"Approaches and Strategies of Social Research"

Essay for Reasearch Methods Class ST700


The Inductive approach to enquiry builds generalisations out of observations of specific events. It starts with singular or particular statements and ends up with general or universal propositions.

It presupposes that explanations about the workings of the world should be based on facts gained from pure, dispassionate and neutral observation, rather than on preconceived notions; that nature will reveal itself to a passively receptive mind.

The Inductive strategy assumes that all science starts with observations which provide a secure basis from which knowledge can be derived and claims that reality impinges directly on the senses, hence there is a correspondence between sensory experiences, albeit extended by instrumentation, and the objects of those experiences. The conclusion of an inductive argument makes claims that exceed what is contained in the premises and so promises to extend knowledge by going beyond actual experience.

The more observations that demonstrate, say, a relationship between phenomena, the higher the probability that the general statement is true. Verification of derived generalizations comes through observations about particular phenomena that appear to support it.

Critics of this approach claim that: it is essentially descriptive and does not really explain anything as it fails to uncover the causes of the generalized conjunctions; there is no purely logical inductive process for establishing the validity of universal statements from a set of singular ones; it is impossible to make the infinite number of observations required to prove the universal statement true in all cases and; is objectivity possible when observations and their analysis are made by people who have some view of the world arising out of their particular discipline?

The Deductive (hypothetico-deductive or falsificationist) approach is the reverse of an Inductive one. It begins explicitly with a tentative hypothesis or set of hypotheses that form a theory which could provide a possible answer or explanation for a particular problem, then proceeds to use observations to rigorously test the hypotheses.

The Deductive argument moves from premises, at least one of which is a general or universal statement, to a conclusion that is a singular statement. Deductive propositions form a hierarchy from theoretical to observational; from abstract to concrete. The Deductivist accepts that observation is guided and presupposed by the theory.

Attempts are made to refute the hypotheses through rigorous criticism and testing. If the data derived by testing the hypothesis is not consistent with the predicted conclusions, the theory must be false. Surviving theories are corroborated, but are never proved true despite withstanding testing and observation. A current theory is superior to its predecessors only because it has withstood tests which falsified its predecessor.

Critics of this approach claim that: where a theory has not been falsified, its acceptance relies on data that lend 'inductive support'; Deductivists are reluctant to deal with the process by which hypotheses come into being; whether Deductivism provides any rational basis for choosing between all unrefuted alternative theories in order to make some practical prediction. The Inductivist position is that the truth of theories could be conclusively established. The Deductivist position claims that while the pursuit of truth is the goal of science, all scientific theories are tentative. Neither Induction or Deduction contributes a single new concept or new idea.

Combined approach:
A scheme has been proposed by Wallace (1971) that combines Inductive and Deductive strategies to capitalise on their strengths and minimise their weaknesses creating a cyclic process that allows for movement between theorising and doing empirical research while using both Inductive and Deductive methods of reasoning.

Retroductive research strategy involves the building of hypothetical models as a way of uncovering the real structures and mechanisms which are assumed to produce empirical phenomena. The model, if it were to exist and act in the postulated way, would therefore account for the phenomena in question. In constructing these models of mechanisms that have usually never been observed, ideas may be borrowed from known structures and mechanisms in other fields.

A phenomena or range of phenomena is identified, explanations based on the postulated existence of a generative mechanism are constructed and empirically tested, and this mechanism then becomes the phenomenon to be explained and the cycle repeats. p168

Peirce regarded Retroduction or 'hypothesis formulation' as being the first stage of an enquiry. It is a process akin to finding the right key for the lock, although the key may never have been observed before. The hypothesis must be tested using both Deduction and Induction; in the second stage of an enquiry, consequences are deducted from the hypothesis and, in the third stage, these consequences are tested by means of Induction. He suggested that a hypothesis must eliminate puzzlement as a necessary first step.

Retroduction/Abduction occurs in the context of ontological, conceptual and theoretical assumptions; the researcher does not start with a blank slate in the manner implied by Inductivisits. Quasi-accessible mechanisms can be discovered from empirical studies of an exploratory kind with input from an associated field of knowledge in which some process is used as an analogy for the one under investigation. p 169.

Retroduction differs from Induction which infers from one set of facts, another set of facts, whereas Retroduction infers from facts of one kind, to facts of another. Unlike Deductive reasoning, Inductive and Retroductive reasoning are synthetic or ampliative because they make claims that do not follow logically from the premises. In addition, neither Induction nor Deduction can produce any new ideas. On the other hand, Retroductive/Abductive reasoning involves making an hypothesis which appears to explain what has been observed; it is observing some phenomenon and then claiming what it was that gave rise to it.

The Abductive research strategy is used by Interpretivism to produce scientific accounts of social life by drawing on the concepts and meanings used by social actors and the activities in which they engage.

Access to any social world is by the accounts given by the people who inhabit it. These accounts contain the concepts that people use to structure their world - the meanings and interpretations, the motives and intentions which people use in their everyday lives and which direct their behaviour.

Abduction/Interpretivism acknowledges that human behaviour depends on how individuals interpret the conditions in which they find themselves and accepts that it is essential to have a description of the social world on its own terms. It is the task of the social scientist to discover and describe this world from an 'insider' view and not impose an 'outsider' view.

A position taken by Douglas rules out experimental situations. Everyday life is studied in its own terms - the members' understanding, and only methods of observation and analysis that retain the integrity of the phenomena should be used.

Abduction is applied when attempting to move from lay accounts of everyday life, to technical, scientific or expert descriptions of that social life. p 177 Abduction is a developing strategy with on-going debate on how best to move from lay language to technical language. There are differences of opinion with regard to retaining the integrity of the phenomena when moving first order constructs (people's views and explanations), to second order constructs (the social scientist's interpretations).

The Abductive strategy has many layers to it. There is some difficulty in preceding to the final stage in which social theories might be generated from the second order constructs or that these social scientific descriptions can be understood in terms of prevailing social theories and perspectives, leading to the possibility of an explanation or a prediction.

Some positions argue that the research should go no further than to sort through, devise categories for and pigeon hole the various constructs provided by the social actors within the study.

The Abductive/Interpretivist approach and has been advocated as either the only approach for social sciences, or an adjunct to other strategies.

On Finding a Suitable Strategy:
I am a novice to the social sciences having had no prior contact with the difficult philosophical arguments pertaining to ontology and epistemology. Ouch! My head hurts. Obviously there wouldn't be so much controversy and vigorous debate over strategies for social enquiry if there was an approach which all agreed was the correct one.

Even amongst the protagonists of a particular strategy, there is, like so many brands of religion, disagreement about what exactly constitutes the process. And yet in this struggle to articulate and refine an appropriate model for all research endeavour, I see body of thought within the prescribed texts which gives credibility to notions that I am very familiar with in my own discipline of animation, multimedia and film making. Such notions as...

"invented or guessed at."
"eliminating the puzzlement."
"finding the right key for the lock."
"the surrender to the insistence of an idea."
"spontaneous conjectures of instinctive reason."
"every discovery contains an irrational element."
"being overwhelmed by an idea that cannot be resisted."
"science is more sloppy and irrational than its methodological image."
"free creations of our own minds, the result of an almost poetic intuition."
"a rapid reciprocation between imaginative conjecture and critical evaluation."
"chaos or opportunism has an important function in the development of theories."

It is here in some of the 'left-over' messy bits of the recognised strategies of social enquiry where I can find approaches that will be useful to my own research. Such notions are not allowed for in the Inductive camp, are partly used but are unacknowledged by Deductivists but find acceptance in Retroductionism.


Description of Research Proposal:

'Clarence Tells All'. This research project will investigate ways in which an animated cartoon character named Clarence, 'inhabiting' an interactive computer environment appears to act intelligently, be aware of its surroundings, remember events that it encounters and is able to recount these observations in fluent spoken word. The project will develop software and animation routines to make this illusion possible. This research will culminate in the production of an interactive game in which the user can define various missions for the animated character and be able 'colour' its reportage in various ways.

The justification for the nature of the project and the area of the research chosen is based on the premise that any explorations, albeit playful ones, within the area of computer interface design, might throw up useful applications for future human/computer interaction.

What I won't be doing:
Finding any one of the common strategies used in social enquiry that will be entirely applicable to my research proposal is somewhat problematical. As you can see, my proposal is not about social research. I will not be observing social behaviour or be involved in collecting data from people, even an informal survey of reaction the project's outcomes. Therefore I believe an Abductive strategy can be safely rejected.

Nor is the proposal purely scientific. I have no 'why' questions and therefore do not need to develop an overriding hypothesis. I have no data to collect from which theories can be drawn. I have no theories to test. It does not need to explain a phenomena. This project does not have phenomena under investigation, rather it seeks to create a phenomenon through investigation!

What I will be doing:
My proposal is by project, not by thesis. From a hunch based on prior experience and knowledge in related fields, from what my own prior experiments suggest and from observations of what has already been achieved by others, I have set myself a specific tangible working outcome which I suspect is possible. How and by what means I shall achieve this is far less certain and will form the research component of my project - my 'what' and 'how' questions.

Research Questions:

• What software mechanisms will be necessary to track and record data within a randomly generated world?
• What kind of software-driven animation routines will be required to depict this process?
• What software mechanisms will be necessary to specify a mission and selectively interpret the acquired data in various ways?
• How can the collected data be related back to the user so that Clarence's spoken account of the day's events seems fluent and natural?

The project does not assume a blank slate. Therefore one research strategy to be adopted will be to review the current state of knowledge, looking specifically into areas such as the generation of random events in computer games and screen savers, speech synthesis and recognition, text based games in which the computer plays psychotherapist, Artificial Intelligence techniques, and the methods used in relational data base software.

The proposal consists of research into artistic, technical and technological issues. The research will primarily be about problem and puzzle solving, particularly in the design of the software which controls the animated events. The prime research strategy will be to discover the means and ways to an end through iterative experimentation, exploration and testing.

In part the project is also testing the current state of the art of computer technology. The thing will either work as I envisage, partly work, or not work at all. Failure to work could be because the project was too ambitious, the technology needed to realise it isn't sufficiently developed yet, it was never going to work anyway or because the researcher was not talented enough. Any of these outcomes are acceptable, bar the last, as I will be able to report why.

Research Methodology:
My project can be divided up into three main areas, for which I can describe an appropriate methodology. They are: Script Design, Animation Design and Software Design.

Script Design:
The methodology here involves the progressive reworking of the script through a number of drafts in order to improve on the screenplay's ideas, define plot points around which the drama swings, strengthen its characters, clarify their motivations and to retain the simplistic essence of the premise in a way that can be rendered using the language of cinema rather than of the written word.

It is considered a useful exercise to encapsulate what a film is about within a simple premise. Harry is in love with Jane but her affections lie with Harry's pet warthog. Harry decides to kill two birds with one stone and over a romantic candle lit dinner, serves Jane pig trotters in wine sauce. As the script develops, it informs the premise and the premise, in turn, keeps the script on the rails.

Like the recipe for a cake, there are ingredients and formulas that are supposed to make a better script. They include such notions as: each character should have distinctly different and distinguishing characteristics; when characters are put in dramatic conflict, made to disagree over something, fight for a possession or desire to win against one another, it makes for more interesting stories; characters who undergo a change in their nature through trial and circumstance, are also interesting. Action content, emotional content, etc.

Invented character histories also part of the methodology of script writing. Although this additional background information does not necessarily play a part in what is to be shown, it can underpin the reasons for a character behaving the way it does, giving it depth and believability which can translate to the screen. Clarence has such a history.

The environment in which Clarence observes events will be depicted as his work place within the records office of some public service department. The information he collects here will be related back to his widowed mother when he returns home.

In order to test the underlying software mechanisms of this project, a rich and varied set of events will have to be scripted. These events will have social, emotional, factual and statistical qualities and will range from the inconsequential to the dramatic. It will not be a strictly narrative script in the traditional sense, as the various events depicted will be randomised by the software in both order and in time to produce a believable cartoon office environment. However each event will need to be prescribed and the larger office subplots that come from the relationships between the characters considered so that when put together in various ways, they provide entertaining viewing.

Animation Design:
An animator's craft skills are based on observations of the way things move in the real world, interpreted for the stylized world of animation. It is a partly intuitive and partly a process of intellectual analysis. The methodology of animation and film making is well tried and tested. This process is always informed by the requirements of the script and it runs something like this:

• Storyboarding - an interpretation of the script using visuals to depict events.
• Animatic - recording these static pictures in time to get a sense of cinematic language and narrative development.
• Design of characters. The design and look of characters is important for the way the audience feels about them. A boss who is drawn big, mean and intimidating through the attitude of its body and build, immediately establishes a stereotype that is useful when we do not have hours in which to tell a story or get to know the characters more intimately. Whether hero or villian, Disney placed great emphasis on character 'appeal' which comes through its design.
• Design of settings including any device a character may need to use, doors, chairs etc.
• Track reading - where dialogue is invovled, the sound track must be carefully analysed.
• Staging the action within the scene, where is a character placed and where it moves to.
• Timing the action - blocking out action in time with reference to plot and dialogue.
• Key poses - drawings of the critical poses that characters will adopt during the sequence. These describe 'what' happens.
• Pencil Testing - recording the key poses to video tape.
• Refining the key poses in the light of feedback from the above and perhaps re-testing.
• Inbetween - a process of linking each key pose with the next. It smooths out the animation and unlike the raw key poses, describes 'how' an action happens.
• Pencil Testing - recording the above drawings to video tape. Review and re-test.
• Clean up and trace off - cleaning up the rough drawings and tracing these off to applying the required style of line-work.
• Colouring

Software Design:
Although software must be logical to its bootstraps if it has any hope of being interpreted by a computer, it is interesting to speculate about two pieces of software coded by different programmers. Although the outcomes of each computer program might be identical in terms of their performance, what they do and how the achieve it, unlike, say, two artist's interpretation of a horse, the approach that each programmer took in its design, could be vastly different. Software programming is a creative task, but one that must work within the constraints of Mechano-like building blocks. Curiously, the creative process at work during its design is usually invisible.

The major component of this project will be based on the design of its software. The research strategy I will adopt will be to develop a number of experimental prototypes commencing with a software only version to test and refine the software and animation mechanisms. Animation routines will then be progressively linked to the software and tested, building further complexity over a series of prototypes.

It would seem that Retroductive reasoning would be a useful approach to determine why the software might fail, as when the theory is true, it no longer produces 'why' questions. If my software code can be said to represent my theory/hypotheses, I'll predict that it will operate in a certain way. I will test the software and observe its behaviour. When it fails I will speculate why and develop another approach. One difference is, however, that I will be directly building the mechanisms (software code) to make this model or prototype work rather than a model which can reveal a mechanism.