Current through Register Vol. 48, No. 39, March 25, 2024
Subpart 1.
Scope.
This part applies to a jurisdiction with more than three
male-dominated classes.
Subp. 2.
Criteria for statistical analysis test.
To pass this test, analysis of the jurisdiction's
implementation report must show:
A. an
underpayment ratio of 80.0 percent or more; or
B. an underpayment ratio less than 80.0
percent, and:
(1) for a jurisdiction with six
or more male-dominated classes and with one or more salary ranges, an average
pay difference which is the same for male-dominated and female-dominated
classes or which does not represent a disadvantage for female-dominated
classes;
(2) for a jurisdiction
with six or more male-dominated classes and with one or more salary ranges, an
average pay difference which represents a disadvantage for female-dominated
classes, and a determination that the difference is not statistically
significant; or
(3) for a
jurisdiction with fewer than six male-dominated classes, and for a jurisdiction
that has no salary ranges for any of its classes, a determination that the
jurisdiction meets the alternative analysis test described in part
3920.0600.
Subp. 3.
Steps in statistical analysis.
For each jurisdiction with more than three male-dominated
classes, the department must conduct a statistical analysis. The analysis
includes determining and analyzing the following data: predicted pay,
underpayment ratio, average pay difference, and statistical significance of the
average pay difference as described in subparts
4 to
9. All operations in this
part are based on unrounded data, except when otherwise specified.
Subp. 4.
Determining predicted pay.
The department must determine predicted pay for each
male-dominated and female-dominated class in the jurisdiction. Predicted pay
means predicted salary for those jurisdictions which do not have different
benefits for male-dominated and female-dominated classes of comparable work
value, as described in part
3920.0300, subpart
6. For those jurisdictions
which do have different benefits for male-dominated and female-dominated
classes of comparable work value, predicted pay means the total predicted
amount of salary plus benefits contribution limits.
The department must determine predicted pay by creating a
window, drawing a regression line within the window, and identifying a
predicted pay point on the regression line. The process described in items A to
D is continued until pay has been predicted for each male-dominated and
female-dominated class in the jurisdiction.
A. Creating a window. The analysis creates a
window around the class. The window defines classes of comparable work value
for purposes of the statistical analysis. Except as provided in item B, each
window represents 20 percent of the total range of job evaluation ratings in
the jurisdiction. The total range of evaluation ratings is determined by
subtracting the lowest rating assigned to any class from the highest rating
assigned to any class. The result is then multiplied by 20 and divided by 100.
In addition, the window must meet the criteria in subitems (1) to (4).
(1) The lower limit of the window is below
the evaluation rating of the class being analyzed by ten percent of the total
range of evaluation ratings, except when the class being analyzed is in the
bottom ten percent or the top ten percent of the total range of evaluation
ratings. The upper limit of the window is above the evaluation rating of the
class being analyzed by ten percent of the total range of evaluation ratings,
except when the class being analyzed is in the top ten percent or the bottom
ten percent of the total range of evaluation ratings.
(a) If the evaluation rating of the class
being analyzed is in the bottom ten percent of the total range of evaluation
ratings, the lower limit of the window is the lowest rating assigned to any
class in the jurisdiction and the upper limit of the window is 20 percent above
the lower limit.
(b) If the
evaluation rating of the class being analyzed is in the top ten percent of the
total range of evaluation ratings, the upper limit of the window is the highest
rating assigned to any class in the jurisdiction and the lower limit of the
window is 20 percent below the upper limit.
(2) The window must include at least three
male-dominated job classes. When analyzing a male-dominated job class, the
class being analyzed is counted as one of the three male-dominated job
classes.
(3) The window must
include at least two male-dominated job classes with different job evaluation
ratings. When analyzing a male-dominated job class, the class being analyzed is
counted as one of the two male-dominated job classes with different
ratings.
(4) The window must
include at least 20 percent of all the male-dominated classes in the
jurisdiction.
B.
Expanding the window. If any of the criteria in item A, subitems (2) to (4),
are not met, the window is expanded in increments of five percent of the total
range of evaluation ratings on both sides of the previous window, except as
provided in subitems (1) and (2). That is, in the first expansion the lower
limit becomes the rating level 15 percent below the class being analyzed and
the upper limit becomes the rating level 15 percent above the class being
analyzed. The window is increased using these five percent increments as many
times as necessary until the criteria in item A, subitems (2) to (4), are met.
(1) If the expansion results in a lower limit
below the lowest rating assigned to any class in the jurisdiction, then the
lower limit is the lowest rating assigned to any class in the jurisdiction. The
upper limit is above the lower limit by the total length of the expanded
window, that is, 30 percent in the first expansion, 40 percent in the second
expansion, and so forth. The window is expanded until the criteria in item A,
subitems (2) to (4), are met.
(2)
If the expansion results in an upper limit above the highest rating assigned to
any class in the jurisdiction, then the upper limit is the highest rating
assigned to any class in the jurisdiction. The lower limit is below the upper
limit by the total length of the expanded window, that is, 30 percent in the
first expansion, 40 percent in the second expansion, and so forth. The window
is expanded until the criteria in item A, subitems (2) to (4), are
met.
C. Drawing a line.
Using conventional statistical regression techniques, the analysis fits a
linear regression line to all male-dominated classes in the window. The line is
weighted to reflect the number of employees in each male-dominated class. The
regression line represents the relationship between job evaluation ratings and
salary, or between job evaluation ratings and salary plus benefits in the case
of jurisdictions with different benefits for male-dominated and
female-dominated classes of comparable work value, as explained in part
3920.0300, subpart
6.
D. Predicting pay. The analysis predicts pay
for the class being analyzed by determining the dollar value on the regression
line which corresponds to the job evaluation rating of the class being
analyzed.
Subp. 5.
Determining underpayment ratio.
The analysis tabulates the number of female-dominated and
male-dominated classes which are paid below predicted pay for their job
evaluation ratings. The analysis then calculates female-dominated classes paid
below predicted pay as a percentage of all female-dominated classes in the
jurisdiction, and male-dominated classes paid below predicted pay as a
percentage of all male-dominated classes in the jurisdiction, as
follows:
A. the number of
male-dominated classes which are paid below predicted pay is divided by the
total number of male-dominated classes, and the result is multiplied by
100;
B. the number of
female-dominated classes which are paid below predicted pay is divided by the
total number of female-dominated classes, and the result is multiplied by 100;
and
C. the result from item A is
divided by the result from item B, and the quotient is multiplied by 100 and
rounded to one decimal place. This is the underpayment ratio.
Subp. 6.
Analyzing underpayment ratio.
If the underpayment ratio is 80.0 percent or more, the
department must find that the jurisdiction passes the statistical analysis
test. If the underpayment ratio is less than 80.0 percent, the department must
continue the compliance review as explained in items A to C.
A. If the underpayment ratio is less than
80.0 percent, and the jurisdiction has fewer than six male-dominated classes,
the department must use the alternative analysis test described in part
3920.0600. The department must not
continue the statistical analysis as described in subparts
7 to
9.
B. If the underpayment ratio is less than
80.0 percent, and the jurisdiction has no salary ranges for any of its classes,
the department must use the alternative analysis test described in part
3920.0600. The department must not
continue the statistical analysis as described in subparts
7 to
9.
C. If the underpayment ratio is less than
80.0 percent, and the jurisdiction has six or more male-dominated classes and
one or more salary ranges, the department must continue the statistical
analysis as described in subparts
7 to
9.
Subp. 7.
Determining average pay difference.
For a jurisdiction described in subpart
6, item C, the department
must determine and analyze the average pay difference. The average pay
difference is the dollar amount of the average difference from predicted pay,
calculated as follows:
A. The number
of employees in each female-dominated class is multiplied by the dollar amount
of the difference from predicted pay for that class. Both positive amounts
(above predicted pay) and negative amounts (below predicted pay) are
included.
B. The sum of the amounts
in item A is divided by the total number of employees in female-dominated
classes and rounded to the nearest whole dollar. The result is the average
difference from predicted pay for female-dominated classes.
C. The process explained in items A and B is
repeated for male-dominated classes. The result is the average difference from
predicted pay for male-dominated classes.
Subp. 8.
Analyzing average pay difference.
The department must evaluate the average pay difference as
explained in items A and B.
A. If the
average pay difference is the same for male-dominated and female-dominated
classes, or if the average pay difference does not represent a disadvantage for
female-dominated classes, the department must find that the jurisdiction passes
the statistical analysis test.
B.
If the average pay difference represents a disadvantage for female-dominated
classes, the department must continue the analysis as described in subpart
9.
Subp. 9.
Significance of average pay difference (t-test).
If the average pay difference represents a disadvantage for
female-dominated classes, a standard test of statistical significance called
the t-test must be applied to this finding. The department must evaluate the
results as explained in items A and B.
A. The t-test of pooled variance is applied
using conventional statistical techniques. Significance is determined at the
five percent level for a one-tailed test. The statistical analysis rounds the
value of t to three decimal places. The sample t table is taken from a standard
statistical text: Blalock, Social Statistics, Second Edition 1972, published by
McGraw-Hill, page 559. The degrees of freedom is the total number of employees
in male-dominated and female-dominated classes, minus two. To be significant,
the value of t for a jurisdiction must be at or above the level listed, except
that for degrees of freedom not listed, the required level of t is taken from a
standard normal distribution table.
Distribution of t (five percent significance)
Degrees of Freedom |
Value of t |
1 |
6.314 |
2 |
2.920 |
3 |
2.353 |
4 |
2.132 |
5 |
2.015 |
|
|
6 |
1.943 |
7 |
1.895 |
8 |
1.860 |
9 |
1.833 |
10 |
1.812 |
|
|
11 |
1.796 |
12 |
1.782 |
13 |
1.771 |
14 |
1.761 |
15 |
1.753 |
|
|
16 |
1.746 |
17 |
1.740 |
18 |
1.734 |
19 |
1.729 |
20 |
1.725 |
|
|
21 |
1.721 |
22 |
1.717 |
23 |
1.714 |
24 |
1.711 |
25 |
1.708 |
|
|
26 |
1.706 |
27 |
1.703 |
28 |
1.701 |
29 |
1.699 |
30 |
1.697 |
|
|
40 |
1.684 |
60 |
1.671 |
120 |
1.658 |
infinity |
1.645 |
B.
If the t-test is not significant, the department must find that the
jurisdiction has passed the statistical analysis test. If the t-test is
significant, the department must find that the jurisdiction has failed the
statistical analysis test and is not in compliance.
Statutory Authority: MS s
43A.04