Chapter 4 Explanation Explanation Chapter 4 The logic of - - PDF document

SLIDE 1

1

Chapter 4 Chapter 4 Research Design Research Design

Chapter Outline Chapter Outline

• Three purposes of research

Three purposes of research

• Exploration

Exploration

• Description

Description

• Explanation

Explanation

• The logic of nomothetic explanation

The logic of nomothetic explanation

• Criteria for nomothetic causality

Criteria for nomothetic causality

• False criteria for nomothetic causality

False criteria for nomothetic causality

• Necessary and sufficient causes

Necessary and sufficient causes

• Units of analysis

Units of analysis

• Individuals

Individuals

• Groups

Groups

• Organizations

Organizations

• Social artifacts

Social artifacts

• Faulty reasoning about units of analysis

Faulty reasoning about units of analysis – – ecological fallacy and reductionism ecological fallacy and reductionism

• The time dimension

The time dimension

• Cross

Cross-

• sectional studies

sectional studies

• Longitudinal studies

Longitudinal studies

• Approximating longitudinal studies

Approximating longitudinal studies

• Elements of a research proposal

Elements of a research proposal

What are the three general purposes What are the three general purposes

• f research?
• f research?
• Although it is worthwhile to distinguish different purposes of

Although it is worthwhile to distinguish different purposes of research, keep in mind that most studies will have more than research, keep in mind that most studies will have more than

• ne purpose.
• ne purpose.
• Three general purposes:

Three general purposes:

• Exploration

Exploration -

• Explore and develop some initial rough understanding of

Explore and develop some initial rough understanding of some phenomenon some phenomenon

• Description

Description – – Precise measurement and reporting of the characteristics Precise measurement and reporting of the characteristics

• f some population or phenomenon under study.
• f some population or phenomenon under study.
• Social sciences descriptions are usually more accurate and preci

Social sciences descriptions are usually more accurate and precise than se than casual observations. Answers the question of “what’s so”? casual observations. Answers the question of “what’s so”?

• Explanation

Explanation – – Discovery and reporting of relationships among different Discovery and reporting of relationships among different aspects of phenomenon under study. aspects of phenomenon under study.

• Answers the question of “why”.

Answers the question of “why”.

What are the two models of What are the two models of explanation again? explanation again?

• Idiographic

Idiographic -

• aims at a complete understanding

aims at a complete understanding

• f a particular phenomena.
• f a particular phenomena.
• Example: Finding out what the Smith family plans

Example: Finding out what the Smith family plans

• n filing for personal bankruptcy. There can be 20
• n filing for personal bankruptcy. There can be 20

different reasons leading to that decision. different reasons leading to that decision.

• Nomothetic

Nomothetic -

• aims at a general understanding

aims at a general understanding

• f a class of phenomena.
• f a class of phenomena.
• Example: Finding out the top 5 reasons why

Example: Finding out the top 5 reasons why people file for personal bankruptcy. people file for personal bankruptcy.

What are the three criteria for What are the three criteria for nomothetic causality? nomothetic causality?

• Variables must be correlated.

Variables must be correlated.

• Example: Education level of parents are found to be correlated t

Example: Education level of parents are found to be correlated to

• education level of their children.

education level of their children.

• Cause takes place before the effect.

Cause takes place before the effect.

• Example: Education level of parents usually is determined BEFORE

Example: Education level of parents usually is determined BEFORE the education level of their children. Thus the education level the education level of their children. Thus the education level of parents

• f parents

is likely to be the cause in this relationship. is likely to be the cause in this relationship.

• Effect cannot be explained in terms of a third variable

Effect cannot be explained in terms of a third variable (nonspurious). (nonspurious).

• Example of a spurious relationship: Ice cream sales are positive

Example of a spurious relationship: Ice cream sales are positively ly correlated with number of people drown at the beach. And ice cre correlated with number of people drown at the beach. And ice cream am eating happens before the drowning. But is eating ice cream caus eating happens before the drowning. But is eating ice cream causing ing people to drown? No, because there is a third factor: hot weathe people to drown? No, because there is a third factor: hot weather. Hot

• r. Hot

weather cause ice cream consumption to go up, and number of peop weather cause ice cream consumption to go up, and number of people le swimming to go up at the same time. Hot weather, in this case, h swimming to go up at the same time. Hot weather, in this case, has as caused a spurious relationship between ice cream consumption and caused a spurious relationship between ice cream consumption and number of people drown. number of people drown.

What are necessary and sufficient What are necessary and sufficient causes? causes?

• Necessary cause represents a condition that must be present

Necessary cause represents a condition that must be present for the effect to follow. for the effect to follow.

• Rule: A must be true for B to be true. Thus A is a necessary cau

Rule: A must be true for B to be true. Thus A is a necessary cause for se for B. B.

• Example: Being pregnant is a necessary cause for giving birth to

Example: Being pregnant is a necessary cause for giving birth to a a child. child.

• Sufficient cause represents a condition that if present,

Sufficient cause represents a condition that if present, guarantees the effect in question. guarantees the effect in question.

• Rule: If A is true, B must be true. Thus A is a sufficient cause

Rule: If A is true, B must be true. Thus A is a sufficient cause for B. for B.

• Example: Not taking the final exam (thus getting 0 point on the

Example: Not taking the final exam (thus getting 0 point on the final) is final) is a sufficient cause for failing this class. a sufficient cause for failing this class.

• Causes that are necessary and sufficient are the most satisfying

Causes that are necessary and sufficient are the most satisfying

• utcome in research. However, in social science research,
• utcome in research. However, in social science research,

these clear these clear-

• cut causal relationships are rare.

cut causal relationships are rare.

SLIDE 2

2

What are units of analysis? What are units of analysis?

• Units of analysis refers to what or whom to study

Units of analysis refers to what or whom to study

• Four types of units of analysis

Four types of units of analysis

• Individuals

Individuals

• Examples: students, voters, female faculty members at the U.

Examples: students, voters, female faculty members at the U.

• Groups

Groups

• Examples: families, communities

Examples: families, communities

• Social organizations

Social organizations

• Examples: corporations, companies, grocery stores

Examples: corporations, companies, grocery stores

• Social artifacts: products or social beings or their behavior

Social artifacts: products or social beings or their behavior

• Examples: newspaper, buildings, cars

Examples: newspaper, buildings, cars

What are the mistakes related to What are the mistakes related to units of analysis? units of analysis?

• The ecological fallacy

The ecological fallacy -

• The danger of making assertions about

The danger of making assertions about individuals when the unit of analysis is a group individuals when the unit of analysis is a group

• Example: A researcher finds that crime rates were higher in cit

Example: A researcher finds that crime rates were higher in cities with ies with large black populations than those with fewer blacks. Thus this large black populations than those with fewer blacks. Thus this researcher draws a conclusion that black people are more likely researcher draws a conclusion that black people are more likely to to commit crime than whites. The mistake here is that the unit of a commit crime than whites. The mistake here is that the unit of analysis nalysis is “city”, not “individuals”. Thus it is possible that it’s the is “city”, not “individuals”. Thus it is possible that it’s the white people white people living in the cities with high black population committing crime living in the cities with high black population committing crimes. We

• s. We

just don’t know. To reach a conclusion like this is ecological f just don’t know. To reach a conclusion like this is ecological fallacy. allacy.

• Reductionism

Reductionism -

• Overly restrictive as to what kind of concepts

Overly restrictive as to what kind of concepts and variables can be used for explanation and variables can be used for explanation

• Example: In explaining juvenile delinquency, a researcher only

Example: In explaining juvenile delinquency, a researcher only considers individual factors (individual level), but not social considers individual factors (individual level), but not social factors factors (peer group, social system, etc.). (peer group, social system, etc.).

What are time dimensions? What are time dimensions?

• There are two major types of time dimensions: cross

There are two major types of time dimensions: cross-

• sectional

sectional

• vs. longitudinal
• vs. longitudinal
• Cross

Cross-

• Sectional Studies

Sectional Studies -

• studies using data at a given time

studies using data at a given time

• Example: 2000 US census data. Visit

Example: 2000 US census data. Visit http://www.census.gov/ http://www.census.gov/

• Longitudinal Studies

Longitudinal Studies -

• designed to permit observations over an

designed to permit observations over an extended period of time. There are three kinds of longitudinal s extended period of time. There are three kinds of longitudinal studies. tudies.

• Trend: Changes in some general groups over time

Trend: Changes in some general groups over time

• Example: US census is done every 10 years. If one combine 1970,

Example: US census is done every 10 years. If one combine 1970, 1980, 1990, 1980, 1990, and 2000 census data, one can study, for example, the population and 2000 census data, one can study, for example, the population growth over growth over time. time.

• Cohort: Changes of specific subgroup, typically, an age group

Cohort: Changes of specific subgroup, typically, an age group

• Example: baby

Example: baby-

• boomers, Generation X, those born during WWII

boomers, Generation X, those born during WWII

• Panel: The same set of people are studied over time

Panel: The same set of people are studied over time

• Example: I interview a group of people in 1998. Then I interview

Example: I interview a group of people in 1998. Then I interview the same the same people again in 2000, then again in 2002. people again in 2000, then again in 2002.

What are the advantages and What are the advantages and disadvantages of cross disadvantages of cross-

• sectional vs.

sectional vs. longitudinal studies? longitudinal studies?

• Longitudinal studies have advantages over cross

Longitudinal studies have advantages over cross-

• sectional studies in terms

sectional studies in terms

• f
• f
• The amount of information contained

The amount of information contained

• The time order clearly established

The time order clearly established

• Longitudinal studies have disadvantages over cross

Longitudinal studies have disadvantages over cross-

• sectional studies in

sectional studies in terms of terms of

• Much higher costs

Much higher costs

• Panel attrition: losing subjects over time and causing statistic

Panel attrition: losing subjects over time and causing statistical biases al biases

• Because longitudinal studies have more information, researchers

Because longitudinal studies have more information, researchers some some times approximate longitudinal studies with cross times approximate longitudinal studies with cross-

• sectional studies by

sectional studies by

• Ask people about their past (such as work history) in cross

Ask people about their past (such as work history) in cross-

• sectional studies

sectional studies

• Examine different age groups in cross

Examine different age groups in cross-

• sectional studies. This is done assuming

sectional studies. This is done assuming when the younger people grow up, they will behave like the older when the younger people grow up, they will behave like the older people in the people in the sample. sample.

An illustration of time dimensions An illustration of time dimensions

Year of Birth Year of Measurement 1970 1980 1990 1960 A-10 B-20 C-30 1950 D-20 E-30 F-40 1940 G-30 H-40 I-50

• If one conducts a study of

If one conducts a study of

• C

C-

• F

F-

• I, meaning that in 1990, one collects data on individuals born i

I, meaning that in 1990, one collects data on individuals born in n the 60s, 50s, and 40s, the study is cross the 60s, 50s, and 40s, the study is cross-

• sectional.

sectional.

• A

A-

• B

B-

• C with different subjects, the study is longitudinal cohort

C with different subjects, the study is longitudinal cohort

• A

A-

• B

B-

• C with the same subjects, the study is longitudinal panel

C with the same subjects, the study is longitudinal panel

• A through I with all subjects, the study is longitudinal trend

A through I with all subjects, the study is longitudinal trend

Think about this … Think about this …

• Cross

Cross-

• sectional studies indicate that IQ

sectional studies indicate that IQ declines with age. Longitudinal studies claim declines with age. Longitudinal studies claim that there is either no decline or less of a that there is either no decline or less of a decline with age than cross decline with age than cross-

• sectional studies

sectional studies

• indicate. Why this difference?
• indicate. Why this difference?

SLIDE 3

3

What are the elements of a research What are the elements of a research proposal? proposal?

• Problem or objective (purpose)

Problem or objective (purpose)

• Literature review (what has been done on this topic)

Literature review (what has been done on this topic)

• Subjects for study (units of analysis)

Subjects for study (units of analysis)

• Measurement (conceptualization+operationalization)

Measurement (conceptualization+operationalization)

• Data

Data-

• collection methods (sample, mode of

collection methods (sample, mode of

• bservation)
• bservation)
• Analysis (statistics or qualitative methods)

Analysis (statistics or qualitative methods)

• Schedule

Schedule

• Budget

Budget

How to apply this to real research? How to apply this to real research? – – An example An example

• Units of analysis

Units of analysis

• Article by Medina et al: individuals as units of

Article by Medina et al: individuals as units of analysis analysis

• Article by Parcel et al: group (families) as units of

Article by Parcel et al: group (families) as units of analysis analysis

• Time dimension

Time dimension

• Article by Medina et al: cross

Article by Medina et al: cross-

• sectional

sectional

• Article by Parcel at al: longitudinal panel

Article by Parcel at al: longitudinal panel

Additional things to do Additional things to do

• Read two research articles:

Read two research articles: (1) Gonzales and (1) Gonzales and Meyers (1993) Meyers (1993) and (2) and (2) Parcel, Nickoll & Dufur Parcel, Nickoll & Dufur (1996). (1996). Please note again that you are not Please note again that you are not expected to understand everything in these two expected to understand everything in these two

• articles. However, please pay attention to the
• articles. However, please pay attention to the

issue of research design, units of analysis, and issue of research design, units of analysis, and time dimension. time dimension.