Content uploaded by Michael McDevitt
Author content
All content in this area was uploaded by Michael McDevitt on Feb 12, 2017
Content may be subject to copyright.
International Journal of Public Opinion Research Vol. 18 No. 1
© The Author 2005. Published by Oxford University Press on behalf of The World Association
for Public Opinion Research. All rights reserved.
doi:10.1093/ijpor/edh079 Advanced Access Published on 1 April 2005
THE PARTISAN CHILD:
DEVELOPMENTAL PROVOCATION
AS A MODEL OF POLITICAL
SOCIALIZATION
Michael McDevitt
ABSTRACT
While recent research highlights the child’s active role in her own political socialization,
this study represents the first test of the premise that political identification proceeds as
adolescents prompt feedback from parents. I propose a model of developmental prov-
ocation in which adolescents’ interest in an election campaign, once stimulated by news
media use, motivates them to engage parents in political conversations. By initiating
discussion, adolescents can generate information from parents as a basis for comparison,
contrast, reflection, and debate—all of these activities might foster party and ideological
identity. Results from survey panel data support the model. The field setting is Lubbock,
Texas, during the final weeks of the 2000 presidential campaign. The stunning aftermath
of the election, including the dispute over ballot recounts in Florida, provided a unique
opportunity to examine political socialization. Child-initiated discussion generated both
parental encouragement and defensive admonitions, reflecting structural changes in family
communication patterns. Despite controls for demographics and parent-initiated discus-
sion, child-initiated conversation and the resulting feedback predicted an increase in the
likelihood that an adolescent would adopt a political identity.
Politics is not on the table for discussion in some homes. When partisan squabbles
do arise, they can elicit comic possibilities of bad behavior, such as the in-law
dustups of Meathead and Archie in the 1970s sitcom All in the Family. Episodes
of bruised relationships have theoretical possibilities as well in terms of political
socialization. A teenager’s desire to unnerve parents might not be cause for
amusement in many homes, but tension in the child–parent dyad is beneficial if
the outcome is the assertion of political identity.
This paper is based on research made possible with the support of the John S. and James L. Knight
Foundation. The author thanks the three anonymous reviewers for their suggestions. The paper was first
submitted to IJPOR July 1, 2004. The final version was received December 29, 2004.
68 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
This paper tests a model of developmental provocation in which adolescents,
once motivated by news media use, solicit responses from parents. The give and
take of child–parent conversation should allow teenagers to practice opinion
expression, to validate views obtained from media, and to express their own
political identity in the home. In terms of developmental growth, however, there
is more at stake than partisan allegiance. The effort to influence parents—and to
outright provoke them in some cases—should be accompanied by a shift in the
structure of family communication, reflecting the child’s desire for autonomy.
While recent research highlights the adolescent’s active role in his or her own
political socialization (Saphir & Chaffee, 2002; McDevitt & Chaffee, 2002), this
study represents the first test of the premise that political identification proceeds
as adolescents generate feedback from their parents.
Data are drawn from a sample of adolescents in Lubbock, Texas. A panel study
encompassed two interview waves during the final weeks of the 2000 campaign.
The stunning aftermath of the campaign, including the dispute over ballot
recounts, provided a unique opportunity to examine political socialization even
as it left constitutional scholars with troubling issues to ponder.
THE GENESIS OF POLITICAL IDENTITY
Early research in political socialization focused on parents as the most important
societal agent that inculcates attitudes in children such as regime support, identi-
fication with political figures, and partisanship (Hyman, 1959; Greenstein, 1965;
Easton & Dennis, 1969). The nuclear family was still the dominant household
unit when research on political socialization emerged in the late 1950s, and in
subsequent studies scholars assumed that parents should be powerful models
because they are available and children depend on them for emotional support.
Partisan identity was viewed initially as an ‘affective, nearly immutable, emotionally
based orientation’ acquired from parents through social learning (Niemi &
Jennings, 1991, p. 970). But skepticism about the apparent imprinting of political
orientations arose when research failed to document strong patterns of corre-
spondence between the political attitudes and opinions of parents and offspring
(Connell, 1972). While parents are not likely to transmit idiosyncratic opinions,
these early studies did document some degree of parental influence on party
identification. Many parents ‘convey to their children quite early that “we” are
Democrats or “we” are Republicans. Party identification is different, then, from
other attitudes in its visibility to the child and its salience to the parents’ (Tedin,
1974, p. 1581).
The related disposition of partisan ideology, typically measured on a liberal to
conservative continuum, has received less attention in socialization research
(Verba, Burns, & Scholzman, 2003). However, ideology is an important indica-
tor of political development given the ideological realignment of the electorate
THE PARTISAN CHILD 69
since the Reagan presidency coupled with the decreasing likelihood that a child
will adopt a parent’s party identification (Abramowitz & Saunders, 1998). Policy
preferences associated with ideology, while often aligned with party preference,
might offer an alternative foothold into the world of politics for many adolescents.
In contemporary approaches, the transfer of political attitudes from parents
to children is often described as a two-step process: The child comprehends the
orientations of parents, creating an image of what parents are like, and subsequently
adapts his or her beliefs to that image (Westholm, 1999; Jennings, Stoker, &
Bowers, 1999). But the assumption of unidirectional influence—evoking images of
heirlooms passed down from generation to generation—is challenged by theorists
who advance a proactive role for the child (Saphir & Chaffee, 2002; McDevitt &
Chaffee, 2002) and by researchers who document the influence of agents exogenous
to the family, including elections and news events (e.g. Sears & Valentino, 1996).
DEVELOPMENTAL PROVOCATION
Despite efforts to refine a transmission model, theorizing is still heavily influenced
by the seemingly reasonable premise that children and younger adolescents—at
least during the onset of partisan awareness—acquire their initial identities from
their parents (e.g. Greene, 2000; Westholm, 1999; Jennings et al., 1999). My
model of developmental provocation, by contrast, asserts that identity formation
is initiated by adolescents when exposed to an upsurge of political stimulation. I
propose that, even during this early stage of attitude formation, the process is largely
driven by cognitive effort in news processing and child-initiated conversation.
The concept of developmental provocation implicates two related outcomes
that should engender attitude formation. The first entails an internal, cognitive
process—that is, the benefits of self-understanding that a child accrues by anti-
cipating conversations with parents. The other outcome occurs at the dyadic and
family-system levels, producing parent feedback and a restructuring of family
communication patterns. I will test the model using two criterion indicators of
political socialization: party identification and partisan ideology.
OUTCOME #1: CHILD-INITIATED DISCUSSION
I expect that situational increases in political interest will foster media use that
leads to conversations with parents. A considerable percentage of children watch
TV news during campaigns (Chaffee, Ward, & Tipton, 1970; Simon & Merrill,
1997), even though national surveys report a declining interest in public affairs
among young people (Bennett, 1997). Televised debates in particular can gener-
ate a great deal of anticipation and drama. Individuals gain substantial knowledge
of candidates’ issue positions during debates (Sears & Chaffee, 1979), and the
viewing of these events is associated with interactive behaviors in the home such
70 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
as co-viewing (Austin & Pinkleton, 1997). I consequently included debate
viewing along with habitual news exposure among the indicators of media use.
Reading papers and watching TV news provide stimulation for initial feelings
and impressions about political objects, which crystallize into partisan and ideo-
logical identification for some adolescents. Atkin and Neuendorf (1980) found in
a survey of elementary school children that exposure to TV news correlates with
liking of political figures. Hawkins, Pingree, Smith, and Bechtolt (1979) reported
that high school students who viewed the 1976 ‘Great Debates’ developed closer
associations of voting preference with candidates’ party affiliations. Much of this
literature describes exposure as leading to attitude formation, assuming that pro-
cessing factors such as integration and retention occur. My model emphasizes an
additional, indirect route in which news processing motivates discussion as a
medium for identification.
Mere exposure to news is not as beneficial for conversation as active processing,
which refers to the mental effort an individual exerts to ‘figure out’ the meaning
of a news story (Kosicki & McLeod, 1990). During a campaign, motivations for
engaging in active processing include a desire to understand the election implica-
tions of news events and to acquire opinions for use in conversation. An adoles-
cent’s expectation that she will participate in political discussions should prompt
her to reflect on the significance of campaign developments and to form tentative
opinions (Kanihan & Chaffee, 1996).
The school represents the other societal agent, besides mass media, that is per-
haps most capable of harnessing stimulation to connect adolescents with electoral
politics. A school setting is potentially consequential in my model not in terms of
didactic learning of government institutions and the inculcation of patriotic alle-
giance. Until the mid 1990s, research on formal instruction generally concluded
that the impact of social studies curriculum is minimal beyond transmission of
textbook knowledge (Niemi & Junn, 1998). More in keeping with my emphasis
on the child’s active role, recent studies show that peer-based discussion at
school engenders multiple dimensions of political cognition and behavior (Center
for Information & Research on Civic Learning & Engagement, 2003). Discussion
of current affairs enables peer-centered learning through socially constructed
knowledge (Torney-Purta, 1995), the establishment of peer norms for political
communication (Kanihan & Chaffee, 1996), and practice in opinion expression
(McDevitt & Chaffee, 2002). In the waning weeks of a campaign, exposure to
political communication at school and through media should foster curiosity
while providing a basis of information to motivate conversations at home.
H1a: Use of news media during a campaign will predict child-initiated discussion
with parents after Election Day.
H1b: School discussion during a campaign will predict child-initiated discussion
with parents after Election Day.
THE PARTISAN CHILD 71
I assessed partisan and ideological identification in both interviews, which
allowed me to trace the genesis of these orientations back to the communication
behaviors. Child-initiated discussion acts as the hinge in this process, linking a
situational increase in political involvement to attitude formation. We have docu-
mented elsewhere how the family, once activated by exogenous influence, can
constitute a domestic sphere in which political communication provides prepara-
tion for the kind of deliberative discourse associated with the public sphere
(McDevitt & Kiousis, 2004). Thus, as older children/young adolescents begin to
imagine their identities as citizens, they can take advantage of the relative safety
of the home as a practice field for political expression. Several recent studies, in
fact, describe the family as a domain in which spontaneous and informal talk,
while falling short of rules-based discourse, is nonetheless translated into opinion
crystallization (Kim, Wyatt, & Katz, 1999; Wyatt, Kim, & Katz, 2000).
Research on the link between conversation and attitudes has found that ‘initial
utterances are sometimes crude but are often followed by more sophisticated
statements as opinions become more concretely formed through conversation’
(Dutwin, 2003, p. 245). In keeping with normative models of discursive democracy,
empirical studies have shown that conversations engender higher quality arguments
(Wyatt et al., 2000), while helping discussants to identify cognitive inconsisten-
cies (Zaller, 1992). Even before parents respond, adolescents are likely to benefit
from the cognitive effort of anticipating and initiating conversations.
OUTCOME #2: FEEDBACK FROM PARENTS
I can illustrate the assertiveness of the child by presenting here the questionnaire
items used to create a measure of initiated discussion:
1. Have you encouraged a parent to pay attention to news that you were
watching?
2. Have you asked a parent any questions about news articles in the last
month or two?
3. Have you suggested that a parent read a news article in the last month or
two?
4. Have you said anything to encourage a parent to vote in the November 7
election?
5. Have you tried to talk a parent into supporting any candidate or party?
Adolescents who engage in these kinds of conversations are clearly trying to
influence parents and to generate feedback. In my model of developmental prov-
ocation, the heart of the growth dynamic is reciprocity energized by exogenous
influence, a process that can be summarized as E → [A ← → (A ← → P)], where E
represents an external factor such as an election campaign, A is the adolescent, and
P signifies the parent. The adolescent initiates this process in political discussion,
72 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
and he or she, in turn, benefits from the stimulation of a newly structured rela-
tionship with his or her parent, a relationship characterized by reciprocity of
information exchange and opinion expression. An infant, a child, and an adoles-
cent all have the capacity to influence parents, but the basis for reciprocal influ-
ence changes as a child grows during adolescence. For example, reciprocal
influence in the infant–parent dyad is based to a large extent on the parent’s
recognition that the child is in need of physical attention and nurturing. An
infant begins to understand that crying elicits the desired response from parents
(Barratt, 1995). An adolescent is more equipped to prompt parental responses
based on the demonstration of cognitive competence in subjects taught in school,
or competence developed from peer groups, computer savvy, or news literacy.
Theorizing on family communication patterns (FCP) provides a rationale for
appreciating the value of feedback as a stimulus for identity formation. McLeod
and Chaffee (1972) developed the FCP construct to assess socializing outcomes
of parent–child communication. Family communication oriented toward public
affairs is concept-oriented, with parents encouraging children to express political
opinions. In socio-oriented families, parents stress to children the importance of
deference and ensure social harmony by insisting that children give in on argu-
ments that might offend others.
A model of developmental provocation suggests several refinements of the
FCP framework as applied to political development. First, FCP research generally
presumes that communication patterns represent stable structures of interaction
determined by parents’ preference. Saphir and Chaffee (2002), however, recently
demonstrated how child-initiated discussion in response to a school intervention
contributes to concept- and socio-orientation.
In fact, the adolescent him- or herself might need to disrupt prior habits of
family communication given the inertia of longstanding relationships and paren-
tal preferences. Families of low socioeconomic status (SES) in particular are
characterized by authoritarian parenting styles that would seem to preclude the
provocations of a partisan child (Dornbusch, Ritter, Leiderman, Roberts, &
Fraleigh, 1987). But, like hatchlings poking through eggshells, some adolescents
should persist nonetheless, motivated by an inevitable desire to express an auton-
omous identity (Meadowcroft, 1986). If this scenario seems overly optimistic, I
can point to prior studies in our research program, which documented that ado-
lescents persist in political conversation during campaigns, even as they prompt
defensive responses from parents (McDevitt & Chaffee, 2002).
As operational scales, socio- and concept-orientation have tended to be uncor-
related in prior studies, but this pattern is based primarily on cross-sectional
studies that might mask fluctuations in the structure of family communication. I
would expect a periodic increase in both concept- and socio-oriented commu-
nication in response to exogenous influence that stimulates adolescent-initiated
discussion. The panel design of Saphir and Chaffee resulted in the observation of
THE PARTISAN CHILD 73
a positive correlation in the change scores of the two variables as measured over a
six-month period. To reconcile this finding with prior studies, the authors sug-
gested that this outcome puts adolescents in a double bind. ‘That is, these parents
said concept-oriented things such as, “Say what you think about politics, even if the
two of us disagree,” and also socio-oriented things such as, “Don’t argue with adults
about politics”’ (Saphir & Chaffee, 2002, p. 103). But if we shift the perspective
from parents to children, this scenario is hardly a double bind. It is a functional and
predictable outcome generated by adolescents as two useful forms of feedback.
Prior studies have shown that concept-orientation is positively correlated and
socio-orientation negatively associated with adolescents’ political knowledge,
attention to news, and conversations about politics. However, most research on
FCP has used cross-sectional surveys to account for incremental effects. From
an episodic perspective, an upsurge of parental defensiveness could signal the
emergence of a partisan child. We expect, consequently, a positive association
between socio-oriented communication and political socialization during a
period of rapid political growth. Specifically, we predict that a child’s assertive
communication will generate both affirmative and defensive feedback.
H2: Child-initiated discussion will account for increases in concept- and socio-
oriented feedback from parents.
Parents could respond in all sorts of ways, from admonitions, to encouragement,
to information seeking in anticipation of future conversations. These responses—
whether negative, positive, or neutral—provide information for the child. An
adolescent might prompt parents so that she or he can learn about the family’s
legacy of partisanship—for example that ‘Grandpa was a union activist but Dad
voted for Reagan’. A daughter might discover an issue that helps her to bond
more closely with her mother. A precocious son, ready to take on Dad in a dinner-
table showdown, might test whether a well-timed remark causes indigestion.
These hypothetical scenarios illustrate that parent responses are useful to the
adolescent in opinion formation as a basis for comparison, for self-understanding,
and for rebelliousness, among other possibilities. Thus, the two outcomes I
have described—the cognitive processing associated with initiating conversa-
tions and parent feedback—should promote political identity.
H3a: Child-initiated discussion will predict party and ideological identification.
H3b: Parent feedback will predict party and ideological identification.
METHOD
The setting for this panel study is Lubbock, located in the high desert plains of
west Texas. The demographic profile of adolescent respondents is primarily
Hispanic and Anglo, with a large percentage of low-SES families. Teachers at
74 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
O. L. Slaton Middle School agreed to distribute questionnaires. I also chose
this location because of the implications for discussion—in this regard a Texas
school in the fall of 2000 would probably benefit from the extra stimulation pro-
vided by George W. Bush running for president as a native son. Thus, the loca-
tion and timing of the study establish field conditions compatible with the
perspective that political identification tends to proceed in periodic as opposed to
incremental fashion, in response to momentous events (Sears & Valentino,
1996). Table 1 provides a time line. Teachers distributed paper-and-pencil ques-
tionnaires to students on October 14, 15 and 16, and once again, after Election
Day, on November 27 and 28. Six weeks separated the first and second interviews,
and during this dramatic period in the nation’s electoral history, the adolescents
were likely to have been exposed to an extraordinary amount of political stimulation.
The final presidential debate, Election Day, and the initial weeks of the Florida bal-
lot dispute occurred between the two interviews. While I relied on the cooperation
of teachers, the purpose of the study is not to document the impact of a specific les-
son plan—the lack of random assignment to classroom conditions prevents me from
conducting a conventional evaluation of curriculum effects. Instead, I will assess
whether several exogenous factors (i.e. media use and school-wide discussion) stim-
ulate changes in the interpersonal dynamics of the child–parent dyad.
The sample size for adolescents who completed both interviews is 620, a
response rate of 77 percent for Wave 2 based on Wave 1 completed questionnaires.
The ethnic profile is 39 percent Hispanic, 36 percent Anglo, 9 percent black, and
the rest Native American, Asian, and ‘other’. Respondents were spread equally
across the seventh, eighth, and ninth grades. I chose to interview children during
early adolescence to test the premise that political identification, even during its
early phase, is grounded in the cognitive effort of news use and initiated discussion.
PLANNED ANALYSES
The first step in the analysis is to assess habitual media exposure, active process-
ing of news, debate viewing, and school discussion as independent variables. I
measured these indicators during the first interview, along with demographics,
to predict child-initiated discussion after the election. Next, I will determine the
influence of initiated discussion on concept- and socio-orientation, incorporating
TABLE 1Study time line
Mid
September
Oct. 14–16 Oct. 17 Nov. 7 November Nov. 27–28
School year
begins
First wave of
interviews
Final
presidential
debate
Election
Day
Florida
ballot
dispute
Second
wave of
interviews
THE PARTISAN CHILD 75
demographics and the benchmark indicators of FCP as Wave 1 controls. Finally,
I will document the influence of child-initiated discussion and parent feedback
on political identity as measured after Election Day. Item wording, coding, and
reliability assessment for each measure are provided in the Appendix.
VALIDITY
This study relied on self-reported indicators of school discussion, media use,
conversations with parents, and political identity. Such measures are problem-
atic if inflated by social desirability bias. I address this possibility in several
ways. First, a randomly distributed upward bias does not affect correlations
across an entire sample, and the statistical analyses control for demographics that
might correlate with a non-random tendency to give desirable answers.1 Second,
I checked the self-reported behaviors against a composite measure of campaign
knowledge. Because answers to knowledge questions cannot be faked, they are not
affected by social desirability bias. The test revealed a consistent pattern of significant
correlations between knowledge and the other indicators of political involvement.
RESULTS
H1a states that media use during the campaign will predict child-initiated dis-
cussion with parents after Election Day. Table 2 reports results from hierarchical
regression in which a demographic block of ethnicity, year in school, grades, and
gender are entered in the first equation, followed by parent-initiated discussion
as an additional control in the second equation. The final block includes the
three media-use measures and school discussion. The standardized betas from
the final equation reveal no appreciable influence of demographics. Parent-initiated
discussion, which represents the standard approach to political socialization as
top-down influence, explains 9 percent of the variance in child-initiated discus-
sion. This finding is hardly surprising given the reciprocal nature of most inter-
personal communication. More importantly for this study, parent-initiated
conversation provides a conceptual basis for evaluating the influences of media use.
In other words, once I control for the parent’s contribution to discussion in the
dyad, I can better assess the child’s own initiative as motivated by media use.
Among these predictors, the amount of news exposure and active processing pro-
duce significant betas, while debate viewing fails as a predictor. Given that prior
research has described TV debates as a strong stimulus for political socialization
(Hawkins et al., 1979), I suspect that the restricted variance of the single-item
measure might account for the meager coefficient.
1The control for student grades alleviates the concern that effects are merely an artifact of the brighter
or more aware adolescents being more likely to recall both school discussion and political communication at home.
76 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
The regression model provides no support for H1b, that school discussion will
stimulate child-initiated conversation at home. Limitations of a single-item
measurement should also be considered when evaluating the poor showing of
this variable. I note as well that the wording of the item—‘In school this fall, how
often has the election been discussed in your classes?’—could be problematic. In
any classroom or combination of classes, there can be tremendous variance in
motivated discussion across students, with active participants and listeners prob-
ably gaining the most. The wording of this indicator, unfortunately, does not
capture these kinds of nuances of differential involvement. Even so, I have
support for H1a and, thus, evidence that adolescents possess cognitive assets
from media to initiate conversations at home. And in partial correlations control-
ling for demographics, school discussion is correlated with news exposure
(r=.17, p<.001), active processing (r=.20, p<.001), and debate viewing (r=.10,
p<.05). I suspect that, in general, media use and school discussion are mutually
supportive as stimulants of learning, family discussion, and attitude formation.2
This would be consistent with research showing that discussion contributes to
2I explored whether exposure to school discussion might interact with the media-use measures to account
for incremental variance in child-initiated conversation. The interaction terms failed to reach statistical
significance.
TABLE 2Influences of media use and school discussion at T1 on child-initiated
discussion at T2 (Hierarchical regression)
aT1 refers to pre-Election Day indicators and T2 refers to the post-campaign effect.
Note: Cell entries are standardized regression coefficients. Increments to variance explained are indicated in
parentheses.
+p<.10, *p<.05, **p<.01, ***p<.001.
Independent variables Equation 1Equation 2Equation 3
Demographics
Ethnicity (Anglo dummy) .09+ .07 .03
Year in school .01 −.02 −.05
Grades .14* .09+ .01
Gender (male dummy) .06 .06 .07
(R2) (.04**)
Parent-initiated discussion .30*** .15**
(ΔR2) (.09***)
Media use and school discussion
Frequency of news exposure .20**
Active processing .18**
Viewing of debates .01
School discussion .01
(ΔR2) (.09***)
(Total adjusted R2) (.22***)
THE PARTISAN CHILD 77
individuals’ ability and willingness to assimilate information from news
(Scheufele, 2002). However, my study design does not allow me to make a defin-
itive assessment of the causal relationship between media use and school discus-
sion as T1 (baseline) factors.
Child-initiated discussion should account for changes in family communica-
tion patterns, as stated in H2. Table 3 reports the results from a hierarchical
regression analysis in which the demographic measures are entered in the first
equation along with the T1 indicators of FCP. The post-Election Day measure
of initiated discussion is entered in the second equation to account for incremental
variance in concept- and socio-oriented communication at T2. This procedure
establishes a rigorous test, in that child-initiated discussion prior to the election
might have already exerted substantial influence on family communication patterns.
Table 3 shows that child-initiated discussion was a strong predictor of parents’
concept oriented responses after the campaign, despite the pre-election control
variables. Child-initiated discussion was also a significant predictor of socio-
oriented responses, although to a lesser degree. Adolescents were certainly gener-
ating a range of reactions from parents, providing evidence in support of H2.
Next I consider whether initiated conversation and feedback help to explain
the origins of political identity (H3a and H3b). As shown in Table 4, demographics
are entered as controls in logistic regression along with the pre-Election Day
indicators of parent-initiated discussion and the adolescent’s partisanship/ideology,
TABLE 3Influences of child-initiated discussion on family communication
patterns (Hierarchical regression)
Note: Cell entries are standardized beta coefficients from the final equations of hierarchical regression. Increments
to variance explained are indicated in parentheses.
*p<.05, ***p<.001.
Independent variables Post-Election Day outcomes
Concept-oriented
communication
Socio-oriented
communication
Pre-election day controls
Ethnicity .09* .02
Year in school .01 .07
Grades .12*** −.07
Gender .05 −.03
Concept-/socio-orientation .36*** .39***
(R2) (.39***) (.20***)
Post-campaign effect
Child-initiated discussion .36*** .21***
(ΔR2) (.11***) (.04***)
Total adjusted R2 (.50***) (.24***)
78 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
news media use, and exposure to school discussion at T1. The demographics are
mostly inconsequential as predictors, although year in school and being female
increase the odds of adopting an ideology. Parent-initiated discussion is also a
weak predictor. Not surprisingly, the Wave 1 measures of identification are asso-
ciated with increased odds of partisan and ideological identity as assessed after
Election Day. Meanwhile, none of the media-use measures help to classify
respondents correctly. We know from Table 2, however, that media use cultivates
child-initiated discussion, and in the current analysis, initiated conversation
increases the odds of a child acquiring a party identification. My interpretation,
which is backed up by path modeling results not shown here, is that media use
promotes partisanship indirectly through its influence on child-initiated discus-
sion. However, both media use and initiated conversation are not associated with
the odds that a child will identify with an ideological stance. School discussion,
while failing to predict party identification, is modestly associated with the odds
of adopting an ideology.
TABLE 4Influences of parent feedback on identification (Logistic regression)
Note: Cell entries are odds ratios from logistic regression. Pseudo R2s are indicated in parentheses. All inde-
pendent variables are standardized.
Party identification: Overall correct classification: 71.2 percent; model χ2=100.20, df =13, p=.0000.
Ideology: Overall correct classification: 81.8 percent; model χ2=102.18, df =13, p=.0000.
+p<.10, *p<.05, **p<.01, ***p<.001.
Independent variables Party identification Ideology
Model 1Model 2Model 1Model 2
Pre-election day measures
Demographics
Ethnicity 0.98 0.93 1.03 1.00
Year in school 1.08 1.10 1.36* 1.40*
Grades 1.07 1.10 1.04 0.95
Gender 1.12 0.97 0.63 0.54*
Parent-initiated discussion 1.01 0.92 0.94 0.83
Party identification/ideology 6.46*** 7.72*** 5.72*** 5.52***
News-media use and school discussion
Frequency of news exposure 1.03 1.01 1.07 1.07
Active processing 0.95 0.90 1.06 1.01
Viewing of debates 1.08 1.04 1.17 1.04
School discussion 0.98 0.93 1.28+ 1.28+
(Pseudo R2) (.22) (.27)
Post-campaign effects
Child-initiated discussion 1.19*** 1.07
Socio-oriented feedback 1.10 1.10
Concept-oriented feedback 1.05 1.17**
(Pseudo R2) (.28) (.32)
THE PARTISAN CHILD 79
Finally, I consider the implications of parent feedback as an aid to children
in identity formation.3 While many prior studies of FCP describe a negative
association between socio-orientation and political socialization, I have sug-
gested that parental admonitions provide adolescents with information that
could be useful in political identification. Table 4 reveals that there is no signific-
ant relationship between socio-orientated feedback and the odds that a child will
adopt a partisan orientation or an ideological identity. However, in partial corre-
lations controlling for demographics, socio-orientation is mildly correlated with
party identification (r=.08, p<.10) and with ideology (r=.08, p<.10). Thus,
while socio-oriented responses appear not to benefit many adolescents directly in
identity formation, they also do not act as a barrier. Defensive reactions from
parents might be a by-product of initiated-discussion engendering identity asser-
tion.4 Anticipating this possibility, we also asked adolescents the following: ‘How
often did a parent get irritated if your views were different from his or hers?’
This measure (a 5-point scale) correlated with child-initiated discussion at T2
after first controlling for demographics (r=.12, p<.05)
Concept-oriented feedback increases the odds of ideological identity but not
of party identification. We consequently have limited support for the final set of
hypotheses: Child-initiated discussion is associated with party identification but
not ideology, while concept-oriented feedback is associated with ideology but not
party identification. A plausible explanation for the findings is that identification
with a concrete institution such as a political party might require less sophistication
than commitment to an ideological orientation. Thus, the cognitive effort associated
with initiated conversation could be sufficient to stimulate party identification
for older children/younger adolescents. This interpretation is consistent with
the argument that party identification is unique as an object of political socializa-
tion in its visibility and salience (Niemi & Jennings, 1991). Additional encour-
agement from parent feedback was apparently needed to engender ideological
identification for many of the adolescents. I do not imagine that adolescents at
3Prior studies on FCP provide evidence of a fourfold typology based on splitting concept- and socio-
orientation at their midpoints, although the issue of dimensionality remains unresolved (Chaffee & Saphir,
1997). The combination of high concept- and high socio-orientation represents the classification of consensual
families. Adopting Meadowcroft’s (1986) approach, I tested the interaction term to see if consensual feedback
adds significant variance to an equation in which concept- and socio-orientation are separate predictors of
political identity (Table 4). The interaction failed as a significant predictor. However, child-initiated
conversation does tend to move the family toward consensual interaction, that is, a combination of concept- and
socio-orientated discussion—the orientations correlated with each other as measured after the election (r=.38,
p<.001).
4In light of the original work on FCP dynamics, socio-oriented feedback might predict mostly congruence in
child and parent identification. In a transmission process, socio-orientation encourages children to adopt
parental perspectives for the sake of harmony. Thus, while socio-orientation in this study is modestly
associated with party ID and ideology, it is probably predicting matching identification with parents. If so, this
scenario would seem to contradict the conception of the active child as described by ‘developmental
provocation’. Still, it is worth pointing out that child-initiated conversation is a strong predictor of socio-
oriented feedback (Table 3). Thus, the child is active in prompting conversation in the dyad even if the
ultimate outcome of this discussion is congruent political identity.
80 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
this age have achieved ideological sophistication; I can only infer that feedback
increased the likelihood that they identified themselves as liberal or conservative.
DISCUSSION
Children are often seen but not heard in perspectives on political socialization—
they frequently reside in theory as quiet and passive receptors of parent influence.
The relationship between parent and child orientations is couched in such stark
terms as inheritance (Glass, Bengtson, & Dunham, 1986), absorption (Verba et al.,
2003), and imprinting (Niemi & Jennings, 1991). In challenging this transmission
paradigm, I developed a theoretical model that explicates how adolescents assert
themselves in identity formation.
At the empirical level, I was unable to detect strong influences from school
discussion despite recent studies showing the benefits of this activity. I suspect that
this non-finding is due to limitations of measurement and the restricted variance
associated with drawing data from one school district in one region of the country.
However, the intrigue of the election campaign, topped off by the fiasco in Florida,
created plentiful stimulation for variance in media use. Frequency of news expo-
sure and active processing promoted child-initiated discussion with parents.
These conversations elicited feedback as manifest in concept- and socio-oriented
responses from parents. Child-initiated discussion predicted party identification
despite a rigorous set of Time 1 controls, while concept-oriented feedback was
associated with the likelihood of adopting an ideological identity.
As I ponder the findings in aggregate, I surmise that adolescents sought to
determine the personal relevance of campaign news as a way to envision their
options for political allegiance. They appeared to acquire this self-understanding
by initiating conversations with parents, a process reminiscent of media
dependency theory. ‘In this perspective a reason to discuss news items is to vali-
date at a personal level the impersonal media content. In this way, people can come
to terms with a private judgement and become more certain of their opinions’
(De Boer & Velthuijsen, 2001, p. 143). By soliciting conversational feedback, ado-
lescents seem to generate information from parents as a basis for comparison, con-
trast, reflection, and debate—all of these activities might foster political identity.
But there appears to be more at stake than information seeking. Adolescents
coaxed and cajoled parents to read newspaper articles, to support certain candi-
dates, and to vote. While the questionnaires did not include a direct measure of
political efficacy, it seems reasonable to conclude that adolescents gained confid-
ence by virtue of their ability to influence parents. Thus, identity formation
might proceed not merely through the cognitive route of prompting information
from parents, but from self-esteem earned in interpersonal political competence.
Whether parental reactions are positive or negative, encouraging or discourag-
ing, these responses are no small achievement. I infer, from the perspective of
THE PARTISAN CHILD 81
developmental psychology (Maccoby & Martin, 1983), that adolescents essen-
tially put parents on notice that they want more active parenting when it comes
to political conversation. They are—at childhood’s end—‘teaching’ parents how
to interact with them (Barratt, 1995).
This approach departs from prior research on political socialization, which has
tended to conceive of identity formation in its early phase as the one-way, affec-
tive influence of parents. The respondents were still in their early adolescence—
they were 12–14 in the fall of 2000. While I do not have data on the specific
content of child-initiated discussions, I can surmise that adolescents backed up
their conversations with active processing of news content (Table 2). The cam-
paign may have been a heady moment for them as they came to realize what it
means to be a citizen, that it involves not just voting, but the ability to interact
with adults as equals.
While emphasizing the capacity of the child to initiate political discussion with
parents, I recognize that a unidirectional, bottom-up perspective would echo
the mistakes of prior theorizing, which promoted a transmission model of influ-
ence in the opposite direction. Growth occurs at the level of the family system,
which allows for reciprocity energized by exogenous stimulation. Both the child
and the parent must contribute to this process. Note, for instance, that parent-
initiated discussion is a strong predictor of child-initiated conversations (Table 2).
It seems likely that parent-initiated discussion influences child-initiated discus-
sion directly as well as indirectly through parental encouragement of the child’s
media use. However, the child can be seen as active in her or his responsiveness
to the stimulation provided by parents as the child seems to eagerly pay more
attention to news. This is itself a noteworthy outcome, given adolescents’ usual
lack of interest in news and public affairs. Further, even as I give the parent a
good deal of credit for spurring the child to initiate discussion, the subsequent
analyses illustrate the efficacy of the child in initiating conversations that alter
family communication patterns (Table 3), and that promote the child’s party
identification (Table 4).
LIMITATIONS AND FUTURE RESEARCH
I have evidence to show that adolescents initiated discussion that led to a
restructuring (at least temporarily) of family communication patterns, but this
begs the question as to whether children otherwise influenced the political
involvement of parents. For a supplemental analysis, I interviewed many of the
parents with post-election questionnaires. Students were instructed to ask one
parent to complete the questionnaire. Unfortunately, this procedure produced a
total of only 230 parent questionnaires, bringing up the issue of selection bias.
Among this group of parents, however, I found that child-initiated discussion—
after first controlling for demographics—was significantly correlated with parents
82 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
viewing the debates. It also predicted parents’ viewing of the debates with chil-
dren and parents’ active processing of news. This evidence—obtained directly
from parents—provides internal validity for the study in that increased political
involvement of parents implies that the adolescents were successful in generating
feedback.
The most obvious concern with respect to external validity is the surreal
circumstances of the 2000 presidential election. However, I suspect that the
voting controversy as an impetus for socialization allowed me to uncover proc-
esses that remained covert in prior studies. Indeed, scholars have argued that
researchers should be highly selective in choosing when and where to observe
political socialization given that some areas of growth do not occur as steady,
incremental progress (Chaffee et al., 1970). For example, Sears and Valentino
(1996) describe attitude crystallization as event-driven: The gains of adolescents
occur only during the period of a campaign itself, during episodes of intense
information flow, and only in attitude domains directly relevant to candidates
and parties during the campaign. The application of developmental provocation
as a theoretical model, consequently, is restricted to family communication
dynamics during situations of high political involvement. Still, a lingering ques-
tion is whether the hypotheses generated by the model would be supported in
routine campaigns.5 Here I can point to research on Kids Voting USA, a K-12
civic curriculum taught during the final weeks of campaigns. Evaluations of
intervention effects have been conducted in the context of several presidential
and non-presidential elections, in several regions of the United States (McDevitt
& Chaffee, 1998, 2002; McDevitt & Kiousis, 2004). In each study, Kids Voting
exposure stimulated student-initiated discussion with parents, which predicted,
in turn, adolescent political identification.
Another concern is that the design lacked contrasting conditions and ran-
dom assignment. This leaves open questions about the relationship between
media use and school discussion as precursors to child-initiated conversation.
The logistics of enacting such a study would be formidable in a field setting.
But this type of design would go a long way toward documenting the anteced-
ents of initiated conversation and identity acquisition. Peer-group discussion
at school, for example, might interact with media use at home to foster family
conversations.
5The sequencing of variables in a model of political communication could change depending on the degree
of exogenous stimulation from a campaign. For example, many empirical and normative models of deliberative
democracy assume that news use initiates the process of opinion formation via discussion (Kim et al., 1999). In
this study, we view both media use and school (or peer-group) discussion as predictors of conversation at
home. However, a highly stimulating campaign, combined with effective civic instruction by teachers, could
engender a sequence in which peer-group discussion at school clearly precedes media use, particularly for
families in which parents did not previously emphasize the value of reading newspapers and paying attention to
the evening news. Consequently, future research is needed to assess the generalizability of the model presented
here with respect to both ordinary and exceptional campaigns.
THE PARTISAN CHILD 83
This study also did not include direct measures of adolescents’ motivation to
elicit feedback. I have inferred that this outcome is an overt manifestation of an
underlying, developmental goal in which children assert autonomy. Thus, there are
short-term and long-term processes at work in political identity that arise out of
changes in the structure of child–parent communication. Table 5 outlines a heuris-
tic for theory building based on identifying variables implicated in developmental
provocation. This framework can also be used to develop hypotheses about the
actual content of child-initiated discussion. In the short-term, an adolescent might
develop curiosity about parents’ emotional response to a touchy political topic; I
suggest the term affective validation for this concept. In a Hispanic family, for
instance, a teenager might wonder how her Catholic parents feel about abortion
given their insistence that ‘we are Democrats’. If discussion occurs, the adolescent
could acquire stronger feelings, or perhaps mixed feelings, about issues that seem to
resonate with deeply held family values. Cognitive validation would proceed when
an adolescent desires to evaluate the attributes of opinions such as their salience,
internal consistency, and persuasiveness. Resulting conversations should lead to
increased sophistication in argument repertoire and attitude crystallization.
Adopting a long-term perspective, scholars of human development highlight two
goals of family interaction: autonomy and connectedness (Peterson, 1995, p. 27).
The dyadic and systemic levels of families are prominent environments in which
feelings of togetherness support the closely related processes of self-concept
development. In the long-term, an adolescent might achieve differentiation by
claiming a partisan loyalty at odds with the preference of parents. In a less optim-
istic perspective, differentiation could be achieved by adolescents adopting a
cynical stance given a contrast between parental insistence on idealism and the
TABLE 5Developmental provocation: A heuristic for future research
Feedback goals Interpersonal communication Expected outcome
Short-term
Affective
validation
Adolescents assess emotional
resonance of opinions
in family.
Strengthened feelings about
political issues and actors.
Cognitive
validation
Adolescents test views for
salience, persuasiveness,
contradictions, etc.
Strengthened argument
repertoire, attitude
crystallization, and
consistency.
Long-term
Differentiation Adolescents provoke
disagreement to assert autonomy.
Incongruence of child and
parent party identification.
Cohesion Adolescents seek agreement
to preserve togetherness
in the family.
Congruence of child and
parent party identification.
84 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
reality of partisan manipulations. Certainly this socialization outcome was pos-
sible in the aftermath of Election Day in 2000. This example illustrates the need
for an accounting of the actual content of family discussion to supplement the
measurement of motivational constructs.
If a son or daughter pursues autonomy outside the context of political commu-
nication, opinion expression might address instead the need to preserve family
cohesion during the tumultuous years of adolescence. In each of these scenarios,
political identity is somewhat inadvertent, involving family functions that tran-
scend political socialization. Nonetheless, these outcomes signify the achievement
of citizenship in its formative phase.
APPENDIX
Demographics, parent-initiated discussion, media use, and school discussion
were measured at T1.
Ethnicity. A single item recorded ethnicity as a dummy variable (white). Cate-
gories included Anglo, Hispanic, black, Native American, and ‘other’.
Year in school. A single item measured the student’s year in school. ‘What
grade are you in at school?’ Coded: 7th =1, 8th =2, 9th =3.
Grades. A single item measured grades received in school. ‘Would you say
your grades are mostly A’s, B’s, C’s or D’s?’ mostly A’s =4, mostly B’s =3,
mostly C’s =2, mostly D’s =1.
Gender. A single item recorded gender as a dummy variable (male).
Parent-initiated discussion. A summed two-item scale assessed frequency of
parent-initiated discussion. The correlation is .35 (p<.001). ‘Has a parent asked
you any questions about news articles in the last month or two?’ No, don’t know
(DK) =1; sometimes =2; yes =3. ‘Has a parent suggested that you read a news
article in the last month or two?’
Use of news media. A six-item scale measured adolescents’ habitual exposure to
news media. I standardized and summed the scores. The alpha is .71. ‘On average,
how many days per week do you read a newspaper?’ 0 days, DK =0; 1 day =1;
2days =2; 3 days =3; 4 days =4; 5 days=5; 6 days=6; 7 days =7. ‘About how
many days per week do you watch a television news program, such as the evening
news?’ ‘Is TV news something you usually turn on, or does someone else turn it
on?’ I turn it on =3; someone else turns it on, DK=1; both =2. ‘Using a scale
from 1 to 5, with 1 meaning ‘none’ and 5 ‘a great deal’, how much attention do
you pay to news about national politics in Washington, D.C.?’ ‘How much attention
do you pay to news about state politics in Texas?’ ‘How much attention have you
paid to news about the presidential campaign?’
Active processing. A summed three-item scale measured news processing. The
alpha is .72. ‘When I see or read a news story about a candidate, I try to figure
out if it is biased.’ Not at all like me, DK=1; somewhat like me =2; a lot like
THE PARTISAN CHILD 85
me =3. ‘When I hear news about politics, I try to figure out what is REALLY
going on’. ‘News about people running for office makes me wonder how they
might change things’.
Viewing of debates. A single item measured viewing of the presidential debates.
‘Have you watched any of the presidential or vice-presidential debates on TV?’
No, DK =1; sometimes =2; yes=3.
School discussion. We asked respondents to estimate the frequency of school
discussion. ‘In school this fall, how often has the election been discussed in your
classes?’ Never =1; rarely =2; sometimes=3; often =4; very often =5.
Child-initiated discussion. I summed responses from five questions to assess
child-initiated discussion at T2. The alpha is .69. ‘Have you encouraged a parent
to pay attention to news that you were watching?’ No, DK =1; sometimes =2;
yes =3. ‘Have you asked a parent any questions about news articles in the last
month or two?’ ‘Have you suggested that a parent read a news article in the last
month or two?’ ‘Have you said anything to encourage a parent to vote in the
November 7 election?’ ‘Have you tried to talk a parent into supporting any can-
didate or party?’
Parent feedback. A summed three-item scale measured concept-oriented
responses. The alpha is .75 at T1 and .79 at T2. ‘How often does a parent
encourage you to say what you think about politics, even if the two of you might
disagree?’ Never =1; rarely =2; sometimes =3; often =4; very often =5. ‘How
often does a parent encourage you to form your own opinions about political
issues?’ ‘How often does a parent ask for your opinions on things?’
A summed two-item scale measured socio-oriented responses. The correlation
is .41 (p<.001) at T1 and .43 (p<.001) at T2. ‘How often does a parent indicate
that political discussions are better if you keep them pleasant?’ ‘How often does a
parent encourage you NOT to argue about politics with adults?’
Party identification. This measure was created by folding over an identification
scale so that Republicans and Democrats are grouped at the high end. It was
assessed at T1 and T2. ‘Do you think of yourself as a Republican, a Democrat, or
something like that?’ Republican, Democrat=1; independent, not a member of
another party, DK =0.
Ideological identification. This measure was created by folding over a scale so
that liberal and conservative identifications are grouped at the high end. It was
assessed at T1 and T2. ‘When it comes to politics, do you ever think of yourself
as liberal or conservative?’ Liberal, conservative=1; no, I’m moderate or ‘middle
of the road’, neither: I don’t think of myself that way, DK=0.
REFERENCES
Abramowitz, A. I., & Saunders, K. L. (1998). Ideological realignment in the U.S.
electorate. Journal of Politics, 60, 634–652.
86 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
Atkin, C., & Neuendorf, K. (1980). Television news exposure and children’s knowledge about
political figures. Unpublished manuscript, Department of Communication, Michigan
State University.
Austin, E., & Pinkleton, B. E. (1997). Parental mediation as information source use: Political
socialization effects. Paper presented to the International Communication Association,
Montreal.
Barratt, M. S. (1995). Communication in infancy. In M. A. Fitzpatrick & A. L. Vangelisti
(Eds.), Explaining family interaction (pp. 5–33). Thousand Oaks, CA: Sage.
Bennett, S. E. (1997). Why young Americans hate politics, and what we should do about
it. PS: Political Science & Politics, 30, 47–52.
Center for Information & Research on Civic Learning & Engagement. (2003). The civic
mission of schools. CIRCLE and Carnegie Corporation of New York.
Chaffee, S. H., & Saphir, M. N. (1997). Family communication patterns: The story so far.
Paper presented to the Association for Education in Journalism and Mass Communi-
cation, Chicago.
Chaffee, S. H., Ward, S., & Tipton, L. (1970). Mass communication and political social-
ization. Journalism Quarterly, 47, 647–659.
Connell, R. W. (1972). Political socialization in the American family: The evidence
reexamined. Public Opinion Quarterly, 33, 323–333.
De Boer, C., & Velthuijsen, A. S. (2001). Participation in conversations about the news.
International Journal of Public Opinion Research, 13, 140–157.
Dornbusch, S. M., Ritter, P. L., Leiderman, P. H., Roberts, D. F., & Fraleigh, M.
(1987). The relation of parenting style to adolescent school performance. Child Devel-
opment, 58, 1244–1257.
Dutwin, D. (2003). The character of deliberation: Equality, argument, and the formation
of public opinion. International Journal of Public Opinion Research, 15, 239–264.
Easton, D., & Dennis, J. (1969). Children in the political system. New York: McGraw-Hill.
Glass, J. L., Bengtson, V. L., & Dunham, C. C. (1986). Attitude similarity in three-
generation families: Socialization, status inheritance, or reciprocal influence? American
Sociological Review, 51, 685–698.
Greene, S. (2000). The psychological sources of partisan-leaning independents. American
Politics Quarterly, 28, 511–537.
Greenstein, F. (1965). Children and politics. New Haven: Yale University Press.
Hawkins, R., Pingree, S., Smith, K., & Bechtolt, W. (1979). Adolescents’ responses to
issues and images. In S. Kraus (Ed.), The great debates: Carter vs. Ford, 1976 (pp. 368–383).
Bloomington: Indiana University Press.
Hyman, H. H. (1959). Political socialization. Glencoe, IL: Free Press.
Jennings, M. K., Stoker, L., & Bowers, J. (1999). Politics across generations: Family trans-
mission reexamined. Paper presented to the American Political Science Association,
Atlanta.
Kanihan, S., & Chaffee, S. H. (1996). Situational influence of political involvement on
information seeking: A field experiment. Paper presented to the Association for Educa-
tion in Journalism and Mass Communication, Anaheim, CA.
Kim, J., Wyatt, R. O., & Katz, E. (1999). News, talk, opinion, participation: The part
played by conversation in deliberative democracy. Political Communication, 16, 359–360.
THE PARTISAN CHILD 87
Kosicki, G., & McLeod, J. (1990). Learning from political news: Effects of mass media
images and information-processing strategies. In S. Kraus (Ed.), Mass communication
and political information processing (pp. 69–83). Hillsdale, NJ: Erlbaum.
Maccoby, E., & Martin, J. A. (1983). Socialization in the context of the family: parent–
child interaction. In P. H. Mussen (Ed.), Handbook of child psychology (pp. 1–101).
New York: Wiley.
McDevitt, M., & Chaffee, S. (1998). Second chance political socialization: ‘Trickle-up’
effects of children on parents. In T. J. Johnson, C. E. Hays, & S. P. Hays (Eds.),
Engaging the public: How government and the media can reinvigorate American democracy
(pp. 57–66). Lanham, MD: Rowman & Littlefield.
McDevitt, M., & Chaffee, S. (2002). From top-down to trickle-up influence: Revisiting
assumptions about the family in political socialization. Political Communication, 19,
281–301.
McDevitt, M., & Kiousis, S. (2004). The discursive bonding of family and school: Spanning
the domestic and public spheres. Paper presented to the International Communication
Association, New Orleans.
McLeod, J. M., & Chaffee, S. H. (1972). The construction of social reality. In J. Tedeschi
(Ed.), The social influence processes (pp. 50–99). Chicago: Aldine-Atherton.
Meadowcroft, J. M. (1986). Family communication patterns and political development:
The child’s role. Communication Research, 13, 603–624.
Niemi, R. G., & Jennings, M. K. (1991). Issues and inheritance in the formation of party
identification. American Journal of Political Science, 35, 970–988.
Niemi, R. G., & Junn, J. (1998). Civic education: What makes students learn. New Haven:
Yale University Press.
Peterson, G. W. (1995). Autonomy and connectedness in families. In R. D. Day,
K. R. Gilbert, B. H. Settles, & W. R. Burr (Eds.), Research and theory in family science
(pp. 20–41). Pacific Grove, CA: Books/Cole.
Saphir, M. N., & Chaffee, S. H. (2002). Adolescents’ contribution to family communica-
tion patterns. Human Communication Research, 28, 86–108.
Scheufele, D. A. (2002). Examining differential gains from mass media and their impli-
cation for participatory behavior. Communication Research, 29, 46–65.
Sears, D., & Chaffee, S. H. (1979). Uses and effects of the 1976 debates: An overview of
empirical studies. In S. Kraus (Ed.), The great debates: Carter vs. Ford 1976 (pp. 223–261).
Bloomington: Indiana University Press.
Sears, D., & Valentino, N. A. (1996). Politics matters: Political events as catalysts for
preadult socialization. American Political Science Review, 91, 45–65.
Simon, J., & Merrill, B. D. (1997). The next generation of news consumers: Children’s
news media choices in an election campaign. Political Communication, 14, 307–321.
Tedin, K. L. (1974). The influence of parents on the political attitudes of adolescents.
American Political Science Review, 68, 1579–1592.
Torney-Purta, J. (1995). Psychological theory as a basis for political socialization research:
Individuals’ construction of knowledge. Perspectives on Political Science, 24, 23–41.
Verba, S., Burns, N., & Schlozman, K. L. (2003). Unequal at the starting line: Creating
participatory inequalities across generations and among groups. American Sociologist,
34, 46–69.
88 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
Westholm, A. (1999). The perceptual pathway: Tracing the mechanisms of political
value transfer across generations. Political Psychology, 20, 525–551.
Wyatt, R. O., Kim, J., & Katz, E. (2000). How feeling free to talk affects ordinary polit-
ical conversation, purposeful argumentation, and civic participation. Journalism &
Mass Communication Quarterly, 77, 99–114.
Zaller, J. (1992). The nature and origins of mass opinion. Cambridge: Cambridge University
Press.
BIOGRAPHICAL NOTE
Michael McDevitt is an assistant professor in the School of Journalism and Mass Com-
munication at the University of Colorado, Boulder. He conducts research in the areas of
political communication, political socialization, and journalism sociology.
Address correspondence to Michael McDevitt, School of Journalism and Mass Com-
munication, 1511 University Avenue, 478 UCB, Boulder, Colorado 80309–0478, USA,
E-mail: michael.mcdevitt@colorado.edu
