��������������������������
All mathematical and many computer-based techniques express relationships quantitatively and can deal only with quantitative information . In particular , computer models that provide a way of understanding , accessing , and manipulating mathematical and logical relationships fall into this group , including the popular methods of Bayesian Networks , Discrete Event Simulation and System Dynamics .
The advantage of these purely quantitative models is that they generate numerical results , so can be relatively easily used to forecast and compare options . Continuing the same example , the qualitative relationship is of limited help in setting prices : what is a ‘ large ’ price increase and how much would sales reduce ? If instead the relationship between price and number of widgets sold is expressed with an equation , it is easy to compare total ( estimated ) profit with different prices . Consider the relationship ∆Q = −∆P / 2 , where ∆ indicates percentage change , Q is quantity sold per week and P is price . Now the seller can estimate the profit for different prices . For example , increasing the price to $ 6 is a 20 % price change so would reduce the number sold by 10 %. This changes the profit from $ 1 to $ 2 for each widget , but only 90 are sold per week . Total profit would be $ 180 per week , a substantial increase from the original $ 100 . However , if the reduction in sales is given by , say , ∆Q = −3∆P , the same price increase would lead to sales of only 40 per week and profits would fall to $ 80 ( with the additional cost of annoyed customers ).
To take advantage of the power of quantitative methods , researchers may measure qualitative information using a quantitative scale and then use the data with a quantitative model . For example , attitudes are routinely assessed by asking survey respondents whether they ‘ strongly agree , agree or strongly disagree ’ with some statement . The measurements may then be treated as quantitative data for statistical analysis . A different type of example is conversion of health states such as ‘ blindness ’ into utility values relative to perfect health . This allows the effect of treatments to be measured in quality adjusted life years . These types of quantification are explicit and have well established validation rules to ensure the conversion of the underlying qualitative information is reasonable .
In contrast , some modeling methods implicitly quantify information through the modeling process or the assumptions inherent in the method . These methods are quantitative in their analysis but operate over qualitative information , without the formal consideration of the effect of quantification . For example , the notion of friendship is inherently subjective , with some friendships stronger than others and two people having different thresholds for who they consider to be a friend . Nevertheless , a social network can be constructed by asking people to identify their friends , and then powerful techniques from Social Network Analysis ( Newman , 2003 ) can analyze , for example , which person is most influential in the group . In a different type of example , similarity between ideas is quantified in Concept Mapping , a method used to group abstract ideas . The method includes a process step where a similarity score is assigned for each pair of concepts by counting the number of people who allocate them to the same group . Agent Based Modeling ( Gilbert , 2008 ) requires logical or mathematical rules to specify how the properties of simulated individuals , resources , and other model entities change over time , but the properties themselves may be qualitative ( such as attitude ) or quantitative ( such as income ). These methods are all quantitative in their application .
Finally , there are qualitative methods that do not make any assumptions about whether the knowledge to be represented is
77