Data serves key functions in everything from business and medicine to politics. Given its importance, a world without data is not one many of us would like to inhabit. It would lack the benefits of modern healthcare, science, and industry.
So, what is the best way to go about interpreting data, specifically in an academic context?
Understanding Data Collection Methods
Firstly, we need to understand the methods that were used to collect it. If our data was collected as part of a survey then we should be asking ourselves these sort of questions:
- How many people were sampled?
- What questions were asked?
- Were participants free to give any response that they preferred or did they select from a range of possible answers?
Characterizing Data
Besides knowing the nature of the sample and the methods used, it’s critical that we are able to characterise our data.
Is it numerical (e.g., consumer satisfaction ratings) or verbal, as is usually the case with interviews and focus groups? Was the data recorded in written form or in an audio recording? Video recordings provide richer data still, and may allow analysis of a greater variety of information (e.g., including participants’ movements).
Data Conversion and Analysis
Data should be converted into a format that is suitable for analysis. Frequently, qualitative data (e.g., recorded interviews) can be converted into quantitative data (i.e., numerical data). Rarely is the converse true. However often it is easier and more cost-effective to collect numerical data.
This makes it well-suited to collection at scale. Indeed, quantitative data can often be analysed more rapidly—the relationships and differences that it contains can be uncovered more quickly—adding, justifiably, to its appeal.
Communicating Data Information
Information about a dataset should be communicated simply and clearly. For numerical data this is frequently helped with a table or figure.
A clearly labelled, well-presented graph can go further in helping the audience understand a dataset’s characteristics—its ‘spread’ for example—than a written description.
Many of the above considerations influence decisions about the analyses that should be run. They also help us determine the applicability and scope of our findings.
If we have collected data on the social views of young Americans, we may be able to generalise much of what we have found in our analyses to the social views of other Americans of the same age.
Making novel inferences based on this data alone could be risky however. It might be unwise to apply the results of our analyses to the economic views of the same demographic, or to the social views of young people in other countries.
Statistical Analysis and Confidence
That said, we’re rarely interested in the sample itself. Usually we hope to make inferences about the larger population from which it has been drawn.
Statistical analysis allows us to do this with confidence. In fact, it allows us to quantify the probability that we are wrong. That is, whether the trends that are present in our dataset are there by chance and should not be generalised.
Recognizing Research Limitations
Lastly, all research is limited. Often these limitations can be traced back to the means by which the data was collected.
By determining its shortcomings we are able to generate ideas about the studies that we should conduct in future. If we have found a relationship between the number of purchases that customers make and their satisfaction ratings, we may wish to collect data that allows us to calculate the association between the amount of money each customer spends, or the types of purchases that they make, and their satisfaction ratings.
If our existing dataset contains a disproportionate number of respondents of a particular demographic we may wish to collect data that targets the undersampled portion of the population.
Knowing the limitations of research is a strength that allows us to plug the gaps in our knowledge.
Indeed, that’s what scientific research is all about: learning what we don’t know and creating a way to find it out, to get the data that allows us to test our ideas. It may not be flashy or charismatic but it is relentlessly effective. Â
If you would like to discuss this topic in any more detail with me then please get in touch!
