Knowledge Base

Calculate Statistics

Calculate Statistics: Procedure

Aim

To calculate descriptive statistics of one or more dependent variables.

Prerequisites

• You have smoothed data (optional).
• You have selected data in a Data profile (optional).

To Calculate Statistics

1. Choose Analysis > Analysis Profile > Open (to open an existing profile) or New (enter a name for the new profile), then click OK.
2. Choose the dependent variable. Click the Add button next to the dependent variable you want to use for analysis. If you do not see that variable name, click the More button. In most cases Distance Moved and Velocity are already selected.
3. In the dependent variable-specific tab, select the properties of the dependent variable. For details on dependent variables, see Dependent Variables in Detail.
4. Choose the Trial statistics. Click the Trial Statistics tab, and choose the statistics you want to calculate for each trial.
5. Choose the Group statistics. Click the Group Statistics tab, choose the statistics you want to calculate as a summary for a group. For example, if you want to calculate the average and standard error of the Total distance moved for Control and Treated subjects, choose Distance Moved as a dependent variable, Total as a Trial statistic, and Mean and Standard Error as Group statistics.
6. Repeat steps 2 to 5 to add more dependent variables.
7. Choose Analysis > Results > Statistics and Charts, then click the Calculate button on the toolbar.
8. The Statistics and Charts table appears on the screen.

Notes

• Some dependent variables are available if you have one or more of the following add-on modules: Social Interaction, Rat/Mouse Behavior Recognition, Trial and Hardware Control.
• You can only select specific body points if your experiment is set to Center-point, nose-point and tail-base detection.
• You can change the properties or delete a dependent variable already selected. You can also add multiple instances of a dependent variable with different settings. See Edit the Analysis Profile.
• You can also calculate statistics in batch mode. See Batch Calculation.

Statistics Available

Statistics are a summary of the values of the dependent variable calculated for all the samples in your tracks (or the segments you have selected in the active Data profile). For each dependent variable you select in the Analysis profile, you can choose Trial Statistics and Group Statistics. Click on the corresponding tab to select the statistics you want to have in the results table.

List of Trial Statistics

Aim

Trial Statistics give you summary statistics of individual trials. If a trial includes two or more tracks (subjects), the statistics are by default shown for each subject separately.

The Data profile filters tracks and/or intervals within tracks. If the Data profile contains groups of tracks, these groups are ignored in the results. To view results per group, choose Results > Group Statistics and Charts.

Cumulative Duration

The cumulative duration of the state variable in the whole track, or in the track segment selected. For example, the time that the animal spent in a zone (for the variable In Zone), or the time that the subject was in the state Mobile. When a state does not occur at all in a track or track segment, the value of Cumulative Duration is zero.

Cumulative Duration (%)

The Cumulative Duration divided by the following, expressed as a percentage:

• If you do not define time bins: The duration of the track (this corresponds to the Recording Duration independent variable, not Trial Duration). This is also the case when you define nesting intervals or results per zone.
• If you define time bins: The duration of a time bin defined.

Cumulative Duration Within Nesting (%)

The Cumulative Duration divided by the following, expressed as a percentage:

• If you calculate results per zone: The total time the animal is in the zone.
• If you defined track segments with Nesting: The total duration of the track segments defined.
• If you defined time bins: The duration of a time bin defined.
• If you combine any of the three above: The duration of the time period shared by all selection criteria (the time periods of each criterion are combined with AND logic).

Frequency

The total number of occurrences of a state (only for state variables).

Latency to First

The time from the start of the track to the first occurrence of the behavior (for example, the first time in the track that the animal is in a zone). Latency is always calculated from the start of the track, even when you define time bins and nesting intervals.

Latency to Last

The time from the start of the track to the last occurrence of the behavior. Latency is always calculated from the start of the track, even when you define time bins and nesting intervals.

Notes About Latency to First/Last

• Latency to First and Latency to Last are calculated from the start of the track. This is the time when EthoVision XT collects the first sample (missing or not). If the animal is not yet in the arena and tracking has started, EthoVision XT collects a number of missing samples. The time that the animal is missing is included in Latency to First/Last, which may overestimate the true latency. To prevent this, make sure that the Trial Control rule includes a Start Track condition so that tracking only starts when the animal is first found in the arena. To calculate the time from one event to another event in the trial, select a Free Interval in the Analysis profile.
• When two or more animals are tested simultaneously in the same arena, the meaning of Latency to First/Last depends on how you set this condition. Consider the following scenarios when you want to calculate the Latency of "In Zone A" for two subjects released at different times:
  • If you set the Trial Control to start tracking when any subject is found in the arena, tracking starts when the first animal is released. Latency is calculated from that point also for the second animal, which enters the arena later, meaning that Latency for this animal will be overestimated.
  • If you set the Trial Control to start tracking when all subjects are found in the arena, tracking starts when both animals are found. Latency is calculated from this point, so Latency for the second animal is correct. However, for the first-released animal, the time it was in the arena before the second was released is not counted in Latency, which may underestimate the result. For this reason, try to release the animals at the same time, or one immediately after the other.

Mean

• For linear variables like Distance Moved, Velocity, Distance to Zone, or Distance Between Subjects, and the variables Turn Angle, Angular Velocity and Meander, the arithmetic mean is used: the sum of the per-sample values divided by the number of samples.
• For the angular variables Heading, Head Direction, and Heading to Point, the mean is calculated according to the circular mean formula, where the circular mean value can also be represented by the angle of the vector resultant R obtained by summing up all the vectors that represent the per-sample angle values.
• For state variables like In Zone, Movement and behaviors scored manually or automatically recognized, the sum of the duration of occurrences of the state divided by the number of occurrences.

Minimum

The lowest value of the variable. For state variables, this is the minimum duration of the instances of a state. When the state does not occur at all in a track or track segment, the value of Minimum is zero.

Maximum

The highest value of the variable. For state variables, this is the maximum duration of the instances of a state. When the state does not occur at all in a track or track segment, the value of Maximum is zero.

Number of Samples

The total number of valid values of that variable (only for numerical variables like Distance Moved).

Standard Error

The standard deviation divided by the square root of the number of samples. This is also the case for angular variables Heading, Head Direction, and Heading to Point.

Standard Deviation

For most variables, including Turn Angle, Angular Velocity and Meander, the sample standard deviation of the mean, where x is the individual variable value, x-bar is the mean of the variable, and N is the number of samples.