Transcript Lecture 22

Statistics 270 - Lecture 22
• Last Day…completed 5.1
• Today Parts of Section 5.3 and 5.4
Example
• Government regulations indicate that the total weight of cargo in a
certain kind of airplane cannot exceed 330 kg. On a particular day a
plane is loaded with 81 boxes of a particular item only. Historically,
the weight distribution for the individual boxes of this variety has a
mean 3.2 kg and standard deviation 1.0 kg.
• What is the distribution of the sample mean weight for the boxes?
• What is the probability that the observed sample mean is larger
than 3.33 kg?
• Statistical Inference deals with drawing conclusions about
population parameters from sample data
• Estimation of parameters:
• Estimate a single value for the parameter (point estimate)
• Estimate a plausible range of values for the parameter
(confidence intervals)
• Testing hypothesis:
• Procedure for testing whether or not the data support a theory
or hypothesis
Point Estimation
• Objective: to estimate a population parameter based on the
sample data
• Point estimator is a statistic which estimates the population
parameter
• Suppose have a random sample of size n from a normal population
• What is the distribution of the sample mean?
• If the sampling procedure is repeated many times, what proportion
of sample means lie in the interval:
• If the sampling procedure is repeated many times, what proportion
of sample means lie in the interval:
• In general, 100(1-a)% of sample means fall in the interval

 



z
,


z
a /2
a /2

n
n 
• Therefore, before sampling the probability of getting a sample
mean in this interval is
• Could write this as:

 

P   za / 2
 X    za / 2   (1  a )
n
n

• Or, re-writing…we get:

 

P X  za / 2
   X  za / 2   (1  a )
n
n

• The interval below is called a
confidence interval for

 

X

z
,
X

z
a /2
a /2

n
n 
• Key features:
• Population distribution is assumed to be normal
• Population standard deviation, , is known
Example
• To assess the accuracy of a laboratory scale, a standard weight
known to be 10 grams is weighed 5 times
• The reading are normally distributed with unknown mean and a
standard deviation of 0.0002 grams
• Mean result is 10.0023 grams
• Find a 90% confidence interval for the mean
Interpretation
• What exactly is the confidence interval telling us?
• Consider the interval in the previous example. What is the
probability that the population mean is in that particular interval?
• Consider the interval in the previous example. What is the
probability that the sample mean is in that particular interval?
Large Sample Confidence Interval for 
• Situation:
• Have a random sample of size n (large)
• Suppose value of the standard deviation is known
• Value of population mean is unknown
• If n is large, distribution of sample mean is
• Can use this result to get an approximate confidence interval for the
population mean
• When n is large, an approximate
for the mean is:
100(1  a )% confidence interval
Example
• Amount of fat was measured for a random sample of 35 hamburgers of a
particular restaurant chain
• It is known from previous studies that the standard deviation of the fat
content is 3.8 grams
• Sample mean was found to be 30.2
• Find a 95% confidence interval for the mean fat content of hamburgers for
this chain
Changing the Length of a Confidence Interval
• Can shorten the length of a confidence interval by:
• Using a difference confidence level
• Increasing the sample size
• Reducing population standard deviation
Sample Size for a Desired Width
• Frequent question is “how large a sample should I take?”
• Well, it depends
• One to answer this is to construct a confidence interval for a
desired width
Sample Size for a Desired Width
• Width (need to specify confidence level)
• Sample size for the desired width
Example
• Limnologists wishes to estimate the mean phosphate content per
unit volume of a lake water
• It is known from previous studies that the standard deviation is
fairly stable at around 4 ppm and that the observations are
normally distributed
• How many samples must be sampled to be 95% confidence of
being within .8 ppm of the true value?
Example
• A plant scientist wishes to know the average nitrogen uptake of a
vegetable crop
• A pilot study showed that the standard deviation of the update is
about 120 ppm
• She wishes to be 90% confident of knowing the true mean within
20 ppm
• What is the required sample size?