دانلود رایگان مقاله انگلیسی مدل رشد دندریتیک در بازاریابی چند سطحی - نشریه الزویر

Abstract

Biologically inspired dendritic network growth is utilized to model the evolving connections of a multilevel marketing (MLM) enterprise. Starting from agents at random spatial locations, a network is formed by minimizing a distance cost function controlled by a parameter, termed the balancing factor bf, that weighs the wiring and the path length costs of connection. The paradigm is compared to an actual MLM membership data and is shown to be successful in statistically capturing the membership distribution, better than the previously reported agent based preferential attachment or analytic branching process models. Moreover, it recovers the known empirical statistics of previously studied MLM, specifically: (i) a membership distribution characterized by the existence of peak levels indicating limited growth, and (ii) an income distribution obeying the 80-20 Pareto principle. Extensive types of income distributions from uniform to Pareto to a “winner-take-all” kind are also modeled by varying bf. Finally, the robustness of our dendritic growth paradigm to random agent removals is explored and its implications to MLM income distributions are discussed.

4. Conclusions

We have described a procedure that models a multilevel marketing scheme based on a dendritic growth mechanism. Our network evolves by following a branching process governed by a parameter bf that balances the wiring and path length costs. We have shown that such dendritic dynamics can create network types with structures having similar morphological signature and capture the essential features of an MLM such as: (i) the limited growth distribution as exhibited by peak levels independent of network size and bf value; and (ii) a wide range of possible distribution of income from uniform to Pareto-governed.

Using an empirical data set taken from an actual MLM company in the Philippines, we have demonstrated that such procedure provides a better fit more accurate than the existing agent based preferential attachment and the analytic branching process models. We have also highlighted the rich type of network behavior that results from scanning the generic parameter bf, the sole driver of our network’s dynamics. We have shown that higher bf values translate to total income diverted to top members (members with ranks closer to the root node) of the network. However, we have to note that we have no firm interpretation of what the bf parameter is in a real MLM company. Nonetheless, we have speculations of its implication such as the ability of potential recruits to choose which cost factor (d or pl) has a dominant effect on their decision to join the network; or the management’s ability to control the growth of the network. If the latter is true, we have shown in our analyses that the management can easily change bf to manipulate the income distribution to their benefit.