Interview Method (IECM)

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 

The researcher’s goal is to gather participants’ knowledge or beliefs and cast it into

cognitive structures pertaining to a specific domain. The task is to access relevant

participants and assist them in articulating their sometimes tacit knowledge or beliefs.

Individuals serve as the data source and the narratives are gathered through interviews

(ranging from unstructured to structured), which are discussed later in this section.

Sampling

One option is to use random sampling, which is particularly useful when engaging in

studies from a social constructionist perspective. From this perspective, expertise is

Table 1. Mental model measurement techniques

Dimension PAN MDS IECM TBCM

Content Fixed and supplied

by the researcher,

low emphasis

Fixed and supplied

by the researcher,

low emphasis

Variable and

supplied by

participant, high

emphasis

Variable and

supplied by

participant, high

emphasis

Structure Associative explicit

linkages, high

emphasis

Associative explicit

linkages, high

emphasis

Causal explicit

linkages, high

emphasis

Causal inferred

linkages, high

emphasis

Researcher

Skill

Low Moderate High High

Participant

Demands

Moderate Moderate High None

Model

Comparisons

Easy Easy Difficult Difficult

Adapted from Mohammed, Klimoski and Rentsch (2000)

PAN = Pathfinder Associative Network; MDS = Multidimensional Scaling; IECM = Interactively

Elicited Causal Map; TBCM = Text-Based Causal Map

Research Context Data Collection Methods

Discovery Unstructured interviews

Evocative Unstructured or semi-structured interviews

Hypothesis Testing Semi-structured or structured interviews

Intervention Structured interviews

Table 2. IECM data collection methods

uniformly distributed and therefore random sampling is an appropriate method of

identifying participants in a study. In expert-anchored studies a snowball technique

(Shanteau, 1987, 1992) with convenience sampling (Stone, 1978) is often used. Snowball

sampling becomes necessary when experts of a domain cannot easily be located by

random sampling or by screening, where domain knowledge (expertise) is important, and

where the members of a domain are known to one another (Simon & Burstein, 1985). The

snowball technique asserts that those individuals closest to a domain are appropriate to

define the experts of that domain (Shanteau, 1987, 1992). An initial participant is chosen

and additional participants are obtained from information provided by the initial participant.

One expert identifies another and that expert identifies another, and so on. Once

identified, each expert is interviewed (Axelrod, 1976; Huff, 1990).

Interview Protocol

The interview process may consist of fairly structured interviews (Bougon, 1983), semistructured

interviews, unstructured interviews depending on the research context. See

Table 2 for a listing of appropriate data collection methods for each research context. An

interview guide is developed by the researcher to facilitate the interview process. When

developing the interview guide the researcher should be cognizant of several factors,

such as the research context, the specific domain under study and the respondent pool.

Readers wishing guidance in developing an interview guide may wish to see: Bradburn

(1979); Kvale (1996); Payne (1951); and Rubin and Rubin (2004). Based on the participant’s

answer to the question, follow-up probes may be asked to elicit further details regarding

the participants’ thought process. The interviews are then transcribed verbatim into a

document format (e.g., Microsoft Word).

Point of Redundancy

Within the CM method, the researcher should interview to the point of redundancy,

which determines the adequacy of the sample size (Axelrod, 1976). In causal mapping

research the point of redundancy, or saturation, represents the point at which further data

collection would not lead to the identification of additional concepts. As the concepts

emerge from the participants rather than being imposed by the researchers, this point

serves as a way of establishing the adequacy of the sample. The point of redundancy

is operationalized by aggregating the concepts mentioned by each participant (Nelson

et al., 2000).

The participant’s text (interview transcript) is reviewed and the number of concepts

elicited is graphed (the X axis is the participant number and the Y axis is the running total

of the number of concepts). The next participant text is reviewed, the number of additional

concepts identified is added to the number from the first text, and the result is graphed.

This process continues until all of the texts have been reviewed and the concepts elicited

are identified. The difficulty is that the point of redundancy is not calculated until after

the interviews have been completed and the classification scheme has been developed.

If redundancy is not reached, additional interviews would have to be conducted. The

same process would be used until redundancy is reached.

For example, if you identify ten concepts for the first participant, a point would be plotted

on the graph at (1,10). If you identify an additional eight concepts for the second

participant a point would be plotted on the graph at (2, 18), and so on. No additional

concepts are elicited from participants 19 and 20, so the point of redundancy is reached

by the 18th participant. See Figure 1 for a graphical representation.

0

10

20

30

40

50

60

70

80

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Participant

Total Number of Concepts

Data for Figure 1

Participant Unique Concepts Identified Total Concepts

1 10 10

2 8 18

3 7 25

4 7 32

5 5 37

… … …

15 1 63

16 2 65

17 1 66

18 1 67

19 0 67

20 0 67

Point of Redundancy

Figure 1. Point of redundancy

Text-Based Method (TBCM)

Text-Based Causal Maps rely on non-invasive data collection techniques that avoid the

recall biases of interviews (Axelrod, 1976). The researcher’s goal is still to gather

knowledge or beliefs and cast it into cognitive structure pertaining to a specific domain.

The task with TBCMs is to determine the appropriate source of information and gather

the data from that source. TBCMs have been found to be more economical in terms of

time and effort required of researchers and subjects (Brown, 1992). Data sources for textbased

causal mapping include any complex text (e.g., annual reports, case analysis, IS

change request documentation, and legal decisions). TBCMs are particularly appropriate

for longitudinal studies because they do not depend upon participants who may not

be accessible, or whose memories may have faded with regard to the event under study

(Narayanan & Fahey, 1990).