Effectiveness of a National Media Campaign to Promote Parent

Abstract

Background. Although there is debate on the effectiveness of youth-focused abstinence education programs, research confirms that parents can influence their children’s decisions about sexual behavior. To leverage parent-based approaches to adolescent sexual health, the U.S. Department of Health and Human Services launched the Parents Speak Up National Campaign (PSUNC) to encourage parent–child communication about sex. Previous experimental studies have found the campaign to be efficacious in increasing parent–child communication. But to date, the actual reach of the campaign and its real-world effectiveness in promoting parent–child communication has not been established. The present study addresses this gap. Method. The authors surveyed 1,804 parents of 10- to14-year-old children from the nationally representative Knowledge Networks online panel. The survey included questions about parents’ awareness of PSUNC ads and parent–child communication behaviors. The authors also analyzed market-level data on campaign gross rating points, a measure of market-level intensity of PSUNC advertising in the United States. Multivariate regressions were used to examine the association between PSUNC exposure and a three-item scale for parent–child communication. Results. Overall, 59.4% of parents in the sample reported awareness of PSUNC. The authors found that higher market-level PSUNC gross rating points were associated with increased parent–child communication. Similar relationships were observed between self-reported awareness of PSUNC and increased frequency of communication and recommendations to wait. These associations were particularly strong among mothers. Conclusions. This study provides the first field-based data on the real-world reach and effectiveness of PSUNC among parents. The data support earlier experimental trials of PSUNC, showing that the campaign is associated with greater parent–child communication, primarily among mothers. Further research may be needed to develop additional messages for fathers.

Keywords

adolescent health parent–child sexual communication social marketing teen pregnancy prevention

Recent data suggest that teen pregnancy rates in the United States may be rising again despite significant declines in both pregnancy and sexual intercourse among teens since the 1990s (Centers for Disease Control & Prevention, 2007; Martin et al., 2006). The primary focus of U.S. government policy and funding for interventions to reduce teen pregnancy has been on sexual abstinence education, under the enactment of Title V, Section 510, of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996. Currently, there is continued debate about the effectiveness of abstinence education programs funded under Title V (Kim & Rector, 2008; Sandfort, Orr, Hirsch, & Santelli, 2008; Trenholm et al., 2007; Weed, Eriksen, Lewis, Grant, & Wibberly, 2008). Some studies suggest that these programs are ineffective and may in fact promote negative attitudes toward condom use and other risk prevention behaviors (Sandfort et al., 2008; Trenholm et al., 2007). Other studies have found that youth who receive abstinence education may have lower rates of sexual activity compared to youth who do not receive abstinence education (Kim & Rector, 2008; Weed et al., 2008). A recent highly publicized randomized control trial found that abstinence-only education in particular may play an important role in early sexual debut among adolescents (Jemmott, Jemmott, & Fong, 2010).

Although there is debate on the effectiveness of youth-based abstinence education approaches, other research overwhelmingly confirms that parents can influence their children’s decisions on sexual behavior and use of contraceptives. For example, numerous studies show that parent–child communication about sex is associated with delays in adolescent sexual initiation (Aspy et al., 2007; Dilorio, Kelly, & Hockenberry-Eaton, 1999; Furstenberg, Morore, & Peterson, 1985; Lehr, Dilorio, Dudley, & Lipana, 2000; Pick & Palos, 1995). Prior research has also shown parent–child communication can be effectively promoted with media-based interventions. For example, DuRant, Wolfson, LaFrance, Balkrishnan, and Altman (2008) showed that exposure to a media campaign promoting parent–child communication about sex was associated with increased communication and parent intentions to communicate with their child. In spite of this evidence, parent-based interventions to reduce teen pregnancy have been seldom used.

To capitalize on the potential of parent-based approaches that use media campaigns, the U.S. Department of Health and Human Services recently launched the Parents Speak Up National Campaign (PSUNC), a media campaign to encourage parent–child communication about sex. The campaign is grounded in social cognitive theory, which predicts a chain of cognitive events that lead to behavioral outcomes and choices (Bandura, 2004). The PSUNC theoretical framework, published in detail elsewhere (Evans, Davis, & Zhang, 2008), hypothesizes that increased parent–child communication will result from the promotion of normative beliefs, parental self-efficacy to communicate, and outcome expectations about the effectiveness of parent–child communication behaviors in promoting future life success for the child.

Various aspects of this framework have been tested and evaluated in previous literature. For example, data from a longitudinal experimental trial of PSUNC showed that outcome expectations that parent–child communication would be successful and norms related to the appropriate age for sexual initiation were both predictive of future parent–child communication behaviors among both mothers and fathers (Davis, Blitstein, Evans, & Kamyab, 2010). Furthermore, this same study showed that experimentally controlled exposure to PSUNC TV ads was associated with positive changes in both of these cognitive variables among mothers. Other previous research also showed that outcome expectations about parent–child communication mediate the effects of social norms on parent–child communication behaviors (Evans, Blitstein, & Davis, 2011). These experimental studies have helped establish the validity of the campaign’s theoretical framework.

The campaign was launched nationally in June 2007 and targeted parents of children 10 to 14 years old, primarily using televised public service announcements (PSAs) supplemented by radio and print advertising. In 2007, the campaign initially relied on PSAs, aired during donated advertising time on local television networks. Increased campaign resources in 2008 and 2009 allowed for national paid advertising in addition to PSAs during those years.

The campaign’s advertisements address common barriers and parent fears about talking about sex and encourage parents to talk “early and often” with their children. Between 2007 and 2009, the campaign used four TV ads, titled “Muffinhead,” “Talk to Me,” “Gadgets,” and “Mimes.” The first two ads were the first ads used during the campaign’s initial airing in 2007 and featured age-appropriate youth letting their parents know that they want to talk to them about sex and that they should talk “early and often.” These ads also depict youth saying to their parents, “Tell me you want me to wait.” The latter two ads were parent-voiced ads that featured parents portraying the barriers and difficulties parents face in initiating conversations about sex with their children. These ads also feature youth saying to their parents, “Tell us to wait to have sex.” The TV ads were supplemented with a radio ad called “S-Word” that also aired in local media markets across the country. This ad also focused on addressing parent barriers to communication about sex and encouraged parents to talk “early and often” with their children. All campaign PSAs also promoted a campaign-specific website that provided parents information and guidance on how to talk with their children about sex.

A recent study using a randomized trial found that parents who were exposed to the campaign under controlled conditions were more likely to initiate conversations about sex with their children (Evans et al., 2009; Evans et al., 2011). Further results from this trial showed that parents’ exposure to the campaign was associated with norms related to appropriate child age for sexual initiation and outcome expectations that communication would be successful (Davis et al., 2010). These cognitions were also found to be predictive of future parent–child communication, validating the campaign’s strategy for behavior change by first targeting cognitive precursors to parent–child communication. A follow-up study of children of parents who participated in the randomized trial confirmed the aforementioned findings, showing that children of parents who received PSUNC exposure were more likely to report having conversations about sex with their parents (Evans, Davis, Silber-Ashley, & Snider, in press). Although these studies strongly suggest that the campaign is efficacious, their external validity is limited because of the controlled nature of parents’ exposure to PSUNC. To date, there have been no field-based studies to measure parents’ exposure to PSUNC and assess its actual effectiveness in the real world.

In the present study, we address this gap in the evidence base for PSUNC by first examining the extent to which parents in the United States were exposed to the campaign. We then explore the relationships between exposure to the campaign and measures of parent–child communication about sex among parents nationwide. Implications of our findings for PSUNC and for parent-targeted communication campaigns in general are discussed.

Method

Survey Data, Sample, and Procedures

This study uses an online survey of parents sampled from the Knowledge Networks (KN) panel, conducted in fall of 2009. The KN panel is designed to be statistically representative of the U.S. population and is the only online panel that uses a dual-frame, probability-based panel recruitment method that covers both online and off-line populations in the United States. KN panelists are randomly recruited by either random digit dialed (RDD) telephone surveys or by self-administered mail surveys through address-based sampling that uses the U.S. Postal Service’s Delivery Sequence File of U.S. mailing addresses. Individual panelists without a computer or Internet access are provided free laptops with Internet in exchange for panel participation. Thus, the KN dual-frame recruitment approach includes both listed and unlisted telephone numbers, landline and cell-only households, and non–Internet users in the United States. The KN panel data are further weighted to match U.S. demographic and geographic benchmarks for parents of 10- to 14-year-old children. Specifically, the sample weights are designed to match distributions found in the Current Population Survey (CPS) for gender, age, race/Hispanic ethnicity, education, census region, urban/rural status, and Internet access among U.S. parents of 10- to 14-year-old children.

The study sample was limited to parents of 10- to 14-year-old children, the target audience of PSUNC. We first identified all adult panelists who met this eligibility criteria (N = 3,386). Mothers and fathers were sampled separately, and parents who did not participate were not replaced by a parent of the opposite gender. Each parent was asked to provide the gender, age, and nickname of one child in the eligible age range for this study. That child was referenced throughout the survey in a number of questionnaire items on the parent’s level of communication with that child and other related survey topics. A total of 2,258 parents responded to the study invitation and were eligible to participate. Among those, 1,804 parents (1,253 mothers and 551 fathers) completed the survey (total cooperation rate = 53.3%). Because response rate data at the overall KN panel recruitment stage were not available, the true survey response rate could not be estimated. Given an assumed 20% current response rate at the panel recruitment stage, which would be consistent with typical random digit dialed–based surveys, the total survey response rate for our study specifically may be approximately 10%. Further data would be needed to calculate the exact survey response rate. Only one parent per household participated in the study. Mothers were oversampled since mothers more often engage in sexual communication with their children compared with fathers (Fisher, 1990; Meschke, Bartholomae, & Zentall, 2002; Raffaelli, Bogenschneider, & Flood, 1998) and PSUNC ads were mother-centric in their messages.

The study questionnaire, consent procedures, and protection of human subjects were reviewed and approved by the sanctioned Institutional Review Board of RTI International. Our study was also reviewed and approved by the Federal Office of Management and Budget prior to the study implementation (Office of Management and Budget control No. 0990-345).

Measures

Parents completed a self-administered online survey that consisted of questions on parent sociodemographics and other individual characteristics, a number of items on the nature and frequency of parent–child communication about sex, and measures of parents’ awareness of specific PSUNC ads. Our analysis focuses on the relationship between parents’ exposure to PSUNC ads and parent–child communication behavior. To model this relationship, we used survey questions on parents’ self-reported awareness of the campaign as well as external media market–level data on the campaign’s advertising reach. Each of these variables is described in greater detail below.

Parent–child communication

Parent–child communication was measured with a three-item scale based on survey questions that assessed the frequency with which parents have engaged in communication with their children about waiting to have sex, a primary focal message of PSUNC ads. Before answering each question, parents were told that the questions are about specific topics they may currently discuss with their child. These items include the following: (a) frequency of initiating conversations with child about sexual activity or waiting to have sex (never, seldom, sometimes, or often), (b) frequency of talking to child about waiting to be sexually active until married (not at all, somewhat, a moderate amount, or a great deal), and (c) whether the parent has asked or recommended to their child to wait to have sex (no or yes). Items (a) and (b) above are measured with a response scale ranging from 1 to 4, corresponding to each of the response frequencies, whereas the third item is measured with a simple response scale of 1 (no) and 2 (yes). We created an overall parent–child communication scale equal to the linear sum of these three items.

Parent awareness of PSUNC ads

Parents’ awareness of specific PSUNC ads was assessed with a validated procedure that measures ad recognition. Each parent was first shown a 30-second version of each PSUNC TV ad, streamed within the online survey. Participants who were not able to view the ad were instead shown a series of screenshots from the ad. A similar procedure was used to assess recognition of the “S-Word” radio ad by playing a streaming audio file within the survey. Participants who were unable to hear the radio ad were provided a text script of the audio from that ad. Once the participants had viewed and listened to each ad, they were asked how frequently they had seen each ad on TV (or heard it on the radio) in the past 6 months, with response options ranging from never, rarely, sometimes, often, and very often. We used this measure to create exposure frequency variables for (a) average frequency of exposure across all PSUNC TV ads assessed in the survey (1 = never to 5 = very often) and (b) frequency of exposure to the PSUNC radio ad (1 = never to 5 = very often). These procedures are a validated method of measuring ad exposure and have been shown to more precisely capture encoded exposure than tests that rely on memory recall alone (Southwell, Barmada, Hornik, & Maklan, 2002).

Media market–level exposure to PSUNC

We used external data from the campaign’s media contractor to measure potential exposure to PSUNC at the market level. These additional data on PSUNC exposure serve two purposes. First, they allow us to examine in more detail the reach of the campaign across media markets in the United States. Second, they provide an alternative measure of exposure to PSUNC that is exogenous to the parent. This helps address the potential limitations of selective attention bias that may be present with self-reported awareness where parents who already communicate at higher levels with their children may also be more attentive to PSUNC ads and thus more likely to indicate ad recognition. Thus, we are able to use these data to check the robustness of campaign effects across different measures of exposure and to examine the validity of individual self-reported measures of exposure. That is, market-level campaign intensity should be correlated with individual-level self-reported exposure.

PSUNC commercials were aired on select TV and radio networks across the United States from 2007 to 2009. Market-level exposure was estimated using gross rating points (GRPs), the industry standard for quantifying reach and frequency of exposure to an advertising campaign in a given media market. GRPs are based on broadcast ratings in the markets where the commercials aired and provide an estimate of the percentage of the target audience that was exposed to the ads and of the frequency with which they were exposed. For example, if 80% of the target audience in a media market saw a commercial four times in a given time period, this would translate into 320 GRPs (i.e., 80 × 4) for that period. GRPs are calculated for each of the 210 designated market areas (DMAs) in the United States. DMAs are standard geographic units comprising U.S. counties that are grouped together within television market viewing areas. U.S. counties are assigned to only one DMA. Because these data represent averages within each media market, GRPs represent parents’ potential exposure to PSUNC. A parent’s actual exposure may be more or less than the average represented by the GRP, based on TV-viewing or radio-listening habits.

In our analysis, we used separate measures for total PSUNC television GRPs and total GRPs for the “S-Word” radio ad. Although these data do not capture other PSUNC advertising activities such as unpaid PSAs and print advertising, GRPs represent the best available data on the overall level of campaign “dose” delivered to each media market. We analyzed television and radio GRPs separately because the television campaign was the most prominent part of PSUNC and was thus expected to have a larger impact than the radio ads alone. Data on total PSUNC GRPs were available for the entire year of 2008 and for January through September of 2009, prior to the fielding of our survey in fall of 2009. Because media exposure is expected to take time to influence targeted behavioral outcomes, we created cumulative measures of total 2008-2009 PSUNC television and radio GRPs. These data were then merged to our parent survey, based on standard DMA geographic identifiers for each parent. Total PSUNC GRPs at the DMA level ranged from 0 to 2,660 with the exception of one outlier DMA that received 12,367 total television GRPs.

The variability in GRPs creates a natural quasi experiment that we use to examine how market-to-market variability in campaign exposure is associated with parent–child communication outcomes at the individual level. This method of merging market-level GRP data to individual-level survey data for analysis of campaign effects on health outcomes has been used in multiple-media–evaluation studies, particularly for large antismoking media campaigns (Davis, Nonnemaker, & Farrelly, 2007; Farrelly, Davis, Haviland, Messeri, & Healton., 2005; Farrelly, Nonnemaker, Davis, & Hussin, 2009; Terry-McElrath et al., 2007).

Potential confounders

Our survey included a number of other measures on parent characteristics and other potential correlates of parent–child communication that we controlled for in our statistical analyses. These included parent marital status, educational attainment, race/ethnicity, parent age, full-time employment status, family structure (one or two parents in home), child gender, child age, child’s access to television or Internet in his or her bedroom, metropolitan statistical area status (urban or rural), and an eight-item scale for parent–child involvement. This scale includes items drawn from previous studies (National Commission on Children, 1999; Scher, Maynard, & Stagner, 2006) that capture the frequency of joint parent–child activities. These items measure past month frequency (never, less often, at least once a month, or at least once a week) of (a) shopping, (b) going to movies or sporting events, (c) watching television, (d) attending religious services, (e) doing homework, (f) attending a party, (g) volunteering, and (h) playing a game or sport. Prior research has shown that these items load into a single scale with high reliability in factor analysis (Evans et al., 2009). We also included a measure of parent religiosity based on agreement with a survey item that asked parents about how important religious beliefs are in their lives. This measure was included in our analysis to control for the potential influence of religion in parent–child communication about sex.

Because PSUNC GRPs vary across media markets in the United States, it is also important to account for factors that may be correlated with both the dose of PSUNC in any given media market and preexisting levels of parent–child communication in that market. For example, high-exposure markets tend to be less rural, be higher educated, and have higher income—all factors that may be associated with parent–child communication. Although we measure urban/rural status at the individual level, failing to account for these types of factors at the aggregate market level may result in an overstatement of the relationship between PSUNC exposure and parent–child communication. We therefore included separate DMA-level variables for three aggregate media market characteristics: (a) median household income, (b) percentage of the media market that are college graduates, and (c) DMA population size.

Statistical Analysis

We first conducted principal factor analysis to determine whether the three parent–communication scale items described earlier form a single scale (Carmines & Zeller, 1979). We examined eigenvalues for indentified factors, loadings for retained factors, and interitem correlations to verify the existence of a single-factor solution. Scale reliability was then assessed by estimating Cronbach’s alpha coefficient for the scale.

Our analyses of the impact of PSUNC involve two primary sets of multivariable regressions. First, we used ordinary least squares regression models to estimate self-reported frequency of exposure to PSUNC ads as function of media market PSUNC GRPs. These models were conducted to establish validity of our self-reported awareness variable (i.e., we expect market-level GRPs to be associated with individual-level frequency of exposure). We estimated this model for parents overall and separately among mothers and fathers.

We then estimated two primary ordinary least squares regression models of the parent–child communication scale outcome as a function of exposure to PSUNC TV and radio ads. Because there are two different response scales (1 to 4 and 1 to 2) among the constituent variables of the parent–child communication scale, we standardized each item to z scores in the scale for purposes of multivariate analyses. This gives appropriate equal weight to each item in the overall scale value regardless of differences in the minimum/maximum values across the items.

The first model specification used total PSUNC television GRPs and total PSUNC radio GRPs as the two primary independent variables for PSUNC exposure. Because GRPs are measured as a continuous variable with a large range of values, GRPs were scaled such that model coefficients represent the change in the parent–child communication scale given an increment increase of 100 GRPs. Furthermore, we explored appropriate functional forms for GRPs given that ever-increasing levels of exposure likely do not generate infinite linear increases (or decreases) in parent–child communication. That is, there should be diminishing campaign effects at higher levels of exposure. Specification tests (Linktest in Stata 11.0) of our models indicated that the square root of PSUNC GRPs fit the data better across all models we estimated. Postestimation analyses of comparisons of model fit statistics (using Akaike’s information criterion and Bayesian information criterion) also indicated that the square root of GRPs was warranted over a simpler linear function (Cohen, Cohen, West, & Aiken, 2003). We therefore used the square root functional form for PSUNC GRPs in these models.

The second specification used our measure of self-reported frequency of awareness of PSUNC TV ads and self-reported frequency of awareness of the “S-Word” radio ad as the two independent variables for PSUNC exposure. Each of these models was estimated separately for mothers and fathers, given that so little is known about father–child communication about sexual activity. Each model was weighted using the KN panel weights and included control variables for each of the confounders described earlier. All analyses were conducted using robust variance estimates in Stata Version 11 statistical software to appropriately account for the survey weights and estimate standard errors (StataCorp, 2009).

Results

Sample Characteristics

Parent demographic characteristics are summarized in Table 1. Most parents were between 35 and 54 years old, with a slightly greater percentage of parents having male children between the ages of 10 and 14. The sample contained a slightly higher proportion of White and higher educated parents relative to the U.S. population, particularly among mothers. Estimates of White, African American, and Other races shown in Table 1 are based on non-Hispanic origin. The weighted distributions of race and education are not significantly different from CPS benchmarks.

Table 1.

Unweighted and Weighted Sample Characteristics

	Overall (N = 1,804)		Mothers (N = 1,253)		Fathers (N = 551)
Demographic Variable	Unweighted (%)	Weighted (%)	Unweighted (%)	Weighted (%)	Unweighted (%)	Weighted (%)
Age (years)
18-34	10.5	22.2	12.0	24.5	7.1	19.0
35-44	48.6	49.4	50.8	51.6	43.4	46.4
45-54	37.1	24.8	35.8	22.7	40.1	27.8
55+	3.8	3.6	1.4	1.2	9.4	6.8
Child gender
Boy	52.3	52.5	52.8	51.3	51.4	54.1
Girl	47.7	47.5	47.2	48.7	48.6	45.9
Race/ethnicity
White^a	76.5	66.8	76.2	63.6	76.9	71.6
African American^a	7.9	11.4	9.2	13.6	5.3	8.3
Hispanic	8.1	15.6	8.4	17.3	7.4	12.9
Other^a	7.5	6.2	6.2	5.5	10.4	7.2
Education
<High school	2.9	10.5	2.6	9.2	3.8	12.3
High school or GED	12.5	27.7	12.3	26.8	13.1	29.0
Some college	37.1	31.8	38.9	35.3	32.9	26.7
>Bachelor’s degree	47.5	30.0	46.2	28.7	50.2	32.0

Non-Hispanic.

Self-Reported Awareness and PSUNC GRPs

Overall, 59.4% of parents in our sample reported seeing or hearing at least one of the PSUNC TV or radio ads we assessed in our survey. Self-reported frequency of awareness of PSUNC also appears to be associated with the media market “dose” of the campaign in our multivariable logistic regressions for self-reported awareness of PSUNC (data not shown). Higher total PSUNC GRPs at the media market level are associated with increased frequency of awareness of any PSUNC ad (b = 0.14, p < .001). This relationship was present among both mothers (b = 0.11, p < .001) and fathers (b = 0.38, p < .05). We did not find a significant association between radio GRPs and frequency of awareness of the “S-Word” radio ad among parents.

Reliability of the Parent–Child Communication Scale

Principal factor analysis of the three parent–child communication variables described earlier retained only one factor, with loadings ranging from .71 to .76 across the three parent–child communication variables (Table 2). Interitem correlation was high, the overall parent–child communication scale had a mean value of 6.99 (SD = 2.14, minimum = 3, maximum = 10), and scale reliability was acceptable (Cronbach’s α =.82). These results suggest that our three-item scale adequately reflects parent–child communication and exhibits sufficient interitem correlation.

Table 2.

Descriptive Statistics and Reliability Analysis of the Parent–Child Communication Scale

Item	Mean	SD	Minimum	Maximum	Factor Loading	Interitem Correlation	α
Initiates conversation with child about sexual activity or waiting to have sex	2.73	0.93	1 (never)	4 (often)	.76	.59	.74
Talks to child about whether to wait to be sexually active until married	2.56	1.08	1 (not at all)	4 (a great deal)	.76	.58	.73
Has asked or recommended that child wait to have sex	1.70	0.46	1 (no)	2 (yes)	.71	.64	.78
Parent–child communication scale	6.99	2.14	3	10	—	.66	.82

Association With Parent–Child Communication Outcomes

Multivariate logistic regressions of the relationship between PSUNC GRPs and parent–child communication outcomes are shown in Table 3. Among mothers, higher PSUNC television GRPs are associated with increased frequency of parent–child communication (b = 0.21, p < .001). Total radio GRPs at the market level were also associated with increased frequency of parent–child communication among mothers (b = 0.31, p < .05). We did not find any significant relationships between PSUNC GRPs and parent–child communication among fathers.

Table 3.

Ordinary Least Squares Coefficients Showing Association Between PSUNC Market-Level GRPs and the Parent–Child Communication Scale

	Outcome Variable = Parent–Child Communication Scale
Independent Variable	Mothers	Fathers
Total PSUNC television GRPs	.21** [0.08, 0.34]	−.13 [−0.33, 0.07]
Total S-Word radio GRPs	.31* [0.04, 0.57]	.12 [−0.31, 0.54]
N	1,253	551

Note. PSUNC = Parents Speak Up National Campaign; GRP = gross rating point; MSA = metropolitan statistical area. All models control for child gender; child age; marital status; highest educational attainment; race/ethnicity; parent age; employment status; family structure; whether child has computer, cable TV, or Internet in his or her bedroom; MSA status, median market-level household income; average market-level college graduation rates; media market population size; parent religiosity; and a scale for parental involvement. Figures in brackets are 95% confidence intervals.

p < .05, **p < .01, ***p < .001.

To illustrate the magnitude the effect of television GRPs, we plotted the predicted model results for the parent–child communication scale against total PSUNC television GRPs, excluding the aforementioned outlier value of 12,367 GRPs (Figure 1). These plots were done separately for mothers and fathers. Based on this plot, the average increase in the parent–child communication scale value per unit increase of 100 GRPs was 0.036 or 0.02 of a standard deviation of the scale.

Figure 1.

Model-predicted relationship between predicted parent–child communication scale and PSUNC GRPs

Self-reported frequency of exposure to PSUNC TV ads was also strongly associated with parent–child communication (Table 4). Increased frequency of exposure to PSUNC TV ads was associated with increased parent–child communication among mothers (b = 0.43, p < .001) and fathers (b = 0.59, p < .01), suggesting the existence of a dose–response relationship between self-reported campaign exposure and parent–child communication among both female and male parents. There were no significant associations between self-reported frequency of exposure to the “S-Word” radio ad and parent–child communication among parents of either gender.

Table 4.

Ordinary Least Squares Coefficients Showing Association Between Self-Reported Frequency of Exposure to PSUNC Ads and the Parent–Child Communication Scale

	Outcome Variable = Parent–Child Communication Scale
Independent Variable	Mothers	Fathers
Self-reported frequency of exposure to PSUNC TV ads	.43*** [0.22, 0.64]	.59** [0.19, 1.00]
Self-reported frequency of exposure to PSUNC radio ads	.10 [−0.11, 0.30]	−.19 [−0.56, 0.17]
N	1,253	551

Note. PSUNC = Parents Speak Up National Campaign; MSA = metropolitan statistical area. All models control for child gender; child age; marital status; highest educational attainment; race/ethnicity; parent age; employment status; family structure; whether child has computer, cable TV, or Internet in his or her bedroom; MSA status; median market-level household income; average market-level college graduation rates; media market population size; parent religiosity; and a scale for parental involvement. Figures in brackets are 95% confidence intervals.

p < .05. **p < .01. ***p < .001.

Discussion

This study adds critical field-based data to the PSUNC evidence base, which to date has been limited to experimental efficacy trials conducted under controlled settings (e.g., Davis et al., 2010; Evans et al., 2009; Evans et al., in press). Until now, the actual reach of the campaign among parents in the real world and its potential impact on parent–child communication outcomes have been unknown. Our study shows that 59.4% of parents in the United States were exposed to at least one PSUNC ad at the time of our survey. However, although the overall proportion of parents who were exposed was relatively high, this may not necessarily translate into a great deal of exposure. Based on the overall level of GRPs that were delivered by the campaign, we estimate that the average person in the target audience received approximately 22 total ad exposures or 1 exposure per month during the 2008-2009 time frame of the campaign. Although this is a modest level of exposure, our results suggest that it was sufficient enough to achieve some measurable effects on parent–child communication.

We further found that both market-level and self-reported measures of exposure to PSUNC were associated with increased frequency of parent–child communication. Results from both sets of multivariate models suggest the existence of a dose–response relationship between campaign exposure and frequency of parent–child communication. These data are thus consistent with the overall findings from previous efficacy trials, lending further confidence to conclusions that PSUNC was likely an effective intervention for increasing parent–child communication about sex.

Our findings are also consistent with previous experimental results (e.g., Davis et al., 2010; Evans et al., 2009) in terms of patterns in campaign effects by parent gender. That is, we find mixed if not weaker evidence of associations between PSUNC exposure and parent–child communication among fathers. Whereas self-reported exposure to PSUNC was associated with parent–child communication among fathers, exogenous market-level GRPs were not. Davis et al. (2010) note that these seemingly weaker effects among fathers may be partially explained by the gender-specific focus of some PSUNC ads as well as potential gender differences in overall attentiveness to such messages. For example, PSUNC messages such as the “Muffinhead” ad are mother-centric. We also found that 63.9% of mothers indicated awareness of any PSUNC ad compared with 52.9% awareness among fathers (p < .05), suggesting that mothers may indeed be more attentive to PSUNC messages. This may be an important determinant of gender-specific campaign effects, given that mothers are also generally more involved in the sexual health education of their children compared with fathers (Fisher, 1990; Meschke et al., 2002; Raffaelli et al., 1998). Taken together, this evidence suggests that fathers may represent an important focal group where additional gains can be made for increasing parent–child communication. Future similar campaigns may need to consider gender-specific message tailoring to enhance effects among fathers.

Results from this study further suggest that the associations between campaign exposure and parent–child communication are driven by PSUNC TV ads. Self-reported exposure to radio ads was the only measure of radio advertising that was found to be associated with increased parent–child communication, and this relationship was only apparent among mothers. No other associations between exposure to radio ads and parent–child communication were observed.

It is instructive to reflect on the plausibility of these findings in the context of evaluation data for other prior national health marketing campaigns as well as in the context of the aforementioned experimental studies on PSUNC. Based on the available PSUNC GRP data and our findings on the proportion of parents who were exposed to at least one ad, we estimate the campaign purchased enough advertising time to achieve approximately 22 total exposures per person in the target audience, or 1 exposure per month during the 2008-2009 time frame of the paid media campaign. In comparison, previous national campaigns such as the national “truth” smoking prevention campaign achieved more than 4 exposures per audience member per month, based on available “truth” campaign GRP data and estimates on that campaign’s reach (Farrelly et al., 2002; Farrelly et al., 2005). Other national campaigns such as the National Anti-Drug Media Campaign invested enough resources to achieve 2.5 ad exposures per week or 10 exposures per month among youth (Hornik, Jacobsohn, Orwin, Piesse, & Kalton, 2008). Hence, PSUNC achieved comparatively lower exposure than some other previous national campaigns. However, in terms of the number of exposures required for campaign effects, our findings are consistent with previous experimental studies of PSUNC (e.g., Davis et al., 2010; Evans et al., 2009) that demonstrated message effects after approximately 5 exposures to PSUNC TV ads over a 6-month period.

Our study is limited by a few factors. First, it should be noted that although we find a significant relationship between television GRPs and parent–child communication among mothers, the overall magnitude of this effect appears to be relatively small. The relationship between television GRPs and parent–child communication shown in Figure 1 suggests that the average increase in the parent–child communication scale per unit increase of 100 GRPs is approximately 0.036 or 0.02 of a standard deviation of the outcome scale. This represents a fairly modest effect size for GRPs. However, it is also worth noting that this pattern of effects for GRPs compared with effects for self-reported awareness is consistent with a number of other similar media evaluation studies (e.g., Davis et al., 2007) that have shown estimated effects for self-reported measures of exposure to be generally larger in magnitude than the effects of GRPs.

Second, our analyses of the relationship between self-reported PSUNC exposure and parent–child communication may be susceptible to selective attention bias. This bias can manifest in the form of overstated campaign effects if parents who have preexisting elevated levels of parent–child communication are more likely to recall the campaign with higher frequency. However, our analysis of the relationship between market-level measures of PSUNC GRPs and self-reported frequency of awareness suggests that selective attention bias is not a concern. Unlike self-reported awareness, market-level GRPs are exogenous to individual characteristics and should thus be uncorrelated with preexisting levels of parent–child communication. Furthermore, we found that market-level GRPs are predictive of individual-level awareness and that both types of exposure measures have similar associations with parent–child communication. This suggests that our self-reported measure of awareness is a valid measure of PSUNC exposure and that findings derived from both types of exposure measures can be effectively triangulated.

A third limitation to our study is that although the KN panel is designed to be nationally representative, our sample contained higher proportions of White and higher educated parents than would be expected in the U.S. population as a whole. This discrepancy was less prevalent among fathers. Thus, our results may not perfectly represent the broader population of parents in the United States. However, the weighted distributions of race and education were very similar to CPS benchmarks. We applied sampling weights to all analyses reported in this study to better reflect national estimates of parent awareness of PSUNC and the associations between exposure to PSUNC and parent–child communication.

A final limitation is that as with any observational study, it is challenging to completely rule out alternative explanations that may explain our findings. For example, despite our attempts to control for market-level characteristics that may be correlated with the campaign’s delivery, the market-level confounders we have controlled for may be imperfectly measured or there may be other unmeasured characteristics that are not accounted for.

In summary, this study offers the first field-based data on the real-world reach and effectiveness of PSUNC among parents nationwide. Our data support earlier findings from experimental efficacy trials of PSUNC involving both parents and children, showing that the campaign is associated with greater rates of parent–child communication, particularly among mothers. Thus, the PSUNC message design and advertising strategy may serve as a model for future similar media interventions. However, questions remain about the impact of media messages on parent–child communication among fathers. Consistent with previous PSUNC efficacy trials, we find mixed and weaker evidence of associations between PSUNC and parent–child communication among fathers. Future research and campaign efforts should explore the development and testing of new messages targeted specifically at fathers. For example, our study did not explore differences in PSUNC effects among fathers by child gender. It is possible that fathers may be less in favor of making recommendations to their child about sexual behavior compared with mothers, particularly when talking to a son. These nuances of parent–child communication with respect to father–child gender pairings should be more fully explored in future research and development for messages targeted to fathers.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article:

This study was supported by the U.S. Department of Health and Human Services, Office of Population Affairs.

References

Aspy

C. B.

Vesely

S. K.

Oman

R. F.

Rodine

Marshall

McLeroy

(2007). Parental communication and youth sexual behaviour. Journal of Adolescence, 30, 449-466.

Bandura

(2004). Health promotion by social cognitive means. Health Education & Behavior, 31, 143-164.

Carmines

E. G.

Zeller

R. A.

(1979). Reliability and validity assessment. Beverly Hills, CA: Sage.

Cohen

West

S. G.

Aiken

L. S.

(2003). Applied multiple regeression/correlation analysis for the behavioral sciences (3rd ed.). Hillsdale, NJ: Erlbaum.

Centers for Disease Control and Prevention. (2008). Youth risk behavior surveillance—United States, 2007. Morbidity & Mortality Weekly Report, 57(4), 1-131.

Davis

K. C.

Blitstein

J. L.

Evans

W. D.

Kamyab

(2010). Impact of a parent-child sexual communication campaign: results from a controlled efficacy trial of parents. Reproductive Health. 7(17). Retrieved from http://www.reproductive-health-journal.com/content/7/1/17

Davis

K. C.

Nonnemaker

J. M.

Farrelly

M. C.

(2007). Association between national smoking prevention campaigns and perceived smoking prevalence among youth in the United States. Journal of Adolescent Health, 41, 430-436.

Dilorio

Kelley

Hockenberry-Eaton

(1999). Communication about sexual issues: mothers, fathers, and friends. Journal of Adolescent Health, 24, 181-189.

DuRant

R. H.

Wolfson

LaFrance

Balkrishnan

Altman

(2008). An evaluation of a mass media campaign to encourage parents of adolescents to talk to their children about sex. Journal of Adolescent Health, 38, 298e.1-298e.9.

10.

Evans

W. D.

Blitstein

J. L.

Davis

K. C.

(2011). Social cognitive mediators of parent-child sexual communication. American Journal of Health Behavior, 35, 428-437.

11.

Evans

W. D.

Davis

K. C.

Ashley

O. S.

Blitstein

Koo

Zhang

(2009). Efficacy of abstinence promotion media messages: findings from an online randomized trial. Journal of Adolescent Health, 45, 409-416.

12.

Evans

W. D.

Davis

K. C.

Silber-Ashley

Snider

(in press). Effects of media messages on parent-child sexual communication. Journal of Health Communication.

13.

Evans

W. D.

Davis

K. C.

Zhang

(2008). Health communication and marketing research with new media: Case study of the Parents Speak Up National Campaign evaluation. Cases in Public Health Communication & Marketing, 2, 140-158.

14.

Farrelly

M. C.

Davis

K. C.

Haviland

M. L.

Messeri

Healton

C. G.

(2005). Evidence of a dose-response relationship between “truth” antismoking ads and youth smoking prevalence. American Journal of Public Health, 95, 425-431.

15.

Farrelly

M. C.

Healton

C. G.

Davis

K. C.

Messeri

Hersey

J. C.

Haviland

M. L.

(2002). Getting to the “truth”: Evaluating national tobacco countermarketing campaigns. American Journal of Public Health, 92, 901-907.

16.

Farrelly

M. C.

Nonnemaker

J. M.

Davis

K. C.

Hussin

(2009). The influence of the National Truth Campaign on smoking initiation. American Journal of Preventive Medicine, 36, 379-384.

17.

Fisher

T. D.

(1990). Characteristics of mothers and fathers who talk to their adolescent children about sexuality. Journal of Psychology & Human Sexuality, 3(2), 53-70.

18.

Furstenberg

F. F.

Jr. Moore

K. A.

Peterson

J. L.

(1985). Sex education and sexual experience among adolescents. American Journal of Public Health, 75, 1331-1332.

19.

Hornik

R. C.

Jacobsohn

Orwin

Piesse

Kalton

(2008). Effects of the national anti-drug media campaign on youths. American Journal of Public Health, 98(12), 2229-2236.

20.

Jemmott

J. B.

Jemmott

L. S.

Fong

G. T.

(2010). Efficacy of a theory-based abstinence-only intervention over 24 months: A randomized controlled trial with young adolescents. Archives of Pediatric & Adolescent Medicine, 164, 152-159.

21.

Kim

Rector

R. E.

(2008). Abstinence education: Assessing the evidence (Heritage Foundation Report No. 2126). Retrieved from http://www.heritage.org/Research/Welfare/bg2126.cfm

22.

Lehr

S. T.

Dilorio

Dudley

W. N.

Lipana

J. A.

(2000). The relationship between parent- adolescent communication and safer sex behaviors in college students. Journal of Family Nursing, 6, 180-197.

23.

Martin

J. A.

Hamilton

B. E.

Sutton

P. D.

Ventura

S. J.

Menacker

Kirmeyer

Mathews

T. J.

(2006). Births: Final data for 2006. National Vital Statistics Report 2009, 57(7). Retrieved from http://www.cdc.gov/nchs/data/nvsr/nvsr57/nvsr57_07.pdf

24.

Meschke

L. L.

Bartholomae

Zentall

S. R.

(2002). Adolescent sexuality and parent-adolescent processes: Promoting health teen choices. Journal of Adolescent Health, 31, 264-269.

25.

National Commission on Children. (1999). Survey of parents and children, 1990. Ann Arbor, MI: Inter-University Consortium for Political and Social Research.

26.

Pick

Palos

P. A.

(1995). Impact of the family on the sex lives of adolescents. Adolescence, 30, 667-675.

27.

Raffaelli

Bogenschneider

Flood

M. F.

(1998). Parent-teen communication about sexual topics. Journal of Family Issues, 19, 315-333.

28.

Sandfort

Orr

Hirsch

Santelli

(2008). Long-term health correlates of timing of sexual debut: Results from a National U.S. Study. American Journal of Public Health, 98, 155-161.

29.

Scher

L. S.

Maynard

R. A

Stagner

(2006). Interventions intended to reduce pregnancy-related outcomes among teenagers. Campbell Systematic Reviews. Retrieved from http://www.campbellcollaboration.org/lib/download/103/

30.

Southwell

B. G.

Barmada

C. H.

Hornik

R. C.

Maklan

D. M.

(2002). Can we measure encoded exposure? Validation from a national campaign. Journal of Health Communication, 7, 445-453.

31.

StataCorp. (2009). Stata Statistical Software: Release 11 [Computer software]. College Station, TX: Author.

32.

Terry-McElrath

Y. M.

Wakefield

M. A.

Emery

Saffer

Szczypka

O’Malley

P. M.

. . . Flay

B. R.

(2007). State anti-tobacco advertising and smoking outcomes by gender and race/ethnicity. Ethnicity & Health, 12, 339-362.

33.

Trenholm

Devaney

Forston

Quay

Wheeler

Clark

(2007). Impacts of four Title V, Section 510 abstinence education programs (Final report). Retrieved from http://aspe.hhs.gov/hsp/abstinence07/

34.

Weed

S. E.

Eriksen

I. H.

Lewis

Grant

G. E.

Wibberly

K. H.

(2008). An abstinence program’s impact on cognitive mediators and sexual initiation. American Journal of Health Behavior, 31, 60-73.

Effectiveness of a National Media Campaign to Promote Parent–Child Communication About Sex

Abstract

Keywords

Method

Survey Data, Sample, and Procedures

Measures

Parent–child communication

Parent awareness of PSUNC ads

Media market–level exposure to PSUNC

Potential confounders

Statistical Analysis

Results

Sample Characteristics

Self-Reported Awareness and PSUNC GRPs

Reliability of the Parent–Child Communication Scale

Association With Parent–Child Communication Outcomes

Discussion

Footnotes

Declaration of Conflicting Interests

Funding

References