The Observer XT 17 - Reliability Analysis - The Duration/Sequence Method in Detail
Last updated: Jul 28, 2026
The Duration/Sequence Method in Detail
Calculation
In the Duration/Sequence method, only events with duration are taken into account. For each observation pair, The Observer XT compares the total duration that the events of the same type overlap in both observations. Gaps between events are only included in the calculation if you select Analyze gaps between events.
Suppose you observe Event B and Event C in two observations, i and j. The agreements (A) are calculated as the sum of the duration that Event B overlaps in observations i and j, plus the duration that Event C overlaps in observations i and j. A is the total of all overlapping durations.
The disagreements (D) represent the total duration that events of the same type do not overlap. If the observations are compared entirely, the events within the observation do not overlap, and both observations have the same duration, then D equals the observation duration minus A.
Agreements and Disagreements
The values A and D are used in calculation of the reliability statistics.
Effect of Behavior Group - Duration/Sequence Method
Reliability Analysis and Gaps
Example
Four behaviors have the following overlapping duration in two observations.
Agreements (A) = 28.8 + 14.1 + 12.3 + 12.5 = 67.7 s
You also find these numbers in the diagonal of the Confusion matrix.
| Behavior | Total Duration Same Event Overlaps in Both Observations (s) |
|---|---|
| Gazing at patient | 28.8 |
| Gazing at nurse | 14.1 |
| Gazing at patients file | 12.3 |
| Other gazing behavior | 12.5 |
The observation duration is 74.0 s. The Disagreements (D) is 74.0 - 67.7 = 6.3 s. You also find these numbers by adding up the numbers outside the diagonal of the Confusion matrix, including the No Records. The sum of these values is 6.4. The small difference with the calculated D is because of a rounding effect.
Gaps
If you select Analyze gaps between events, a gap is treated as an event, so the time nothing is scored in both observations is treated as agreement. Mind that this may lead to a high agreement between observations if the gaps are large compared to the scored events.
The Frequency Method in Detail
Calculation
For each observation pair, The Observer XT compares the total number of occurrences of each event. Gaps between events are not included in the calculation.
Suppose you observe Event B and Event C in two observations i and j.
The agreements (A) are calculated as follows:
A = min[Bi, Bj] + min[Ci, Cj]
Where Bi and Bj are the total number of occurrences of Event B in observation i and observation j. Ci and Cj are the total number of occurrences of Event C in observation i and observation j. A (Agreements) is the sum of the minimum occurrences of events B and C.
The disagreements (D) are calculated as follows:
D = (max[Bi, Bj] + max[Ci, Cj]) - A
The values A and D are used in calculation of the reliability statistics.
Example
You observe the following total numbers of occurrences of four behaviors in two observations.
Agreements (A) = 10 + 7 + 5 + 5 = 27
You also find these numbers in the diagonal of the confusion matrix.
| Behavior | Observation 1 Total Number | Observation 2 Total Number |
|---|---|---|
| Gazing at patient | 10 | 12 |
| Gazing at nurse | 7 | 7 |
| Gazing at patients file | 5 | 5 |
| Other gazing behavior | 5 | 6 |
Disagreements (D) = (12 + 7 + 5 + 6) - 27 = 3
You also find these numbers when you add up the Total Error cells of the confusion matrix.
The Duration Method in Detail
Calculation
In the Duration method only events with duration are taken into account. For each observation pair, The Observer XT compares the total duration of each event type, which is the sum of the durations of the individual events. Gaps between events are not included in the calculation.
Suppose you observe Event B and Event C in two observations.
The agreements (A) are calculated as follows:
A = min[Bi, Bj] + min[Ci, Cj]
Where Bi and Bj are the total durations of Event B in observation i and observation j. Ci and Cj are the total durations of Event C in observation i and observation j. A (Agreements) is the sum of the minimum durations of events B and C.
The disagreements (D) are calculated as follows:
D = (max[Bi, Bj] + max[Ci, Cj]) - A
The values A and D are used in calculation of the percentage and proportion of agreements, Kappa, Rho, and other statistics.
Example
You observe the following total durations of four behaviors in two observations.
Agreements (A) = 29.7 + 14.8 + 13.1 + 14.3 = 71.9 s
You also find these numbers in the diagonal of the Confusion matrix.
| Behavior | Observation 1 Total Duration (s) | Observation 2 Total Duration (s) |
|---|---|---|
| Gazing at patient | 29.7 | 30.4 |
| Gazing at nurse | 14.8 | 16.1 |
| Gazing at patients file | 13.1 | 13.1 |
| Other gazing behavior | 15.2 | 14.3 |
Disagreements (D) = (30.4 + 16.1 + 13.1 + 15.2) - 71.9 = 2.9 s. You also find these numbers when you add up the Total Error cells of the confusion matrix. The value 3.0 instead of 2.9 is due to a rounding effect.
The Reliability Analysis Result
How to Display the Results
- Choose Analyze > Reliability Analysis > New.
- Select the settings and click OK.
The Results Windows
The Reliability Analysis results consists of the following windows:
- Statistics
- Confusion Matrix
- If you selected the Frequency/Sequence comparison method, the Reliability Analysis Result also contains the Comparison List.
To Save a Result
To save a result, click the Archive button on the toolbar of the result window, or from the Analyze menu, select Archive Analysis Results. To re-open a saved result, click the Open Archive button on the toolbar or choose Analyze > Reliability Analysis > Open Archive. For more information on saving, opening and exporting results, see Manage Analysis Results in File Management.
To Print a Result
To print a result, choose File > Print.
Note: You cannot edit the content of the Reliability Analysis Results sheet.