenses to an average of $ 100 per salesman per day. What will be H0 and Ha?
  • 31. Steps in Hypotheses TestingFormulation of the null and alternate hypothesisDefinition of a test statisticDetermination of the distribution of the test statisticDefinition of critical region of the test statisticTesting whether the calculated value of the test statistic falls within the acceptance region.
  • 32. 1: Formulation of H0The Null hypothesis assumes a certain specific value for the unknown population parameter.Defined as an inequality – greater than or less than.For example, if the mean of a population is considered, thenH0: μ ≤ μ0H0: μ = μ0H0: μ ≥ μ0
  • 33. 2: Formulation of HaThe alternate hypothesis assigns the values to the population parameter that is not contained in the null hypothesis.For example,Ha: μ > μ0Ha: μ ≠ μ0Ha: μ < μ0The null hypothesis is accepted or rejected on the basis of the information provided by the sample.
  • 34. 3: Definition of a Test StatisticA test statistic must be defined to test the validity of the hypothesis.The test statistic is computed from sample information.A number calculated to represent the match between a set of data and the expectation under the null hypothesisA test that uses the z-score as a test statistic is called a z-test.
  • 35. 4: Determination of the distribution of the test statisticThe probability distribution of the test statistic depends on the null hypothesis assumed, the parameter to be tested, and the sample size. Commonly used ones are the Normal, Student’s “t”, Chi-square and F-distributions.
  • 36. 5: Definition of the critical region for the test statisticThe set of values of the test statistic that leads to the rejection of H0 in favour of Ha is called the rejection region or critical region. Depends upon whether the testing is one-sided or two-sided.
  • 37. 6: Decision ruleA decision rule is used to accept or reject the null hypothesis.P- value P < α Reject the null hypothesis Statistically significantTest statistic Test statistic (calculated value) < Table value of α Accept H0 Statistically insignificant
  • 38. 7: OutcomeThe acceptance or rejection of the hypothesis will lead to the following possible outcomes:To accept H0 when H0 is true- Correct decisionTo reject H0 when H0 is false- Correct decisionTo reject H0 when H0 is true- Type I errorTo accept H0 when H0 is false- Type II error
  • 39. 8: Error probabilitiesThe inferences made on the basis of the sample information would always have some degree of error. α = Type I error = Rejecting H0 when H0 is true β = Type II error = Accepting H0 when H0 is falseRejecting a null hypothesis (Type I error) is more serious than accepting it when it is false. Therefore, the error probability α is referred to as the significance level.
  • 40. Parametric & Non Parametric TestsParametric TestStatistical procedures that use interval or ratio scaled data and assume populations or sampling distributions with normal distributions.Non Parametric TestStatistical procedures that use nominal or ordinal scaled data and make no assumptions about the distribution of the population.
  • 43. t-TestUsed when sample size is ≤ 30.Given by W.S.Gosset (pen name Student)Also called Student’s t distribution.Based on t distribution.The relevant test statistic is t.
  • 44. t-TestConditionsSample should be small.Population standard deviation must be unknown.AssumptionNormal or approximately normal population.
  • 45. t DistributionCharacteristicsFlatter than normal distribution.Lower at mean, higher at tails than normal distribution.As degrees of freedom increase, t-distribution approaches the standard normal distribution (df=8 or more)Degrees of Freedom: No. of observations minus the no. of constraints or assumptions needed to calculate a statistical term.
  • 47. t-TestThe specimen of copper wires drawn from a large lot to have the following breaking strength (in Kg weight) 578, 572, 570, 568, 572, 578, 570, 572, 596, 544Test whether the mean breaking strength of the lot may be taken to be 578 kg, at 5% significance level.
  • 48. t-TestGiven a sample mean of 83, a sample standard deviation of 12.5, & a sample size of 22, test the hypothesis that the value of population mean is 70 against the alternative that is more than 70. Use 0.025 significance level.
  • 49. z -TestBased on the normal distribution.Mostly used for judging the significance level of mean.The relevant test statistic is z. The value of z is calculated & compared with its probable value.If calculated value is less than table value- accept H0
  • 50. z -TestThe mean of a certain production process is known to be 50 with a standard deviation of 2.5. The production manager may welcome any change in mean value towards higher side but would like to safeguard against decreasing values of mean. He takes a sample of 12 items that gives a mean value of 48.5. What inference should the manager take for the production process on the basis of sample results? Use 5% significance value.z -TestA 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 & standard deviation 1.30 inches? Test at 5% level of significance.
  • 51. z -TestHinton press hypothesizes that the average life of its largest web press is 14,500 hours. They know that the standard deviation of press life is 2100 hours. From a sample of 36 presses , the company finds a sample mean of 13,000 hours. At a 0.01 significance level, should the company conclude that the average life of the presses is less than the hypothesized 14,500 hours?
  • 52. F-TestBased on F-DistributionUsed to compare variance of 2 independent samples.Relevant Test statistic is F.Larger the F value, greater the possibility of having statistically significant results.
  • 53. F-TestCharacteristicsFamily of distributions .Has 2 degrees of freedom.
  • 54. F-TestTwo random samples drawn from two normal populations areTest using variance ratio of 5% and 1% level of significance whether the two populations have the same variance.