3-D Analysis


3-D analysis refers to the utilization of several dimensions (methods) of analysis to identify key indications of the query target. The analytical model demonstrates the interaction of multiple method analysis (see figure).


3-d.jpg


























NSA-1 refers to non-state actor 1. NSA-1, in the above diagram, is analyzed on the basis of four indicators and four role assessment perspectives. NSA-1 analysis would consist of 16 cubes demonstrating the interaction between each indicator and each role assessment. Each cube would have an indicator value, a role assessment value and an interaction value (i.e. I-1 value A=R-1 value B...see diagram). Indicator interactions are then assessed and compared to reveal patterns in NSA indicators and NSA roles. These patterns can then be used to generate a indicator/role relationships. These relationships would then be the source of building a predictive model for the role of non-state actors in Sub-Saharan Africa.
cube2.jpg


The original 3D Analysis graphic was confusing so I created the following “Qubert” graphic to help the understanding of the concept.

Qubert2.jpg

The concept is truly simpler than it looks. Simply stated, by comparing each indicator with each method of analysis the analyst is able to identify patterns and correlations to determine the most efficient method of analysis. Furthermore, the design is conceptual in nature. It can be represented in more difficult for humans, but easier for computers to understand formats, such as Excel.

Correlation/Pattern analysis would be simple at this point. If high values of indicator one (I1), (such as % of population infected with HIV) are consistent with high value of “influential role” produced by all four analytical techniques (such as link analysis, financial analysis, citation analysis, and analysis of competing hypotheses) then I1 indicates that analysts in general will assess the target NSA to have a high value for “influential role.” Like wise if high values of role assessment 1 (R1) (such as link analysis), are consistent with multiple indicators (such, % of pop below poverty level, distribution of wealth, telecommunication assets etc) then it may be possible to develop an algorithm to predict the outcome of link analysis in terms of the target NSA. To be sure, poverty levels great than 30% matched with distribution of wealth lower than 10% and few telecommunications assets may indicate that tribal bounds are strong in the target country or NSA. Analysis across NSAs may develop reliable indicators and algorithms that produce effect analysis. Simple equations such as I1/R1= X can be used to produce comparable data. For instance: if I1= 30% and R1= 8 then I1/R1= .333/8=.0416. Values for multiple NSAs can be compared. If values for I1/R1 generally fall between .03 and .05 then the prediction can be made that I1 correlates to R1 regardless of the variation in the individual values for I1 and R1. Therefore if I1 for another NSA equaled 50% then we may predict that R1 would equal 12.019 (I1= .5 so .5/.0416= 12.019). Likewise if R1 for another NSA equaled 6 we may predict that I1 would equal 25% (R1= 6 so 6•.0416= .2496). Lets us say that, from a random sample, we developed a mean value of .0416 for I1/R1 with a standard deviation of .009 at a .05 significance level. We may then say if: NSA X’s I1= 25% then R1= 6 with a variation of R1= 4.94 to 7.669 (I1/.0416 +/- 1SD) within one standard deviation of the mean with a .05 error.

I believe such a method would answer both of the decision-maker’s CIRs: 1) Determine the role of non-state actors in Sub-Saharan Africa; 2) Use multiple analytical methods and keep notes (journals) for review after the assessment. Implied CIR: determine the efficacy and viability of different analytical methods. Furthermore, the scope of the CIR “the role of NSAs in Sub-Saharan Africa” is large and difficult to tackle. Therefore, by using a method that can identify indicators and test them for accuracy, the task becomes somewhat manageable.

Perhaps the best way to conceive the 3D analysis model is to consider it a multi-attribute/multiple target regression analysis. It not only allows the analyst to measure the interaction between different attributes (indicators) and analytical methods but also allows the analyst to determine the relationship between multiple targets and attribute/method interaction, ergo 3 dimensional analysis. Excel can easily be programmed to produce such data; furthermore we have access SPSS through the Mercyhurst Psychology Department. 3D analysis may seem daunting but given the scope of our CIR it is likely the most efficient process of fulfilling the requirement.
moz-screenshot-3.jpg