Proposed Strategy

К оглавлению
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 
136 137 138 139 140 141 142 143 144 145 146 

Our approach to the analysis of methodological validity of the RCMs is anchored in

Rescherian epistemology. Specifically, we adopt a four-stage process as shown in Figure

1(b).

1. First, we engage the applicative cycle, by creating a text of what constitutes OO

expertise to isolate concepts in the language of the practitioner. Semi-structured

interviews were conducted with OO software development experts to gather their

knowledge regarding the concepts that constitute OO development.

2. Second, through a conscious process of choice, we choose a theoretical framework

to interpret the concepts, thus engaging the theoretical cycle. As stated above,

three theoretical frameworks were initially chosen as candidates. Additional OO

experts were asked to assess the congruence of the emergent concepts with each

Knowledge

Representation

by Experts

Interviews

Empirically Derived Structure of Concepts

Factor Analysis of Survey Responses

Theoretical Structure of Concepts

Card Sort by Experts

Revealed Causal Maps

Reconstructed Maps

Bootstrapping

Empirical

Control

Theoretical

Control

Basis of Comparison

CONCEPTS

1

2

3

4

Figure 1. Rescherian epistemology applied to validation

(b) Four stage validation process for RCM

theoretical framework. Each developer performed a card sort for each of the three

theoretical frameworks. Based on the results of the card sort, we chose theoretical

Framework III.

3. Third, we employ the interpretive framework to engage an alternate methodology

(survey) for empirical validation. We developed and validated an instrument from

the concepts that emerged from the interviews. The validated instrument was then

given to a large sample of software developers. Factor analysis identified five

factors that explain the pattern of correlations within the data.

4. Finally, we attempt a rapprochement of the tensions between theoretical and

applicative cycles through a system of bootstrapping: We use the empirical

evidence to recast the interpretive framework to contrast the outputs of theoretical

and applicative cycles. The factor structure is used as another framework (Framework

IV) to interpret the interview data. Each causal statement is re-coded using

Framework IV, and an additional causal map is produced. The two maps, one using

Framework III and the other using Framework IV are then contrasted.