Issues in Scientific Explanations

1
Issues in Scientific Explanations

• The phenomenon
• What is "explained" by the theory?
• What is the "contrast space"?
• Is the theory itself tested/expanded, or used to
  describe some phenomenon?
• What are the structural presuppositions? what is
  given in the explanation?
• Psychological processes the theory invokes to
  explain the phenomenon
• What are the core social and/or psychological
  concept(s) or process?
• What "level of explanation" is appropriate?
• When does a different level of explanation alter
  the nature of the phenomenon?
• Pragmatic criteria for explanation
• necessary sufficient cause
• articulation with accepted principles

2
explanations

• Descriptions of causal relations among the terms
  of the theory
• What does "cause" mean? material, efficient,
  formal, final
• Mediators, moderators, interactions....
• Measurement v. experimental v. intervention
  designs
• Linear v. non-linear relation between cause and
  effect.
• Prerequisites and/or boundary conditions around
  the causal model/theory
• Presuppositions or untestable assumptions
  implicit in the theory or explanation
• Boundaries on the phenomenon groups, settings,
  time (moderators).
• How much does the theory permit or encourage
  enlargement to incorporate additional explanatory
  concepts and/or a broader range of phenomenon
• Open v. closed system.
• Comparison of theory with competing or
  alternative explanation
• Mutually exclusive?
• Complementary?

3
The main tasks in my idiosyncratic view

• 1. What is the phenomenon? What exactly will
  your research do?
• Will it simply describe something? What?
• Will it explain something?
• Will it test an existing or new theory?
• The adequacy of the theory itself?
• A derivation of the theory?
• The power of theory-based variables relative to
  variables derived from another theory?
• Will it use a theory to explain some concrete
  events?
• Will it use a theory to construct and test an
  intervention?

Convergent research / use of theory
Divergent research
4
Some Key terms

• 2. Contrast space
• What is being compared to what -- what are we
  actually trying to explain?
• At what level are we explaining it?
• Products
• Direct effect and Mediator analyses
• Explanatory theory of how the phenomenon works
• How do alcohol drugs increase sexual risk among
  gay men?
• drugs v. other causes?
• gay v. non gay?
• sexual v. other risks
• increase v. decrease?

5
Some Key terms

• 3. Boundary conditions
• What are the conditions under which this theory
  applies / this hypothesis is supported?
• Under what conditions might the hypothesis be
  reversed?
• McGuire
• Counterfactuals as hypothesis generators
• No hypothesis is false
• Products
• Moderator analysis
• Explanatory theory of larger variables that
  control when the phenomenon occurs.

6
Explanatory frame The object to be explained
the form of explanation.

• What does it mean to explain something? What,
  exactly, is being explained?
• Constant v. variable terms what needs
  explaining?
• Why is BSB so weird?
• BSB is this way because the architect wanted a
  dramatic building to win awards
• why are there weird buildings?
• why is BSB one of them?
• why this form of weirdness
• Why did the rabbit get eaten by the fox?
• Micro question (this rabbit by this fox)
  proximity to tree, degree of light, time of day
  the rabbit was in the wrong place, the fox was
  hungry...
• mid-macro question fluctuation in relative
  fox/rabbit populations, explain current rabbit
  predation rate (or fox population!)
• Macro question co-evolution ? rabbit predation
  by foxes generally
• Why do people take drugs?
• ...to be more happy, to be popular, etc. (All
  are K, but drug taking is not)
• ...why do people take crack cocaine? (Dont ask
  me why I smoke, ask me why I smoke Winstons...)

7
Explanatory frame (cont.)

• At what level are we explaining the phenomenon?
• Are we explaining its occurrence?
• how it works?
• how to fix it?
• When is a different level of explanation actually
  explaining a different phenomenon?
• When are two explanations consistent with each
  other?
• When are two explanations irrelevant to each
  other
• Complimentary wave particle explanations of
  light
• better than another?
• actually explaining something different?

8
Explanatory frame (cont.)

• How do we know what causes something?
• Material causes
• Efficient cause
• Formal cause
• Final cause
• How do we decide if something is really the
  cause?

The concrete stuff something is made of.
Simple co-occurrence / correlation
The larger structure or system the outcome is
embedded in
The purpose or larger meaning of a process.
9
Causality the case of the biting dog

• I kicked the dog and he bit me. Why?
• Simple covariate
• Every time I kick a mean looking 4-legged animal
  I get bit.
• Material (reductionism)
• Dogs are equipped with teeth and a defensive
  biting reflex.
• Functional or efficient
• I kicked him hard enough to get him mad at me
  the kick directly caused the bite.
• Formal or structural
• The dog and I have a really sick relationship
  where I keep kicking him and he keeps biting me.
• Dogs have evolved over millennia to protect their
  territory, and I aggressively invaded his.
• Final
• Dogs bite to mold human behavior (make us not
  kick).

Eternal question Are these causes all
explaining the same thing?
10
What caused the Challenger space shuttle to crash?

• Highly vulnerable tile design
• Falling foam from the booster
• Damaged insulation tiles
• Hot plasma in the wheel well
• Loss of control of the vehicle at reentry
• Habituation to tile falling (and other
  anomalies) loss of recognition of debris as a
  problem
• Poor communication between engineers and
  management
• Poor decisions under powerful political pressure
  to show success for expensive manned space program

11
Causality the Challenger space shuttle crash
"Challenger, like Columbia, was an institutional
failure. That is, it wasn't just a matter of the
decision-
making structure. It had to do with the entire
organization and its culture, and the critical
parts of that really didn't get changed after
the Columbia failure." -- Diane Vaughan,
author of The Challenger Launch Decision
12
10 of the U.S. population is depressed. Why?

• Simple covariate / descriptive (e.g.,
  epidemiology)
• Lower socio-economic status women and upper class
  youth get depressed.
• Material
• Serotonin depletion or high re-uptake rates
  underlie depression.
• Genetic / brain -based negative affectivity.
• Functional or efficient
• After adverse events or stress people tend to get
  depressed.
• Negative thoughts make people depressed.
• Depression occurs when a variant on the 5-HTT
  gene gets switched on by stress.
• Formal or structural
• Depression reflects the mismatch between human
  evolution and the evolution of our physical /
  technological / economic environment.
• Depression is the expression of a persons
  social class position.
• Final
• The purpose of (transient) depression is to help
  us transition from one stable array of
  reinforcers to another.

• Each of these
• Are subtly different questions, that dictate
  different contrast spaces
• different theories
• different hypotheses
• different research strategies
• Are these different / competing / complimentary
  explanations of the same thing?
• Suggest different prevention / treatment
  strategies

13
Explanatory relativity and contrast spaces Dog
bites man

• When I kicked the dog he bit me. Why?

• How many possible contrast spaces are there in
  this empirical question?
• Each contrast space defines an explanatory
  space
• An independent / dependent variable
• A term to hold constant / boundary condition.

14
Explanatory relativity What are we explaining?
What is the contrast space?

• Why do people get depressed instead of calm,
  affectless, violent
• Material structural theories of brain function,
  affect and social learning.
• Functional / efficient theories of social
  structure / resources, stress and coping.
• Why / how do these people instead of others
  get depressed?
• Descriptive / covariance epidemiological
  questions.
• Functional / efficient individual difference
  theories of cognitive vulnerability, social
  support
• Structural theories experience and genetic
  expression, temperament.
• When / where / how does depression occur instead
  of then?
• More structural developmental or gender-based
  theories.
• Functional theories within-person differences in
  psychosocial variables.
• Specification of possible boundary conditions.
• Why is she not depressed now whereas others
  still are?
• Application of functional or structural theories
  to developing and testing interventions.

15
Some random terms

• Explanatory frame The object to be explained
  the form of explanation.
• Structural presuppositions.
• Nature versus nurture what of major depression
  is (alcoholism, smoking) genetic?
• Reductionism and Reducability
• When is a reductionist explanation actually
  addressing a different phenomenon?
• What is love?
• Merging of soul mates
• Search for meaning and intimacy
• Economic / reproductive contract
• Displacement of arousal
• Evolutionary response to similarity
• Neuro-chemical trigger event
• (my love is chemical)

What is being explained? human attraction at
all attraction to this person intensity,
duration, action potential of attraction pair
bonding cultural differences economic forces
on social behavior Different explanations or
different phenomena?
16
Some useful (?) distinctions

• Three steps in convergent research taking a
  phenomenon or empirical relation and developing
  or applying a theory to explain it, or testing
  the relative adequacy of diverse theories.
• Find / specify a phenomenon stressed people get
  sick a lot.

Then specify a basic explanatory theory Stress ?
Illness effects are caused by immune
suppression.
17
Convergent research, 2

• Consider other theories that may explain the
  phenomenon or data pattern. (That may also lead
  to different levels of explanation).
• What other mediating variable(s) (beside or in
  addition to immune function) may account for the
  effect of stress on illness?
• What other psychosocial variables may lead to
  physical illness?
• What (more exogenous) variables may control both
  your predictor and your mediator?

18

• Convergent use of theory application test
  diverse possible mediators

19

• Convergent theory application testing several
  possible theories

? Stress, Learned helplessness negative affect
Immune function
Illness
Resilience Heartiness, optimism
High (physiological) arousability
20

• Conceptualizing / testing a structural exogenous
  variable

Socio-economic status
Stress, negative affect
Immune function
Illness
Of course each link is a core research question
21

• Adding two levels of endogenous mediating
  variables to a structural exogenous variable

Immune function
Stress, negative affect
Negative health behavior
Illness
Socio-economic status
Exposure to pathogens
Structural cultural barriers to health care
Typical psychosocial conceptual frame
More public health approach
22
Convergent research.

• 3. Turn the relation on its head
• Under what conditions might stress ? health?
• What moderators create boundary conditions to
  the theory or even reverse a common effect?

Stress negative affect
Interaction of stress by ? resources
Health status
Immune function
Resilience heartiness, social support
23
Divergent research

• Take an established theory and apply it to new
  and novel contexts.
• Simple extension to new domains
• Does motivational enhancement work for
  non-problem behaviors?
• Can cybernetic models of information behavior
  relations be applied to self-regulation of
  health?
• Can a physical fatigue like model explain
  failure of self-regulation over time?
• Shifts in levels of explanation
• Can genetic theories explain individual,
  contextual, or cultural differences on mood?
• Can political events or socio-cultural based
  stress explain risky decision making?

24
McGuire Types of theories

• Categorical
• Clustering of phenomena
• Types of social support (emotional, practical,
  etc.)
• Affective clusters NA
• Clustering of people
• Diagnostic categories
• Personality big 5, introversion extroversion
• Process
• Flow-chart like perspectives
• Steps in persuasion
• Stages of Change
• Developmental theories

25
types of theories

• Axiomatic
• Write predictions from a highly plausible or
  tautological axiom
• protection motivation theory
• Axiomatic that people want protection from threat
• Write predictions about
• Origins of threat perception
• Mediators of responses to threat
• Health belief / Illness cognition models
• Axiom peoples thoughts / understanding of
  illness (or illness threat) ? key behaviors
• Predictions
• Relevance of specific cues to action
• Short v. long -term thoughts, etc.
• Affective motivation (? analysis)

26
types of theories

• Guiding Light or heuristic theories larger
  model of man
• Clear, top-down principles or axioms
• Self-Efficacy / Learned helplessness
• Cognitive-social information processing models
• Basic learning theories
• Deci autonomous control as core motive
• Sense of coherence
• Controllable,
• coherent,
• comprehensible
• Reductionist / physiological (?)
• Behavioral constraints of neuro-anatomy
• Transmitter mediated models of depression, etc.
• Theory itself not open to test, only to
  application or test of derivation

27
McGuire Heuristics for developing hypotheses or
empirical questions
I. Heuristics Simply Calling for Sensitivity to
Provocative Natural Occurrences A. Recognizing
and Accounting for the Oddity of Occurrences 1.
Accounting for deviations from the general
trend 2. Accounting for the oddity of the general
trend itself B. Introspective Self-Analysis 3.
Analyzing ones own behavior in similar
situations 4. Role playing ones own behavior in
the situation C. Retrospective Comparison 5.
Extrapolating from similar problems already
solved 6. Juxtaposing opposite problems to
suggest reciprocal solutions D. Sustained,
Deliberate Observation 7. Intensive case
studies 8. Participant observation 9. Assembling
propositional inventories
28

• II. Heuristics Involving Simple Conceptual
  Analysis (Direct Inference)
• E. Simple Conversions of a Banal Proposition
• 10. Accounting for the contrary of a trite
  hypothesis
• 11. Reversing the plausible direction of
  causality
• 12. Pushing an obvious hypothesis to an
  implausible extreme
• 13. Imagining the effects of reducing a variable
  to zero
• 14. Conjecturing interaction variables that
  qualify a relation
• F. Multiplying Insights by Conceptual Division
• 15. Linguistic explorations
• 16. Alternative manipulations of the independent
  variable
• 17. Dividing the dependent variable into
  subscales
• 18. Arranging output subcomponents into a
  sequence
• G. Jolting Ones Conceptualizing Out of its Usual
  Ruts
• 19. Shifting attention to an opposite pole of the
  problem
• 20. Alternating preferred with non-preferred
  research styles
• 21. Expressing ones hypothesis in multiple
  modalities
• 22. Disrupting ordinary states of consciousness

29
III. Heuristics Calling for Complex Conceptual
Analysis (Mediated Inference) H. Deductive
Reasoning Procedures 23. Generating multiple
explanations for a given relation 24. Alternating
induction and deduction 25. Identifying
counterforces obscuring an obvious relation 26.
Hypothetico-deductive sets of postulates I. Using
Thought-Diversifying Structures 27. Using an
idea-stimulating checklist 28. Constructing
provocative complex generating structures 29.
Formalizing explanatory accounts J. Using
Metatheories as Thought Evokers 30. The
evolutionary functionalism (adaptivity)
paradigm 31. Transferring conceptualizations
analogously 32. Quixotic defense of a theory
30

• IV. Heuristics Demanding Reinterpretations of
  Past Research
• K. Delving into Single Past Studies
• 33. Accounting for irregularities in an obtained
  relation
• 34. Decomposing non-monotonic into simpler
  relations
• 35. Deviant-case analysis
• 36. Interpreting serendipitous interaction
  effects
• L. Discovery by Integrating Multiple Past Studies
• 37. Reconciling conflicting outcomes or
  non-replications
• 38. Bringing together complementary past
  experiments
• 39. Reviewing and organizing current knowledge in
  an area

31

• V. Heuristics Necessitating Collecting New or
  Reanalyzing Old Data
• M. Qualitative Analyses
• 40. Allowing open-ended responses for content
  analysis
• 41. Participating actively in the research
  routine
• 42. Exploring a glamorous technique
• 43. Including low-cost interaction variables in
  the design
• 44. Pitting confounded factors against one
  another
• 45. Strategic planning of programmatic research
• N. Quantitative Analyses
• 46. Multivariate fishing expeditions
• 47. Subtracting out the effect of a known
  mediator
• 48. Computer simulation
• 49. Mathematical modeling