The priorities for elements in each cluster of the objectives hierarchy, as well as the priorities derived for the alternatives with respect to each covering objective, are referred to as "local" priorities.
The options available on the "Cluster Priorities" page depend on if the inconsistency ratio is hidden or shown.
Local Results when Inconsistency Ratio is hidden
If the Project Manager has specified that the inconsistency ratio for your judgments not be shown, you will see the priorities for the judgments elements you have just made on a screen that looks like:
You can click on any heading to sort by that column.
If you think the priorities are not reasonable (i.e., are not intuitive), then click the 
button.
You can then (1) select a pair of elements: one element that you think may have too high a priority and the other element too low a priority.
(2) After selecting the pair and clicking "Re-evaluate," you will be taken to the screen where you can enter or revise the judgment comparing these two elements.
After doing so and clicking Next, you will be taken back to the screen showing the revised cluster priorities.
Local Results when Inconsistency Ratio is shown
If the Project Manager has specified that the inconsistency ratio of your judgments is shown (See Math of AHP and Inconsistency Ratio), then the priorities of the elements - as well as the inconsistency ratio - will be displayed on a screen like the following:
As a very rough rule of thumb, the inconsistency ratio should be .10 or less. However, there are reasons for accepting results even if the inconsistency ratio is as high as .2 or .3. (See Inconsistencies, or Decision by Objectives on Professor Forman's Website or at Amazon.) It is more important that the priorities be reasonable to you than to have a low inconsistency ratio. You should NOT change judgments just because of inconsistencies. You should re-examine judgments because of high inconsistency and change only those judgments that you feel were incorrectly recorded or for which you have a change of opinion -- regardless of inconsistency.
If you feel that either the priorities are not satisfactory or would like to review the judgments to address a high inconsistency ratio, click the above button. Doing so will produce the following screen:
Clicking the "Click here to review your judgments" will take you through the first page of the evaluation for the given cluster.
Clicking the "Click here if you think the inconsistency is too high" button will result in a screen showing the judgment matrix (discussed below).
Clicking the "Click if you would like to redo a judgment for one pair of elements" will take you through the sequence explained at the top of this page.
Judgment Matrix
The judgment matrix will be displayed when you click the second button 
The judgments shown in the cells of the matrix indicate how much more important or preferable the row element corresponding to a judgment cell is, rather than the column element corresponding to the judgment cell.
Red judgments mean that judgments for an element in the column are more important or preferable than an element in a row.
Blank cells in the upper diagonal of the matrix represent judgments that were not elicited or entered.
The radio button 
makes the intersecting cells clickable. Clicking on any one of the cells will redirect you to the step displaying the pairwise comparison for the row and column elements corresponding to that cell.
Make changes on the judgment matrix
If you wish to make or investigate possible changes to judgments in the matrix itself, click on the 
button and enter or change the judgment.
The judgments are shown numerically in this matrix, regardless of whether they were made in the verbal or numerical/graphical modes.
You can type in judgments and then press enter to save.
To invert judgments (change from black to red or red to black), press either the - or i keys. Inverting is saved automatically.
Red judgments mean that judgments for an element in the column are more important or preferable than an element in a row.
By default, the elements (objectives or alternatives) are sorted by original order in the model as shown above.
You can sort the elements by priority (descending) by clicking 
.
You can easily notice that elements are sorted by priority by looking at the priority bars below the elements:
You can hover on the element cell to see its priority as shown above.
You can sort back the elements based on their original order by clicking 
You can then click Sort by priority to see how the sorting changed after altering the judgments.
Conditions for consistent judgments when the matrix is sorted by priority.
If the judgments were perfectly consistent, they would be increasing (or more precisely non-decreasing) as you look at them:
a) from left to right in each row, and
b) from bottom-up in each column
In the above figure, which corresponds to a "reasonably" low inconsistency ratio of 0.15, the most inconsistent judgment is the 5 in the top row, which violates condition a) above. Changing this judgment to a 1 (less than the 2 to its right) will decrease the inconsistency to 0.04. However, changing judgments just to reduce the inconsistency is NOT A GOOD IDEA. Judgments should be changed only when the evaluator feels that the judgment itself was an error or is no longer warranted. If this judgment (Financials vs. Improve Organizational efficiency) was to be changed from a 5 to a 1, the ranks of the elements also change as shown below:
If the evaluator doesn't believe that a judgment of 1 (equal) is appropriate, or that the change in ranks of the objectives/alternatives is correct, then pressing the 
button will abandon all changes and return to the original judgments.
Looking at the original matrix and without knowing what the participant's thinking is for these judgments, it is more likely that changing the 2 (Leverage Knowledge vs. Maintain Serviceability) to something higher, such as 8 ("very strong to extreme" if these were verbal judgments), would be more appropriate -- it violates both conditions a) and b) above.
Even though this inconsistency (.08) is more than that resulting from the first change considered (.04), it is more logical. But once more, we must say it is NOT A GOOD IDEA to change judgments just to lower the inconsistency ratio.
Click on the 
to return to the results screen.
Note: Comparion doesn't calculate inconsistency for a group of evaluators because the group geometric average reduces the consistency considerably, so the simulations that were done for individual sets of judgments do not apply for priorities computed from geometric averages.
Rank and Best Fit
You can display the inconsistency rank and the best fit by checking 
and 
respectively.
This will show small numbers on each cell with judgments.
The Rank by Inconsistency is the number at the upper right (blue), while the best fit is at the lower right (black or red (inverted)) of the actual judgment.
The Rank by Inconsistency is the order of inconsistency of that judgment with the other judgments. So, for example, the cell with a judgment of 5 (strong in the verbal mode) and a 1 in the upper right corner of the cell is the most inconsistent judgment; and the judgment of 2 with a 2 in the upper right of the cell is the second most inconsistent judgment.
The best-fit judgment is not necessarily the best judgment, but rather it is the judgment that fits best with all of the other judgments that were made.
Best Fit suggests what changes should be made to the judgments to reduce inconsistency from a mathematical perspective. It is not advisable to change judgments to the "best fit" values, but rather to use the "best fit" values to give you an idea of which judgments you might want to reconsider in order to reduce the inconsistency. As a decision-maker, you must determine if this is logical and what changes in judgment should be made versus what is recommended.
