Other's On-Line Resources (Chapter 3)
|
Scatter Diagrams |
- Description:
You'll see 4 scatter diagrams from a real study involving people
who were measured in terms of two variables: isometric strength
and job performance.
- What to Do:
- Click on the colored title of this on-line resource: "Scatter
Diagrams."
- Examine the 4 scatter diagrams, carefully noting (in each
case) what 2 variables were involved.
- Try to correctly answer, on your first try, the multiple-choice
question that's located immediately after the final scatter
diagram
- Sky Huck's Puzzle Question:
The 2nd of the 4 scatter diagrams involves the variables "SIMS"
and "ARM." Do you think the correlation would be higher or lower
if all data points were to be deleted except those that involve
positive SIM scores combined with ARM scores greater than 80?
|
|
Guessing Correlations |
- Description:
In this interactive on-line resource, you'll be given several
scatter diagrams and are asked to guess the size of the correlation
coefficient for each set of data points. Your answers are scored
for each set of 4 scatter diagrams you examine, with a cumulative
score displayed if you look at more than 4 sets of data.
- What to Do:
- Click on the colored title of this on-line resource: "Guessing
Correlations."
- Scroll down to the section called "Step by Step Instructions."
- Look at the scatter diagram and then choose one of the 5 possible
values for r shown beneath the scatter diagram.
- Click on the "New Data" button to see a new scatter diagram with a new
set of possible values for r.
- Repeat steps 3 and 4.
- Sky Huck's Puzzle Question:
Select one of this site's scatter diagrams, and determine which of the 5 options
has the correct value of r. Next, imagine that each dot's score along the
abscissa (i.e., the horizontal axis) is cut in half. Likewise, imagine that each
dot's score along the ordinate (i.e., the vertical axis) is cut in half. Now, if the
correlation were to be computed for the revised data, would r be half as
large as it was for the original data?
|
|
Pearson's Product-Moment Correlation |
- Description:
In this interactive on-line resource, you'll first enter pairs
of scores on 2 variables and then quickly be able to see the resulting
value of r for the data you entered. (The scores you enter can
be "real," in the sense that they come from a set of people, animals,
or things actually measured by you or someone else; or, your scores
can be "artificial," in the sense that they are created by you
solely for the purpose of "playing" with data to see how changes
in data influence the size of Pearson's r.)
- What to Do:
- Click on the colored title of this on-line resource: "Pearson's
Product-Moment Correlation."
- In the small box that appears on your screen, enter the
number of pairs of scores in your data set. (The word "undefined"
that initially appears in the small box will be replaced by
your value for "n" once you type in the number of pairs of
scores you have.)
- On the next screen that appears, scroll down until you get
to a section called "Data Entry." In that section, enter your
paired scores in the X and Y columns under the heading "Data
cells." (Do not put anything in the column called "Residuals"
or in the box called "Import/Export Box.")
- Click on the "Calculate" button positioned beneath the Y
scores you just entered.
- Scroll down slightly, just past where you're provided with
the standard deviations for your X and Y scores. Look to see
(a) the numerical value of the correlation coefficient (r)
for your data and (b) the coefficient of determination (r2).
- Sky Huck's Puzzle Question:
Using these 10 whole numbers [1, 2, 3, 4, 5, 6, 7, 8, 9, and 10],
create 5 pairs of scores in an effort to generate the largest
possible positive value of r. There are 2 little rules for this
game: (1) each of the 10 numbers must be used once and only once
as you go about the task of creating your 5 pairs of numbers,
and (2) the 2 numbers that form any pair cannot together add up
to more than 14. If you accept this little challenge, can you
obtain an r equal to +.90? Can you obtain an r that's higher than
this?
|
|
Spearman's Rho |
- Description:
This interactive on-line resource will compute, for the data you
enter, Spearman's rank-order correlation coefficient. You have
the option of entering your data in the form of ranks or in the
form of raw scores. If you enter raw scores, they will be converted
into ranks for you. (The raw scores or ranks you enter can be
"real," in the sense that they come from a set of people, animals,
or things actually measured by you or someone else; or, your data
can be "artificial," in the sense that they are created by you
solely for the purpose of "playing" with data to see how changes
in data influence the size of rs.)
- What to Do:
- Click on the colored title of this on-line resource: "Spearman's
Rho."
- In the small box that appears on your screen, enter the
number of pairs of scores in your data set. (The word "undefined"
that initially appears in the small box will be replaced by
your value for "n" once you type in the number of pairs of
scores you have.)
- On the next screen that appears, scroll down until you get
to a section called "Data Entry." In that section, enter your
data in one of these two ways:
- If you have data in the form of unranked raw scores,
enter your data in the two columns labeled "Raw data for
X and Y." The top numbers in those columns would be your
1st pair of scores, the 2nd numbers down would be your
2nd pair of scores, etc.
- If your data are in the form of ranks, enter those
ranks in the two columns labeled "Ranks for X and Y."
The top numbers in those columns would be your 1st pair
of ranks, the 2nd numbers down would be your 2nd pair
of ranks, etc.
- Click on the "Calculate" button directly beneath the data
you just entered.
- Just below the "Calculate" button, look at the numbers displayed
in the "n" and "rs" windows. The first of these
number should match up with the number of pairs of scores
you had. The number in the "rs" window is the value
for Spearman's rho.
- Sky Huck's Puzzle Question:
First set n equal to 5. Next, consider yourself free to enter
the numbers 1 through 5 in any order you'd like in the
two columns labeled "Ranks for X" and "Ranks for Y." And now here
comes your two-fold task:
- Can you arrange the ranks such that rs turns out
equal to +.50?
- Can you arrange the ranks such that rs turns out
equal to +.75?
|
|
