Identifying Objects

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 

Figure 5.Construct level causal map

Behavior

Structure

Object Oriented

Development

Systems

Object Oriented

Modeling

The individual maps are then aggregated (Axelrod, 1976; Bougon et al., 1977). The

aggregation is performed at both the concept level and the construct level. The

aggregation process consists of combining the causal maps of each participant into a

single aggregate map.1

Validation of Maps

Once the maps have been created, they should be validated. The validation method is

determined by the data collection method (IECM or TBCM). As a source of validation

for an IECM a member check may then be performed using the aggregated maps to ensure

accurate and comprehensive representation (Lincoln & Guba, 1985). Each participant (or

as many as you have access to) is shown the aggregated maps and asked if the maps

accurately reflect the concepts, linkages and constructs. It should not be unexpected that

as the participant walks through the map he or she will be surprised. The map reflects

multiple causal relationships and most individuals do not consciously perceive the

causality of concepts in terms of a network. The key is to engage the participant so he

or she can reflect on the map you constructed based on the interviews. After a thorough

discussion of the map with the participant, any discrepancies should be reported. For

example, Nelson et al. (2000) fed the maps back to the organization to get feedback on

maps.

When using TBCMs (e.g., archival data) validation becomes a more complicated process

because there is no one to confirm your results. With TBCMs one commonly used method

of validation is triangulation with other sources. For example, if the researcher is using

change request data to track software development productivity, additional data may be

gathered from departmental annual reports or individual annual reviews. In another

example, Nadkarni and Narayanan (in press) validated the causal maps they constructed

from annual reports of firms with the firms K-10 statements. Both internal and external

sources can be used if available.