SlideShare a Scribd company logo
1 of 84
Chapter 10 Estimating with Confidence
10.1 	Confidence Intervals:		The Basics
Definitions Statistical Inference The method of drawing conclusions about a population based on a sample The last Major Topic for this last statistics course The mindset: we are looking at data that comes from a random sample (or an experiment) and inferring characteristics of the population
The Basic Plan 	In the absence of other data, the sample estimate is the estimate of the parameter We would like an interval estimation State the parameter that is being estimated Check to see if the data can be Normalized Compute the interval using area under the Normal curve Write  a nice conclusion
What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05)
What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) Confidence Level C
What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) Confidence Interval (CI)
What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) Lower Bound Upper Bound
What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) EX2We are 90% confident that the proportion that support this law is 0.74 ± 0.08
What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) EX2We are 90% confident that the proportion that support this law is 0.74 ± 0.08 Confidence Level C
What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) EX2We are 90% confident that the proportion that support this law is 0.74 ± 0.08 Confidence Interval (CI)
What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) EX2We are 90% confident that the proportion that support this law is 0.74 ± 0.08 Margin of Error (ME) Point Estimate
Confidence Level The value of the parameter is fixed before we start our sampling Either the parameter is ‘in’ our interval or it’s not. There is no probability involved here “There is a 95% probability the parameter is in the interval”
Confidence Level The value of the parameter is fixed before we start our sampling Either the parameter is ‘in’ our interval or it’s not. There is no probability involved here “There is a 95% probability the parameter is in the interval” F   A   I   L
Confidence Level Remember that our sample is just one of many samples that could have been taken If good sampling technique is used, 95% (for example) of the samples would contain the parameter We just don’t know if our sample is part of the 95% or the 5%
Confidence Level Interpretation of “90% Confidence Level” “We are 90% confident that our CI contains the value of the parameter” “90% of CI’s computed with this technique will contain the parameter” “There is a 90% probability our CI contains the parameter” WRONG INTERPRETATION
Confidence Level Interpretation of “90% Confidence Level” “We are 90% confident that our CI contains the value of the parameter” “90% of CI’s computed with this technique will contain the parameter” “There is a 90% probability our CI contains the parameter” WRONG INTERPRETATION
PANIC When constructing Confidence Intervals for the AP test, there is a definite checklist that readers look for Use the acronym P.A.N.I.C. to help memorize the steps to construction a confidence interval P = state the Parameter A = check that Assumptions are satisfied N = state is the Name of the interval I = compute numerical values of the Interval C = write a Conclusion for your findings
PANIC for Means We are going to start with Means, we’ll save proportions for later Parameterdefine what  measuresdefine what x-bar measures “ = mean length of all Great White Sharks‘ flukes x-bar = mean length the Great White Shark flukes in our sample of 35 sharks”
Assumptions for CI’s of Mean SRS- the data must have come from an SRS Independence- the population size must be more than 10 times the sampleN> 10n (Independence)
Assumptions for CI’s of Mean SRS- the data must have come from an SRS Independence- the population size must be more than 10 times the sampleN> 10n (Independence) This is a condition that must be met since we are sampling without replacement and our formula for std dev needs to hold true
Assumptions for CI’s of Mean SRS- the data must have come from an SRS Independence- the population size must be more than 10 times the sampleN> 10n (Independence) The sampling Distribution is approximately Normal (Normality)
More on Normality We are looking for justification that the sampling distribution is Normal If the population is Normal, then the samp dist is also Normal If n> 30 and the sample data is not given, state “The Central Limit Theorem guarantees that the sampling distribution is Normal” For smaller samples, check the following:(1) the Histogram is single peak, symmetric with 	no outliers(2) the Normal Probability Plot is approx Linear
Name of the Interval Currently, the only Interval we will worry about is “z-interval for sample means” This will always be the name of the distribution used
Confidence Interval Computation
Confidence Interval Computation Margin of Error
Confidence Interval Computation Standard Error Always the std dev of the sampling distribution Critical Value -from the Normal (z) distribution
Critical Value The area between –z* and +z* = Confidence Level Because they are used frequently, there is a shorthand method in table C The last row gives different confidence levels Keep critical values to 3 decimal places  The row above the CL row has the z* In the AP test, they will call this z
A General z* curve
z* for a 80% CL
Computing a CI Let’s compute the 95% CI for a sample of 35, with a mean of 5.38 and a population std dev =0.74 (1) locate the critical value z* = 1.960 (2) Compute S.E. and M.E. SE = 0.74/(35) = 0.1251ME = (1.960)(0.1251) = 0.2452
Computing a CI State the CI = 5.38 ± 0.25 Interval estimate:(5.13, 5.63) Notice that the point estimate is the average of the upper and lower bounds
Conclusions We are 95% confident that the mean length of a great white shark fluke is 5.38 ± 0.25 ft. 	OR We are 95% confident that the mean length of all great white shark flukes is in the interval  (5.13 ft, 5.63 ft).
Calculators TI83/TI84
Calculators TI83/TI84 Your TI is very efficient for finding these intervals!  This doesn’t excuse you from the mathematics, of course. [stat] ->  “TESTS” -> “Zinterval” Inpt: “Stats” (if your data is in L1, you can use “Data” Enter values for , x-bar, n, and C-Level. Viola!
Calculators TI89 Run the “Stats/List Editor” APP [2nd] -> [F2] (F7) -> “Zinterval” Input Method = “Stats”Choose Data if all the observations are in a List Enter values for , x-bar, n, and C-Level. Viola!
Behavior of Margin of Error ME = z* (/n) In practice, we would like to minimize ME. (1) decrease z* This also means decrease our CL! (2) increase nthis is usually a trade off, since obtaining large samples could be more expensive/time consuming (3) decrease Not really an option.  is a known quantity.
Sample Size By using algebra, we can get a formula to compute minimum sample size for a given ME You should always round the sample size up in this calculation Note: ME is produced by sampling variability, it has nothing to do with “sloppy work”
10.2 Estimating a Population 	Mean
Student’s t-distribution The confidence intervals computed in the previous section assumed that we knew . It doesn’t seem likely that we would know  and not know the value of  When  is unknown, we can no longer use the Normal distribution
Student’s t-distribution The t-distribution is to be used when  is unknown. The t-distribution is very similar to the Normal distribution with a key difference The shape of the distribution changes based upon the sample size/degrees of freedom (df) Large samples have a tall peak and thinner tails Small samples have a small peak and thicker tails.
Student’s t-distribution Degrees of Freedom = n -1
Using the t-distribution The CI is found using PANIC The Assumptions are the same as a z-interval Since sample sizes tend to be small, you will most likely need to check the histogram (symmetric, no outliers) and Normal prob plot (approx linear) We cannot use the t-distribution when there are outliers! The Name of the interval is “1-sample t-interval for mean”
Using the t-distribution Interval is computed with df = n – 1 t* is from table C or your calculator s is the sample’s standard deviation Uses the sample std dev to approximate 
Upper Tail Area
Upper Tail Area This is the Upper Tail Area
Using the t-distribution
Using the t-distribution Using TI84 	[2nd] -> [vars](DIST) -> “invT” “invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“−invT(Upper Tail Area, df)”
Using the t-distribution Using TI84 	[2nd] -> [vars](DIST) -> “invT” “invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“−invT(Upper Tail Area, df)”
Using the t-distribution Using TI84 	[2nd] -> [vars](DIST) -> “invT” “invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“−invT(Upper Tail Area, df)” Don’t forget the negative!
Using the t-distribution
Using the t-distribution Using TI89 From home screen: [catalog] -> [F3](FlashApps) -> inv_t…TIStat “tistat.invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“-tistat.invT(Upper Tail Area, df)”
Using the t-distribution Using TI89 From home screen: [catalog] -> [F3](FlashApps) -> inv_t…TIStat “tistat.invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“-tistat.invT(Upper Tail Area, df)”
Using the t-distribution Using TI89 From home screen: [catalog] -> [F3](FlashApps) -> inv_t…TIStat “tistat.invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“-tistat.invT(Upper Tail Area, df)” Don’t forget the negative!
Using the t-distribution ALTERNATIVE TI89 titanium [APPS] -> “Stat/List Editor” -> [F5] (distrib) -> “Inverse” -> “Inverse  t…”  “Area: 1- Upper Tail” Upper Tail = (1 – CL)/2 “degrees of freedom, df: df” This takes longer to get to, but the menu “guides” you through
Table C Occasionally, you will need to fine the t* for a df that does not appear in the chart. When this happens, you are to use the nearest greatest  df in the same column This usually means “use the t* that is in the line above where the desired t* should be”
Example: 1 sample t-interval 	Problem 10.30 	The amount of Vitamin C (mg/100g) in CSB for a random sample of 8 are given as: 	26, 31, 23, 22, 11, 22, 14, 31 	Construct a 95% Confidence interval for the amount of vitamin C in CSB.
Example: 1 sample t-interval Parameter  = the mean amount of vitamin C in CSB produced at the factory x-bar = the mean amount of vitamin C in a sample n = 8
Example: 1 sample t-interval Assumptions SRS ‘Our problem states that we have a random sample’ Independence ‘10n = 80 < N; we can infer that more than 80 CSB is produced in the factory’
Example: 1 sample t-interval Assumptions Normality ‘The histogram is symmetric w/ no outliers’ ‘The Norm Prob Plot is approx linear’ ‘We have good evidence that our sampling distribution is approximately Normal’ x 10     15    20    25    30    35 Norm Prob Plot Histogram z
Example: 1 sample t-interval Name of Interval 	‘We will compute a 1-sample t-interval for a mean’
Example: 1 sample t-interval Name of Interval 	‘We will compute a 1-sample t-interval for a mean’ Interval Calculation
Example: 1 sample t-interval Name of Interval 	‘We will compute a 1-sample t-interval for a mean’ Interval Calculation Conclusion 	‘We are 95% confident that the mean amount of vitamin C in a unit of CSB produced in the factory is between 16.487 and 28.513 mg/100 g’
10.3 Estimating a population Proportion
Estimating a Proportion Like with means, we are going to estimate the population proportion based on the proportion in a sample x = # of positive responsesn = total number of responses We will again use the PANIC procedures Note: nothing is averaged.  We are not looking at the average proportion from many samples
Parameter Some typical parameters: ‘p = prop of people in CA who support the proposition 	p-hat = proportion of people in a sample n = 35 who support the proposition’ ‘p = prop of students at THS who ride the bus 	p-hat = propotion of students in  a sample (n = 14) from THS who ride the bus
Assumptions Simple Random Sample SRS must be either stated or inferred Independencebecause you are usually sampling without replacement:N> 10n Normalityn·p-hat > 10n·q-hat > 10
Name of the Interval “1-proportion z interval” Unlike for means, you will always use the Normal curve when dealing with proportions!
Interval Calculation z* is calculated as before Use table C 	OR “ − invNorm( (1 – C) /2 )”This calculation is similar to the calculations for the last section
Interval Calculation z* is calculated as before Use table C 	OR “ − invNorm( (1 – C) /2 )”This calculation is similar to the calculations for the last section Margin of Error (ME)
Interval Calculation z* is calculated as before Use table C 	OR “ − invNorm( (1 – C) /2 )”This calculation is similar to the calculations for the last section Standard Error(SE)
Conclusion Some Examples: We are 90% confident that the proportion of voters in CA who support the proposition is 0.34 ± 0.03 We are 95% confident that the proportion of students at THS who ride this bus is in the interval (0.39, 0.44)
Sample Size The relevant formula (from ME) for the sample size is: p* and q* are guessed values of the proportion If there is no previous data or study, use p* = q* = 0.5 (this will maximize the error and sample size) As before, you are to round the sample size up to the nearest integer
TI83/84
TI83/84 [stat] -> “TEST” -> “1-PropZInt” “x: number of successes” this can be computed with “p-hat x n” “n: number of people in sample” “C-Level : confidence level” “Calculate” and you are done You still need to fully write up “PANIC” procedures
TI 89
TI 89 [APPS] -> “Stat/List Editor” [2nd] -> [F2](F7) -> “1-PropZInt” “Successes x: # of successes” “n: number of people in sample” “C-Level : confidence level” “Calculate” and you are done You still need to fully write up “PANIC” procedures
Example 1-PropZInt 	The 2004 Gallup Youth Survey asked a random sample of 439 US teens aged 13 to 17 whether they though young people should wait to have sex until marriage.  246 sad “yes.”  Let’s construct a 95% confidence interval for the proportion of all teens who would say “Yes”
Parameter p = the proportion of all teens in the US aged 13-17 who would answer “Yes” to the survey p-hat = the proportion of teens in the survey of 439 who answered “Yes” to the survey
Assumptions SRS: We are told in the problem that the survey was random sample. Independence: there are than 10(439) = 4390 teens aged 13-17 in the US Normalityn x p-hat = 246, n x q-hat = 193The sampling distribution is approx. Normal (Note that this is just #successes and #failures)
Name of Interval We are constructing a “1 proportion Z interval”
Construction of Interval
Conclusion “We are 95% confident that the true proportion of 13-17 year olds who would answer “yes” when asked if young people should wait to have sex until they are married is between 0.514 and 0.606.”
Stats chapter 10

More Related Content

What's hot

Identifying Test Objective (Assessment of Learning) - CES report 011114
Identifying Test Objective (Assessment of Learning) - CES report 011114Identifying Test Objective (Assessment of Learning) - CES report 011114
Identifying Test Objective (Assessment of Learning) - CES report 011114mcdelmundo
 
Barriers to Technology Integration _ CLARITO
Barriers to Technology Integration _ CLARITOBarriers to Technology Integration _ CLARITO
Barriers to Technology Integration _ CLARITOLhean Clarito
 
Elements of essay lesson plan
Elements of essay lesson planElements of essay lesson plan
Elements of essay lesson planElaine Moran
 
Quarter 1 summative test -english-7
Quarter 1 summative test -english-7Quarter 1 summative test -english-7
Quarter 1 summative test -english-7MarkLesterLacanilaoS
 
Spelling bee contest
Spelling bee contestSpelling bee contest
Spelling bee contestJEWEL TECSON
 
TYPES OF CURRICULUM
TYPES OF CURRICULUMTYPES OF CURRICULUM
TYPES OF CURRICULUMParvathy V
 
Common test techniques
Common test techniquesCommon test techniques
Common test techniquesMaury Martinez
 
Lesson Plan Oral Language Grade 1112.2
Lesson Plan Oral Language Grade 1112.2Lesson Plan Oral Language Grade 1112.2
Lesson Plan Oral Language Grade 1112.2Chad Cornwell
 
Ict and assessment of learning
Ict and assessment of learningIct and assessment of learning
Ict and assessment of learningerwin marlon sario
 
Characteristics of outcomes based assessment
Characteristics of outcomes based assessmentCharacteristics of outcomes based assessment
Characteristics of outcomes based assessmentWilliam Kapambwe
 
Kinds Of Learning Targets
Kinds Of Learning TargetsKinds Of Learning Targets
Kinds Of Learning TargetsLaura Chambless
 
Assessment 1 writing general and specific objectives
Assessment 1   writing general and specific objectivesAssessment 1   writing general and specific objectives
Assessment 1 writing general and specific objectivesGerald Diana
 
1.3 tests of grammar and vocabulary: workshop CTS-Academic
1.3 tests of grammar and vocabulary: workshop CTS-Academic1.3 tests of grammar and vocabulary: workshop CTS-Academic
1.3 tests of grammar and vocabulary: workshop CTS-AcademicSeltAcademy
 
Teaching strategies their use and effectiveness as perceived by
Teaching strategies their use and effectiveness as perceived byTeaching strategies their use and effectiveness as perceived by
Teaching strategies their use and effectiveness as perceived byBabyJaneAnay
 
Multiple choice test items
Multiple choice test itemsMultiple choice test items
Multiple choice test itemslorne27
 
Arnold Serrano (Chapter 8)
Arnold Serrano (Chapter 8)Arnold Serrano (Chapter 8)
Arnold Serrano (Chapter 8)Jaypee Tan
 
Sources of expected student learning outcome
Sources of expected student learning outcomeSources of expected student learning outcome
Sources of expected student learning outcomedark191919
 

What's hot (20)

Identifying Test Objective (Assessment of Learning) - CES report 011114
Identifying Test Objective (Assessment of Learning) - CES report 011114Identifying Test Objective (Assessment of Learning) - CES report 011114
Identifying Test Objective (Assessment of Learning) - CES report 011114
 
Barriers to Technology Integration _ CLARITO
Barriers to Technology Integration _ CLARITOBarriers to Technology Integration _ CLARITO
Barriers to Technology Integration _ CLARITO
 
Achievement test
Achievement testAchievement test
Achievement test
 
Elements of essay lesson plan
Elements of essay lesson planElements of essay lesson plan
Elements of essay lesson plan
 
Quarter 1 summative test -english-7
Quarter 1 summative test -english-7Quarter 1 summative test -english-7
Quarter 1 summative test -english-7
 
Spelling bee contest
Spelling bee contestSpelling bee contest
Spelling bee contest
 
Completiontestppt
CompletiontestpptCompletiontestppt
Completiontestppt
 
TYPES OF CURRICULUM
TYPES OF CURRICULUMTYPES OF CURRICULUM
TYPES OF CURRICULUM
 
Common test techniques
Common test techniquesCommon test techniques
Common test techniques
 
Lesson Plan Oral Language Grade 1112.2
Lesson Plan Oral Language Grade 1112.2Lesson Plan Oral Language Grade 1112.2
Lesson Plan Oral Language Grade 1112.2
 
Ict and assessment of learning
Ict and assessment of learningIct and assessment of learning
Ict and assessment of learning
 
Characteristics of outcomes based assessment
Characteristics of outcomes based assessmentCharacteristics of outcomes based assessment
Characteristics of outcomes based assessment
 
Kinds Of Learning Targets
Kinds Of Learning TargetsKinds Of Learning Targets
Kinds Of Learning Targets
 
Technology in Teaching
Technology in TeachingTechnology in Teaching
Technology in Teaching
 
Assessment 1 writing general and specific objectives
Assessment 1   writing general and specific objectivesAssessment 1   writing general and specific objectives
Assessment 1 writing general and specific objectives
 
1.3 tests of grammar and vocabulary: workshop CTS-Academic
1.3 tests of grammar and vocabulary: workshop CTS-Academic1.3 tests of grammar and vocabulary: workshop CTS-Academic
1.3 tests of grammar and vocabulary: workshop CTS-Academic
 
Teaching strategies their use and effectiveness as perceived by
Teaching strategies their use and effectiveness as perceived byTeaching strategies their use and effectiveness as perceived by
Teaching strategies their use and effectiveness as perceived by
 
Multiple choice test items
Multiple choice test itemsMultiple choice test items
Multiple choice test items
 
Arnold Serrano (Chapter 8)
Arnold Serrano (Chapter 8)Arnold Serrano (Chapter 8)
Arnold Serrano (Chapter 8)
 
Sources of expected student learning outcome
Sources of expected student learning outcomeSources of expected student learning outcome
Sources of expected student learning outcome
 

Similar to Stats chapter 10

Estimating population values ppt @ bec doms
Estimating population values ppt @ bec domsEstimating population values ppt @ bec doms
Estimating population values ppt @ bec domsBabasab Patil
 
Point and Interval Estimation
Point and Interval EstimationPoint and Interval Estimation
Point and Interval EstimationShubham Mehta
 
MLlectureMethod.ppt
MLlectureMethod.pptMLlectureMethod.ppt
MLlectureMethod.pptbutest
 
MLlectureMethod.ppt
MLlectureMethod.pptMLlectureMethod.ppt
MLlectureMethod.pptbutest
 
Lesson04_Static11
Lesson04_Static11Lesson04_Static11
Lesson04_Static11thangv
 
Mca admission in india
Mca admission in indiaMca admission in india
Mca admission in indiaEdhole.com
 
Lesson04_new
Lesson04_newLesson04_new
Lesson04_newshengvn
 
Bca admission in india
Bca admission in indiaBca admission in india
Bca admission in indiaEdhole.com
 
Confidence Interval ModuleOne of the key concepts of statist.docx
Confidence Interval ModuleOne of the key concepts of statist.docxConfidence Interval ModuleOne of the key concepts of statist.docx
Confidence Interval ModuleOne of the key concepts of statist.docxmaxinesmith73660
 
Business Analytics _ Confidence Interval
Business Analytics _ Confidence IntervalBusiness Analytics _ Confidence Interval
Business Analytics _ Confidence IntervalRavindra Nath Shukla
 
Business Statistics Chapter 8
Business Statistics Chapter 8Business Statistics Chapter 8
Business Statistics Chapter 8Lux PP
 
Lecture-3 inferential stastistics.ppt
Lecture-3 inferential stastistics.pptLecture-3 inferential stastistics.ppt
Lecture-3 inferential stastistics.pptfantahungedamu
 
Section 7 Analyzing our Marketing Test, Survey Results .docx
Section 7 Analyzing our Marketing Test, Survey Results .docxSection 7 Analyzing our Marketing Test, Survey Results .docx
Section 7 Analyzing our Marketing Test, Survey Results .docxkenjordan97598
 

Similar to Stats chapter 10 (20)

Estimating population values ppt @ bec doms
Estimating population values ppt @ bec domsEstimating population values ppt @ bec doms
Estimating population values ppt @ bec doms
 
Chapter09
Chapter09Chapter09
Chapter09
 
Chapter10 Revised
Chapter10 RevisedChapter10 Revised
Chapter10 Revised
 
Chapter10 Revised
Chapter10 RevisedChapter10 Revised
Chapter10 Revised
 
Chapter10 Revised
Chapter10 RevisedChapter10 Revised
Chapter10 Revised
 
Point and Interval Estimation
Point and Interval EstimationPoint and Interval Estimation
Point and Interval Estimation
 
MLlectureMethod.ppt
MLlectureMethod.pptMLlectureMethod.ppt
MLlectureMethod.ppt
 
MLlectureMethod.ppt
MLlectureMethod.pptMLlectureMethod.ppt
MLlectureMethod.ppt
 
Lesson04_Static11
Lesson04_Static11Lesson04_Static11
Lesson04_Static11
 
Mca admission in india
Mca admission in indiaMca admission in india
Mca admission in india
 
Chapter 8
Chapter 8Chapter 8
Chapter 8
 
Stats chapter 9
Stats chapter 9Stats chapter 9
Stats chapter 9
 
Lesson04_new
Lesson04_newLesson04_new
Lesson04_new
 
Bca admission in india
Bca admission in indiaBca admission in india
Bca admission in india
 
Confidence Interval ModuleOne of the key concepts of statist.docx
Confidence Interval ModuleOne of the key concepts of statist.docxConfidence Interval ModuleOne of the key concepts of statist.docx
Confidence Interval ModuleOne of the key concepts of statist.docx
 
Business Analytics _ Confidence Interval
Business Analytics _ Confidence IntervalBusiness Analytics _ Confidence Interval
Business Analytics _ Confidence Interval
 
Business Statistics Chapter 8
Business Statistics Chapter 8Business Statistics Chapter 8
Business Statistics Chapter 8
 
Lecture-3 inferential stastistics.ppt
Lecture-3 inferential stastistics.pptLecture-3 inferential stastistics.ppt
Lecture-3 inferential stastistics.ppt
 
Section 7 Analyzing our Marketing Test, Survey Results .docx
Section 7 Analyzing our Marketing Test, Survey Results .docxSection 7 Analyzing our Marketing Test, Survey Results .docx
Section 7 Analyzing our Marketing Test, Survey Results .docx
 
Estimating a Population Mean
Estimating a Population Mean  Estimating a Population Mean
Estimating a Population Mean
 

More from Richard Ferreria (20)

Chapter6
Chapter6Chapter6
Chapter6
 
Chapter2
Chapter2Chapter2
Chapter2
 
Chapter3
Chapter3Chapter3
Chapter3
 
Chapter8
Chapter8Chapter8
Chapter8
 
Chapter1
Chapter1Chapter1
Chapter1
 
Chapter4
Chapter4Chapter4
Chapter4
 
Chapter7
Chapter7Chapter7
Chapter7
 
Chapter5
Chapter5Chapter5
Chapter5
 
Chapter9
Chapter9Chapter9
Chapter9
 
Chapter14
Chapter14Chapter14
Chapter14
 
Chapter15
Chapter15Chapter15
Chapter15
 
Chapter11
Chapter11Chapter11
Chapter11
 
Chapter12
Chapter12Chapter12
Chapter12
 
Chapter10
Chapter10Chapter10
Chapter10
 
Chapter13
Chapter13Chapter13
Chapter13
 
Adding grades to your google site v2 (dropbox)
Adding grades to your google site v2 (dropbox)Adding grades to your google site v2 (dropbox)
Adding grades to your google site v2 (dropbox)
 
Stats chapter 14
Stats chapter 14Stats chapter 14
Stats chapter 14
 
Stats chapter 15
Stats chapter 15Stats chapter 15
Stats chapter 15
 
Stats chapter 13
Stats chapter 13Stats chapter 13
Stats chapter 13
 
Stats chapter 12
Stats chapter 12Stats chapter 12
Stats chapter 12
 

Recently uploaded

"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr LapshynFwdays
 
Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!Manik S Magar
 
SAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptxSAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptxNavinnSomaal
 
Developer Data Modeling Mistakes: From Postgres to NoSQL
Developer Data Modeling Mistakes: From Postgres to NoSQLDeveloper Data Modeling Mistakes: From Postgres to NoSQL
Developer Data Modeling Mistakes: From Postgres to NoSQLScyllaDB
 
My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024The Digital Insurer
 
DevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache MavenDevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache MavenHervé Boutemy
 
Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 3652toLead Limited
 
Unraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfUnraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfAlex Barbosa Coqueiro
 
The Future of Software Development - Devin AI Innovative Approach.pdf
The Future of Software Development - Devin AI Innovative Approach.pdfThe Future of Software Development - Devin AI Innovative Approach.pdf
The Future of Software Development - Devin AI Innovative Approach.pdfSeasiaInfotech2
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek SchlawackFwdays
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Mattias Andersson
 
Install Stable Diffusion in windows machine
Install Stable Diffusion in windows machineInstall Stable Diffusion in windows machine
Install Stable Diffusion in windows machinePadma Pradeep
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationRidwan Fadjar
 
Vertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering TipsVertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering TipsMiki Katsuragi
 
"ML in Production",Oleksandr Bagan
"ML in Production",Oleksandr Bagan"ML in Production",Oleksandr Bagan
"ML in Production",Oleksandr BaganFwdays
 
Dev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebDev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebUiPathCommunity
 
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)Wonjun Hwang
 
Human Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR SystemsHuman Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR SystemsMark Billinghurst
 

Recently uploaded (20)

"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
 
Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!
 
SAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptxSAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptx
 
DMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special EditionDMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special Edition
 
Developer Data Modeling Mistakes: From Postgres to NoSQL
Developer Data Modeling Mistakes: From Postgres to NoSQLDeveloper Data Modeling Mistakes: From Postgres to NoSQL
Developer Data Modeling Mistakes: From Postgres to NoSQL
 
My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024
 
DevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache MavenDevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache Maven
 
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptxE-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
 
Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365
 
Unraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfUnraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdf
 
The Future of Software Development - Devin AI Innovative Approach.pdf
The Future of Software Development - Devin AI Innovative Approach.pdfThe Future of Software Development - Devin AI Innovative Approach.pdf
The Future of Software Development - Devin AI Innovative Approach.pdf
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?
 
Install Stable Diffusion in windows machine
Install Stable Diffusion in windows machineInstall Stable Diffusion in windows machine
Install Stable Diffusion in windows machine
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 Presentation
 
Vertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering TipsVertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering Tips
 
"ML in Production",Oleksandr Bagan
"ML in Production",Oleksandr Bagan"ML in Production",Oleksandr Bagan
"ML in Production",Oleksandr Bagan
 
Dev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebDev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio Web
 
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
 
Human Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR SystemsHuman Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR Systems
 

Stats chapter 10

  • 1. Chapter 10 Estimating with Confidence
  • 3. Definitions Statistical Inference The method of drawing conclusions about a population based on a sample The last Major Topic for this last statistics course The mindset: we are looking at data that comes from a random sample (or an experiment) and inferring characteristics of the population
  • 4. The Basic Plan In the absence of other data, the sample estimate is the estimate of the parameter We would like an interval estimation State the parameter that is being estimated Check to see if the data can be Normalized Compute the interval using area under the Normal curve Write a nice conclusion
  • 5. What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05)
  • 6. What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) Confidence Level C
  • 7. What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) Confidence Interval (CI)
  • 8. What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) Lower Bound Upper Bound
  • 9. What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) EX2We are 90% confident that the proportion that support this law is 0.74 ± 0.08
  • 10. What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) EX2We are 90% confident that the proportion that support this law is 0.74 ± 0.08 Confidence Level C
  • 11. What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) EX2We are 90% confident that the proportion that support this law is 0.74 ± 0.08 Confidence Interval (CI)
  • 12. What does an interval estimation look like? EX1We are 95% confident the mean is in the interval (9.11, 12.05) EX2We are 90% confident that the proportion that support this law is 0.74 ± 0.08 Margin of Error (ME) Point Estimate
  • 13. Confidence Level The value of the parameter is fixed before we start our sampling Either the parameter is ‘in’ our interval or it’s not. There is no probability involved here “There is a 95% probability the parameter is in the interval”
  • 14. Confidence Level The value of the parameter is fixed before we start our sampling Either the parameter is ‘in’ our interval or it’s not. There is no probability involved here “There is a 95% probability the parameter is in the interval” F A I L
  • 15. Confidence Level Remember that our sample is just one of many samples that could have been taken If good sampling technique is used, 95% (for example) of the samples would contain the parameter We just don’t know if our sample is part of the 95% or the 5%
  • 16. Confidence Level Interpretation of “90% Confidence Level” “We are 90% confident that our CI contains the value of the parameter” “90% of CI’s computed with this technique will contain the parameter” “There is a 90% probability our CI contains the parameter” WRONG INTERPRETATION
  • 17. Confidence Level Interpretation of “90% Confidence Level” “We are 90% confident that our CI contains the value of the parameter” “90% of CI’s computed with this technique will contain the parameter” “There is a 90% probability our CI contains the parameter” WRONG INTERPRETATION
  • 18. PANIC When constructing Confidence Intervals for the AP test, there is a definite checklist that readers look for Use the acronym P.A.N.I.C. to help memorize the steps to construction a confidence interval P = state the Parameter A = check that Assumptions are satisfied N = state is the Name of the interval I = compute numerical values of the Interval C = write a Conclusion for your findings
  • 19. PANIC for Means We are going to start with Means, we’ll save proportions for later Parameterdefine what  measuresdefine what x-bar measures “ = mean length of all Great White Sharks‘ flukes x-bar = mean length the Great White Shark flukes in our sample of 35 sharks”
  • 20. Assumptions for CI’s of Mean SRS- the data must have come from an SRS Independence- the population size must be more than 10 times the sampleN> 10n (Independence)
  • 21. Assumptions for CI’s of Mean SRS- the data must have come from an SRS Independence- the population size must be more than 10 times the sampleN> 10n (Independence) This is a condition that must be met since we are sampling without replacement and our formula for std dev needs to hold true
  • 22. Assumptions for CI’s of Mean SRS- the data must have come from an SRS Independence- the population size must be more than 10 times the sampleN> 10n (Independence) The sampling Distribution is approximately Normal (Normality)
  • 23. More on Normality We are looking for justification that the sampling distribution is Normal If the population is Normal, then the samp dist is also Normal If n> 30 and the sample data is not given, state “The Central Limit Theorem guarantees that the sampling distribution is Normal” For smaller samples, check the following:(1) the Histogram is single peak, symmetric with no outliers(2) the Normal Probability Plot is approx Linear
  • 24. Name of the Interval Currently, the only Interval we will worry about is “z-interval for sample means” This will always be the name of the distribution used
  • 27. Confidence Interval Computation Standard Error Always the std dev of the sampling distribution Critical Value -from the Normal (z) distribution
  • 28. Critical Value The area between –z* and +z* = Confidence Level Because they are used frequently, there is a shorthand method in table C The last row gives different confidence levels Keep critical values to 3 decimal places The row above the CL row has the z* In the AP test, they will call this z
  • 29. A General z* curve
  • 30. z* for a 80% CL
  • 31. Computing a CI Let’s compute the 95% CI for a sample of 35, with a mean of 5.38 and a population std dev =0.74 (1) locate the critical value z* = 1.960 (2) Compute S.E. and M.E. SE = 0.74/(35) = 0.1251ME = (1.960)(0.1251) = 0.2452
  • 32. Computing a CI State the CI = 5.38 ± 0.25 Interval estimate:(5.13, 5.63) Notice that the point estimate is the average of the upper and lower bounds
  • 33. Conclusions We are 95% confident that the mean length of a great white shark fluke is 5.38 ± 0.25 ft. OR We are 95% confident that the mean length of all great white shark flukes is in the interval (5.13 ft, 5.63 ft).
  • 35. Calculators TI83/TI84 Your TI is very efficient for finding these intervals! This doesn’t excuse you from the mathematics, of course. [stat] -> “TESTS” -> “Zinterval” Inpt: “Stats” (if your data is in L1, you can use “Data” Enter values for , x-bar, n, and C-Level. Viola!
  • 36. Calculators TI89 Run the “Stats/List Editor” APP [2nd] -> [F2] (F7) -> “Zinterval” Input Method = “Stats”Choose Data if all the observations are in a List Enter values for , x-bar, n, and C-Level. Viola!
  • 37. Behavior of Margin of Error ME = z* (/n) In practice, we would like to minimize ME. (1) decrease z* This also means decrease our CL! (2) increase nthis is usually a trade off, since obtaining large samples could be more expensive/time consuming (3) decrease Not really an option.  is a known quantity.
  • 38. Sample Size By using algebra, we can get a formula to compute minimum sample size for a given ME You should always round the sample size up in this calculation Note: ME is produced by sampling variability, it has nothing to do with “sloppy work”
  • 39. 10.2 Estimating a Population Mean
  • 40. Student’s t-distribution The confidence intervals computed in the previous section assumed that we knew . It doesn’t seem likely that we would know  and not know the value of  When  is unknown, we can no longer use the Normal distribution
  • 41. Student’s t-distribution The t-distribution is to be used when  is unknown. The t-distribution is very similar to the Normal distribution with a key difference The shape of the distribution changes based upon the sample size/degrees of freedom (df) Large samples have a tall peak and thinner tails Small samples have a small peak and thicker tails.
  • 43. Using the t-distribution The CI is found using PANIC The Assumptions are the same as a z-interval Since sample sizes tend to be small, you will most likely need to check the histogram (symmetric, no outliers) and Normal prob plot (approx linear) We cannot use the t-distribution when there are outliers! The Name of the interval is “1-sample t-interval for mean”
  • 44. Using the t-distribution Interval is computed with df = n – 1 t* is from table C or your calculator s is the sample’s standard deviation Uses the sample std dev to approximate 
  • 46. Upper Tail Area This is the Upper Tail Area
  • 48. Using the t-distribution Using TI84 [2nd] -> [vars](DIST) -> “invT” “invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“−invT(Upper Tail Area, df)”
  • 49. Using the t-distribution Using TI84 [2nd] -> [vars](DIST) -> “invT” “invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“−invT(Upper Tail Area, df)”
  • 50. Using the t-distribution Using TI84 [2nd] -> [vars](DIST) -> “invT” “invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“−invT(Upper Tail Area, df)” Don’t forget the negative!
  • 52. Using the t-distribution Using TI89 From home screen: [catalog] -> [F3](FlashApps) -> inv_t…TIStat “tistat.invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“-tistat.invT(Upper Tail Area, df)”
  • 53. Using the t-distribution Using TI89 From home screen: [catalog] -> [F3](FlashApps) -> inv_t…TIStat “tistat.invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“-tistat.invT(Upper Tail Area, df)”
  • 54. Using the t-distribution Using TI89 From home screen: [catalog] -> [F3](FlashApps) -> inv_t…TIStat “tistat.invT(1-Upper Tail Area, df)” Where “Upper Tail Area” = (1-CL)/2 This is the area to the left of the right crit. Value. ALTERNATIVELY, you may use“-tistat.invT(Upper Tail Area, df)” Don’t forget the negative!
  • 55. Using the t-distribution ALTERNATIVE TI89 titanium [APPS] -> “Stat/List Editor” -> [F5] (distrib) -> “Inverse” -> “Inverse t…” “Area: 1- Upper Tail” Upper Tail = (1 – CL)/2 “degrees of freedom, df: df” This takes longer to get to, but the menu “guides” you through
  • 56. Table C Occasionally, you will need to fine the t* for a df that does not appear in the chart. When this happens, you are to use the nearest greatest df in the same column This usually means “use the t* that is in the line above where the desired t* should be”
  • 57. Example: 1 sample t-interval Problem 10.30 The amount of Vitamin C (mg/100g) in CSB for a random sample of 8 are given as: 26, 31, 23, 22, 11, 22, 14, 31 Construct a 95% Confidence interval for the amount of vitamin C in CSB.
  • 58. Example: 1 sample t-interval Parameter  = the mean amount of vitamin C in CSB produced at the factory x-bar = the mean amount of vitamin C in a sample n = 8
  • 59. Example: 1 sample t-interval Assumptions SRS ‘Our problem states that we have a random sample’ Independence ‘10n = 80 < N; we can infer that more than 80 CSB is produced in the factory’
  • 60. Example: 1 sample t-interval Assumptions Normality ‘The histogram is symmetric w/ no outliers’ ‘The Norm Prob Plot is approx linear’ ‘We have good evidence that our sampling distribution is approximately Normal’ x 10 15 20 25 30 35 Norm Prob Plot Histogram z
  • 61. Example: 1 sample t-interval Name of Interval ‘We will compute a 1-sample t-interval for a mean’
  • 62. Example: 1 sample t-interval Name of Interval ‘We will compute a 1-sample t-interval for a mean’ Interval Calculation
  • 63. Example: 1 sample t-interval Name of Interval ‘We will compute a 1-sample t-interval for a mean’ Interval Calculation Conclusion ‘We are 95% confident that the mean amount of vitamin C in a unit of CSB produced in the factory is between 16.487 and 28.513 mg/100 g’
  • 64. 10.3 Estimating a population Proportion
  • 65. Estimating a Proportion Like with means, we are going to estimate the population proportion based on the proportion in a sample x = # of positive responsesn = total number of responses We will again use the PANIC procedures Note: nothing is averaged. We are not looking at the average proportion from many samples
  • 66. Parameter Some typical parameters: ‘p = prop of people in CA who support the proposition p-hat = proportion of people in a sample n = 35 who support the proposition’ ‘p = prop of students at THS who ride the bus p-hat = propotion of students in a sample (n = 14) from THS who ride the bus
  • 67. Assumptions Simple Random Sample SRS must be either stated or inferred Independencebecause you are usually sampling without replacement:N> 10n Normalityn·p-hat > 10n·q-hat > 10
  • 68. Name of the Interval “1-proportion z interval” Unlike for means, you will always use the Normal curve when dealing with proportions!
  • 69. Interval Calculation z* is calculated as before Use table C OR “ − invNorm( (1 – C) /2 )”This calculation is similar to the calculations for the last section
  • 70. Interval Calculation z* is calculated as before Use table C OR “ − invNorm( (1 – C) /2 )”This calculation is similar to the calculations for the last section Margin of Error (ME)
  • 71. Interval Calculation z* is calculated as before Use table C OR “ − invNorm( (1 – C) /2 )”This calculation is similar to the calculations for the last section Standard Error(SE)
  • 72. Conclusion Some Examples: We are 90% confident that the proportion of voters in CA who support the proposition is 0.34 ± 0.03 We are 95% confident that the proportion of students at THS who ride this bus is in the interval (0.39, 0.44)
  • 73. Sample Size The relevant formula (from ME) for the sample size is: p* and q* are guessed values of the proportion If there is no previous data or study, use p* = q* = 0.5 (this will maximize the error and sample size) As before, you are to round the sample size up to the nearest integer
  • 75. TI83/84 [stat] -> “TEST” -> “1-PropZInt” “x: number of successes” this can be computed with “p-hat x n” “n: number of people in sample” “C-Level : confidence level” “Calculate” and you are done You still need to fully write up “PANIC” procedures
  • 76. TI 89
  • 77. TI 89 [APPS] -> “Stat/List Editor” [2nd] -> [F2](F7) -> “1-PropZInt” “Successes x: # of successes” “n: number of people in sample” “C-Level : confidence level” “Calculate” and you are done You still need to fully write up “PANIC” procedures
  • 78. Example 1-PropZInt The 2004 Gallup Youth Survey asked a random sample of 439 US teens aged 13 to 17 whether they though young people should wait to have sex until marriage. 246 sad “yes.” Let’s construct a 95% confidence interval for the proportion of all teens who would say “Yes”
  • 79. Parameter p = the proportion of all teens in the US aged 13-17 who would answer “Yes” to the survey p-hat = the proportion of teens in the survey of 439 who answered “Yes” to the survey
  • 80. Assumptions SRS: We are told in the problem that the survey was random sample. Independence: there are than 10(439) = 4390 teens aged 13-17 in the US Normalityn x p-hat = 246, n x q-hat = 193The sampling distribution is approx. Normal (Note that this is just #successes and #failures)
  • 81. Name of Interval We are constructing a “1 proportion Z interval”
  • 83. Conclusion “We are 95% confident that the true proportion of 13-17 year olds who would answer “yes” when asked if young people should wait to have sex until they are married is between 0.514 and 0.606.”