Psychological Test Toolbox
  Full example controlling Acquiescence and Social Desirability


Full example controlling Acquiescence and Social Desirability


Read Data

When the application is successfully initialized, you have to selected “Data” tab. Once in this window, you have to click “Read Data” button. Also you can import data from File tab and selecting “Open Data”. A new window will be opened, displaying some options:

First of all you’ve to select the data file containing data to be analyzed. You can provide data from raw data file or from a dispersion matrix (both options from .dat file in ASCII code or excel file). When you select the file by browse button, it automatically display the number of participants and items if raw data is provided or only the number of items if you provide a dispersion matrix. If data contains missing values, you’ve to assign a unique value to these (for example: 999) and specify in this window the option “The data file contains missing values” and define the previously determined value.

Additionally, you can provide a file containing the item labels (.txt file, on every row define the label from one item) and semi-specified rotation target matrix (.dat file in ASCII code). Both of these optional files have to be consistent in the number of items from the data file. Additionally, if you’ve imported data from .xls or .xlsx file which contains headers, you can use it as labels.

Once everything is well defined, click “OK” button in order to import the provided data. If some inconsistency is found in the process, it’s possible that some error message shows up.

In any case, the number of participants and variables must been defined, even when a dispersion matrix is provided, in order to adjust some analysis by the original sample size.

Dataset analyzed

The example data was extracted from a sample of OPERAS (Vigil-Colet, Morales-Vives, Camps, Tous & Lorenzo-Seva, 2013) responders. Only includes the items related to extraversion, agreeableness and social desirability. More precisely, this sample includes 599 participants and contains 19 items: 1 dummy item, 7 extraversion related items, 7 agreeableness related items and also 4 social desirability markers. The response format is 5-point Likert, and we will consider a multidimensional solution (Agreeableness, Extraversion and Social Desirability).

Compute Descriptive Statistics

Once back at the main window, you can see the number of participants and items of the selected data and you can also see all the items in the included listbox. If you want to exclude some items from the analysis, you can do it by pressing the “>” button, and if you want to re-include some, use “<”. Also, you can check your data on the next tab “View data”. In this example we excluded item number 1 because it is a dummy item, and the content of this item is irrelevant.

In the next tab “Descriptive Statistics” you can select which type of dispersion matrix will be used in the following analysis. Additionally, you can select if parallel analysis will be computed checking the “Optimal Implementation of Parallel Analysis (PA)” option, and if you check it, it’s possible to select the option of plotting a graphic of the percentage of explained variance of the real data and the generated with Parallel Analysis.

When you’re finished configuring data, one optional step is computing now the descriptive statistics by clicking “Compute”. It is optional because in the final analysis, when the factor analysis will be computed, the descriptive statistics will also be computed. The time required to perform the analysis is very variable, depending on the selected options and the provided data. Once the analysis is complete, you can see the output in the subwindow, launch the output in a new window, save the output in plaintext (.txt), latex file (.tex) or Rich Text Format (.rtf).

Controlling Social Desirability

The first response bias that we will configure is Social Desirability. First of all, go to Social Desirability tab and enable this function by checking the “Enable Social Desirability control function” option. When this option is enabled you will not be able to modify the excluded or included items in “Data” tab.

In this tab you can see the number of previously excluded items, the number of content items and the number of Social Desirability markers. You’ve to select at least 3 SD items by clicking “>” button. In this example, the SD items are: item number 3, 6, 10 and 14.

Controlling Acquiescence

Now it is time to configure the other bias, so you have to select Acquiescence tab and enable Acquiescence control function. When this option is enabled you will not be able to modify the excluded or included items in “Data” tab and/or in the SD items.

This tab will display the content items defined in the Social Desirability tab. In order to control properly the Acquiescence impact, each factor have to be formed by a balanced pair number of items, half positively worded, and the other half negatively worded. If there are some factor defined by a different number of positive worded items than negative worded, it can be solved by selecting as unbalanced items the extra items. In the example data, there are 14 content items, 7 measuring one factor, which is Extraversion, and the other 7 measuring Agreeableness. In both cases there are 4 positively worded items and only 3 negatively worded, so, in order to obtain a balanced set of items, we have to select one positively worded item from each scale and move them to the unbalanced items column by pressing “>” button. In our example we select items number 2 and 7.

If some factor have an odd number of items, we recommend doing one Exploratory Factor Analysis without selecting any unbalanced item and check the Acquiescence factor loadings searching for the item with lower Acquiescence loading. When you have detected which item is the less influenced by Acquiescence bias, we recommend to use it as unbalanced on the next analysis.

Computing Factor Analysis

Once you are done configuring the items, go to Response bias tab. On this window you will see a summary of current data configuration, including which items are excluded from the analysis and which are content items. You have to define the number of content factors and if you want to compute factor scores for each participant. You are able to select the option “Compute all possible bias analysis combinations”, which it can be useful if you want to appreciate the bias control effect. You can also select the factor rotation that best suits:

  • Orthogonal: Varimax method.
  • Oblique: Promin method.
  • Semi-specified oblique target rotation (if you’ve provided it).
  • Semi-specified orthogonal target rotation (if you’ve provided it).

When you’re ready, press Compute button. Once the analysis is complete, you can see the output in the subwindow, launch the output in a new window, save the output in plaintext (.txt), latex file (.tex), Rich Text Format (.rtf) and save factorial scores if you have request it previously.

Additionally, you’re able to save the current analysis status (meanwhile some data is available) at any time in a .mat file, in order to load it whenever you want and resume the analysis or reanalyze with some new configuration.