Internet
Demonstrations
Up Sampling Dist. Ans.
| |
You can link to a sampling distributions applet to create from scratch
the sampling distribution of the mean, and also observe how this distribution
changes with shifts in population shape and sample size.
- Read the Instructions (printing a hard copy, if possible, for easy
reference.)
- To simulate from scratch the sampling distribution of the mean, follow the
directions in the "Basic operations" paragraph of the
Instructions. Always click the "Fit normal" button to superimpose
a normal distribution over the simulated sampling distribution.
- Click on "Begin" to start this simulation, and when finished, return
to this page.
Question: This simulation illustrates two of the three important
properties of the sampling distribution of the mean discussed in Chapter 13.
Which are they? Answer
- To simulate the effect of the central limit theorem, use the pop-up menus
as follows: specify "Skewed" as the shape of the population;
specify N = 5 for the one sampling distribution of means (third histogram);
and specify the mean and N=25 for the second sampling distribution of means
(fourth histogram); and click on both "Fit normal" buttons.
Then, click the "Animated sample" button to observe
precisely how the process works. To speed things up, take 10,000 samples at a time.
Refer to the left margin and use measures of skew
("lopsided-ness") and kurtosis ("peaked-ness"), which
each should equal zero when the distribution is normal, to decide whether
the sampling distribution with N = 25 approaches a normal distribution more
rapidly than the sampling distribution with N = 5, as predicted by the
central limit theorem. (If, just by chance, results seem to contradict
the central limit theorem -- because measures of skewness or kurtosis for
N=25 are slightly larger than for N=5 -- continue taking 10,000 samples,
without clearing, and note any changes with very large numbers of
samples.)
- Repeat this process after changing the shape of the population to
"Uniform" or, to create your own shape with the mouse, specify
"Custom".
|