Post on 15-Apr-2017
SUMMARY OF MY DOCTORATE THESIS AND CELLULAR AUTOMATON MODEL – RUBENS ZIMBRES rubens.zimbres@eclipso.de
This document is a brief explanation of the steps of my Doctorate thesis. Samples of results are given. The Cellular automaton model is presented.
The thesis is available here: http://pt.slideshare.net/RubensZimbres/thesisrubens-‐zimbres In my doctorate thesis I studied 4 constructs: role clarity, involvement, service quality and behavioral intentions in the following nomological network. The research was diadic and dynamic, made with clients and service providers.
For each one of the constructs, I did a comprehensive research where knowledge acquired was inserted in a matrix, as seen next page.
Then, indicators cited in literature were placed in another matrix, correlating authors.
Then, the research phases are presented.
After that, I did qualitative interviews to develop the pre-‐test questionnaire. Interviews were transcripted, contente analysis was done, compared to literature and quantitative pre-‐test questionnaire was developed. Then I did a statistical analysis and I finally had the quantitative questionnaire, below. An ordinal scale of 5 options was chosen to allow respondents to have a neutral opinion, without being forced to choose between two opposites, what happens in a scale with 6 options. Next page there is a small part of the questionnaire.
A comprehensive statistical analysis was done. Here are some results: Graphic: Time as a cliente of the servisse provider
A descriptive analysis was done:
Normality tests were done:
Also, collinearity tests were done:
Correlations with factor 1 and 2
Factor analysis and Cronbach's Alpha (for clients).
And the regression model, with principal components:
Finally, the cellular automaton model was developed to simulate interactions between clients and service providers. I did two quantitative researches with interval of 4 months. Research A and B. My idea was to find a cellular automaton that could be applied to research A (first) and generate an outcome of the simulation similar to research B (second, after 4 months).
I had to find a rule in a space of 1 in 10 to the power of 80 possibilities. This would take more than 200 years with a regular computer, so I did a randomly guided search. The cellular automaton is a tool where you have three cells. The center cell is updated according to the opinion of the cells on the left and on the right. Each CA rule generate on specific outcome.
Next page, the cellular automaton model workflow:
The accuracy of the cellular automaton model to simulate interactions was 73.80% case by case and 99% considering mean of indicators. This is greater than any linear regression found in service quality literature.