The first piece of this article discussed an overview of triangulation, how it works, its output and its benefits. So, let’s continue the discussion by explaining the triangulation types.
Triangulation can take several forms; data triangulation, methodological triangulation, theoretical triangulation, investigator triangulation and data analysis techniques triangulation.
This type involves using two or more data sources to study the phenomenon being researched. Basically, the information areas required for the research are the same but come from multiple sources. For example, collecting the primary and operational data to understand the reasons behind the declining sales
In this type, mixed methods are employed to study a unique phenomenon, including research and data collection methodologies. For instance, when quantitative and qualitative methods are used to study and understand the phenomenon.
When multiple theories and hypotheses are used to conduct the research, that’s called theoretical triangulation. For example, formulate different hypotheses to explain the reasons behind the declining sales. The churn-out rate last quarter increased due to a new tariff plan launched by the competitors, while our brand does not respond to this dynamic. This is one of the hypotheses. Another one can be: the bad quality of our network is the main reason; hence the research design can be developed taking into consideration these different hypotheses.
This type replies on the human for applying triangulation simply uses different researchers or data analysts in the study without prior discussion and collaboration between them. The idea is to remove the bias that might occur across the research process, whether in the designing or analysis stage.
Data analysis triangulation
In this type, a mix of data analysis techniques is applied in a study and conclude the outcome based on the output of these analysis techniques.
Triangulation in the research realm
To put theory into practice, I will demonstrate a case study from my previous work experience where data analysis triangulation has been employed. It was a concept design test. The study was about exploring the preference of several design concepts (3 design concepts). The three concepts are constructed in picture forms to be shown to the respondents. To achieve the objective effectively with high accuracy, a composite measure has been developed using latent and observed variables model. Preference Index (latent variable), the index comprises multiple observed variables (likability and purchase intention).
The observed variables have been measured using a ratio scale where respondents give their rating using a score from 0 to 100. A randomization plan is prepared to clear the order bias across the tested concepts. In the analysis stage, data analysis triangulation is employed using the following steps:
Firstly, descriptive analysis is done, and the mean score is calculated for all concepts for the observed variables. Preference index has calculated using a formula: Y = (ML + Mp)/ n
Y: preference index (index range is between 0 to 100, the higher, the better preference)
ML: mean score of likability.
Mp: mean score of purchase intention.
n: number of observed variables.
As shown in Table-1, it is very clear that concept-3 is the most preferred one. However, to increase confidence in the results and conclusion the second step is done.