Manuals methodology

How to apply generated knowledge to address actual problems?

Applied questions are asked to solve a specific social, political, organizational or commercial problem. Applied questions do not aim at generating general knowledge. Some elements belonging to ‘scientific research designs’ are therefore not relevant for and not included in ‘applied science project descriptions’. Of course, applied questions can only be answered systematically if some empirical research questions have been answered first. Otherwise there is nothing to apply.

The systematic discussion of research designs and data collection methods in introductory text books occur in the context of ‘generating general knowledge’ and ‘inference’. The area of applied science is less standardized. On this website we distinguish between several types of applied research: descriptions, evaluations and predictions, remedy selections and designs.

Descriptions and interpretations

Examples: what are the needs of our clients? How do people feel about our company? What happened during the decision making process that led to the adoption of the law banning smoking from public places?

Descriptive questions in the context of applied research are similar to descriptive questions in empirical research and oftentimes involve ‘inference’ too, because we often want to say rather abstract things (‘quality’) about abstract units (‘the school’). The aim of this type of research is to describe characteristics of a limited (finite) set of units (clients, people, companies). In this type of research you start with a conceptualization of the variables you intend to use. In the literature you can find models, typologies, or classifications (sometimesd confusingly called ‘theories’) to describe your units. In the data analysis section, you may want to break up the results to different sub-sets (man/women; different units within the company). Note, however, that questions of inference are still relevant. Does the way you measure your central variable(s) allow you to make a general statement; and does the study of a subset of the units allow you to say something about all units? Theory can be relevant too. If you want to describe and understand why something happened in a particular case, you need a theory to decide what aspects of reality to look at.

Predictions, ex post and ex ante evaluations

Examples: Did the policy work as intended? Will the decision work out as intended? What are our market prospects?

Evaluations can be asked before (ex ante) or after (ex post) a decision is made. If the evaluation is asked before the actual decision, the only research strategy available is to use existing (correlational or causal) knowledge about the known effects of some course of action to explain why the decision may or may not work as intended.

Example: suppose that a manager wants to know whether the strategy to target a specific set of customers will work out, given some known market conditions. This means the manager has to turn to some theory helping her to understand the possible consequences of this type of action.

If the evaluation is asked after the actual decision is made and after possible effects may have occurred, a researcher can also use an interrupted time series design to study the possible effects of the decision. In that case the design is similar to ‘standard’ explanatory research designs.

More information on evaluation research can be found here.


Examples: which of a set of strategies is best under the current circumstances?

If a set of potential remedies is available, you have to select these remedies on the basis of a set of criteria including availability, costs, and effectiveness. The basis of this comparison is a list of all potential remedies, a set of criteria and a ‘score card’. If you put the potential remedies in the rows and the criteria in the columns of a matrix, the scores can be put in the cells. To establish the effectiveness of the various remedies, one needs (the results of) empirical research, showing that a specific remedy is better with some criterion than some other remedy. Techniques are developed in cost-benefit analysis and multi-criteria analysis to assess the score cards.


Example: what should we do about this problem?

Everyone can build a house, but will it meet the formal requirements, will it stand a storm, and will it be affordable? Everyone can design a policy to reduce the amount of litter on streets, but are you allowed to implement that policy, how much will it cost and will it help? Some people will argue that this is not a scientific question, but scientifically established knowledge can be useful to design something new.

If you plan to design systematically, you first have to establish the goals and to define a set of criteria and demands which have to be met. Secondly, existing designs (when available) have to be evaluated against this set of demands. Thirdly, a new design is made using as much information as you can. This design is made individually (by just being creative) or in a group (as in a ‘brain storm session’). Fourthly, the news design is evaluated using the pre-established set of criteria, (compared with existing designs if available) and tested. Of course, this is an on-going process and may include a repetition of step 3 and 4 several times. The ‘testing phase’ is similar to the aforementioned evaluation studies and may even include a proper experiment. In the literature the steps in a design process are presented in various ways, but include all four steps mentioned here.

Basic readings

Babbie, Earl (2004). The Practice of Social Research (12th edition). Belmont: Wadsworth/Thomson. Chapter 12.