METHODOLOGY OF SOCIAL SCIENCESEARCH

DILIP M. NACHANE

DIRECTOR, IGIDR

15 JUNE 2009

PURPOSE OF SCIENCE

• 1. UNDERSTANDING//EXPLANATION OF OBSERVED EVENTS/PHENOMENA

• 2. BASED ON THIS UNDERSTANDING TO MAKE PREDICTIONS FOR THE FUTURE

• 3. IN SOME SCIENCES, APART FROM FORECASTING THE FUTURE, THERE MAY BE NEED TO ADVISE OR SUGGEST ON POLICY MATTERS (ECONOMICS, POLITICS, ENGINEERING, LAW ETC.)

TYPE OF PHENOMENA

• 1. REGULAR OCCURRENCE (USUALLY NATURAL PHENOMENA):

• WHY DOES IT RAIN IN MUMBAI IN JULY?

• 2. EPISODICALLY OCCURRING EVENTS:

• NATURAL : TSUNAMI, EARTHQUAKE

• SOCIO/POLITICAL : WAR

TYPES OF EXPLANATIONS

• WHY DID JULIUS CAESAR DIE ?• 1. HE WAS A HUMAN BEING AND ALL

HUMAN BEINGS DIE.• 2. POOR CHAP. ROTTEN LUCK !• 3. GOD PUNISHED HIM FOR HIS MANY

CRUEL DEEDS IN ROME AND GAUL.• 4. HIS WIFE TOLD HIM NOT TO GO OUT

THAT DAY (BECAUSE SHE HAD A BAD DREAM) BUT HE DID NOT LISTEN TO HER.

• 5. HE WAS AUTOCRATIC AND WANTED TO ASSUME TOTAL POWER IN ROME. REPUBLICANS IN ROME DID NOT LIKE THIS AND HENCE CONSPIRED TO KILL HIM.

• MOST OF US WOULD ACCEPT THE LAST EXPLANATION AS THE MOST SATISFYING.

FROM EXPLANATION TO A GENERALIZATION

• POLITICIANS AND MILITARY GENERALS OFTEN BECOME TOO AMBITIOUS AND TRY TO USURP POWER BY UNSCRUPULOUS MEANS. THEY THEN COME INTO CONFLICT WITH PEOPLE’S WILL. THE LATTER FINDS EXPRESSION IN REVOLT OFTEN LEADING TO FATL CONSEQUENCES.

CASE STUDIES

• CHARLES I OF ENGLAND, LOUIS XVI OF FRANCE, NAPOLEON, CZAR NICHOLAS II, HITLER ETC. THEN ALL BECOME CASE STUDIES WHICH FIT THE GENERALIZATION ABOVE BUT WITH WIDELY DIFFERING DETAILS.

FROM GENERALIZATION & CASE STUDIES TO A THEORY

• POLITICIANS CANNOT IGNORE PEOPLE’S WILL FOR LONG. SINCE THIS IS BEST EXPRESSED IN A DEMOCRATIC FRAMEWORK, THE IDEAL FORM OF GOVERNMENT IS A DEMOCRACY (OR A CONSTITUTIONAL MONARCHY)

PROGRESS OF HUMAN KNOWLEDGE (AUGUSTE COMTE)

• COMTE – 1842 (COURS DE PHILOSOPHIE POSITIVE)

• THREE SEQUENTIAL STAGES IN PROGRESS OF HUMAN KNOWLEDGE

• 1. THEOLOGICAL

• 2. METAPHYSICAL

• 3. POSITIVE

POSITIVE STAGE

• 1. ACCORDING TO COMTE, A FIELD OF ENQUIRY QUALIFIES AS SCIENCE WHEN IT ATTAINS THE THIRD STAGE (POSITIVE).

• 2. THE THIRD STAGE IS CONCERNED WITH THE DISCOVERY OF LAWS OF CO-EXISTENCE, ASSOCIATION AND CONJUNCTION.

CO-EXISTENCE, ASSOCIATION AND CONSTANT CONJUNCTION• LOTS OF PHENOMENA CO-EXIST

WITH EACH OTHER. ASSOCIATION IS ESTABLISHED WHEN CO-EXISTENCE IS TOO FREQUENT TO BE ATTRIBUTED TO CHANCE ALONE. CONSTANT CONJUNCTION IS ESTABLISHED WHEN THE PRESENCE OF ONE PHENOMENON ALMOST ALWAYS IMPLIES THE OTHER.

AN EXAMPLE

• SOME LEFT-HANDED PEOPLE CAN BE EXTREMELY INTELLIGENT (CO-EXISTENCE OF GENIUS AND LEFT-HANDEDNESS) BUT THIS IS NOT AN ASSOCIATION.

• A SUFFICENT NUMBER OF POOR PEOPLE ARE SICK TO ESTABLISH AN ASSOCIATION BETWEEN POVERTY AND BAD HEALTH BUT NOT ALL POOR ARE SICK, SO CONSTANT CONJUNCTION IS NOT PRESENT.

• EXAMPLES OF CONSTANT CONJUNCTION E.G. VIRUS & FEVER, FIRE AND SMOKE

COMTE’S VIEWS OF SCIENCE

• • UNITY OF SCIENCE PRINCIPLE (USP)• ALL SCIENCES (NATURAL, SOCIAL,

HISTORICAL) ARE METHODOLOGICALLY SIMILAR, BUT NATURAL SCIENCES ARE USUALLY AT A HIGHER STAGE OF EVOLUTION THAN SOCIAL SCIENCES. HENCE ALL SCIENCES SHOULD FOLLOW THE SAME METHODS OF ENQUIRY.

ERKLAREN SCHOOL

• 1. FIRM BELIEF IN USP

• 2. TASK OF SCIENCE IS TO UNDERSTAND PHENOMENA BY SEEKING TO FIND INSTANCES OF CO-EXISTENCE, ASSOCIATION AND CONSTANT CONJUNCTION.

DIFFERENCES WITHIN THE ERKLAREN SCHOOL

• THOSE WHO BELIEVE THAT SCIENCE CANNOT GO BEYOND CONSTANT CONJUNCTION (COMTE, HUME ETC.)

• THOSE WHO BELIEVE THAT THE AIM OF SCIENCE IS THE DISCOVERY OF CAUSAL LAWS (BERKELEY, KANT, J.S. MILL ETC.).

KANT’S VIEWS

• In his Critique of Pure Reason (1787), Kant tried to demonstrate that the principle of causality—namely, "everything that happens, presupposes something upon which it follows by rule," (1965 ed., p. 218)—is a precondition for the very possibility of objective experience.

CRITICISM OF ERKLAREN SCHOOL

• THE USP WAS METHODOLOGICALLY CHALLENGED BY A GROUP OF GERMAN PHILOSOPHERS (DROYSEN, DILTHEY, AND RICKERT) IN THE LATTER HALF OF THE 19TH CENTURY.THEY ARGUED FOR METHODOLOGICAL AUTONOMY OF THE SOCIALSCIENCES, IN WHICH PHENOMENA WERE TO BE UNDERSTOOD RATHER THAN EXPLAINED IN TERMS OF PRIOR CAUSES. IN SUCH SCIENCES, ABSTRACT CONCEPTS LIKE MEANING, LANGUAGE, INTUITION AND EVEN RHETORIC PLAYED AN IMPORTANT ROLE

VERSTEHEN

• The term “verstehen” was introduced to denote an admixture of these abstract concepts. While difficult to pinpoint with precision, there is no denying that the verstehen element strongly permeates the social sciences.

SOME EXAMPLES

1. WHY DID THE APPLE FALL TO THE GROUND ?

2. WHY DID VESUVIUS ERUPT IN LAVA EXPLOSION IN AD 79?

3. WHAT CAUSED WORLD WAR II?

• 1. ANSWER TO THE FIRST QUESTION IS IN TERMS OF A SINGLE CAUSE (GRAVITY) – MONO-CAUSAL EXPLANATION

• 2. ANSWER TO SECOND QUESTION CAN BE GIVEN IN TERMS OF A FEW BASIC CAUSES (POLY-CAUSAL)

• ANSWER TO THE THIRD QUESTION CANNOT BE REDUCED TO A FEW BASIC CAUSAL FACTORS.

DETERMINISM

• THE HISTORY OF SCIENCE, RIGHT UPTO THE PREVIOUS CENTURY, WAS CHARACTERIZED BY A FIRM BELIEF IN DETERMINISM, NOT ONLY IN THE NATURAL SCIENCES BUT ALSO IN THE SOCIAL AND HISTORICAL SCIENCES. GOETHE EXPRESSED THIS IN HIS INCOMPARABLY BEAUTIFUL LINES,

"GREAT, ETERNAL, UNCHANGEABLE LAWS PRESCRIBED

THE PATHS ALONG WHICH WE ALL WANDER"

• AND THUS WHETHER ONE TALKED ABOUT NEWTON'S THREE LAWS OF MOTION, OR OF RICARDO'S LAW OF RENT OR OF MARX'S LAWS OF FALLING PROFIT THEY WERE ALL UNDERSTOOD IN THIS DETERMINISTIC SENSE. IT WAS OF COURSE REALIZED THAT QUITE OFTEN THE LAWS FAILED IN THEIR PREDICTIONS - SUCH FAILURES WERE ATTRIBUTED TO CHANCE OR TO THE IMPOSSIBILITY OF HUMAN KNOWLEDGE.

THREE PRINCIPLES OF DETERMINISM

• A BELIEF THAT ALL PHENOMENA AND EVENTS OF THE WORLD OBEY UNCHANGING CAUSAL LAWS;

• A CONFIDENCE IN THE POSSIBILITY OF DISCOVERING THESE LAWS, AT LEAST IN PRINCIPLE;

• AN UNCONDITIONAL BELIEF IN THE METHOD OF FORMAL LOGIC (DEDUCTION), PRIMARILY MATHEMATICS, FOR UNDERSTANDING THE EXTERNAL WORLD.

CHALLENGES TO DETERMINISM: NATURAL SCIENCES

• MENDEL (1870) FORMULATED LAWS OF HEREDITY VIA AN EXPLICITLY STOCHASTIC MECHANISM, BOLTZMAN (1910) GAVE A STATISTICAL INTERPRETATION TO THE THEORY OF HEAT IN PHYSICS. THE DEVELOPMENT OF QUANTUM MECHANICS GAVE A FURTHER FILLIP TO THE PROBABILISTIC VIWEPOINT.

CHALLENGES TO DETERMINISM

• QUETLET (1869) USED THE CONCEPT OF PROBABILITY IN DESCRIPTIONS OF SOCIAL PHENOMENA.

QUETLET’S CONTRIBUTION TO CRIMINOLOGY

• FOR MOST OF HUMAN HISTORY DEVIANT BEHAVIOR HAS BEEN ATTRIBUTED TO POSSESSION BY DEMONS OR SPIRITS.

• FOR MUCH OF HUMAN HISTORY ATTEMPTS TO CORRECT DEVIANCE WERE BASED ON THE BELIEF THAT EXTREME PUNISHMENT COULD “EXORCISE” THE POSSESSED PERSON.

• EVEN AS SUPERSTITION BEGAN TO FADE WITH WITH EMERGENCE OF THE “AGE OF REASON” IN THE 1700’S, HARSH PUNISHMENT WERE OFTEN METED OUT TO DEVIANTS AND CRIMINALS. PUNISHMENTS INCLUDED “TRIAL BY FIRE,” “QUARTERING,” “THE RACK,” “THE STOCKS,” HANGING, AND “DUNKING.”

1. BY THE LATE 1800'S AND EARLY 1900'S, THE SWAY OF POSITIVISM AND THE FOCUS ON THE USE OF THE SCIENTIFIC METHOD HAD LED TO THE EMERGENCE OF NEW OUTLOOK ON CRIME AND EXPLANATIONS FOR CRIME. THIS NEW OUTLOOK WAS PIONEERED LARGELY BY SOCIOLOGISTS.

PROBABILISTIC MODES OF EXPLANATION

• The phenomenal growth of biological, psychological and social sciences occurring in the early parts of this century made new demands on probabilistic modes of explanation. The way statistical considerations featured in different systems could vary.

• Some systems would exhibit statistical regularity so that the underlying probability distributions could be identified, or at least reasonably approximated empirically e.G. Mendel's law, einstein-smoluchowski theory of brownian motion etc. We call such systems as statistically regular systems.

PROBABILISTIC MODES OF EXPLANATION (CONTD.)

• SYSTEMS IN BIOLOGY AND SOCIAL SCIENCES, ARE NOT DESCRIBED BY WELL-DEFINED AND STABLE STATISTICAL DISTRIBUTIONS. STATEMENTS SUCH AS "SMOKING INCREASES THE RISK OF HEART ATTACKS" OR "EXPOSURE TO SCREEN VIOLENCE LEADS TO JUVENILE AGGRESSION" ETC. ARE STATISTICAL STATEMENTS, BUT NO DEFINITE PROBABILITIES CAN BE ASSIGNED TO THEIR VALIDITY-- NOT DUE TO OUR IGNORANCE OF THE SYSTEMS OR FAULTY DATA, BUT IS AN INHERENT FEATURE OF THOSE SYSTEMS ATTRIBUTABLE TO A MULTIPLICITY OF INFLUENCING FACTORS (WHICH CANNOT BE AVERAGED OUT BY THE LAW OF LARGE NUMBERS). WE REFER TO SUCH SYSTEMS AS UNCERTAIN SYSTEMS.

PROBABILISTIC MODES OF EXPLANATION (CONTD.)

• The formal theory of philosophical explanations startS with the seminal 1948 paper of Hempel and Oppenheim, who advocated two distinct models of inference - the Deductive-Nomological (D-N) model for deterministic systems and the Inductive-Statistical (I-S) model for non-deterministic systems. In his later contributions, Hempel later introduced a third category viz. the Deductive-Statistical (D-S) model for analysing what we have called Statistically Regular Systems, reserving the I-S models for Uncertain Systems. Several limitations of Hempel's analysis were brought out by Salmon who also suggested a new model - the Statistical-Relevance (S-R) model - to replace the I-S model.

DETERMINISTIC SYSTEMS : HEMPEL’S D-N MODEL

• An event e has occurred and this has to be explained in terms of

• Antecedent conditions and,• Causal laws.• E is called the explanandum and the q antecedent

conditions (say c1…..Cq) together with the n causal laws (l1…..Ln) are called the explanans. Given the explanans, the explanandum follows with the force of logical necessity (nomic is used to express this necessity). Schematically we may set out the d-n model as follows:

• C1……cq (antecedents) • L1….....Ln (causal laws)• ____________________• E (explanadum)

AN EXAMPLE OF THE D-N MODEL

• E COULD BE THE EVENT, THAT THE PENNY AND FEATHER DROPPED IN AN AIRLESS TUBE FROM THE TOP OF A TOWER, HIT THE GROUND AT THE SAME INSTANT. THIS COULD BE EXPLAINED IN TERMS OF L1, THE LAW OF GRAVITY AND ANTECEDENTS C1 (GRAVITATIONAL CONSTANT), C2 (HEIGHT OF TOWER) AND C3 (MASSES OF THE TWO BODIES).

STATISTICALLY REGULAR SYSTEMS :

HEMPEL’S D-S MODEL

• The model itself looks as follows:

• C1……Cq (antecedents)

• L1….....Ln (causal laws, at least one of which is probabilistic)

• ____________________

• E (Explanadum)

AN EXAMPLE OF THE D-S MODEL

• CONSIDER A FAMILY OF 4 CHILDREN BORN TO A WHITE MOTHER AND BLACK FATHER. THE EVENT E COULD BE THAT TWO CHILDREN HAVE BROWN COMPLEXIONS, YET ANOTHER ONE IS BLACK AND THE REMAINING CHILD IS WHITE. THE ANTECEDENTS COULD BE THE FAMILY HISTORY OF THE WHITE MOTHER AND THE BLACK FATHER, WHEREAS THE PROBABILISTIC LAW WOULD BE MENDEL'S LAW.

UNCERTAIN SYSTEMS : I-S MODEL

• This looks as follws:

• C1……Cq (antecedents)

• L1….....Ln (causal laws - includes at least one

• which is statistical (in a vague sense))

• ============================= [r]

• E (Explanadum)

THE DOUBLE LINE INDICATES THAT THE INFERENCE IS INDUCTIVE AND [r] IS INDUCTIVE PROBABILITY IN

THE SENSE OF CARNAP (1950) OR KEYNES (1921) OR "RATIONAL

DEGREE OF BELIEF" IN JEFFREYS (1939) SENSE. ONE OF THE

REQUIREMENTS THAT HEMPEL IMPOSED ON R WAS THAT IT

SHOULD BE FAIRLY HIGH.

PROBLEMS WITH THE I-S MODEL OF EXPLANATION

• Unlike deductive inference, inductive inference is not transitive. If A → B is used to denote the fact that B follows deductively from A, then A → B and B → C together result in A → C. But for inductive inference the situation is different. Let us agree that we would say that A inductively implies B if the relevant inductive probability exceeds 0.95. Let A (r) B denote the fact that A inductively implies B with probability r = 0.95. Now if A(r1) B and B(r2) C where r1 = 0.95 and r2 = 0.97 say, then the inductive probability associated with an inference from A to C is only 0.93 which falls below our agreed cut off level, thus precluding an inductive chain from A to C via B.

PROBLEMS WITH THE I-S MODEL OF EXPLANATION (CONTD.)

• If A → B in a deductive system then adding C to the underlying antecedent A, does not affect the deduction i.e. (A & C ) → B. For example from the law that "all men are mortal" (A) we can deduce that "Socrates is a mortal" (B). Now suppose we have additionally the statement that "all men are selfish". The addition of C to A does not affect the mortality of Socrates at all. The situation changes drastically with induction. If statement A refers to the fact that "Mr.X leads a very regular and clean life" and B to the prediction that "Mr.X will live upto a ripe old age" then from A it is reasonably inductive to infer B. Howevr if C states that "Mr.X's parents died early of hereditary diseases", then (A & C) together need not lead to the inductive inference B.

PROBLEMS WITH THE I-S MODEL OF EXPLANATION (CONTD.)

The requirement of a high value for the inductive probability r is neither a necessary nor a sufficient condition for valid inference (see Salmon (1984)).

A cure for the common cold is announced which involves one week's treatment. Then people adopting this treatment would report high rates of recovery, but this could well be because most colds last anyway for one week. Thus a high value of r is not sufficient for a correct inductive inference.

Artho-sclerosis usually strikes the elderly. Hence it is perfectly legitimate to talk of old age as the cause of artho-sclerosis, even though the proportion of old people actually getting the disease is fairly small. Thus a high value of r is also not necessary for inductive inference.

• SALMON (1977) SUGGESTED THE S-R (STATISTICAL RELEVANCE)MODEL OF SCIENTIFIC EXPLANATION WHICH OVERCOMES SEVERAL FLAWS OF HEMPEL'S MODEL.

• HOWEVER, WE HAVE NOW STRAYED VERY FAR AFIELD INTO METAPHYSICAL TERRAIN.

•THANK YOU