SIREN: A Hybrid CFA-EFA R Package for Controlling Acquiescence in Restricted Factorial Solutions

The **siren** package implements a two-step procedure that allows restricted (confirmatory) factor analytic (FA) solutions to be fitted in data matrices that have been previously 'cleaned' of the biasing effects of acquiescent responding (AR) by using an unrestricted (exploratory) FA specification. So, the procedure which is implemented is hybrid: i.e. (a) an unrestricted acquiescence (ACQ) factor is first fitted to the data, (b) the residual data (or covariance) matrix after the impact of ACQ has been partialled-out is obtained, and (c) a restricted FA solution is fitted to the residual matrix. Although the basic foundations of the procedure are known, it contains new methodological developments that are, all of them, implemented in the package. So, provided that fully or partially balanced scales are available, the researcher will be able to: (a) calibrate a multidimensional CFA solution which is free from AR, (b) assess the goodness of model-data fit of this solution, and (c) obtain individual score estimates in the content as well as in the ACQ factors. The functioning of the program is assessed by means of a simulation study, and illustrated with a toy example. Its usefulness is also demonstrated by using an illustrative example in the personality domain. **siren** is submitted to be a valuable tool for use in item CFA applications when AR is expected to be operating.