Proper data analysis techniques are required to obtain relevant and significant results from any statistical or any ‘natural’ experiment. To effectively transform data into information, one needs three elements to come together, namely Data, Analysis, and Results Presentation. A good combination of the three would naturally include a holistic explanation of the data collection method and its associated advantages and disadvantages, the methodology used for the analysis of the data, and the ideas that come out of the results to the person who will use the information that the data, its analysis and its interpretation yield.
Raw data is data that has been collected from a data source that has not been subject to processing or manipulation. Raw data may include anomalies, oddities or inaccurate/wrong inputs, and must therefore be checked before analysis. The quality of the data can be checked through checks constructed in accordance with the nature of the data, visual inspections, Box-Whisker Plots, descriptive statistics (mean, median, standard deviation), normality (skewness, kurtosis), correlation, Cook’s distance statistics, leverage values and sample size. Variables may sometimes need some form of transformation before being subjected to analysis. However, it is always important to make sure that the transformation being effected is not one that will yield the results that the analyst would like to see, as such results would be spurious. There are several data analysis techniques that can be applied to different business, social, economic and scientific domains.
Data analysis is often dependent on data capture equipment, data storage through databases and data warehouses, and statistical analysis and Business Intelligence programs such as Oracle, MS SQL, MS Excel, SPSS, Stata and EViews. These processes, backed up by programs to automate them, allow the statistician to mine the data, to prepare it in the required form, and to subsequently analyse it to draw inferences and produce the required results. Through these results, the researcher can accept or reject an initial hypothesis on the basis of the collected data and statistical tests run on it. When carrying out data analysis on a sample that has been drawn out of a population, there is always a degree of error that the statistical analyst has to live with. The Statistical Significance of the result will determine whether the observed results have occurred due to mere chance or to a resilient statistical relationship, and the initial hypothesis will, on that showing either be rejected or accepted. Other statistical tests can be carried out depending on the type of data at hand, its characteristics and the initial hypothesis to be tested.
Once the data is analysed, useful information that supports decision-making and suggests viable issue resolutions is obtained. Clear and accurate reporting from the data analysis process is a vital step in deducing these results. It is also paramount to ask the right questions to be able to come up with the relevant answers. After observing the data, the researcher must report the statistical findings in a comprehensive manner using presentable formats that allow the reader to understand how the indicators work, their caveats, their relevance to the situation being analysed and their implications.
The team at Equinox Advisory can apply their combined statistical and economic skills to serve your needs by:
- Collecting primary data, that is, data which is not already available;
- Applying and assessing currently-available economic data to the specific circumstance(s) at hand through the understanding of its applications and limitations;
- Identifying specific changes and trends in the data and ascertaining their significance;
- Measuring relationships and associations between two or more variables;
- Applying data mining techniques to analytically process large amounts of data while detecting patterns and relationships in the data through the use of algorithms;
- Using data analysis and statistics software to interpret economic data including cross-sectional, time-series and panel-data; and
- Interpreting different data sets using assumptions and rationales that are coherent with economic theory.