Scribd
Upload
72 views

Lecture (Chapter 14) : Elaborating Bivariate Tables: Ernesto F. L. Amaral

The document discusses elaborating bivariate tables by controlling for a third variable. It explains that controlling for a third variable allows researchers to observe how the third variable affects the relationship between the original two variables. Partial tables display the relationship between the original two variables for each value of the control variable. There are three main patterns that can emerge: direct relationships where the control has little effect; spurious/intervening relationships where the control explains the original relationship; and interactive relationships where the association between the original two variables changes across categories of the control variable. Examples are provided to illustrate these patterns.

Uploaded by

Marvin Yebes Arce
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
72 views

Lecture (Chapter 14) : Elaborating Bivariate Tables: Ernesto F. L. Amaral

The document discusses elaborating bivariate tables by controlling for a third variable. It explains that controlling for a third variable allows researchers to observe how the third variable affects the relationship between the original two variables. Partial tables display the relationship between the original two variables for each value of the control variable. There are three main patterns that can emerge: direct relationships where the control has little effect; spurious/intervening relationships where the control explains the original relationship; and interactive relationships where the association between the original two variables changes across categories of the control variable. Examples are provided to illustrate these patterns.

Uploaded by

Marvin Yebes Arce
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 28

Lecture (chapter 14):

Elaborating bivariate tables

Ernesto F. L. Amaral

April 16–18, 2018


Advanced Methods of Social Research (SOCI 420)

Source: Healey, Joseph F. 2015. ”Statistics: A Tool for Social Research.” Stamford: Cengage
Learning. 10th edition. Chapter 14 (pp. 380–404).
Chapter learning objectives
• Explain the purpose of multivariate analysis in
terms of observing the effect of a control variable
• Construct and interpret partial tables
• Compute and interpret partial measures of
association
• Recognize and interpret direct, spurious or
intervening, and interactive relationships
• Compute and interpret partial gamma
• Explain limitations of elaborating bivariate tables

2
Controlling for a third variable
• Social science research projects are multivariate

• One way to conduct multivariate analysis is to


observe the effect of third variables, one at a
time, on a bivariate correlation

• The elaboration technique extends the analysis


of bivariate tables and associations

3
Partial tables
• We observe how a control variable (Z) affects
the relationship between X and Y

• To control for a third variable, the bivariate


relationship is reconstructed for each value of
the control variable

• Tables that display the relationship between X


and Y for each value of Z (a third variable) are
called partial tables

4
Focus on three basic patterns
• Direct relationships

• Spurious or intervening relationships

• Interaction

5
Direct relationships
• In a direct relationship, the control variable has
little effect on the relationship between X and Y
• The column percentages and Gammas in the
partial tables are about the same as the
bivariate table
• This outcome supports the argument that X
causes Y
• Also referred to as replication

X Y

6
Spurious relationships
• In a spurious relationship, X and Y are not
related, both are caused by Z
• In a spurious relationship, the Gammas in the
partial tables are dramatically lower than the
gamma in the bivariate table, perhaps even
falling to zero
• Also referred to as explanation
X
Z
Y
7
Intervening relationships
• In an intervening relationship, X and Y are not
directly related to each other but are linked by Z,
which “intervenes” between the two

• Also referred to as interpretation

Z
X Y

8
Interaction
• Interaction occurs when the association between
X and Y changes across the categories of Z
– X and Y could only be related for some categories of Z

Z1 Y
X
Z2 0.00
– X and Y could have a positive association for one
category of Z and a negative association for others
Z1 +
X Y
Z2 –
9
Summary
• Possible results when controlling for third variables

Source: Healey 2015, p.389. 10


Partial Gamma
• Partial Gamma indicates the overall strength of
association between X and Y after the effects of
the control variable (Z) have been removed
– Compare Partial Gamma (Gp) to the Gamma (G) for the
bivariate table to see if the relationship has changed
∑ 𝑁& − ∑ 𝑁)
𝐺" =
∑ 𝑁& + ∑ 𝑁)
– Ns is the number of pairs of cases ranked in the same
order across all partial tables
– Nd is the number of pairs of cases ranked in different
order across all partial tables
11
Example 1
• Association between
– Number of memberships in student organizations
• X, independent variable
– Satisfaction with college
• Y, dependent variable

Source: Healey 2015, p.381. 12


Interpretation
• Comparing the conditional distributions of Y (the
column percentages), we find a positive
relationship
– This direction is confirmed by the sign of Gamma
(+0.40)

• College students with at least one membership


in a student organization are more likely than
students with no memberships to have high
satisfaction with college

13
GPA as a control variable
• Associations remain positive

Source: Healey 2015, p.383. 14


Association still positive
• The relationship between integration and
satisfaction is the same in the partial tables as it
was in the bivariate table
– This is evidence of a direct relationship

High GPA Low GPA


Ns = (29)(54) = 1566 Ns = (28)(55) = 1540
Nd = (28)(24) = 672 Nd = (28)(24) = 672

∑ 𝑁& − ∑ 𝑁) 1566 + 1540 − (672 + 672)


𝐺" = = = 0.40
∑ 𝑁& + ∑ 𝑁) 1566 + 1540 + (672 + 672)
Source: Healey 2015, p.390. 15
Class standing as a control
• There is no more association
– Upperclass students: seniors and juniors
– Underclass students: sophomores and freshmen

Source: Healey 2015, p.385. 16


Association disappears
• The original bivariate relationship between
memberships and satisfaction disappears in the
partial tables
– When the association disappears, we have either a
spurious or an intervening relationship
Spurious
X Intervening
Z
Y Z
Membership X Y
Class
standing Satisfaction
17
Spurious relationship
• Decision about whether the association is
spurious or intervening is based on
– Temporal (timing) or theoretical grounds
• A spurious relationship makes more sense
– Class standing likely predicts the number of
memberships, and not the other way around
• Partial Gamma supports our conclusion (reduced to zero)

Upperclass Underclass
Ns = (8)(97) = 776 Ns = (49)(12) = 588
Nd = (32)(24) = 768 Nd = (24)(24) = 576
∑ 𝑁& − ∑ 𝑁) 776 + 588 − (768 + 576)
𝐺" = = = 0.01
∑ 𝑁& + ∑ 𝑁) 776 + 588 + (768 + 576)
Source: Healey 2015, p.390. 18
Example 2
• Relationship for 50 immigrants between
– Length of residence: X, independent variable
– English fluency: Y, dependent variable

• Gamma = +0.67
– Strong and positive association
– As length of residence increases, English fluency also
increases
Source: Healey 2015, p.398, problem 14.1. 19
Sex as a control variable
• Associations remain positive

• Gm = 0.78

• Gf = 0.65

Source: Healey 2015, p.398, problem 14.1. 20


Partial Gamma
𝑁& − 𝑁) 80 − 10
𝐺6 = = = 0.78
𝑁& + 𝑁) 80 + 10

𝑁& − 𝑁) 70 − 15
𝐺7 = = = 0.65
𝑁& + 𝑁) 70 + 15

∑ 𝑁& − ∑ 𝑁) 80 + 70 − (10 + 15)


𝐺" = = = 0.71
∑ 𝑁& + ∑ 𝑁) 80 + 70 + (10 + 15)

Source: Healey 2015, p.398, problem 14.1. 21


Sex has no effect
• While the two Gammas for the partial tables
(0.78 and 0.65) differ slightly
– They both indicate a strong and positive association
between length of residence and English fluency

• Comparing Partial Gamma (0.71) to the original


Gamma (0.67), we find little difference

• We have evidence of a direct relationship


– Controlling for sex does not affect the association
between length of residence and English fluency for
immigrants
Source: Healey 2015, p.398, problem 14.1. 22
Example 3
• Relationship for 78 juvenile males between
– Academic record: X, independent variable
– Delinquency: Y, dependent variable

• Gamma = –0.69
– Juvenile males with better academic records have
lower delinquency
Source: Healey 2015, p.392. 23
Area of residence as a control
• Associations differ across partial tables

Source: Healey 2015, p.392. 24


Interpretation
• Gamma for urban areas is –0.05
– No association between academic record and
delinquency

• Gamma for nonurban areas is –0.89


– Strong and negative association between academic
record and delinquency

• Associations between X and Y differ across


partial tables
– This is an indication of interaction

25
Origin of control variables
• Control variables are based on theory

• Research projects are anchored in theory, so


control variables come mainly from theory

• Understanding a spurious relationship


(explanation) or an intervening relationship
(interpretation) cannot be based on statistical
grounds or inspecting the partial tables

26
Limitations of partial tables
• Basic limitation: Sample size
– Greater the number of partial tables, the more likely to
run out of cells or have small cells

• Potential solutions
– Reduce number of cells by collapsing categories
(recoding)
– Use very large samples
– Use techniques appropriate for interval-ratio level

27

You might also like