A+ A- Reset
Home Resources Data Analysis and Reporting
  • Resources management groups
  • Resources management groups
  • Resources management groups
  • Resources management groups
Data Analysis and Reporting

Data Analysis and Reporting


Data analysis comprises of a collection of methods to deal with data/information obtained through observations, measurements, surveys or experiments about a phenomenon of interest. The aim and purpose of data analysis is to extract as much information as possible that is pertinent to the subject under consideration. The nature of the subject and also the purpose of analysis may vary greatly. The subject could be physical, social or economic and the purpose of the analysis could be purely academic or practical. Due to the great diversity of statistical data, the methods of analysis and the manner of application differ significantly from situation to situation. One cannot possibly expect a single unified system of techniques to be applicable to all cases. However, we have several formal methods of analysis are more or less mutually related and have been successfully applied to most, if not all statistical data.

We can classify data into several types according to some few criteria; qualitative or quantitative in line with the property of the observation or measurement, univariate or multivariate according to whether one or more observations or measurements are taken per subject respectively and finally cross-sectional or longitudinal according to whether measurements are taken at a point in time or over a period of time. Thus each different type of data will require a different type of analysis technique.

Reporting of the results will be largely guided by the purpose of the study undertaken and also the audience to which the results are targeted.

Data analysis, modelling and reporting is generally carried out through a set of steps. These steps are however interrelated and the process of data analysis reviews each step on the basis of both the previous and the subsequent step. We can list the steps under five headings as follows;

  1. Data preparation for analysis
  2. Data summarization and visualization
  3. Data analysis and modelling
  4. Discussion and report writing

Data preparation for analysis

Firstly, before any analysis is carried out, data is entered into a computer system (if not done already) and has to be checked, updated and validated. Next, it is crucial to read and re-read your data in order to know it inside-out. Get completely familiar with your data before start of any analysis.

Secondly, there follows data manipulation into a structural form suitable for analysis and which the software to be used in the analysis can upload easily. This could involve copying, selecting subsets of the data, transforming and merging the data at various levels. Transforming and deriving new variables from existing ones. This might involve coding and categorizing existing variables. Coding can simply be defined as a process

Thirdly, it is crucial to identify the data structure, sometimes known as data layers of the data/information to be analysed. For example, you may have farms, plots within farms or farms, animals within farms or forests, species within forests etc.

Fourthly, the unit of measurement has to be clearly defined. For example in a survey that involves farms, information taken will be at individual or at farm level. At the farm level, we may record number of trees, number of persons, size of the farm in acres and at persons level we could record their ages. Further, we should consider which of our data are quantitative and which are qualitative. [Good statistical practice for natural resources research, Chapter 15, http://www.ssc.rdg.ac.uk/software/sscstat/helpfile/stepsinanalysis.htm http://www.ssc.rdg.ac.uk/software/sscstat/helpfile/gpguides.htm]

Data summarization and visualization

Various methods to achieve data summarization and visualization include; Tabulations, X-Y scatter plots, Boxplots, Category-value plot, Normal probability plots and density estimates among others. The aim of data summarization and visualization is to bring about the main features of the data and to guide in choosing the appropriate statistical techniques. Furthermore, these procedures will help in spotting errors and unusual values in the data matrix allowing corrective steps to be taken early in the analysis.

The process of data summarization and visualization can be appropriately linked to the concept of exploratory data analysis (EDA) whose approach and philosophy is to allow the data to dictate the process and techniques of analysis to be employed. EDA is carried out through a variety of mostly graphical techniques. It allows the data analyst to obtain maximum insight into a set of data by revealing the underlying structure, spotting outliers and anomalies as well as testing underlying assumptions among other things. [The green book, chapter 4.7
Good statistical practice for natural resources research, Chapter 15, http://www.ssc.rdg.ac.uk/software/sscstat/helpfile/stepsinanalysis.htm, http://www.ssc.rdg.ac.uk/software/sscstat/helpfile/gpguides.htm, http://www.itl.nist.gov/div898/handbook/eda/section1/eda11.htm]

Data analysis and modelling

Following data exploration through descriptive statistics and data visualization that are guided by objectives of the study, we identify appropriate statistical analysis techniques (if any) needed to do further data investigation that will assist in answering the objectives of the research study in order to draw sound conclusions. Assumptions associated with the identified statistical techniques require to be confirmed that they are met. Data may arise in the context of two broad areas of statistical inquiry, experimentation and surveys. Methods of analysis chosen usually reflect the kind of inquiry from which the data arose. Furthermore the type of measurements (ordinal, nominal, continuous and ratio) taken during data collection dictate the kind of statistical techniques and models to be employed in the analysis. For example, for survey data see Statistics in Survey Analysis, Steps in Survey Analysis, Approaches to the analysis of survey data.

Multivariate methods, allow the simultaneous analysis of a dataset to explore its overall structure, to measure redundancy in the measurements, to summarize the salient features of a study and to form groups of objects or individuals with common characteristics. Principal component analysis, factor analysis, correspondence analysis, cluster analysis, discriminant analysis and canonical analysis are the most widely used multivariate methods. The methods are commonly used as exploratory tools to help data analysts find hidden structure in their data. They are also useful for data reduction. [Modern Methods of Analysis].

Other useful resources include [Analysing data from participatory on-farm trials, Analysing ranking and rating data from participatory on-farm trials, Confidence and Significance, The Green Book, Chapter 4.7, http://www.ssc.rdg.ac.uk/software/sscstat/helpfile/stepsinanalysis.htm

If and as when faced with difficulties it is recommended that you seek help from an experienced statistician.

Discussion and report writing

It is important that the results of analysis are conveyed or communicated to other parties in clear and unambiguous manner. Technical jargon should be avoided as much as possible. Almost always, results of an analysis are used to make decisions such as to repeat the study (with modifications), to perform another study altogether or even to influence other actions (eg. policy). Report writing is the process through which we share the findings and/or results of data analysis, ofcourse we can also share the same verbally. Ideally, the report should begin with a brief outline of the main point of the study or experiment that was conducted and it’s purpose. It is also advisable to indicate time and place where the research was carried out. A systematic outline of the procedures undertaken during the data analysis including modelling should be stated clearly. The report should also capture the main findings of the study and should point out the next course of action. [Writing up research: a statistical perspective. Statistical Services Centre, Scientific writing for agricultural research scientists]

Useful Reference:

Arifin, B.,; Swallow,B.; Suyanto; Coe, R. World Agroforestry Centre (ICRAF), Nairobi (KEMYA) 2008. A conjoint analysis of farmer preferences for community forestry contracts in the Sumber Jaya watershed, Indonesia. --Nairobi, Kenya: World Aggroforestry Centre (ICRAF) (CRAF Working Pater n. 63, 36,[B5527] 556.51 AZE

Coe,R.; Lusiana,B. World Agroforestry Centre (ICRAF), Nairobi (Kenya) 2008. Review of methods for researching multistrata systems. --Nairobi, Kenya: World Agroforestry Centre (ICRAF) ICRAF Working pater No. 75, 8p.