Resources for Improving Evidence Presentation

As you have seen in the “From the Editor” that Administrative Science Quarterly (ASQ) published in June 2017, the editors of ASQ strongly encourage that authors show the data in their manuscripts, by using graphical approaches to give an indication of the most important features of the data and their theoretical explanation before estimating models. Preferably this should be done as early as the introduction in order to spur the reader’s interest and give an indication of why the paper is valuable. Such use of graphical methods is rare in organizational theory and management research more generally, so we will gradually introduce methods of graphical analysis that can be used by researchers.
 
Graphical methods for showing the data are integrated into Stata, the most common software used by management researchers, and the Stata commands offer a good blend of simplicity and flexibility. Nevertheless, they need some training, especially because statistical training is model-focused in many schools, and highly variable in how well graphical methods are taught. Here are some resources that can be useful:
 
Here is a simple example do-file and data. The data are from published work (Greve and Song, 2016), but is only a small sample of the data in use. Here are the graphs produced by the do-file and the data.

Data source:
Greve, Henrich R., and Seo Yeon Song. 2017. "Amazon Warrior: How a Platform Can Restructure Industry Power and Ecology." Advances in Strategic Management 37:299-335.
 
Here is a simple introduction to some important methods, including scatterplots, lineplots, bar graphs, box plots, and kernel (full distribution) plots.
 
Here is an example of more advanced programming, which is needed because stata does not (yet) have a simple way of showing a grouped bar graph with error bars, which is an important graph for taking a first look at group differences.
 
Displaying statistics on a map can be very helpful for any kind of research involving spatial relations. Here is an introduction to spmap, which is an add-on procedure for producing mapped data displays. Earlier such mapping required changing to different software and exporting data, which is both time consuming and a potential source of errors.
 
Here is an introduction to the coefplot function, which is a graphical display of coefficient magnitudes, and a very informative way of giving a comparative view of a full regression model, or parts of it, in a compact graph. The Greve and Song file above gives an example, but this function has an advanced set of options. An important issue in using it is that it is usually better to standardize the covariates (to a standard deviation of unity) for easy comparison of effect size. This was not done in Greve and Song because their covariates were counts, so unit changes were meaningful and comparable.