LISREL, AMOS or PLS does not deliver what you need for your Ph.D. studies?

Discover how researchers explore nonlinearities and interactions in their cause-effect models and innovate their research fields with exciting new discoveries.

By Frank Buckler, PhD (Cologne, Germany) - PDF Version Printable version

Trial

Why you should read this article?

How will you feel when you discover something really exciting in your research, unlike thousands of other researchers before you? Won't you be elated? Think of how you'll feel when your (doctoral) supervisor is astounded by your latest breakthrough cause-effect insight discovered using NEUSREL! You realize that such discoveries were simply impossible before and how fortunate you are to have access to such a powerful and easy to use causal modeling suite.

As you read this article, you'll learn how to make your thesis/dissertation stand out impressively from "the crowd." 


Why causal analysis methods like LISREL or PLS are now of limited help

Structural Equation Modeling (SEM) methods represent a major improvement over classical statistical methods for multivariate analysis. They are designed for testing theoretically supported linear and additive causal models. For related applications they deliver excellent results. SEM methods will undoubtedly play an increasingly important role in the social sciences into the future.

There exists a serious problem with SEM methods however, when their application assumptions are not met. In practice, it is usually the exception rather than the rule that the existing theory indicates clearly to us which paths in the model we should select and keep. An open secret these days is that, using such tools as LISREL and PLS, causal models of one's data are often constructed a posteriori following fairly extensive exploratory analyses. This type of "pruning process" is the result of an extensive and selective literature search. Finally, your results are represented in the form of a priori hypotheses about your data. This confirmatory veneer underlying many LISREL/PLS/AMOS studies  conducted over the past few decades has resulted in a loss of trust and confidence in quantitative research using linear SEM methods.

Furthermore, classic causal analysis methods not only assume that all relationships in the data are linear, but that they are also independent of one another (i.e. that no moderating effects exist). Even considering more elaborate experimental methods, the main challenge remains: if one cannot describe the relationships among variables in an a priori manner, then a new method is sorely needed to help researchers learn more about the causal relationships in their data. 
Professor Hennig-Thurau and Dr. Buckler took a deeper look at four arbitrarily chosen datasets from articles published in two distinguished scientific journals: The Journal of Marketing Research and The Journal of Marketing. In each study we found clear indications of the existence of previously unknown nonlinear data relationships and interactions. If the world's leading marketing researchers are not currently deploying state-of-the-art causal analysis methods (NEUSREL), how can we expect doctoral students to do so? 

 

If problems are so obvious, why has nobody developed a solution?

The short answer is: The solution is not obvious because ...

  • Linear SEM methods are simply unsuitable for solving exploratory problems.
  • The SEM research community is dominated by a confirmatory research approach. Many researchers do not accept methods that draw structural conclusions  from their data.
  • While SEM was fully developed in the 1960's and 1970's, modern multivariate and exploratory methods such as Artificial Neural Networks (ANN) experienced major developments in recent years.
  • ANN's however are also, generally speaking, not suitable for causal modeling as they suffer from the Black Box Problem: they increase predictive performance but are unable to explain or convey the reasons for the models derived from them.

Imagine a solution,

… that your research field of Customer Confusion is heavily influenced by a number of nominal scale variables. You apply an advanced causal analysis and retrieve the model shown in the figure just below. All paths that terminate with a dot represent interacting influences in your data. For every interaction, your analysis shows plots like the second one below. It indicates that the effect of education on customer confusion is only valid for medium income respondents. The higher the education, the lower the confusion, but only for medium incomes. All others have probably developed simpler decision heuristics: the poor spend the least possible for them, the wealthy spend the most.

Have you (or your supervisor) ever seen such findings using a linear SEM analysis? Using NEUSREL it's more likely that your supervisor will recommend your graduation with honors!

CC Model+

CCInteraktion

Imagine …

… that your research field is service quality within service centers. You build a theoretical model, and estimate it using AMOS – but you are neither satisfied with the fit nor with some contradictory results. You apply an alternative method and obtain the results shown in the figure just below. You discover that many paths are digressive (NLd). This makes perfect sense to you, since a number of variables are known from the literature as “hygienic factors” – a specific level is needed but not more than that. Other paths turn out to be progressive (NLp) indicating that they may be viewed as “satisfaction factors.”

Most interestingly you find new, meaningful paths. Linear SEM modeling of your data indicates that satisfaction influences the factors Word-Of-Mouth, Cross-Buying and Willingness-To-Pay, directly and independently. You discover that willingness is required for Word-Of-Mouth in order to foster Cross-Buying, and that you need Cross-Buying-Willingness in order to enable Willingness-to-Pay. Re-analyzing your data in AMOS, you learn that indeed your fit has increased when introducing this new structure. Despite having used an unsystematic and ad-hoc trial and error search, you nevertheless achieved your results using a systematic approach that only considers theoretically supportable paths. Imagine how achieving these results will make you stand out relative to other researchers who are only using linear SEM methods.

DLCQ

Imagine, …

… a retail store chain is sponsoring your research survey. With the collected data you build an advanced causal model. You conclude that perceived relationship investment is a major prerequisite for repeat purchases. With your analysis you demonstrate that …

Interactions

… excellent interpersonal communications with customers is the major ingredient of success. Expensive “tangible rewards” (especially free gifts such as shoe polish) are just one alternative but a less effective tool. Simply by eliminating this expense item, you reduce overall costs by 1.5%, which in turn boosts profits by almost 30%. Your client firm will now likely offer you an attractive employment contract or at the least, renewed research support!

Which methodology can do all this?

Neusrel Buch CoverThe answer: Universal Structural Modeling (USM). The foundation for USM was developed during a five year research project conducted in cooperation with Harun Gebhardt. Together we developed a stock forecasting system based on neural networks. In 1999 we launched Profit-Station.com, which has been successfully used since then by dozens of individuals and companies.

In that same year, Dr. Buckler began his doctoral studies with the ambitious goal of reinventing Structural Equation Modeling - the crown jewel of the social sciences. In 2001, he published the book entitled "NEUSREL." This introduced a new causal analysis method based on the same neural networks that made Profit-Station successful. Since then, NEUSREL has been continually refined and used in a large number of research and consulting projects. Furthermore, it has matured as a result of extensive scientific discussions with several world class researchers. Important improvements were stimulated as a result of an ongoing dialog with Professor Hennig-Thurau. As a result of all this, the methodological group "USM for NEUSREL" was also formed. 

How does USM work? Cause-effect networks are constructed in two steps:

  1. At the measurement level, survey data is compressed into latent variables,
  2. and
  3. At the structural level, cause-effect relations among latent variables are analyzed.

At the measurement level, USM uses principal components analysis to compute and determine the latent variables. At the structural level a specific neural network we developed is trained for every dependent latent variable, thereby determining in turn the influence of all the other latent variables. The neural network that we use ensures that irrelevant effect paths are eliminated. The black box problem is solved mainly through a methodology introduced by Plate in 1998. This allows for the visualization of each causal effect.

If you'd like to learn more about NEUSREL, please read Dr. Buckler's latest scientific article published in “Marketing – Journal of Research and Management” co-authored with Professor Hennig-Thurau. 

Please request the article (.pdf) with an email message to  USM(at)neusrel.com  
Please provide your name, phone number, and insitution or employer.

How you profit from USM?

A number of site visitors have asked us how USM can assist them in their research projects. A good way for you to begin your project would be to use our very fast, efficient, and cost-effective analysis service. We provide you an Excel spreadsheet template to insert your data and option settings. You then email your spreadsheet to us. We then run the calculations and email the results back to you. For frequent users of USM-NEUSREL we provide you with a software license. 

What the experts say about USM:

  • “I had the chance to read the book NEUSREL in 2001 as an early draft.
    Within the scientific tradition of data-mining, I believe that both NEUSREL and USM add a powerful instrument to uncover hidden, more complex, and perhaps meaningful relationships among variables.”
        -Prof. Dr. Rene Weber, University of California at Santa Barbara, USA.
  • “I use USM whenever I am working on a problem that falls within its capabilities, for example, to estimate structural equation models with many nominal variables such as gender. In the field of customer confusion we found that confusion is particularly prevalent among medium-income consumers, whereas low- and high-income consumers employ buying heuristics that shield them from confusion. A simple finding, however one we would have never found without USM.”
        -Prof. Dr. Gianfranco Walsh, Strathclyde Business School, University of
          Glasgow and University of Jena.
  • “We are planning to apply USM for communication control and planning in the advertising-intensive food industry. We estimate we will save companies a considerable part of their communication spending.”
        -Prof. Dr. Holger Buxel, University of Applied Science, Muenster.
  • “In contrast to classical methods of linear structural modeling NEUSREL offers three advantages: Exploration capabilities, nonlinear relations and arbitrary interactions between constructs are considered and allowed. Models are more realistic as a result of these factors. Results from classical covariance-based methods might possibly be better, but only if the data perfectly conform to all the (quite restrictive) assumptions of the method.”
         -Prof. Dr. Volker Trommsdorff, Technical University of Berlin.
  • “USM allows for exploratory modeling of structural equation models. With this quasi-confirmatory method, new paths, unknown nonlinearities and interactions may be discovered, described and quantified.”
         -Professor Dr. Rolf Weiber, University of Trier.
  • “With NEUSREL Dr. Buckler introduces us to an outstanding contribution to marketing research that has the potential to close a major research gap.”
        -Professor Dr. Klaus-Peter Wiedmann, University of Hanover.
  • “Best wishes as you expand the influence of this exciting software.”
         -Christopher P. Blocker, Ph.D., Assistant Professor, Hankamer School
           of Business, Baylor University.
  • “[The inventor of PLS] Wold talked about a dialog between the researcher and the data, facilitated by the method.  ... I think a tool like NEUSREL brings PLS closer to Wold's original intent for PLS.”
        -Prof. Edward E. Rigdon, Department of Marketing, Georgia State
          University.
  • “I very much enjoyed the MJRM article about NEUSREL and I am particularly intrigued by the nonlinear/interaction capabilities.”
        -Prof. Dr. Claes Fornell, University of Michigan.
NEUSREL user testimonials:

“We are convinced about NEUSRELs capabilities.”
     -Mag. DI Ryffel GFK Swiss.

“... congratulations on creating a wonderful product. I am going to be recommending it at places that I already have connections with.”
    -John Steele, MS, ABD, Kansas State University and Army Research Institute(ARI).

"... with the aid of NEUSREL we were able to uncover important nonlinear effects in the field of psychological brand impact."
    -Gregor Waller, lic. phil., Head of Research, Brandezza AG.

"The program provides very interesting diagnostics which give me a lot of clues to dive into more insightful investigation of the data."
    -Jae Cha, Chief Research Scientist, CFI Claes Fornell International.

"I have applied the NEUSREL software designed by Dr. Buckler to customer satisfaction and loyalty data, and found that it provides some very desirable features. I have been happy about its ease of use, functionality, and new and desirable features such as the ability to identify nonlinear and interaction effects in the model.”
    -Kunal Gupta, Ph.D., Vice President, Burke, Inc.

"We used NEUSREL for exploring product adoption drivers. Thanks to NEUSREL's capability to include all kinds of variables (e.g. moderators or categorial variables such as gender)  into our analysis, we were finally able to avoid spurious findings and to derive some meaningful recommendations concerning our proposition design and go-to-market strategy."
    -Daniel Klein, Senior Manager, T-Mobile, Inc.

Let's summarize what USM-NEUSREL provide to users:

Today's causal analysis methods are designed to test existing theories, rather than to explore new paths, or previously unknown/unanticipated nonlinearities and moderating effects. These three items however are exactly what is required in order for causal modeling to be maximally successful in practical applications.

Prior to the development of USM, the scientific community generally speaking, did not address the abovementioned practical causal modeling issues. Clearly, it became necessary to pursue a new approach. The foundation of this new approach has been developed over the last 25 years or so - a foundation that enabled us to bring forth USM-NEUSREL.

USM is a new type of causal analysis that uses artificial neural network technology and has the following key features ...

  1. Exploratory analysis: USM requires less a priori information-knowledge.
  2. Nonlinearity: USM allows the user to explore (even unknown) nonlinear relationships.
  3. Interactions: USM reveals, displays and quantifies interactions among causes.
  4. Universality: USM makes use of arbitrarily distributed variables, in particular nominally scaled variables such as gender, profession, brand name, etc. Furthermore, USM enables one to model circular causal networks – eliminating the need to distinguish between endogenous and exogenous variables.
  5. Quantification: USM allows quantification of every significant property in one's data, including path strength, linear path coefficient, interaction strength, and significance level. 
  6. Simplicity – USM is very easy to use - no need for elaborate option settings.

We have received numerous success stories indicating the tremendous value that USM delivers to both researchers and the business community. With the aid of our analysis service and a trial usage of the software license, you have the opportunity to experience the vast potential of USM to reveal previously undiscovered cause-effect relationships in your own data. Perhaps you are analyzing broadband usage patterns, or consumer preference. Whatever type of data you are exploring, USM-NEUSREL could be your next great step toward amazing scientific discoveries!

Please contact us to explore together how USM-NEUSREL may significantly benefit your research and/or corporate projects.  

Contact Dr. Buckler (email): Buckler at neusrel.de

P.S. New: Within the "NEUSREL Ph.D. Program" a limited number of doctoral students will be afforded the opportunity to use NEUSREL at no cost.

disclaimer

 

Published in:
MJRM

German »
Sitemap »
Disclaimer »