Knowledge Base

Analyze Data

Click the + symbol next to the behaviors to expand the chart and view the modifiers.

Notes

• All the system events are states of Start-stop type (see Set up your Project in The Observer XT Help for information on this event type).
• There is usually an empty area in the chart from the observation start time (Relative time 0) to the start of the first event (Instructions in the example above). This is the time from when the E-Run starts to when the first instructions are shown on the screen. This time may vary between observations, depending on how long it takes to fill in the information in the dialog boxes that appear at the start of the experiment (Subject ID, Session Number etc.)

The E-Prime Events in the Coding Scheme

Subjects

Behaviors

The name of a behavior depends on the name of the corresponding E-object set in E-Prime (see Prepare the experiment in E-Studio).

System events and user responses are stored in separate Start-Stop behavior groups, System Behaviors and User Behaviors.

• System Behaviors contains events named Show followed by the E-object name. The Show Stimulus behavior has a modifier group attached, named Media. This group contains the names specified in the Trial List under Stimulus.
• User Behaviors contains a behavior named Answer. The Answer behavior has two modifiers attached, Response and Correct.

Nominal Modifiers

In this example, the following modifier groups have been imported, with their elements:

• Response (nominal). Contains Response: followed by the character(s) entered by the test participant.
• Media (nominal). Contains the filenames of the stimuli, or other attributes you may want to import.
• Correct (nominal). Contains Correct: followed by the correct response for the corresponding stimulus (that is set in E-Prime's Trial List).

Numerical Modifiers

If you specified in E-Prime that the answer was numerical, they are added as a numerical modifier group to the Coding Scheme. See Example 3: To import responses as numerical modifiers.

Note: You can modify the names of coding scheme elements at any time, however we advise you not to do so. If you import new data, these will have the names set in E-Prime, and added to the coding scheme. The same event type could therefore be represented by two or more different subjects, behaviors and modifiers, leading to wrong analysis results.

Analysis

The analyses in The Observer XT allow you to calculate most statistics on the stimuli and the user responses.

Nominal (Default) Responses

Nominal responses are analyzed with a Behavior Analysis. Choose Analyze > Behavior Analysis > New and click Calculate.

Example 1: Correct vs. Incorrect Responses

In the default Behavior Analysis result, each cell contains the statistics of a specific combination of behavior and modifiers. For example, the table lists the statistic Total Number for the events of type Answer and a specific combination of the modifiers Correct and Response.

In this example, the test participant gave three times a correct response: one time when Correct= 1 and Response= 1, and two times when Correct =2 and Response =2 (first two lines). In one case the test participant gave an incorrect response, when choosing 2 when the Correct response was 1 (third line). Therefore, the percentage of correct responses is 3/(3+1)= 75%.

In the table, the response "SPACE" was given by the user to start the procedure and should be ignored. "Any Modifier" is the number of events with any combination of modifiers.

Example 2: Response Time (RT)

There are two main cases:

• If the stimulus ends exactly when the user gives a response, the response time is equal to the stimulus presentation time. In this case the response time is the duration of the state event Show Stimulus.
• If the stimulus duration is independent of the user's response, follow the steps below.

1. In the Data profile, choose Select Intervals - By manual selection and define an interval from the start of the event Show Stimulus to the start of the event User Answer.
2. Choose Analyze > Behavior Analysis > New.
3. Click the Statistics button and under Intervals select Analyzed interval duration.
4. In the results table, look for the column named Analyzed interval duration.

Example 3: Response Time Grouped per Response

In this example, Response time (RT) is calculated for when subjects gave a particular response. For simplicity, the solution shown here calculates RT only when subject responses were correct.

For example:

• Correct response was "1" - User Response was "1" - RT was ...
• Correct response was "2" - User Response was "2" - RT was ...

1. Define an interval that captures the response time, as in Example 2. Choose Intervals - By manual selection and select From System Show Stimulus to User Answer. This is the least selective interval, because it captures all instances of response times, even those for incorrect responses. The following steps add more interval criteria, to select only the instances of response time when Correct =1 and Response =1, OR Correct =2 and Response =2.
2. Create two branches that point to two result containers (to create a new result container, click Result Containers - Result). One branch will select Correct = 1, Response = 1, the other Correct = 2, Response = 2. Name the result boxes R1 C1 and R2 C2, respectively.
3. In each branch, define intervals (Select intervals - By manual selection) that include a correct answer. Intervals are defined From 2.00 s before start of User Answer to 0.00 seconds after end of User Answer. However, intervals can only be defined with one modifier (either Correct or Response). Therefore, define two intervals in each branch. For the branch ending with R 1, C 1 (Correct response 1 when Response was 1), the first interval is based on Correct = 1, the second is based on Response = 1. By placing the two Interval boxes one after the other, define an interval that meets both criteria, that is, an interval for when Response 1 was Correct. Create a similar selection in the other branch, for Correct =2 and Response =2.

   Note: Choose "2 seconds before User Answer" to make sure that the time around the answer is included in the selection. Choosing 0 seconds for a point event like User Answer would make the selection invalid.

   The intervals defined in this step reduce the total number of instances of response time, because they are set after the first, less selective Interval box made in step 1. Only the time segments corresponding to the most selective criterion will be in the Result Container.

4. Use a Filter to choose which data to analyze. This removes all events in the selected intervals that are not of interest. This box should be placed at the end of the branch.
5. The Data profile is now configured. For details about selecting events and intervals for analysis, see Select Data for Analysis in The Observer XT Help.
6. Select Analyze > Visualize Data. The two charts at the top show the intervals (response times) selected for R 1, C 1. The other charts show the same for R 2, C 2. You can see that the first instance of response time is not selected because of the incorrect response given (Response = 1, Correct = 2).
7. Run Behavior Analysis and look at the statistics that refer to a specific response (this is indicated with the Result Container column header).
8. To group results, select Layout > de-select Intervals. Look at the Mean duration statistic for the mean response time.

Notes:

• This data profile is also valid when the response does not terminate the stimulus presentation.
• You can create more complex data profiles by adding branches that select instances of incorrect responses (for example: Response was "1", Correct was "2").

Numerical Responses

If you specified in E-Prime that the answer was numerical, for example to rate the stimulus, you may want to calculate numerical statistics such as the mean value or median. See Example 3: To import responses as numerical modifiers for how to specify the response as numerical. In The Observer XT, carry out a numerical analysis. Choose Analyze > Numerical Analysis > New and click the Calculate button.

By default the minimum value, maximum value and mean value of the numerical modifiers are calculated. To select other statistics, click the Statistics button on the toolbar and select the statistic under Numerical Modifiers.

For more information, see Calculate Statistics in The Observer XT Help.

Lag Sequential Analysis

With Lag Sequential Analysis you can quickly calculate how often a stimulus triggers a specific response.

Choose Analyze > Lag Sequential Analysis > New.

Example: How Often a Stimulus Produces a Correct Response

In this example, a stimulus picture is presented for a few seconds, and the test participant must enter a response. The response ends the presentation of the stimulus. For this reason a State lag sequential analysis can be carried out (a Time lag analysis is also possible).

The stimulus presentation is recorded as the behavior Show Stimulus and its modifier Media. The test participant's response and the correct response are stored as modifiers Response and Correct of the behavior Answer, respectively.

1. Click Settings and under Lag Type choose State lag.
2. Because the stimulus and the response events are events scored for different subjects (System and User), de-select Restrict lag to Subject-Behavior group combination.
3. Run analysis.

In the Lag Sequential Analysis result, locate the part of the table that contains the transitions from the behavior Show Stimulus (on the rows) to the behavior Answer (on the columns).

In this example, the stimulus BlueCar.bmp was followed two times by a correct response (Response= 2 and Correct= 2). The stimulus RedCar.bmp was followed by one correct (Response= 1 and Correct= 1) and one incorrect response (Response= 2 when Correct= 1).

For more information, see Lag Sequential Analysis in The Observer XT Help.

Advanced Settings

Use Advanced Settings to change basic settings like the port number for communication between The Observer XT PC and the E-Prime PC, and the location of the E-Prime software and experiments.

If some events are not imported correctly to The Observer XT, see Prepare the experiment in E-Studio.

Procedure

1. On the E-Prime computer, start Noldus E-Prime Server (Apps > Noldus > Noldus E-Prime Server).
2. Click Advanced Settings.

IP Address

This is the IP address of the E-Prime computer. It is shown for information purposes only, and cannot be changed. This number must be entered in the Noldus E-Prime Client software.

Port

The port number should always be the same as that set in The Observer XT. The default is 155.

E-Run Location

This is the location on the E-Prime PC where the E-Run.exe software is installed. If it has been installed on another location, specify that location here.

The default is C:\Program Files (x86)\PST\E-Prime 3.0\Program.

For E-Prime 2.0, the default location is C:\Program Files (x86)\PST\E-Prime 2.0\Program.

Experiment Text Output

This is the folder where the E-Prime output text file is stored. By default, this is the same folder as where the experiment file (*.ebs3 / *.ebs2) you are using is stored.

Notes:

• If you change the Port number after clicking Start, this has no effect on the software. Click Restart to activate communication on the new port.
• If you set the Experiment text output location in E-Prime different from that of the experiment file (*.ebs3 / *.ebs2), you must specify this new location also in Noldus E-Prime Server.
• When clicked, the Start button of Noldus E-Prime Server does not start the E-Prime experiment; it activates the E-Prime Controller to accept messages from The Observer XT.

Troubleshooting

E-Prime Error Messages

Runtime Error: "Unable to obtain a valid refresh rate"

This occurs on Windows 10 computers with Windows version 1703 or later (Windows Creators).

Please see the workaround on the following web page: https://support.pstnet.com/hc/en-us/articles/115011182147

---

Source: The Observer XT 17 - Technical Note - E-Prime