| |
When selecting a method for analysis you have to consider the
chain of operations that your work shall be part of, cf. Planning a Research Project. What is the ground
from which the analysis shall start, that is, the "input" side of
analysis? Another crucial question is, how will you manage the
analysis so that its "output" will optimally serve the intended
use of the results?
The above two questions about input and output become slightly more complicated because of the fact that research always operates simultaneously in two "worlds": in the world of theory and in the empirical world. Your work will be connected to both of these environments on both its input and output edges. We thus have altogether four logical connections between your research project and its context, and each of these bonds can present its own requirements to your methods of analysis. Each of these logical associations shall be discussed below, and the cardinal questions regarding each of them are:
All these questions will be relevant when you select the optimal method of analysis and we shall discuss each of them in turn.
Theoretical input. When selecting the method of analysis it is advisable to consider whether you can base your work on a theoretical model that is already known. Sometimes a model, even a preliminary one, could help your work decisively, on the condition that you can handle it with a suitable method of analysis. Three usual approaches (which are discussed in more detail on another page) in the use of models are:
If you choose an existing model as a starting point, its format will somewhat restrict your freedom in selecting the method of analysis. For example, written models are most easily handled with the methods of case study or comparison, while mathematical models consisting of variables require quantitative methods for analysis.
Moreover, your method will depend on the time perspective of the selected model. For "cross-sectional" (synchronic) studies of static invariances which include no temporal dimension there are a great number of methods, but if you want to do "longitudinal" analysis of dynamic invariances, that is, processes, change and development, you must select one of the diachronic (or historical) methods. --- All these alternatives and appropriate methods for each of them will be summed up in Logical Structures of Analysis, below.
Theoretical output aims at enlarging our knowledge about the object of study. On all fields of science the existing system of knowledge, or theory, is arranged as consistent logical patterns and new findings of research are expected to conform to this existing configuration; typical such patterns are descriptions of objects or phenomena and explanations of processes. One usual theoretical aim of a research project is to create grounds for predicting the future behavior of the object of study.
If you have used a theoretical model as the input of your study, your theoretical output will probably have the same format and thus it will not present any additional requirements regarding the method of analysis.
Empirical input. Whichever aspects or variables of the objects of study you have chosen to collect and analyze, your logical method and tools of analysis must be able to handle them. When selecting the method, a very important criterion is the number of cases or specimens that will be studied. In this respect there are two main approaches which require completely different methods of analysis:
Empirical output is one purpose of most research projects (excepting basic research which only aims at theoretical results). If you think that the outcome of your study will be applied to an empirical situation, it is usually easy to point out those aspects or variables which will be important in the future application and thus also in your analysis.
Quite often the empirical target is to help in ameliorating the condition of the study object, for example reducing known problems in the daily lives and work processes of people, developing an activity or assisting in product development and creating grounds for new design. "Ameliorate" and "develop" are concepts that include evaluations about how the study object should be. This type of research is called normative and the methods of data gathering and analysis must be capable of dealing with evaluations.
On the basis of considering the theoretical and empirical inputs and outputs (see above) of your project you can select a method for the analysis of your empirical data by using the following table.
| Cross sectional study (no time perspective): |
One single case will be studied, or a few similar cases. Holistic view. | Case Study: - Exploratory Case Study - Case Study Based on Earlier Theory - Normative Case Study |
| A few different cases will be studied. Holistic view. | Comparative Study - Informative Comparison - Normative Comparison | |
| A large number of different cases will be studied. | Classification - Exploratory Classification - Classification Into Given Classes - Normative Classification | |
| Variables or measurements from a large number of cases are analysed. | Quantitative Analysis - Analysing Individual Variables - Analysing Relationships between Variables - Normative Study of Variables | |
| Diachronic, or historical, study: | Holistic study of the evolution of individuals or specimens. | Analyzing Development - Describing Development - Explaining a Development - Normative Study of Development |
| Study of the development of variables. | - Study of Time Series |
Note that once the analysis is finished, and before reporting its results, you should assess their validity.
In those projects which aim at practical targets, the study will often be continued by launching practical action for the betterment of the object of study, like developing an activity or developing a new product.
The goal of analysis is to arrange the collected material so that the answer to the initial problem of the project reveals itself. The problem dictates what kinds of information has to be analyzed, and on the type of information depends which tools can be used to handle it.
If you are doing descriptive research you can usually choose the problem to be studied, and select also the types of information you want to collect and analyze. If you want to stay out of difficulties you can select all the types of your material from only one row of the table below. The situation is different in normative research, i.e. when studying practical problems, you cannot omit "awkward" aspects if they are essential in the problem.
When a scientist today starts selecting tools for analysis in a new research project, almost inevitably the computer comes first in mind. Indeed, modern computers are powerful tools for analysis, but you have to remember that they have severe restrictions, too. What the machine demands, above all, is that the material it receives be suitable for electronic manipulations. For almost any type of information there are specific computer programs which can store and manipulate exactly that genre of information, but usually no other types of material. What is specially disappointing to a researcher, is that programs quite often refuse to analyze relations between different types of information, or they accept just some sorts of relationships for analysis, others not.
The most usual classes of information that you will deal with in the research and development of products include:
| "Language" or mode of information | Computer programs suitable for analyzing information in this language of presentation | Method of analyzing relations between facts given in different languages of presentation |
|---|---|---|
| 1. Measurements, in other words, quantitative study |
Spreadsheet programs like Excel. Statistics programs. |
The researcher must first "operationalize" all the factors, making them measurable. |
| 2. Classifications (presented often as codes or tabulations) | Spreadsheet or data-base programs. | Classification can handle all types of data, but their relationships only superficially. |
| 3. Verbal (written or spoken) information, in other words, qualitative study. |
Word processing programs. These have scanty tools for analysis, but you can mark with codes recurring items in the text and then classify these like on line 2, above. | Word processing does seldom help doing analysis, but it is excellent for reporting its results as text with illustrations. |
| 4. Tacit knowledge and skill of the artisan or of the user of a product. Other mental patterns like attitudes and preferences. | Choice of computer program depends on the language which you use when making explicit the tacit information. Usually it will be one of the first three above. | |
| 5. Shapes e.g. the visual forms of products. | There are many computer programs for storing and manipulating images, but their abilities in analysis are poor. | Make the analysis manually and report it as text with illustrations. |
| 6. Patterns of action, e.g. the various ways a a product is used. | You can use a computer program for storing and manipulating video strips, but it cannot do analysis. See also Developing an Activity | Make the analysis manually and report it as text with illustrations. |
As the table above already indicates, it quite often happens that you will find no computer program that could handle all the types of data that you want to analyze. In such a situation you should consider if you can "operationalize" or transform the inconvenient class of your material into one of those formats that your program of analysis can handle. This operation, which you normally have to do manually, means for example
Finally, do not forget the alternative of analyzing your data without a computer. Methods which work nearly always, include the following:
When contemplating the merits and weaknesses of various methods and tools the crucial point is certainly their ability in uncovering the hidden relationships, the invariances in the source data. Nevertheless you should not forget that the analysis tool must also present the findings that became revealed during the analysis. Many computer programs can present their results as beautiful logical models like graphs, trees or tables. However, if the analysis program is unable to produce graphics, or if you could not use a computer for the job at all, you perhaps can present the final results as diagrams made by hand or you can use a specific drawing program. Some word-processing programs also include simple drawing tools.
After the analysis phase most research projects include an important procedure: assessing the results, their factuality and perhaps also their usefulness. Quantitative statistics programs contain specific tools for this task. For qualitative analysis such tools do not exist, and the assessment is mostly done by contemplating standard lists of certain critical questions, some of which are presented elsewhere under the titles Assessing the Outcome of Literature Study, Assessing Qualitative Observations, Assessing Theoretical Output and Assessing Practical Output.
Some research projects include still special procedures like forecasting the future development of the object of study or developing an activity or an industrial product. Available tools for them are discussed on separate www-pages.
The final task in a research project, reporting, necessitates a word processing or publishing program because you will usually want to complement the results of analysis with lengthy verbal comments and explanations, and you will then need powerful tools for layout, cross-referencing, indexing, making the table of contents etc. Modern publishing programs accept the graphics produced by the usual analysis programs; if not, you can insert the illustrations eventually by hand.
| |
January 18, 2005. Original location:
http://www2.uiah.fi/projects/metodi
Comments to the author:
