Abstract
Partial least squares (PLS) path modeling has been adopted as part of the statistical toolbox of many information systems (IS) scholars, particularly when dealing with survey data. Since these data are susceptible to common method variance, several statistical approaches for diagnosing and controlling for this undesirable feature have been developed. While most of these statistical techniques are only applicable to structural equation modeling (SEM), Liang, Saraf, Hu, and Xue (2007) proposed how one of these techniques can be used with PLS analysis. Since this was the first time that a method for controlling common method variance had been made available for PLS users, the method of Liang et al. quickly gained popularity in IS journals. However, recent analysis on the Liang et all approach shows that the method does neither detect nor control for common method variance. In this paper, we propose an alternative PLS marker variable approach for analyzing data contaminated with method variance and provide simulation evidence for the validity of this new approach.
Recommended Citation
Rönkkö, Mikko and Ylitalo, Jukka, "PLS marker variable approach to diagnosing and controlling for method variance" (2011). ICIS 2011 Proceedings. 8.
https://aisel.aisnet.org/icis2011/proceedings/researchmethods/8
PLS marker variable approach to diagnosing and controlling for method variance
Partial least squares (PLS) path modeling has been adopted as part of the statistical toolbox of many information systems (IS) scholars, particularly when dealing with survey data. Since these data are susceptible to common method variance, several statistical approaches for diagnosing and controlling for this undesirable feature have been developed. While most of these statistical techniques are only applicable to structural equation modeling (SEM), Liang, Saraf, Hu, and Xue (2007) proposed how one of these techniques can be used with PLS analysis. Since this was the first time that a method for controlling common method variance had been made available for PLS users, the method of Liang et al. quickly gained popularity in IS journals. However, recent analysis on the Liang et all approach shows that the method does neither detect nor control for common method variance. In this paper, we propose an alternative PLS marker variable approach for analyzing data contaminated with method variance and provide simulation evidence for the validity of this new approach.