Comparing Inference
Intervals
Process Z - Interval
P Parameter –population measurement of interest Population – Who are you trying to reach/talk about
A Conditions: SRS & Large Sample Size (normal curve) Verify: by CLT
How was the sample taken and/or given…
N 1 Sample Z-Interval For Population Means
I
C From the data we are _______ % confident the the true mean, “context of real mean,” lies between
___________ and ____________ (provide units)
n³30
Population³10n
Process T- Interval
P Parameter –population measurement of interest Population – Who are you trying to reach/talk about
A Conditions: SRS & Large Sample Size (t curve)
Verify: 15<n use t-procedures if data is close to normal but no outliers
15 < n<40 use t-procedures except in presence of outliers and strong skewness
40 >n normal t-procedures, even if clearly a skewed distribution
How was the sample taken and/or given…
N 1 Sample t-Interval For Population Means
I
C From the data we are _______ % confident the the true mean, “context of real mean,” lies between ___________ and
____________ (provide units)
Process T- Interval
P Parameter –population measurement as a difference
Population –Difference within the understanding of who you are trying to reach/talk about
A Conditions: SRS & Large Sample Size (t curve)
Verify: 15<n use t-procedures if data is close to normal but no outliers
15 < n<40 use t-procedures except in presence of outliers and strong skewness
40 >n normal t-procedures, even if clearly a skewed distribution
Independent ~ the way trials were taken, one did not effect the other
How was the sample taken and/or given…
N Matched Pairs t-Interval For Population Means
I
C From the data we are _______ % confident the the true mean, “context of real mean,” lies between ___________ and
____________ (provide units)
x±t* s
Comparing Inference Test
Process Z - Test
P Parameter –population measurement of interest Population – Who are you trying to reach/talk about
H H0 : No Change Ha: Change
A Conditions: SRS & Large Sample Size (normal curve) Verify: by CLT
How was the sample taken and/or given…
N 1 Sample Z-Test For Population Means
T
O Probability of z score (Area To Left)
M Fail to Reject Ho Or Reject Ho
S From the data we have/or do not have significant evidence at the_______ significance level, talk about decision and evidence in terms of Ha
n³30
Population³10n
z= x-m
s n
Process T - Test
P Parameter –population measurement of interest Population – Who are you trying to reach/talk about
H H0 : No Change Ha: Change
A Conditions: SRS & Large Sample Size (t curve)
Verify: 15<n use t-procedures if data is close to normal but no outliers
15 < n<40 use t-procedures except in presence of outliers and strong skewness 40 >n normal t-procedures, even if clearly a skewed distribution
How was the sample taken and/or given…
N 1 Sample T-Test For Population Means
T
O Probability of t score (Area To RIGHT), Know degrees of freedom and significance level
M Fail to Reject Ho Or Reject Ho
S From the data we have/or do not have significant evidence at the_______ significance level, talk about decision and evidence in terms of Ha
t= x-m
s n
Process T - Test
P Parameter –population measurement as a difference
Population –Difference within the understanding of who you are trying to reach/talk about
H H0 : No Change Ha: Change
A Conditions: SRS & Large Sample Size (t curve)
Verify: 15<n use t-procedures if data is close to normal but no outliers
15 < n<40 use t-procedures except in presence of outliers and strong skewness 40 >n normal t-procedures, even if clearly a skewed distribution
Independent ~ the way trials were taken, one did not effect the other How was the sample taken and/or given…
N Matched Pairs T-Test For Population Means
T
O Probability of t score (Area To RIGHT), Know degrees of freedom and significance level
M Fail to Reject Ho Or Reject Ho
S From the data we have/or do not have significant evidence at the_______ significance level, talk about decision and evidence in terms of Ha
t= x-m
s n
m = 0
Example #1: Workers at a factory asked their supervisor to provide music during their shift. The supervisor wanted to know whether the music really helped improve the worker’s performance. Because all the workers were assigned to different rooms at random, the supervisor randomly selected one room. For one week, he provided music in this room, and for one
week he provided none. He flipped a coin to determine which week to provide music. Afterward, he recorded the productivity of the workers, using the average number of items assembled per day:
Estimate the difference between the workers’ performance in the
Example #1: Workers at a factory asked their supervisor to provide music during their shift. The supervisor wanted to know whether the music really helped improve the worker’s performance. Because all the workers were assigned to different rooms at random, the supervisor randomly selected one room. For one week, he provided music in this room, and for one
week he provided none. He flipped a coin to determine which week to provide music. Afterward, he recorded the productivity of the workers, using the average number of items assembled per day:
Comparing Two Means
Comparative studies are more convincing than single single-sample investigations, so one-sample inference is not as common as
comparative (two-sample) inference.
In a comparative study, we may want to compare two treatments, or we may want to compare two populations. In either case, the samples must be chosen _________________ and _________________ in order to perform statistical inference.
Since matched pairs are NOT chosen independently, we will NOT
use two-sample inference for a matched pairs design. For a
matched pairs design, apply the one-sample t procedures to the observed differences.
Otherwise, we may use two-sample inference to compare two
treatments or two populations.
Before you begin, check your assumptions! For comparing two means,
both samples must come from an _______ and must be chosen
_______________. Also, both populations must be
__________________________. (Always check for ____________ or
If each population is __________, it turns out that the difference of
distributions will be __________; and if the populations are
approximately __________, then the difference will be approximately
__________. Of course, this all hinges on knowing the values of σX and
σY, which (in all likelihood) we will not know. So what do we do?
When we did that earlier in this chapter, we stopped looking for at the
distribution of , and instead started looking at the
distribution of
Naturally, that's what we're going to do here! The statistic is now
OR The more common notation
t
=
x
-
m
xs
xn
x
t= x1 - x2
s12
n1 + s22
n2
1-C
2 t=
x1 - x2 s12
n1 + s22
There's just one problem—this statistic doesn't have a
__________________________.
Fortunately, it's close. We can still use the t distribution to answer
probability questions (and conduct inference) for this new variable.
Degrees of Freedom
Back before computers and calculators were common, there was a
conservative rule of thumb: use the smaller sample size (minus one) as the df. This is called a conservative approach, because it tends to fail to reject the null hypothesis more often than it should.
Now that we have calculators and computers, we can use a different approach.
Confidence Intervals
A Level C confidence interval for the difference of means is
__________________________________, where t* is the upper
critical value from the t distribution with k degrees of freedom.
1
-
C
Hypothesis Testing
The null hypothesis is that there is no difference between the two parameters.
The alternative hypothesis could be as follows:
One-Sided Two-Sided
Process T - Test
P Parameter –population measurement for comparison between two groups
Population – Who we are trying to reach/ talk about, through comparing groups
H H0 : No Change Ha: Change
A Conditions: SRS & Large Sample Size (t curve)
Verify: 15<n use t-procedures if data is close to normal but no outliers
15 < n<40 use t-procedures except in presence of outliers and strong skewness 40 >n normal t-procedures, even if clearly a skewed distribution
Independent ~ the way trials were taken, one did not effect the other How was the sample taken and/or given…
N 2 Sample T-Test For Population Means
T
O Probability of t score (Area To RIGHT), Know degrees of freedom and significance level
M Fail to Reject Ho Or Reject Ho
S From the data we have/or do not have significant evidence at the_______ significance level, talk about decision and evidence in terms of Ha
t= x1-x2
s12
n1 + s22
Example #1: A study in Stockholm concerned the work conditions of the city's bus drivers—in particular, a program was designed to help relieve stress on the job by changing the bus routes. Drivers were randomly assigned to either an old route (control), or a new route (intervention). The heart rates of the drivers were measured after driving the routes.
Is there any evidence that the improved routes have changed heart rates?
Here are the variables:
μC = mean heart rate for drivers of the ______________
μI = mean heart rate for drivers of the ______________
H0: Ha:
There are some requirements to conduct this test:
(a) SRS?
Each population variable must have a
________________________________. This is not given, so…
Since ____________ are
unknown,______________________________________________.
Find the test statistic:
Calculate the P-value:
Decision:
Example #2: Some research has been conducted comparing the leg strengths of males and females.
Here are the data (Force, in Newtons) for a random sample of males:
Here are the data for a random sample of females: