Tuesday, February 16, 2016

Universal Structure Modeling approach

Universal Structure Modeling approach without the need to install and learn a new software. Read more below (or consult our website at www.neusrel.com/free-analysis )

# FREE FOR ACADEMIC RESEARCH:
INCREASE EXPLAINED VARIANCE OF YOUR MODELS by identifying hidden NONLINEARITIES and INTERACTIONS in your data.

Add your data and your model in an Excel file and let us do the calculation with the NEUSREL Software that is based on a Universal Structure Modeling (USM) approach. Your data remain 100% anonymous and no data will be stored or analyzed after you have received your results.

WHAT IS UNIVERSAL STRUCTURE MODELING (USM)?

USM works somehow similar to PLS-Path Modeling, but it is neither relying on parametric modeling nor on the assumptions of linearity and independent observations. Instead, USM builds on semi-parametric modeling performed by Bayesian Neural Networks – a universal regression technique. Dedicated simulation algorithms analyze the “neural net black box” by quantifying path effects and by visualizing nonlinearities and interactions. For more details, please contact schmidt@neusrel.com  or consult our publications:

Ø  Journal Article by Buckler/Hennig-Thuraus http://goo.gl/Psryv6
Ø  Other Journal Articles http://goo.gl/PP52YT

3 STEPS to YOUR USM Model

Ø  DOWNLOAD the Excel input sheet here. (http://www.neusrel.com/NEUSREL-input.xlsx ). Copy your data matrix (cases in rows) into the tab “data”. Label and define your latent variables as well as the path model in the tab “Project”. Consult this video to see how it works https://www.youtube.com/watch?v=8bL4z4pl0OI  .
Ø   SEND your Excel to schmidt@neusrel.com  together with your complete contact details – Name, Address, Telephone, Email. Feel free to add some background information.
Ø   RESULTS will be send to you in that same Excel including important plots of nonlinearities and interactions. Also, we will include a very brief summary of our takeaways and recommendations.

CONDITIONS of the PROGRAM

Modeling will be performed without bootstrapping and most other special features of NEUSREL. The intention of this free service is to pilot-test the possibilities of USM, not to finalize a research project. When demand exceeds our capacities, your request may either delayed or even cancelled. Models with anonymized variable labels will not be accepted. For a sound modeling, we need at least some knowledge about the data.

Frank Buckler, PhD

NEUSREL Causal Analytics

buckler@neusrel.de



No comments: