Pollock, Essentials Political Analysis part1

The Essentials of Political Analysis Philip H. Pollock, 4th edition, Sage Pub., 2012. (1st Part) Chapters 1-5 By Luigi Cino

SUMMARY 1. The Definition and Measurement f the Concept 2. Measuring and Describing Variables 3. Proposing Explanations, Framing Hypothesis, and Making Comparisons

4. Research Design and the Logic of Control 5. Making Controlled Comparisons

1. The Definition and Measurement of Concepts (1/3) • A concept is an idea or a mental construct that represents phenomena in the real world role of Political Analysis (PA hereinafter) is to describe and analyse relationship between them

•A

conceptual definition describes clearly the concept’s measurable properties and the units of analysis to which the concept applies

• An

operational definition describes the instruments for measuring the concept in the real world operationalization

• First,

it is necessary to clarify a concept. Its properties have two characteristics: they must be concrete and they must vary.

• Brainstorming polar opposites

drop the non essential traits, don’t use a concept to define another concept (vague), pay attention to different dimensions of the same concept (multidimensional concept, ex. democracy)

1. The Definition and Measurement of Concepts (2/3) • TEMPLATE FOR WRITING A CONCEPTUAL DEFINITION:

“The concept of _____ is defined as the extent to which ____ exhibit the characteristics of _____ “

• ex: the concept of economic liberalization is defined as the extent to which

individuals exhibit the characteristic of supporting government spending for social programs.

• This definition communicates three things: 1. The variation within a measurable characteristic or set of characteristics (“as the extent to which”)

2. The subjects or group to which the concept applies (“Individuals” = unit of analysis

3.

can be individual level or aggregate level)

How the characteristic is to be measured (“supporting etc…”)

1. The Definition and Measurement of Concepts (3/3) • An OPERATIONAL DEFINITION describes explicitly how the concept is to be measured empirically

• Risk of measurement error that would measure unintended characteristics: 1. Systematic measurement error operational readings that consistently mismeasure the characteristic the researcher is after

2. Random measurement

haphazard, chaotic distortion into the measurement

process

• Reliability of a measurement is the extent to which it is a consistent measure of a concept (various tests to evaluate both reliability and validity)

• Validity of a measurement is the extent to which it records the true value of the intended characteristic and does not measure any unintended characteristic

2. Measuring and Describing Variables (1/2) • A VARIABLE is an empirical measurement of a characteristic. Every variable has one name and at least two VALUES.

• Es: Variable “marital status” has different values: married, divorced, etc. • Variables have different level of measurements: 1. Nominal Variables: separates people into different categories (es. Married, divorced in the case of marital status)

2. Ordinal Variables: Have values that can be ranked. For ex. Support: strongly support, support, oppose, strongly oppose

ordinal scale, indexes

3. Interval Variables: communicates exact difference between units of analysis. --> quantitative values, for ex. Age.

2. Measuring and Describing Variables (2/2) • Describing Variables: central tendency and dispersion • Central tendency: 1. Mode (N-O-I var): the most common value of the variable 2. Median (O-I var): the value of a variable that divides the cases right down the middle (check cumulative percentage – 50° percentile)

3. Mean (only interval var): the arithmetic average. • Dispersion: the variation or spread of cases across its value. The greatest amount is when the cases are equally spread among all values of the variable, conversely the lowest. Skewness: (asimmetria)

1. Positive when mean > median (right-hand skinnier tail) 2. Negative when mean < median (left-hand skinnier tail)

3. Explanations, Hypothesis and Comparisons (1/6)

• Measurement defines “what is” a concept, but political research wants to answer the question “why” a phenomenon happens

causal explanation

• An explanation suggest an hypothesis that could be tested with empirical data • An hypothesis is then a testable statement about the empirical relationship between cause and effect

conditional statement

• The dependent variable represents the effect • The independent variable represent the causal explanation

3. Explanations, Hypothesis and Comparisons (2/6) • Proposing explanations: explanations are never vague. A good explanation connects two variables and provides detailed description of the causal linkage, asserting the tendency and a testable empirical relationship.

• Framing hypothesis: it makes an explicit testable comparison, telling us how different values of the indep var are related to different values of dep var

• TEMPLATE FOR WRITING HYPOTHESIS:

“in a comparison of [unit of analysis] , those having [one value on the independent variable] will be more likely to have [one value on the dependent variable] than will those having [a different value on the independent variable].”

3. Explanations, Hypothesis and Comparisons (3/6) • A good hypothesis makes states a clear relationship, makes an explicit comparison and states the tendency of such relationship.

• A good explanation describe a causal process about the linkages between

dep and indep var. These linages can suggest the existence of other testable relationships intervening variables that acts as mediator between dep and indep var by describing how they are related.

• Making comparisons: If dep and indep variable are categorical (nominal or

ordinal level) then cross tabulation comparison is performed; if indep var is categorical and dep var is interval-level then mean comparison is performed in order to test the hp.

3. Explanations, Hypothesis and Comparisons (4/6) • CROSS TABULATION

It is a table that shows the distribution of cases across the values of a dependent variable for cases that have different values on an indep var.

• Three rules in cross tab: 1. Categories of indep var defines the columns, values of dep var the rows 2. Always calculate percentages of the categories of the indep var 3. Interpret cross tab by comparing percentages across columns at the same value of the dep var

• MEAN COMPARISON

A mean comparison is a table that shows the mean of a dependent var for cases that have different values on an indep var

3. Explanations, Hypothesis and Comparisons (5/6) MEAN COMPARISON

CROSS TABULATION

3. Explanations, Hypothesis and Comparisons (6/6) • When the independent variable is measured at a ordinal or interval level, then we can have two types of relationship:

1. Direct (or positive) relationship: at the increase of the indep var also the dep var increases ( + , + )

2. Indirect (or inverse or negative) relationship: the increase of the independent variable is associated with a decrease in the dependent variable ( + , - )

• This can be easily see through bar or line charts, where the independent variable goes on the x axis, while the dependent variable on the y axis

• Relationship can also be linear (negative or positive) or non-linear (curvilinear) when it changes direction from positive or negative to negative or positive

• Particular pattern the V(or U)-shaped relationships (or inverted U-shape rel)

4. Research Design and the Logic of Control (1/4) • There could be rival explanations for a relationship, thus it is necessary to control • Es: test group and control group in experiments (one does not receive the cure) • Laboratory/field experiments; random assignment to avoid selection bias • Internal validity: effect of the indep var on dep var is isolated from other plausible explanations

• External validity: the results of a study can be generalized • A controlled comparison is accomplished by examining the relationship between indep and dep var while holding constant other var suggested by rival explanations or hypothesis.

• “how else..?” questions

compositional difference

4. Research Design and the Logic of Control (2/4) • X: independent var; Y= dependent var; Z= control var (rival causal) • Neutralizing the effects of a rival cause by holding it constant • Three scenarios: 1. Spurious relationship: coincidental relationship between X and Y, not causal 2. Additive relationship: Z is a causal explanation for Y, but defines a small compositional difference across values of X. X and Z relat. is weak

3. Interaction relationship: relationship between X and Y depends on the value of Z. For some value of Z, X-Y relationship can be strong, for others weak.

4. Research Design and the Logic of Control (3/4) • In a spurious relationship line chart, the

distance between the lines of the Z values are constant (K) and it represent the effects of Z on Y.

• In an additive relationship lines are parallel:

the effect of X is the same for every value of Z (Z is not K)

4. Research Design and the Logic of Control (4/4) •

In a interaction relationship the distances of the two lines are different for every value of X: the relationship X-Y depends on the values of Z. Many patterns can be identified (figure)

5. Making Controlled Comparisons (1/3) • CROSS TABULATION ANALYSIS

The values of the dependent variable define the rows, the values of the independent variable define the columns

• Zero-order relationship: a difference obtained from a simple comparison ( also known as gross relationship or uncontrolled relationship)

• How else?

controlled comparison table: present a cross tabulation between an independent var and a dep var for each value of the control var

• It reveals the controlled effects of the indep var on the dep var for every value of the control var; it permits to describe the effects of the control var on dep var for every value of the indep var

5. Making Controlled Comparisons (2/3) •

Partial effect (or relationship): summarizes a relationship between two variables after taking into account rival variables

•

Rule of direction for nominal relationships: the value of the variable that defines the left-most column of a cross-tab is the bas category

5. Making Controlled Comparisons (3/3) • IDENTIFYING THE PATTERN: 3 QUESTIONS

(applies both to cross-tab and mean comparison)

1.

Afterholding the control variable constant (Z=K), does a relationship exist between the indep and the dep vars within at least one value of Z? if no, the rel is spurious. If yes, then go to question 2.

2. Is the tendency of the rel between the indep and dep vars the same at all values of the control var? if no, interaction is taking place. If yes, go to question 3.

3.

Is the strenght of the rel between the indep and dep vars the same (or very similar) at all values of the control var? if yes, the relationships are additive. If no, interaction best characterizes the relationship. END