Knowledge Base

Calculate Statistics

Introduction

The Observer XT has a large number of descriptive statistics that give a full numerical overview of your data.

For any analysis, you can restrict the data to analyze in a data profile (see Select Data for Analysis). If you use the default data profile with all data, The Observer XT calculates statistics on all observations in your project. It is therefore important to select the correct data profile before you carry out any analysis.

The Observer XT has the following descriptive analysis types:

Behavior Analysis

To calculate statistics for events scored in the observations, like the duration or number of occurrences of an event. See Statistics for Behavior Analysis in Detail for a detailed description.

• Example 1 – The average duration and standard deviation of the behavior Play for the subject Child.
• Example 2 – The number of times the subject Test Participant 1 requested help.

Numerical Analysis

To calculate statistics for numerical modifiers and external data in the observation, like the average value. See Statistics for Numerical Analysis in Detail for a detailed description.

• Example 1 – Calculate the average value of the numerical modifier Speed.
• Example 2 – Calculate the average heart rate of the test participant.

To analyze numerical modifiers as events, for example to calculate their duration or frequency, carry out a Behavior Analysis instead of a Numerical Analysis. See To Analyze Numerical Modifiers. See also Analyze Modifiers.

General Procedure

To calculate descriptive statistics for your observations.

Prerequisite: You carried out at least one observation.

1. Select the data profile specifying the data you want to analyze. To do so, click it in the Project Explorer.
2. Choose Analyze > Behavior Analysis or Numerical Analysis > New. A table appears with question marks.
3. Click the Statistics button on the toolbar and select the statistics.
4. Click the Calculate button on the toolbar. The Statistics Result table appears.

The Statistics Result

In the statistic result table, you can distinguish the following groups of cells. The exact position depends on the choices you made in the Layout window. See Customize the Results.

• A – Statistics: Cells containing a statistic for a combination of Results Container, Observation, Event Log, Subject, Behavior, and Modifier.
• B – Independent Variables (optional): Cells containing the value of an independent variable linked to a certain observation, event log, or subject. Shown if you have selected Independent Variables on the Layout page.
• C – Main Headers (dark gray area): The names of the main categories: Data Selection Result Containers, Intervals, Observations, Event Logs, Subjects, Behaviors and their Modifiers, and Variables.
• D – Secondary Headers (light gray area): Show the single element values, for example the name of the behaviors or independent variables.

Modifiers

A behavior may be associated with several modifier values. Statistics are shown for the modifiers attached to a number of communication types.

Result Containers

The Result Containers column or row shows the name of the Results boxes specified in the currently active data profile (see Select Data for Analysis). If you have specified more than one Results box, their names are shown in the Result Containers row or column. Statistics are calculated for each Results box separately, if you selected Result Containers in the Layout window. See Group Elements.

Intervals

The Intervals row or column shows the start and end time of the time segments analyzed if you have specified intervals or time bins in your data profile. To display intervals, make sure that Result Containers and Intervals are selected in the Layout window. See Group Elements.

The format of the intervals depends on what you chose as time format. See Time Formats.

What Happens at the Interval Borders?

The border of two connecting time intervals can only be used for analysis in one of the two time intervals. Otherwise the data for this data point would be duplicated. The Observer XT uses the left borders (start) of the time intervals for analysis, while the right borders (end) are used in the analysis of the next interval. In the example below the exact data point 0.0 is used in the first interval, 10.0 in the second, 20.0 in the third, and 30.0 is not used in any interval. If your project contains external data, the sample at the exact end of an interval is also not included in the analysis.

Events Scored at the Observation Stop Time

Events scored at the exact stop of the observation are not included in the analysis. Instantaneous samples cannot be scored at the exact stop of the observation.

<Missing Subject> and <Missing Behavior>

The <Missing Subject> and <Missing Behavior> rows/columns show the statistics for records you have scored with no subject or no behavior. For example, when you score a behavior with no subject, or when you scored comments only.

You can only calculate the frequency, rate per minute, and latency of Missing Behaviors and no durations. Durations are only calculated for State behaviors that are defined in the Coding Scheme.

To hide <Missing Subject> or <Missing Behaviors>, choose View Settings > Subjects or Behaviors and deselect <Missing Subject/Behaviors>.

Group Elements

To group all elements in one category and obtain average or total values for that category. For example, obtain the mean value of the duration of a behavior for all subjects, or for all observations.

Prerequisite: You carried out at least one observation.

1. Follow the General Procedure to obtain a statistics results table.
2. Click the Layout button on the toolbar.
3. Deselect the element you want to group. For example, deselect Observations to calculate the total number of times an event was scored across all observations.

Examples

• Example 1 – Grouping Observations: When Observations is selected in the Layout window, statistics are displayed per observation. When Observations is deselected, the statistics are given for all observations grouped.
• Example 2 – Grouping Modifiers: Deselect Modifiers to calculate statistics for the behaviors, independent of which modifiers were scored.

Notes on Grouping

• Important: Do not group categories if you want to calculate statistics for external data. Doing so may give unexpected results.
• You cannot group Result Containers and Behaviors. To analyze behaviors as a group, merge them first in the data profile. See Merge Data.
• If you want to group some elements in one category and others not, use Merging (see Merge Data) in a data profile and carry out an analysis while this profile is active.

Statistics with Merged or Grouped Data

Merged Data: If you have merged data in the data profile (see Merge Data), the analysis results show the merged results. Consider the following example: Behaviors A and B both have a duration of 5 seconds and they overlap for 1 second. After merging, the frequency of the merged behavior is 1 and its duration is 9 seconds. The percentage of time the merged behavior occurs is 100%.

Grouped Data: If you have grouped data by deselecting a category in the Layout window (see Group Elements), the Number and Duration used to calculate the statistics are summed across all elements in that category.

Example: In two observations of 10 minutes each, behavior A is scored for 5 minutes in the first and 5 minutes in the second. When deselecting Observations in the Layout window, the frequency of behavior A is 2. The total duration is 10 + 10 = 20 minutes. The percentage of time behavior A occurred is ((5+5) / (10+10)) * 100 = 50%.

Analyze Modifiers

To Analyze Numerical Modifiers

Numerical modifiers can be analyzed in two ways:

• Numerical Analysis: This gives numerical statistics on numerical modifiers, like the mean value, minimum or maximum value, or the median.
• Behavior Analysis: This analyzes numerical modifiers as events. Do this, for example, to analyze the number of times that the modifier 0.1 was scored, or the average duration of the numerical modifier 0.5. Do not carry out behavior analysis if you defined a modifier range instead of predefined values and scored many values. Also do not carry out behavior analysis on external data values.

To Analyze Merged Numerical Modifiers

Merging numerical modifiers in a data profile makes a nominal group, which means you cannot carry out a numerical analysis on the group to calculate mean, maximum, and minimum values. You can analyze the group as an event, for example to calculate the frequency or duration.

You get separate results for the group and for the modifiers outside the group. For example, if you have a numerical modifier Speed with predefined values 1, 2, 3, 4, 5, 6, 7, 8 and you created a group Slow containing the modifiers 1, 2, 3, a Behavior Analysis gives results for the group Slow and the modifiers 4, 5, 6, 7, 8.

To Analyze Nominal Modifiers

Nominal modifiers are analyzed in the same way as behaviors.

To Analyze Multiple Nominal Modifiers

If you have attached more than one modifier to a specific behavior, there are two options to analyze them in behavior analysis:

• In specific combinations as they were scored: For example, calculate the number of times this combination occurred: Play (behavior), Duet (modifier 1), Constructive (modifier 2), etc. To do so, select Show Combined Modifiers.
• Independently of other modifiers scored in the same event: For example, calculate the number of times that Duet occurred independent of other modifiers. To do so, clear the Show Combined Modifiers option.

Example: In a study of interactions among children, the type of play is coded with two modifier groups: Play 1 to specify whether the subjects play as a Duet or In Parallel, and Play 2 to specify whether play is Manipulative, With Rules, etc. A typical event scored is, for example, Play – Duet – With Rules.

• If you select Show Combined Modifiers, the statistics are calculated for each combination of modifiers scored. For example: Play–Duet–With Rules Frequency = 4.
• If you do not select Show Combined Modifiers, the statistics are calculated for each modifier separately, regardless of which other modifiers were scored with it. For example: Play–Duet Frequency = 3; Play–With Rules Frequency = 2.

Notes on Show Combined Modifiers

• The results obtained with Show Combined Modifiers are not necessarily the sum of the separate results. In the example above, Play–With Rules may also refer to an event where With Rules was scored with a modifier other than Duet of the Play 1 group.
• The Show Combined Modifiers option is only available in Behavior Analysis.

Customize the Results

• Customize the table layout
• Show missing values as
• Show/hide elements not scored
• Choose number of decimals
• Sort the table
• Select elements
• Set default analysis settings
• Edit statistics
• Save statistics
• Hide rows and columns

Customize the Table Layout

1. Follow the General Procedure to obtain a statistics results table.
2. Click the Layout button on the toolbar.
3. Select the element you want to view on rows, columns, and separate sheets.
4. Clear the selection for the elements you do not want to include in the table. See also Group Elements.