Other's On-Line Resources (Chapter 6)
| Flipping
Coins |
- Description:
This interactive on-line resource allows you to guess whether
a coin-flip will turn out "heads" or "tails." After you make your
guess, the coin is flipped and you get to see the outcome. Results
are then tabulated as you flip the coin multiple times.
- What to Do:
- Click on the colored title of this on-line resource: "Flipping
Coins."
- In the screen that pops up, locate the phrase "Guess below!"
and then click on "Head" or "Tail" to indicate your prediction
as to how the 1st coin toss will turn out.
- Make additional guesses, each time noting your cumulative
"success rate."
- If your final position in the graph does not match the "P(head)
value of 0.05, then recognize that you have--over the collection
of flips in your sample--a sampling error.
- Sky Huck's Puzzle Question:
If you play this little game by flipping 30 times, do you think
you'll have a better chance of ending up on the red line (thus
indicating no sampling error) with "P(head)," the population
parameter, set equal to 0.5 or 0.9? After making your guess, try
it out (both ways) and see what happens.
|
| Sampling
Distributions |
- Description:
Through this interactive exercise, you'll see (1) how a sampling
distribution is created, (2) how the standard error decreases
if n is increased, and (3) how the sampling distribution will
be approximately normal even if the population is nonnormal.
- What to Do:
- Click on the colored title of this on-line resource: "Sampling
Distributions."
- After clicking on "Begin," click "Animated Sample" in the
next screen. Watch what happens. Then click on "Animated Sample"
some more. This will show you how sampling distributions are
built.
- Click on "5 Samples" (or "1,000 Samples" or "10,000 Samples"
to look at the sampling distribution that's created quickly
for lots of samples.
- Change the n, the population's shape, and the statistical
focus (from the mean to the median) to see how these things
affect sampling distributions.
- Sky Huck's Puzzle Question:
In which of these two cases is the standard error smaller? (a)
When n = 10 and the sample's median is computed, or (b) When n
= 5 and the sample's mean is computed.
|
| Confidence
Intervals (A) |
- Description:
You're able to see a visual display of 100 confidence intervals,
each built for a sample of a given size extracted from a population
having M = 50 and SD = 10. You'll see which of the 100 CIs "overlap"
the population mean . . . and which ones do not. And you'll also
see a simple frequency count of how many of the 100 CIs "catch"
mu.
- What to Do:
- Click on the colored title of this on-line resource: "Confidence
Intervals (A)."
- After clicking on "Begin," click "sample" in the new screen.
Clicking "sample" again adds 100 new CIs to your 1st set.
- Sky Huck's Puzzle Question:
Set the sample size equal to 10. Then, after hitting "sample"
10 times in a row, observe the "Proportion Contained" results
in the bottom 2 windows. Now, if you repeat these steps but this
time set n = 20, will the "Proportion Contained" results go up,
go down, or remain the same?
|
| Confidence
Intervals (B) |
- Description:
Using this interactive on-line resource, you'll get to see a picture
showing 50 CIs, each computed for the mean of a different random
sample drawn from the same population. The picture will show how
many of these CIs overlap the population mean. You can control
the level of confidence and the number of CIs that are computed.
- What to Do:
- Click on the colored title of this on-line resource: "Confidence
Intervals (B)."
- On the screen that pops up, read the instructions, look
at the "alpha bar," and examine the set of CIs displayed in
the picture.
- Click the "More Intervals" button to get a new batch of
50 CIs.
- Move the black triangle to change alpha from .05 to .01,
and then click on the "New Alpha" button to produce a brand
new set of 50 CIs.
- Set alpha equal to either .05 or .01, click the "New Alpha"
button, and then repeatedly click the "More Intervals" button
until you can see the combined results for 1,000 CIs.
- Sky Huck's Puzzle Question:
If we could somehow change the word "alpha" in this on-line resource
to "confidence," what changes (if any) would we need to make in
the numbers inside the "Alpha" bar?
|
|
