Trident DOC670 2020 January Module 4 SLP Latest

DOC670 Applied Statistics for Research

Module 4 SLP

DESIGNING THE TRIANGULATION APPROACH TO ADDRESS THE RESEARCH PROBLEM

In DOC640 you discussed using Dedoose to do content analysis on qualitative data such as interviews. Content analysis can also be done on reports, statements, etc., from secondary data. Using the Strategic Plan from the United Way of Orange County’s website this Word Cloud was produced showing the most frequent words used in the Strategic Plan (the Strategic Plan was posted on the Web and then the text was copied to a Word document).

word cloud

Some of the words are clearly a part of their strategic mission and could be used to analyze transcripts from interviews later after you receive IRB approval.

Here is an example using Dedoose to analyze transcripts on priorities in Information Security by Information Security Specialists. Here the scenario for this research is described to show you results from the Dedoose analysis.

There were eleven propositions for insider threat for computer systems in an organization, and the propositions were evaluated in terms of importance and timing within an insider risk program. Seven volunteers participated in the research. The interviews were conducted over Zoom with the audio portion of the discussions recorded. Each of the participants was presented with the eleven propositions and discussed each in the interview. While the respondents are anonymous, Table 1 shows their demographics and role in setting up insider threat programs.

Table 1

Participant demographics

Table 1

The volunteers were recruited from the LinkedIn business networking platform. Over 70 individuals received invitations. Each invitation contained information about the purpose of the study, the fact that there would be a recorded interview, and that the research is carried out under applicable laws that govern research. The Informed Consent document was explained to prospective participants. Once they accepted the LinkedIn invitation, the respondents received a link to a Qualtrics survey containing the Informed Consent document. After they signed the Informed Consent form, it was possible to continue the research using the recorded interview. Of the 73 invitations sent, 25 of the individuals responded in the affirmative. Six withdrew their offer of participation, and others either did not sign the electronic copy of the Informed Consent or could not participate for some other reason. Seven volunteers were the final count of participants.

The central discussion of the interviews was dictated by the research question and the eleven propositions. Those eleven propositions were predetermined as the overarching themes of the interviews. Each proposition was presented to the volunteers one at a time, and then the subject stated their judgments. By signing the informed consent, the subject agreed to the Zoom interview and to have the audio of the interview recorded. Recording the discussions allowed detailed analysis of the conversations using the Dedoose qualitative analysis tool.

Transcription of audio data from the structured interviews occurred in three phases. In the first phase, the audio files underwent automatic transcription to text using a tool called Trint. The second transcription phase consisted of a transcriber listening to the MP3 audio files while reviewing the transcribed text and making corrections to inaccuracies. The third and final step was a second sweep of each text file to ensure that all the corrections had been made. The researcher and one assistant reviewed the transcribed text. The files underwent final validation by the researcher. After transcription of the audio and double-verification for accuracy, the text-based files were placed into Dedoose. Additionally, the demographic data regarding the participants was added as well.

After putting the files into Dedoose, an inductive method was applied to the transcribed data for the data coding. The image below shows a visual example of the coding process. Each transcript was first open coded as the transcriber listens to the audio of the interview. Subsequent passes allowed for linking related ideas in the axial coding part of the analysis. Figure 5 shows an example of the coding process.

Figure 5. Example of Open coding and Axial coding process in Dedoose.

The responses to the 11 propositions were coded according to two criteria, the first criterion is the importance of the proposed practice, and the second criterion is the timeline of implementing the practice when setting up a new insider threat program. Coding of the remainder of the text files from each of the interviews consists of a three-step process. The steps were open coding, axial, and lastly selective coding.

For this project, the 11 propositions provided a pre-existing framework of themes attached to each of the 11 propositions. One set deals with the importance of each proposition and another set deals with the timeliness of each proposition. These two frameworks were entered into Dedoose.

Each of the codes for the two types or characteristics were weighted in a custom fashion. Because there are 11 proposed processes, each code got a weight range of 1 to 11. In Dedoose, the higher the number, the greater the weight. When coding the interview, the researcher can evaluate the mood, context, and emphasis when adding a code weight to an excerpt in the text of each discussion. Figure 6 shows a sample of how each code can be assigned a weight range.

Figure 6. Weight range options set for coding in Dedoose, minimum weight 1, maximum weight 11.

Analysis of the transcripts with the coding yielded these results in Dedoose in Table 2.

Table 2

These weights then can be used to interpret the interviews and to produce a graphical analysis.