Most appropriate solution is to use mixed method employing both quantitative and qualitative methods.  for example, case study method.  then we must able to understand the results of quantitative methods more meaningfully.

Hi,

firstly I doubt that the "majority ...employs random sampling". Secondly, no, random sampling would not reduce the bias that results from ignoring the context. The context may a) function as a moderator that influences the effects of independent variables on dependent variables. When you ignore the context in this situation, you don't get a bias but an average effect across different contexts. Otherwise (b) the context may act as a confounder that acts as an common cause of the indepentent and dependent variable. In this situation, you need to control for the confounder/context in order to get meaningfull results. 

HTH

Holger

decision to adopt random sampling or non random sampling depends on your research requirements. for example if you think that each  business entity is embedded in a different context then it is best to go for multiple case studies  and if the outcome is in coherence in certain context, you an combine them later on and conduct a major study.

Take the explaining model of Liv Kirsten Jacobsen as the basic. Analysis in entrepreneurship research are case studies. Quantitative methods will fail unless you will use QDT based models (e.g. V.I.Yukalov). For qualitative analysis tri-partite model setup will give sufficient results.

Sure, you can specify your research question and the research setup and immunise the model via ceteris paribus conditions and constrains or  limit the valid possibilities. 

Upananda, 

I think you posted a very interesting and often ignored question. I also agree with Holger that many studies are not actually using random sampling. 

Including contextual factors in your study is always tricky as you have to first determine what a contextual factor is in your study. Are you looking to include macroeconomic factors such as unemployment or economic growth? Are you more interested in institutional effects? As you can imagine, finding solid datasets for each contextual factor is hard. As mentioned before, using mutliple case studies might help here. 

Another way you can approach this is to use agent-based modelling. That is a simulation-based approach in which you reconstruct a system through a set of autonomous agents. You have a wide set of parameters to choose from (including contextual effects) and can therefore run multiple scenarios. Agent-based modelling has not been that widely used in entrepreneurship yet, but it can be a good alternative to integrate different (often incomplete) datasets. 

I hope that helps.

Best,

Xaver

Article Contextualizing Entrepreneurship - Conceptual Challenges and...

Article Growing Artificial Entrepreneurs: Advancing Entrepreneurship...

Conference Paper How To Employ Agent-Based Modelling And Simulation As A Viab...