Hi Muhammad

Common method bias (CMB) happens when variations in responses are caused by the instrument rather than the actual predispositions of the respondents that the instrument attempts to uncover. In other words, the instrument introduces a bias, hence variances, which you will be analysing. Consequently, the results you get is contaminated by the 'noise' stemming from the biased instruments. 

One of the simplest ways to test if CMB is of concern in your study, you may want to use Harman's single factor score, in which all items (measuring latent variables) are loaded into one common factor. If the total variance for a single factor  is less than 50%, it suggests that CMB does not affect your data, hence the results. Note that, Harman's approach is to test for CMB, but not to control for CMB. 

For more detailed instructions to run Harman's single factor score, you can have a look at the attached publications. Page 42 of  'Introduction to SPSS' guides you on steps to check for CMB using Harman's single factor score.

Should you encounter potential CMB (total variance for one factor is in excess of 50%), you may want to remedy the issue based on suggestions in these articles:

Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879-903. doi: 10.1037/0021-9010.88.5.879

Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012). Sources of method bias in social science research and recommendations on how to control it. Annual Review of Psychology, 63(1), 539-569. doi: doi:10.1146/annurev-psych-120710-100452

I hope these can be helpful for you.

Saiyidi

http://dl.acm.org/citation.cfm?id=2843182

Article Introduction to SPSS

Hi Muhammad

Common method bias (CMB) happens when variations in responses are caused by the instrument rather than the actual predispositions of the respondents that the instrument attempts to uncover. In other words, the instrument introduces a bias, hence variances, which you will be analysing. Consequently, the results you get is contaminated by the 'noise' stemming from the biased instruments. 

One of the simplest ways to test if CMB is of concern in your study, you may want to use Harman's single factor score, in which all items (measuring latent variables) are loaded into one common factor. If the total variance for a single factor  is less than 50%, it suggests that CMB does not affect your data, hence the results. Note that, Harman's approach is to test for CMB, but not to control for CMB. 

For more detailed instructions to run Harman's single factor score, you can have a look at the attached publications. Page 42 of  'Introduction to SPSS' guides you on steps to check for CMB using Harman's single factor score.

Should you encounter potential CMB (total variance for one factor is in excess of 50%), you may want to remedy the issue based on suggestions in these articles:

Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879-903. doi: 10.1037/0021-9010.88.5.879

Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012). Sources of method bias in social science research and recommendations on how to control it. Annual Review of Psychology, 63(1), 539-569. doi: doi:10.1146/annurev-psych-120710-100452

I hope these can be helpful for you.

Saiyidi

http://dl.acm.org/citation.cfm?id=2843182

Article Introduction to SPSS

Thanks Saiyidi for your response. It was indeed very helpful. I will surely look into the references you have suggested. 

Yeah use Podsakoff.

You can use the Harman's one factor test or the CFA based test as a more advanced version of it on SEM.

For example:

https://www.youtube.com/watch?v=w7zZCBlRXog

Best of luck

In addition to Podsakoff (2003) you may also like to read this panel discussion: Brannick, M. T., Chan, D., Conway, J. M., Lance, C. E., & Spector, P. E. (2010). What is method variance and how can we cope with it? A panel discussion. Organizational Research Methods.

Hope this will be helpful.

Arif ul Haq

Two comments: (1) We cannot automatically assume that just because data originates from the same type of respondents, so, there will automatically be bias in that data.

Hence, comments of reviewers about CMB are usually over-stated. See this recent paper

Fuller, C. M., Simmering, M. J., Atinc, G., Atinc, Y., & Babin, B. J. 2016. Common methods variance detection in business research. Journal of Business Research, 69(8): 3192-3198.

(2) Harman's one factor-test is not a good way of measuring common method bias. CFA is a more sophisticated way.

You can learn how to conduct this test using the video shared by Rohail.

Thanks every one for your very valuable comments and suggestions. 

I agree with the advise of Saiyidi Mat Roni. However  the decision to use Harman's single factor score, in which all items (measuring latent variables) are loaded into one common factor is not pschometrically convincing.

Harman's single factor test is diagnostic. From a prescriptive perspective, it is better to  measure the dependent and criterion variables using different methods. For example if the dependent variable is innovation outcomes, you may want to look at patent filings rather than include questions in the survey itself. Patent filings can be viewed as a reliable "other" source.

Many thanks for your responses... It's helpful and constructive.

Harman's single factor test become obsolete. so we should use CFA common factor

i recieved a comment that authors should test common method bias since the research used self-reported data.what should i do

What is meant by Common Method Bias? How do we test and control it?

You can also refer to the following RG links:

https://www.researchgate.net/post/Does_anyone_know_what_is_the_purpose_of_common_method_bias_in_structural_equation_modeling

https://www.researchgate.net/post/Can_anyone_comment_on_the_Common_Method_Bias_testing_in_structural_model

https://www.researchgate.net/post/How_can_I_detect_Common_method_bias_in_PLS-SEM

https://www.researchgate.net/post/common_method_bias

I found Prof. Gaskin's video very useful. If you are using Amos 24, then there is a plugin for the same. Check this-

https://www.youtube.com/watch?v=Y7Le5Vb7_jg

The reviewer has a valid point.

It might be worth mentioning that the most popular approaches to measuring CMV vastly underestimate the extent of such bias, especially for the popular psychological theories. Here are two papers that show that one theory derived from two of the most popular theories in psychology is little other than CMV:

Sharma, R., et al. (2009). "Estimating the effect of common method variance: The method—method pair technique with an illustration from TAM Research." MIS quarterly: 473-490.

Gefen, D. and K. R. Larsen (2017). "Controlling for Lexical Closeness in Survey Research: A Demonstration on the Technology Acceptance Model " Journal of the Association for Information Systems 18(10): 727-757.

Worth reading Articles Thanks Larsen

If you use PLS, this paper is useful.

"The occurrence of a VIF greater than 3.3 is proposed as an indication of pathological collinearity, and also as an indication that a model may be contaminated by common method bias. Therefore, if all VIFs resulting from a full collinearity test are equal to or lower than 3.3, the model can be considered free of common method bias."

Kock, N. (2015). Common method bias in PLS-SEM: A full collinearity assessment approach. International Journal of e-Collaboration (IJeC), 11(4), 1-10.

See this video, how to do

https://www.youtube.com/watch?v=J7eeu4O80_M

Dear @Tanikan Pipitwanichakarn

That is a very useful manuscript about collinearity. Recommended reading.

Common method bias (CMB) happens when responses systematically vary because of the use of a common scaling approach on measures derived from a single data source.

Four remedies to resolve the problem of CMB?

1. Using other sources of information for some of the key measures (ex-ante)

2. Mixing the order of the questions to using different scale types (ex-ante)

3. Complicated specifications of regression models. Specifically, respondents are unlikely to be guided by a cognitive map that includes difficult-to- visualize interaction and non-linear effects. This is less likely the more complicated the model (ex-post)

4. Statistical methods to detect and control for any possible CMB (ex-post)

To resolve the problem of CMB

1. Harman’s single factor test

2. Common latent factor

3. Marker variable

For more about CMB please refer to the following articles:

• Rönkkö, M., & Ylitalo, J. (2011). PLS marker variable approach to diagnosing and controlling for method variance.
• Jakobsen, M., & Jensen, R. (2015). Common method bias in public management studies. International Public Management Journal, 18(1), 3-30.
• Examining the Impact and Detection of the "Urban Legend" of Common Method Bias
• Fuller, C. M., Simmering, M. J., Atinc, G., Atinc, Y., & Babin, B. J. (2016). Common methods variance detection in business research. Journal of Business Research, 69(8), 3192-3198.

You can read Podsakoff et al. 2003,2011 on this topic, and also you can go through all my papers since I reported CMV in all of my papers

Dear Muhammad,

As previous respondents have identified, the Harman's Single Factor test is the obvious and easy way to undertake a crude test for an indication of the presence or absence of CMV/CMB - it is not a proof.

A weaker test, which many report in the literature, is to undertake an exploratory factor analysis with the number of factors set at the number of constructs in your full model. If the factor that accounts for the largest proportion of variance has a value less than the benchmark value of 0.5, this is also considered to be a persuasive, yet crude, test for CMV/CMB.

If you are using a CBSEM, you can use the Marker Variable technique to control for this bias, if present, provided you have an appropriate and unused indicator (see Lindell and Whitney (2001) Accounting for Common Method Variance in Cross-Sectional Research Designs, Journal of Applied Psychology, 86(1):114-121).

As this method requires access to the individual indicators, it is not appropriate for a PLS-SEM. Approximate alternatives have been proposed but none are currently accepted in the literature.

Good luck with your paper,

James

Helpful papers. Tq dear Kai R. Larsen

CMV and the resulting common method bias are a consequence of data collection method variance associated with your collection instrument and/or method of collection. When this is large, it can make a major contribution to the measured effect. The seminal paper by Podsakoff et al. ((2003) Journal of Applied Psychology, 88(5): 879–890) explains this in detail and suggests various remedies. There is a school of thought that believes this is usually a small effect but it is necessary to, at least, check for the possibility of this being present. The crude way to do this is to undertake a one factor analysis with all the interval and ratio data variables in your model, present. Harman suggests that this single factor should not account for 50% or more of the variance present - this is a test for the degree of common variance. Some authors also look to no single factor accounting for more than 50%. If this benchmark is exceeded, there is an indication of a problem due to the data collection approach. If you are fitting a structural equation model, and the sample size is large enough to use a covarianced model, you can use a marker variable approach to control for this but only if you have suitable additional data which usually requires previous design in your data collection instrument.

Hi Muhammad. The reviewer may be inaccurate in his/her language when discussing CMB as demand characteristics (which could be part of CMB). Here are some definitions:

"Measurement items should be designed to eliminate item ambiguity, demand characteristics, positive/negative affectivity, and social desirability to the extent possible (Podsakoff et al. 2003)" {Wakefield, 2008 #5259, p. 452}

"Caution must still be taken to counter concerns that students may be subject to demand characteristics that make them more likely to provide faulty data to "help" the researcher" {Compeau, 2012 #5260, p. 1102}

"Response bias Demand characteristics bias Description Arises when respondents figure out the purpose of the study, which can influence the participants and their responses (Orne, 1962)." {Srivastava, 2015 #5261, p. 395}.

Methods folks have come up with ways to evaluate CMB, which in my humble opinion do not actually work, but will get you through peer review. You should read the papers listed on this page and then get back to us for more specific questions.

CMB can be detected by conducting Harman's single-factor analysis. It means that you can run an unrotated factor analysis with your data to see if one component loads more than 50% of the variance, it can happen due to CMB.

However, I recommend the following publication for the full picture of the CMB in PLS-SEM.

Kock, N., 2015. Common method bias in PLS-SEM: A full collinearity assessment approach. International Journal of e-Collaboration (IJeC), 11(4), pp.1-10.

There are lots of reason why a single factor could account for 50% or more of the variance, including having successfully captured the underlying concept behind the measured variables.

Dear colleagues,

First of all, very traditional approaches such as the Harman test, honestly, are useless for all kind of methods. Already in 2003, Podsakoff et al. (2003, p. 889) explicitly pointed out the following about Harman's single factor-test: “Therefore, despite the fact this procedure is widely used, we do not believe it is a useful remedy to deal with the problem and turn our attention to other statistical remedies that we feel are better suited for this purpose.” More recently, MacKenzie and Podsakoff (2012, pp. 543-544) reaffirmed themselves by describing techniques such as Harman's test in the following terms: "the weaker, conceptually flawed statistical control procedures" (pp. 543-544). Finally, in a research that I have participated with researchers of the Louisiana State University, we have finished verifying that this test neither detects nor corrects the common method bias (CMB) (Schwarz et al., 2017).

Secondly, a test that has had enough predicament to detect CMB in the PLS field was the adaptation made by Liang et al (2007) of the famous unmeasured latent marker construct (ULMC) approach to the PLS field. As there was no other procedure, this test also had its development and is still widely quoted today. However, as Chin et al. (2012) demonstrated, this method neither detects nor corrects CMB.

Third, the most robust method in PLS is the Measured Latent Marker Variable Approach described by Chin et al. (2013). The only problem with this method is necessary to introduce previously this marker variable in the questionnaire. If you have not done so, it cannot be applied. In that case, you could apply that approach suggested by Ned Kock (2012, 2015, 2017).

In conclusion, from my point of view, the issue of the CMB must be mainly dealt from a priori perspective at the time of the design and application of the questionnaire. I recommend to read Podsakoff and MacKenzie (2012) and MacKenzie and Posakoff (2012) for this point.

References:

Chin, W. W., Thatcher, J. B., Wright, R. T., & Steel, D. (2013). Controlling for Common Method Variance in PLS Analysis: The Measured Latent Marker Variable Approach. In H. Abdi, W. W. Chin, V. Esposito Vinzi, G. Russolillo, & L. Trinchera (Eds.), (Vol. 56, pp. 231–239). New York, NY: Springer New York. http://doi.org/10.1007/978-1-4614-8283-3_16

Chin, W. W., Thatcher, J. B., & Wright, R. T. (2012). Assessing Common Method Bias: Problems with the ULMC Technique. MIS Quarterly, 36(3), 1003-A11.

Kock, N. (2015). Common method bias in PLS-SEM: A full collinearity assessment approach. International Journal of E-Collaboration, 11(4), 1–10. http://doi.org/10.4018/ijec.2015100101

Kock, N. (2017). Common Method Bias: A Full Collinearity Assessment Method for PLS-SEM. In Partial Least Squares Path Modeling (pp. 245–257). Cham: Springer International Publishing. http://doi.org/10.1007/978-3-319-64069-3_11

Kock, N., & Lynn, G. S. (2012). Lateral Collinearity and Misleading Results in Variance-Based SEM : An Illustration and Recommendations. Journal of the Association for Information Systems, 13(7), 546–580.

Liang, H., Saraf, N., Hu, Q., and Xue, Y. 2007. “Assimilation of Enterprise Systems: The Effect of Institutional Pressures and the Mediating Role of Top Management,” MIS Quarterly (31:1), pp. 59 - 87.

MacKenzie, S. B., & Podsakoff, P. M. (2012). Common Method Bias in Marketing: Causes, Mechanisms, and Procedural Remedies. Journal of Retailing, 88(4), 542–555. http://doi.org/10.1016/j.jretai.2012.08.001

Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: a critical review of the literature and recommended remedies. The Journal of Applied Psychology, 88(5), 879–903. http://doi.org/10.1037/0021-9010.88.5.879

Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012). Sources of Method Bias in Social Science Research and Recommendations on How to Control It. Annual Review of Psychology, 63, 539–569. http://doi.org/10.1146/annurev-psych-120710-100452

Schwarz, A., Rizzuto, T., Carraher-Wolverton, C., Roldán, J. L., & Barrera-Barrera, R. (2017). Examining the Impact and Detection of the “Urban Legend” of Common Method Bias. ACM SIGMIS Database: The DATABASE for Advances in Information Systems, 48(1), 93–119. https://doi.org/10.1145/3051473.3051479