Test 2 - PHIL105 Flashcards
Specific Claims - Definition
Claims about particular instances of things
Universal Claims - Definition
Claims about entire classes of things
Inductive Generalisations - Definition
Making universal claims based on specific claims
Note: Inductive Arguments = not valid since premises do not guarantee conclusion (unless complete enumeration)
Deductive reasoning - Definition
Making specific claims based on universal instances
Benefits of Inductive Generalisations:
Helps us figure out how things are are going to be, based on how things have been in the past
Problems with Induction
Justification for using induction is itself an inductive argument
Argument in favour of using induction is circular (begs the question)
Logical Asymmetry?
Any no. of positive instances of a universal claim cannot guarantee its truth (unless you have every possible instance of the universal claim)
Thus, one negative instance of a universal claim is enough to show it to be false
Complete Emuneration
When entire class of things is small and accessible, all possible claims can be checked
Three rules for good induction
This is the exam question - “three rules for good generalisation”
- Sample should be sufficiently numerous and various
- Should look for disconfirming as well as confirming cases of generalisation
- Consider whether a link b/w the two classes is plausible, given other knowledge that we possess
Causal Claims - Definition
Assertions that events of one type are followed by events of another type
Two types of causal claims:
General: A causes B
Particular: A caused B
General causal claim:
Means that another event of the same kind as A, would, in similar circumstances, produce another event of the same kind as B
Particular causal claim:
Presupposes a general causal law (i.e. shirt is stained because I spilt coffee on it)
Mills Methods: All 5
- Method of Agreement
- Method of Difference
- Joint Method of Agreement & Difference
- Method of Residues
- Method of Concomitant Variation
Method of Agreement
Look for common factor that is present in all cases in which effect occured
Method of Difference
Look at antecedent circumstances when E (event) occurs and compare these to antecedent circumstances where E fails to occur
Joint Method of Agmt and Diff
Requires that there be some Agreement, but also that there is at least one different case where the proposed cause isn’t present and where effect is also not present
Method of Residues
If we already know cause of part of the effect, we can subtract that to figure out what causes the rest of the effect
Method of Concomitant Variation
If quantitative changes in a phenomenon are associated with quantitative changes in another phenomenon - likely causal connection b/w them.
Idea = change in the strength of the effect should correspond to a change in the strength of the cause
Limitations of Mills Methods
Does not tell us how to determine whether the antecedent circumstances have been properly analysed
Does not tell us which antecedent circumstances to investigate
Can never provide a certain, demonstrative, or conclusive proof of a causal claim
Does not provide a mechanical or automatic way to discover causal connections.
Necessary v Sufficient Conditions
Necessary = condition must have occurred to guarantee result Sufficient = conditions is enough to guarantee result
Necessary v Sufficient: Method of Agmt & Diff
Agmt: provides evidence that the proposed cause is sufficient for the effect
Diff: gives us necessity
Statistical Propositions
Propositions that present quantitative evidence about a category of things
Statistical Properties can be divided into two categories … those being?
Values: numerical measure or category to which particular variables can be assigned
Variable: a kind of property a thing can have
Variables can be
Qualitative or Quantitative
Types of Statistical Info: Definitions Totals Ratios Frequency Distribution Average (mean) Median
Totals; adding up a set of units
Ratios; proportion of total
Frequency; how many things in a class have a certain property - Have absolute (actual number of P's that are Q's) and relative (proportion of P's that are Q's)
Distribution; how many things have each property (requires categories)
Average (mean); central value of S’s on a quantitative variable
Median; Central value in set of quantitative values (middle value)
Average v Median
Average = Sensitive to extreme values in ways that medians are not
Note: General rule - if you only have extremes in one direction, the median is often a better measure
Selection Bias
Bias introduced by selection of individuals, groups or data for analysis in such way that proper randomisation is not achieved
Testing Bias
- Occur due to the way questions are worded/phrased or even the questions themselves are presented
- Participants can easily be primed so that they are more likely to answer one way or another
Priming
Phenomenon whereby exposure to one stimulus influences a response to a subsequent stimulus, without conscious guidance or intention
How to generalise from a sample: (steps)
- Was MoE reported? If so, how large?
- How was the sample selected? Was it done in such a way that we can ensure it’s representative?
- How was info about the sample acquired? Any suggestions of unreliability
Correlation: Defintion
Specific type of relationship b/w two variables
Correlation & Causation:
KEY: Correlation does not imply causation!
Experimental v Observational Studies: Note
Although can control some variables, we cannot control them all, due to either:
(a) the inability to control said variable
(b) controlling the variable would be wildly impractial or immoral
Therefore, need to rely on an observation of things that exist.
Observational Studies - Note?
With these studies, we need to rely on observation of things that already exist, that do not involve random assignment to control and experimental groups.
Thus, if a correlation is found, may be due to a 3rd unknown variable = CONFOUNDING VARIABLE
Internal Validity
How sure we can be about cause & effect in regard to the actual numbers of the group that was studied
External Validity
Whether the generalisation holds for a wider group
Proxy variables
Sometimes we use other variables as a proxy for what we wish to measure
However, if you find a correlation b/w a cause and a proxy variable, need to assume that the proxy is in fact a good measyre of the variable you are interested in
Statistical Fallacies: Examples?
Mistaking correlation for causation
Gamblers Fallacy
Mistaking statistical significance for clinical significance
Base rate fallacy
Conjunction Fallacy
thinking that the conjunction of two events is more likely than a single general event
Mistaking statistical significance for clinical significance
Stat. signif. means that there is a low probability results are due to random chance
Base Rate Fallacy
When presented with base-rate info (general) and info about a specific case, we tend to ignore the general info and focus on the specific case
Truth Inflation: Two effects
File-drawer Effect:
- Researchers who do not find any effect just put on their work away and never submit for publication
Publication Bias:
- Research reporting that there is some effect is far more likely to be published than research that shows there is no effect
Conc:
- end up with much higher probability that a statistically significant effect reported was just due to random chance
Repression to the Mean:
IF first measurement of an effect is extremely high or low compared ti the mean, it will usually be closer to the mean on a subsequent measurement and vice versa
This is simply due to natural variation.
Demarcation Problem
How to distinguish science from non-science from pseudoscience?
Characteristic Features of Pseudoscience
a) Hostility towards scientific criticism
b) Trying to move ‘burden of proof’ away from themselves
c) Rendering claims unfalsifiable
d) Claims easy solutions for complex problems
e) Failing to consider all hypotheses
f) Fundamental principles are often based on a single case
g) Making a virtue of ignorance
h) Working backward from a conclusion
i) Cherry picking data
j) Failure to challenge core assumptions
k) Failure to engage with scientific community
l) Utilising scientific-sounding but ultimately meaningless language
m) Claiming to be many years/decades ahead of the current research community
n) Rely on poor forms of evidence
Standard Forms:
Universal Assertion (All S are P) Particular Assertion (Some S are P)
Universal Denial (No S are P) Particular Denial (Some S are not P)
Contraries v Contradictories
Contraries = Do NOT Exhaust logical space Contradictories = DO exhaust logical space
Connectives
Negation = Not = ~ Conjunction = And = & Disconjunction = Or = v Conditional = if ... then ... = horseshoe Biconditional = If & only if = three lines horizontal
