Friday, April 7, 2017
PLS Applications Symposium; 5 - 7 April 2017; Laredo, Texas
(Abstract submissions accepted until 10 February 2017)
*** Only abstracts are needed for the submissions ***
The partial least squares (PLS) method has increasingly been used in a variety of fields of research and practice, particularly in the context of PLS-based structural equation modeling (SEM). The focus of this Symposium is on the application of PLS-based methods, from a multidisciplinary perspective. For types of submissions, deadlines, and other details, please visit the Symposium’s web site:
*** Workshop on PLS-SEM ***
On 5 April 2017 a full-day workshop on PLS-SEM will be conducted by Dr. Ned Kock, using the software WarpPLS. Dr. Kock is the original developer of this software, which is one of the leading PLS-SEM tools today; used by thousands of researchers from a wide variety of disciplines, and from many different countries. This workshop will be hands-on and interactive, and will have two parts: (a) basic PLS-SEM issues, conducted in the morning (9 am - 12 noon); and (b) intermediate and advanced PLS-SEM issues, conducted in the afternoon (2 pm - 5 pm). Participants may attend either one, or both of the two parts.
The following topics, among others, will be covered - Running a Full PLS-SEM Analysis - Conducting a Moderating Effects Analysis - Viewing Moderating Effects via 3D and 2D Graphs - Creating and Using Second Order Latent Variables - Viewing Indirect and Total Effects - Viewing Skewness and Kurtosis of Manifest and Latent Variables - Conducting a Multi-group Analysis with Range Restriction - Viewing Nonlinear Relationships - Conducting a Factor-Based PLS-SEM Analysis - Viewing and Changing Missing Data Imputation Settings - Isolating Mediating Effects - Identifying and Dealing with Outliers - Solving Indicator Problems - Solving Collinearity Problems.
Tuesday, March 7, 2017
How do we interpret the results of a model with an endogenous dichotomous variable? Let us use the model below to illustrate the answer to this question. In this model we have one endogenous dichotomous variable “Success” that is significantly caused in a direct way by two predictors: “Projmgt” and “JSat”. The direct effect of a third predictor, namely "ECollab", is relatively small and borderline significant.
Let us assume that the unit of analysis is a team of people. The variable “Success” is coded as 0 or 1, meaning that a team is either successful or not. After standardization, the 0 and 1 will be converted into a negative and a positive number. The standardized version of the variable “Success” will have a mean of zero and a standard deviation of 1.
One way to interpret the results is the following. The probability that a team will be successful (i.e., that “Success” > 0) is significantly affected by increases in the variables “Projmgt” and “JSat”.
In version 6.0 of WarpPLS, users will be able to calculate conditional probabilities as shown below, without having to resort to transformations based on assumed underlying functions, such as those performed by logistic regression. In this screen shot, only latent variables are used, and they are all assumed to be standardized.
In the screen shot above, we can see that the probability that a team will be successful (i.e., that “Success” > 0), if “Projmgt” > 1 and “JSat” > 1, is 52.2 percent. Stated differently, if “Projmgt” and “JSat” are high (greater than 1 standard deviation above the mean), then the probability of success is slightly greater than chance.
A probability of 52.2 percent is not that high. The reason why it is not higher, in the context of the conditional probabilistic query above, is that we are not including the variable "ECollab" in the mix. Still, it does not seem like “Projmgt” and “JSat” being high are sufficient conditions for success, although they may be necessary conditions.
Consider a different set of conditional probabilities. If a team is successful (i.e., if “Success” > 0), what is the probability that “Projmgt” and “JSat” are low for that team. The answer, shown in the screen below, is 1.3 percent. That is a very low probability, suggesting that “Projmgt” and “JSat” matter as necessary but not sufficient elements for success.
These are among the conditional probabilistic queries that users will be able to make in version 6.0 of WarpPLS, which should be released in a few months. Bayes’ theorem is used to produce the answers to the queries.
Tuesday, September 13, 2016
WarpPLS users can assess measurement invariance in PLS-SEM analyses in a way analogous to a multi-group analysis. That is, WarpPLS users can compare pairs of measurement models to ascertain equivalence, using one of the multi-group comparison techniques building on the pooled and Satterthwaite standard error methods discussed in the article below. By doing so, they will ensure that any observed between-group differences in structural model coefficients, particularly in path coefficients, are not due to measurement model differences.
Kock, N. (2014). Advanced mediating effects tests, multi-group analyses, and measurement model assessments in PLS-based SEM. International Journal of e-Collaboration, 10(3), 1-13.
For measurement invariance assessment, the techniques discussed in the article should be employed with weights and/or loadings. While with path coefficients researchers may be interested in finding statistically significant differences, with weights/loadings the opposite is typically the case – they will want to ensure that differences are not statistically significant. The reason is that significant differences between path coefficients can be artificially induced by significant differences between weights/loadings in different models.
A spreadsheet with formulas for conducting a multi-group analysis building on the pooled and Satterthwaite standard error methods is available from WarpPLS.com, under “Resources”. As indicated in the article linked above, this same spreadsheet can be used in the assessment of measurement invariance in PLS-SEM analyses.
Nonlinear analyses employing the software WarpPLS allow for the identification of linear segments emerging from a nonlinear analysis, but without the need to generate subsamples. A new article is available demonstrating the advantages of nonlinear over data segmentation analyses. These include a larger overall sample size for calculation of P values, and the ability to uncover very high segment-specific path coefficients. Its reference, abstract, and link to full text are available below.
Kock, N. (2016). Advantages of nonlinear over segmentation analyses in path models. International Journal of e-Collaboration, 12(4), 1-6.
The recent availability of software tools for nonlinear path analyses, such as WarpPLS, enables e-collaboration researchers to take nonlinearity into consideration when estimating coefficients of association among linked variables. Nonlinear path analyses can be applied to models with or without latent variables, and provide advantages over data segmentation analyses, including those employing finite mixture segmentation techniques (a.k.a. FIMIX). The latter assume that data can be successfully segmented into subsamples, which are then analyzed with linear algorithms. Nonlinear analyses employing WarpPLS also allow for the identification of linear segments mirroring underlying nonlinear relationships, but without the need to generate subsamples. We demonstrate the advantages of nonlinear over data segmentation analyses.
Among other things this article shows that identification of linear segments emerging from a nonlinear analysis with WarpPLS allows for: (a) a larger overall sample size for calculation of P values, which enables researchers to uncover actual segment-specific effects that could otherwise be rendered non-significant due to a combination of underestimated path coefficients and small subsample sizes; and (b) the ability to uncover very high segment-specific path coefficients, which could otherwise be grossly underestimated.
Thursday, September 1, 2016
An article is now available explaining how WarpPLS users can test hypotheses based on confidence intervals, contrasting that approach with the one employing P values. A variation of the latter approach, employing T ratios, is also briefly discussed. Below are the reference, link to PDF file, and abstract for the article.
Kock, N. (2016). Hypothesis testing with confidence intervals and P values in PLS-SEM. International Journal of e-Collaboration, 12(3), 1-6.
E-collaboration researchers usually employ P values for hypothesis testing, a common practice in a variety of other fields. This is also customary in many methodological contexts, such as analyses of path models with or without latent variables, as well as simpler tests that can be seen as special cases of these (e.g., comparisons of means). We discuss here how a researcher can use another major approach for hypothesis testing, the one building on confidence intervals. We contrast this approach with the one employing P values through the analysis of a simulated dataset, created based on a model grounded on past theory and empirical research. The model refers to social networking site use at work and its impact on job performance. The results of our analyses suggest that tests employing confidence intervals and P values are likely to lead to very similar outcomes in terms of acceptance or rejection of hypotheses.
On Table 1 in the article, each T ratio and confidence interval limits (lower and upper) are calculated through the formulas included below. Normally a hypothesis will not be supported if the confidence interval includes the number 0 (zero).
T ratio = (path coefficient) / (standard error).
Lower confidence interval = (path coefficient) - 1.96 * (standard error).
Upper confidence interval = (path coefficient) + 1.96 * (standard error).
Here is a quick note to technical readers. The P values reported in Table 1 in the article are calculated based on the T ratios using the incomplete beta function, which does not assume that the T distribution is exactly normal. In reality, T distributions have heavier tails than normal distributions, with the difference becoming less noticeable as sample sizes increase.
Wednesday, June 15, 2016
Among the many innovative features of WarpPLS are those that deal with identification of Simpson’s paradox and modeling of nonlinear relationships. A new article discussing various issues that are important for the understanding of the usefulness of these features is now available. Its reference, abstract, and link to full text are available below.
Kock, N., & Gaskins, L. (2016). Simpson’s paradox, moderation, and the emergence of quadratic relationships in path models: An information systems illustration. International Journal of Applied Nonlinear Science, 2(3), 200-234.
While Simpson’s paradox is well-known to statisticians, it seems to have been largely neglected in many applied fields of research, including the field of information systems. This is problematic because of the strange nature of the phenomenon, the wrong conclusions and decisions to which it may lead, and its likely frequency. We discuss Simpson’s paradox and interpret it from the perspective of path models with or without latent variables. We define it mathematically and argue that it arises from incorrect model specification. We also show how models can be correctly specified so that they are free from Simpson’s paradox. In the process of doing so, we show that Simpson’s paradox may be a marker of two types of co-existing relationships that have been attracting increasing interest from information systems researchers, namely moderation and quadratic relationships.
Among other things this article shows that: (a) Simpson’s paradox may be caused by model misspecification, and thus can in some cases be fixed by proper model specification; (b) a type of model misspecification that may cause Simpson’s paradox involves missing a moderation relationship that exists at the population level; (c) Simpson’s paradox may actually be a marker of nonlinear relationships of the quadratic type, which are induced by moderation; and (d) there is a duality involving moderation and quadratic relationships, which requires separate and targeted analyses for their proper understanding.
Saturday, June 11, 2016
Recently an interview was conducted for the 3rd Conference on Information Systems in Latin America. In it, Dr. Ned Kock was interviewed by Dr. Alexandre Graeml. The topics covered include: structural equation modeling (SEM), partial least squares (PLS) and related techniques, PLS-based SEM, covariance-based SEM, factors versus composites, nonlinear analyses, and WarpPLS.
WarpPLS and its application to research in business and information systems
The link below is for the Conference’s web site.
ISLA 2016 - Information Systems in Latin America Conference