Knowledge Base

Lag Sequential Analysis

What Is Lag Sequential Analysis?

Transitions Between Events

In The Observer XT, Lag Sequential Analysis calculates the frequency of transitions between pairs of events within a certain lag.

Lag Sequential Analysis allows you to answer questions like:

• How many times is the event Mother Smiles followed by the event Baby Smiles?
• How likely is the dog's behavior Ignores Trainer preceded by each type of command by the trainer?
• How often is Courtship Song by a fruit fly male followed by Mating?

Important Terms in Lag Sequential Analysis

Criterion – Of a pair of behaviors, the first one is called Criterion.

Target – Of a pair of behaviors, the second one is called Target.

State Lag – State lags are transitions between events that directly follow each other or are separated by a specific number (lag order) of other events. For example, from an event to the next one (lag order +1), or to the second next (lag order +2), the third next, etc. You can calculate transitions with a lag order from -9 to +9. The time between the Criterion and the Target does not influence your results.

Time Lag – Time lags are transitions between events within a specific time window, independent of how many events are between them. You can calculate transitions, for example, from an event to those events occurring in the next 1 minute. Since there may be more than one Target event in that 1 minute window, Time Lag analysis generates multiple transitions from the same Criterion.

General Procedure

Aim: To analyze how often or how likely an event is followed or preceded by other events.

Prerequisites:

• Your license for The Observer XT includes the Advanced Analysis Module.
• You carried out at least one observation.

Procedure:

1. Make sure that the data profile specifying the data you want to analyze is active, that is, highlighted in blue in the Project Explorer.
2. Do one of the following:
   • Choose Analyze > Lag Sequential Analysis > New.
   • In the Project Explorer, click the Analyses folder. In the window that appears, choose Analyze Data > Lag Sequential Analysis.
   • In the Project Explorer under Analyses, click Lag Sequential Analyses and then New Lag Sequential Analysis.
3. A table with question marks appears. Click the Calculate button on the toolbar to perform the analysis.

Note: By default, the table contains a State Lag - Order 1 analysis with frequencies.

Lag Sequential Analysis Options

Aim: To choose the type of lag sequential analysis and other settings.

Prerequisites:

• Your license for The Observer XT includes the Advanced Analysis Module.
• You carried out at least one observation.
• You opened the Lag Sequential Analysis window (Analyze > Lag Sequential Analysis).

Procedure: Click the Settings button on the toolbar and select the following options:

• Lag Type
• Restrict Lag to Subject-Behavior Group
• Frequency or Probability
• Missing values and number of decimals

Lag Type

Select one of the following:

• State Lag – Enter the lag order you want analysis to be based on. For example, to analyze transitions from each event to the next one, enter 1. If you want to calculate transitions from each event to the second next, enter 2. If you want to calculate transitions from each event to the previous third event, enter -3. By default, 1 is selected.
• Time Lag – Enter the limits of the time window. This time window will be applied to each Criterion event. By default, From 0 to 1 sec is selected, that is, transitions are counted from each event to any event occurring up to 1 second later.

You have the following additional options:

• Ignore Recurring Criteria – Select this option if you want to exclude instances of a criterion event that occur in the same time lag.
• Ignore Recurring Targets – Select this option if you want to exclude second, third, etc. instances of a target event that occur within the same time lag.

Restrict Lag to Subject-Behavior Group

Select this option to calculate the transitions within subject and behavior groups. For example, select this option to calculate transitions within the behavior groups Locomotion and Social Behavior, that is, from an event of Locomotion to another event of Locomotion, and from an event of Social Behavior to another event of Social Behavior, not from an event of Locomotion to an event of Social Behavior (or vice versa).

If you leave this option cleared, transitions are calculated regardless of which group events belong to. Leave the option deselected to calculate transitions between the behavior of one subject (for example Mother) and another subject (for example Child).

Frequency or Probability

• Frequency – To calculate the number of transitions from the criterion to the target.
• Probability – To calculate the number of transitions for a particular combination of criterion and target divided by the total number of transitions from that criterion. The sum of the transition probabilities for each criterion event equals 1. Select the Only Show When Greater Than option if you want to display only the transition probabilities that are greater than a specific threshold.

Missing Values and Number of Decimals

• Show Missing Values As – Choose the character you want to have displayed for missing values, that is, cells in which transitions are not calculated. For example, transitions between behaviors of different groups when you have selected the Restrict Lag to Subject-Behavior Group option.
• Number of Decimals – Choose the number of decimals in the matrix cells. This option is only available if you have chosen to display lag probabilities.

Note: If you defined a range of numerical modifiers instead of predefined values, many different values may have been scored. If you subsequently carry out a Lag Sequential Analysis, this may result in many cells in the analysis result. This may especially happen if you have numerical modifiers with several decimals, or if you have imported external data as numerical modifiers. Deselect Modifiers in the Layout window, or create a Data Selection to reduce the data to analyze.

The Lag Sequential Analysis Result

Results Table

The Lag Sequential Analysis results table contains the following:

• A – Cells containing the values of independent variables.
• B – Cells containing counts or probabilities of transition. Each row represents a criterion event and each column a target event. Each cell contains, depending on what you have chosen in your analysis settings:
  • If you have chosen Frequency, the number of transitions from the criterion (on the row) to the target event (on the column).
  • If you have chosen Probability, the probability of transition from the criterion event to the target event.
• C – Main headers for categories (for example, Observations, Subjects, Behaviors, etc.).
• D – Secondary headers for the elements within a category.

Dependent on your choices for the layout, some of the options may not be shown.

Understanding X0 and Y0

Transitions are also counted when no criterion can be found for a target or vice versa. Such transitions can be found in the cells of the matrix indicated by X0 (last row) or Y0 (last column).

X0 refers to transitions where a target has been identified but the criterion is not found. This happens in the following cases:

• If the lag is positive – At the start of an observation (or interval defined within an observation). If the first event is 2.00 Sit, this event is considered to be the target of no event. A transition is counted from X0 to 2.00 Sit.
• If the lag is negative – At the end of an observation or interval, where the last event is considered to be the target of no event.
• Special case – In Time Lag analysis, when an event is preceded by a gap longer than the time lag of the previous event. For example, the event Answer occurs after a gap. If the time lag is shorter than the time between Answer and any preceding event, then Answer is considered to be the target of no event. A transition is counted from X0 to Answer.

Y0 refers to transitions where no target is found. This happens in the following cases:

• If the lag is positive – At the end of an observation or interval, where the last event is considered to be the criterion of no event.
• If the lag is negative – At the start of an observation or interval, where the first event is considered to be the criterion of no event.
• Special case – In Time Lag analysis, when there is no event within a time lag. For example, the event Play Music is not followed by any other event in the next 10 seconds. With a time lag of 5 seconds, the time lag of Play Music does not cover any target event. A transition is counted from Play Music to Y0. Note: when at least one event is found within a time lag, the gap is not considered, so the transition from Play Music to Y0 is not counted. Instead, the transition from Play Music to that event is counted.

Notes:

• X0 and Y0 transitions are included in the probability calculations to quantify the probability that an event is not preceded nor followed by any events.
• For instantaneous sampling – X0 and Y0 do not indicate transitions from/to samples not scored. In lag sequential analysis, samples not scored are ignored.

Save a Result

To save a lag sequential analysis result, click the Archive button on the toolbar of the result window. To re-open a saved result, choose Analyze > Lag Sequential Analysis > Open Archive.

Notes:

• You cannot edit the content of the statistics result.
• To print a result table, choose File > Print. We advise you to select Landscape mode prior to printing.

Customize the Analysis Results

Aim: To change the layout of the analysis results table.

Prerequisites:

• Your license for The Observer XT includes the Advanced Analysis Module.
• You carried out at least one observation.
• You opened the Lag Sequential Analysis window (Analyze > Lag Sequential Analysis).

Procedure: Click the Layout button on the toolbar. Specify what the resulting transition matrix will look like. Select the type of elements you want to view on rows, columns, and separate sheets. Clear the selection for any element type you do not want included in the matrix.

• Criteria – Are displayed on rows. Select the element types to include.
• Targets – Are displayed on columns. Select the element types to include.
• On Sheets – Select the element types you want to view in separate sheets.

Show Elements Not Scored

Select this option if you want to have all elements of your coding scheme shown in the matrix, including those you have not scored in any observation. For not scored elements, cells will contain zeros or the missing value symbol. If you keep this option cleared, the matrix lists the elements scored in at least one observation.