You're facing conflicting opinions on data analysis methods. How do you choose the right approach?

Identifying the business needs is critical in terms of technique selection. The goal of the analysis, the target audience, the problem we are looking to solve all govern this decision making. Follow up questions and trying to identify the actual answer being seeked often leads to avoiding running down the wrong path.

Choosing the right approach for data analysis amidst conflicting opinions is challenging. However a thorough understanding of the context and objectives of the research helps to choose the required analytical approach.

You can also consult with peers and other experts by reviewing existing literature and industry standards, and guidelines to see what methods are commonly accepted and recommended for similar analyses. Colleagues and other domain experts who have experience with similar analyses can provide insights based on practical experience. If possible, present your reasoning and chosen method to peers or experts for feedback before finalizing your analysis. Their input can help validate your approach.

Literature review obviously is very important and provides you valuable information in the area of your interest and the research you work on. However, we often ignore ways to collect resources in geographies where the research work is very less. In such cases we need to build strong networks and collaborations locally through 'chain-referrals' till we meet the entity from whom we get the relevant information. In information scarce areas such as Gilgit-Baltistan, we need to focus on networking, physical visits, cataloging the resources, evaluating and then summarizing and synthesizing only then can we produce a proper research.

In my work, I have learnt that keeping the tool as simple as possible makes it easy for analysis and the results are more practical and covers wide range of audience. The tool and the subsequent analysis plan are interlinked and keeping the tool lean and simple improves the efficiency and resourcefulness of the research.

I would also be wary of checking the various assumptions associated with each technique. For example if utilizing linear regression we want to check for the linearity assumptions by utilizing a residuals plot or validate the homogeneity of variance using a spread location plot etc. Most statistical or analytical methods rely on assumptions or boundary conditions which need to be met for the results to interpreted accurately.

In my experience, pilot testing of a research tool is very important or rather a key element for operationalization of a tool. We have research experience in the communities of remote villages in the Gilgit-Baltistan region of Pakistan and found out that the tools require proper contextualization which is not possible without involvement of the participants or population segment that we are conducting research on. An important dimension of conducting pilot testing is that this makes the tool more comprehensive and empowers the researcher with in depth understanding on the local variables that most of the time are lost in desk-work.

If feasible, conduct sensitivity analyses or robustness checks using different methods to see how sensitive your conclusions are to the choice of method. Whichever method you choose, ensure it is well-documented. Clearly outline your rationale for selecting the method, including any trade-offs or limitations. Data analysis is often iterative. You can start with one method, evaluate results, and if necessary, consider alternative methods based on the initial findings or feedback.

Recently, we conducted a research on the validation of a survey conducted by the government line department in Baltistan region of Pakistan and found out that education is still a dream for the mountain communities in remote and inaccessible parts of northern Pakistan. Endowed with rich natural resources, the valleys in Baltistan considerably lag behind in terms of education and health metrics. There is considerable room for researchers to add to the knowledge on the dynamics of health and education at the conflux of climate change in remote geographies such as Basha valley in the upper Karakorams.