Sie sind auf Seite 1von 4

Creating Confidence Intervals using Excel 2010

Create Confidence Intervals using Excel 2010

11/17/2013 V0d

Create Confidence Intervals using Excel 2010

Create Confidence Intervals


Using Excel 2010
by
Milo Schield
Member: International Statistical Institute
US Rep: International Statistical Literacy Project
Director, W. M. Keck Statistical Literacy Project
Slides at: www.StatLit.org/pdf
/Create-Confidence-Intervals-Excel2010-6up.pdf

Start
here:

Create Confidence Intervals using Excel 2010

Using Excel to Build


One-group Margin of Error
For one-group confidence intervals:
11. Use Excel toolpak Descriptive-Statistics
summary. This is simplest, but it does not
automatically update if data changes.
12. Use CONFIDENCE function. It automates
the Z or T inverse.
13. Use Excel basic statistical functions:
average, standard deviation, sample size,
and Z or T inverse.

Create Confidence Intervals using Excel 2010

11a Margin of Error


using Descriptive Statistics

11b Margin of Error


using Descriptive Statistics

From Tools tool-bar, select


Data Analysis at far right.

Select Descriptive Statistics.

You must enter a confidence level (95)


to get the margin of error (Confidence).

Create Confidence Intervals using Excel 2010

Confidence Level: 95% Margin of Error =T*StdDev()/Sqrt(n)

Create Confidence Intervals using Excel 2010

12a. Margin of error


using Confidence Functions
CONFIDENCE.NORM: Pop. Stdev known; Z-scores
CONFIDENCE.T: Pop. std dev. unknown; T-scores

12b. ME using CONFIDENCE


for Measurements
.

All CONFIDENCE functions return a margin of error


and have three arguments: alpha, SD and sample size
=Confidence.Norm (alpha, Pop. std. deviation, sample size)
=Confidence.t (alpha, Sample std. deviation, sample size)
=Confidence (alpha, Pop. Std. deviation, sample size)
If CL=95%, = .05. Confidence.T = (T/Z)*Confidence.Norm

Use Confidence.Norm for proportions.


Confidence is Confidence.Norm in Excel 2003.

www.StatLit.org/pdf/Create-Confidence-Intervals-Excel2010-6up.pdf

Creating Confidence Intervals using Excel 2010

Create Confidence Intervals using Excel 2010

12c. ME using CONFIDENCE


for Proportions

11/17/2013 V0d

Create Confidence Intervals using Excel 2010

13a. Margin of Error


using Excel components
Margin of Error
= Z*StdDevP(data)/Sqrt(n) = Z*Sqrt(p(1-p)/n)
= T*StdDev(data) / Sqrt(n)

Use Z if population standard deviation is known:


Z = NormSInv(0.5+CL/2)
Use T if only the sample standard deviation is known:
T = TINV(, n-1) where = (1-CL)/2

Create Confidence Intervals using Excel 2010

13b. Margin of Error


using Excel components

Create Confidence Intervals using Excel 2010

10

Using Excel to Build


Two-group Margin of Error
For two-group confidence intervals of stacked data (one
subject per row) without pre-sorting:

21. Use pivot tables, descriptive statistics and Z or T inverse


22. Use array IF function along with basic statistical
functions: average, standard deviation, sample size and
Z or T inverse
23. Use array IF function along with CONFIDENCE and
other basic statistical functions
If stacked data is pre-sorted into contiguous groups, use
one-group approaches

Create Confidence Intervals using Excel 2010

11

Unsorted left; sorted right


Sort makes contiguous ranges

Create Confidence Intervals using Excel 2010

12

21. Build Margin of Error


using Pivot Table Statistics

www.StatLit.org/pdf/Create-Confidence-Intervals-Excel2010-6up.pdf

Creating Confidence Intervals using Excel 2010

Create Confidence Intervals using Excel 2010

11/17/2013 V0d

Create Confidence Intervals using Excel 2010

13

22a. Build ME for Averages


from AverageIF and Stdev(IF)
.

14

22b. Build ME for Proportions


from AverageIF and Stdev(IF)
.

Create Confidence Intervals using Excel 2010

Create Confidence Intervals using Excel 2010

15

23a. Build ME for Averages


using CONFIDENCE

16

23b. Build ME for Proportions


Using CONFIDENCE

Create Confidence Intervals using Excel 2010

17

Survey 95% Margin of Error


Surveys typically give the size of the 95% margin of
error. E.g., 3 percentage points.
This is the most conservative 95% margin of error for
the entire survey. It is computed for =50% so it is the
broadest for that sample size.
A 25% subset of the group will have a 95% margin of
error that is twice as big all other things being equal.
A question for the whole group having p ~ 1/16 will
have a 95% margin of error that is half the size of the
survey margin of error calculated for =50%.

Create Confidence Intervals using Excel 2010

18

Statistical Significance
and Confidence Intervals
If two 95% confidence intervals do not overlap, then
the difference in means is statistically-significant. But
the converse may be false: those cases that overlap may
still have differences that are statistically significant.
If the 95% confidence interval for one group does not
overlap the mean for the second, then the difference
in means is statistically-significant at the 0.05 level.
In this case, the converse is also true: those cases where
the confidence interval overlaps the mean are not
statistically significant. See Conrad Carlbergs article:
www.quepublishing.com/articles/article.aspx?p=1717265&seqNum=3

www.StatLit.org/pdf/Create-Confidence-Intervals-Excel2010-6up.pdf

Creating Confidence Intervals using Excel 2010

Create Confidence Intervals using Excel 2010

11/17/2013 V0d

Create Confidence Intervals using Excel 2010

19

Statistical Significance
and Confidence Intervals

20

11a Descriptive Statistics


(with Confidence)

Instead of looking for non-overlapping confidence


intervals from two independent samples, the traditional
approach is to calculate the confidence interval on the
difference between the two sample means.
If that 95% interval does not include zero, then the
difference is reported as being statistically significant.
If that 95% interval does include zero, then the
difference is reported as being statistically insignificant.

From Tools tool-bar, select Data Analysis at far right.


Select Descriptive Statistics.

You must enter a


confidence level (95)
to get a the margin of
error (Confidence).

Create Confidence Intervals using Excel 2010

21

Confidence Interval Overlap


and Statistical Significance
Generating two-group 95% confidence intervals
and looking for the presence or absence of overlap
is a useful first-step in determining if the difference
in sample means is statistically-significant.
The lack of overlap is generally sufficient to infer
the difference is statistically-significant. The
converse is not true: overlapping 95% confidence
intervals may still be statistically-significant.

www.StatLit.org/pdf/Create-Confidence-Intervals-Excel2010-6up.pdf

Das könnte Ihnen auch gefallen