14 September 2005, 01:29
Social Marketing in Rostov: Student's Rights
Theodore P. Gerber and Sarah E. Mendelson
This paper addresses the results of the June 2004 regional survey assessing the impact of the student's rights campaign on residents of Novocherkassk and Rostov City, both in the Rostov oblast. While there are other measures of the impact, we address here the quantitative results from the resurvey in Rostov and the control group Stavropol. Novocherkassk did an impressive job reaching its public while Rostov City did a less impressive job. The message tested here, however, was not always clear to residents. We do see evidence of the desired effect in Novocherkassk. Results should be taken with some caution since the sample size is small. That said, we find Novocherkassk activists mounted a campaign that does show some shift in thinking about the rights of young people. These findings, along with the Ryazan activists' campaign, suggest that social marketing holds promise even in regions with restricted media and political space.
Exposure and Reaction to the "Sovmesti..." campaign (Table 1)
Finding: The Novocherkassk campaign reached 30% of its residents, and of these, about a third understood the campaign to be related to human rights. The Rostov City campaign reached few residents.
Activists used two different slogans in the campaigns implemented in Novocherkassk and Rostov city. However, we were only given one slogan, "sovmesti svoi prava s zhizn'iu" by the activists to test. We therefore report only on recognition of this slogan, and have no data to measure exposure and reaction to the "my s pravami na ty, a ty?" campaign.
Our results suggest wide exposure to the "sovmesti" materials in Novocherkassk: according to our best estimate, one third of the city's population recalls seeing the materials at least once (Panel C). By contrast, exposure was quite low in Rostov city: only 7% recall seeing the materials. On the one hand, this is not surprising, since Rostov is a much larger city than Novocherkassk. Yet, when we compare exposure in Perm city to the "Ya chelovek" campaign materials, the Rostov city campaign must be judged disappointing. Also, we note that 7% of respondents in the rest of Rostov oblast - where there was no campaign to speak of -- say they saw materials. This is more reason for disappointment in the results of the Rostov city campaign. Notably, only a handful of the respondents in Stavropol oblast said they had seen the materials. In sum, the data suggest that the campaign managed to reach a broad audience in Novocherkassk but not in Rostov city.
But if proportionately more residents in Novocherkassk saw the "Sovmesti" materials, they were much less likely to correctly identify the theme of the campaign: only 29% said the materials deal with human rights (Panel D). The most common response was uncertainty about the subject of the materials, chosen by 45% of respondents. Another 29% thought the posters and calendars were part of an election campaign. This is by far the weakest result in terms of interpretation of the campaign theme that was obtained in any of the project regions. Rostov city fared much better in this regard: 84% correctly identified the campaign theme.
Reactions to the campaign in Novocherkassk also suggest the message did not get through to very many. Few respondents - only 13% of those who saw the materials - agreed that it made them think about violence in schools (Panel E). More - 43% of those who saw the materials - agreed that it made them think about the right to education, but this is still less than a majority (Panel F). In Rostov city, the reaction was, if anything, even more disappointing. While 24% agreed the materials made them think about violence in schools, a mere 5% reported that they evoked concern about the right to high-quality education.
Overall, the campaign materials do not appear to have elicited the desired reaction for most people who saw them. Based on these results we must conclude that the clarity and the effectiveness of the campaign materials could have been improved.
Who saw the "Sovmesti" campaign and who reacted as hoped? (Table 2)
Finding: Young people are slightly more likely to report having seen the campaign.
We ran logistic regressions to analyze whether exposure to the campaign varied by demographic characteristics in Novocherkassk and Rostov city. These results must be interpreted carefully because of the relationship between cohort and education. We found no differences between college educated and high school educated respondents, so we specify the effect of education with a single dummy variable denoting less than high school education. Of course, many of the respondents with less than high school education are very young respondents who still are in high school. So, when we find, for example, that those under 30 are less likely to report seeing materials than those over 30 (Panel A), this must be weighed against the finding that those with less than high school are more likely to have seen the materials. The size of the "less than high school" effect exceeds the size of the negative "under 30" effect, so overall we conclude that high-school age respondents are slightly more likely to have seen the materials, while young adults (in their twenties) are less likely to have seen them. Exposure did not vary by gender. Rostov city residents were significantly less likely to see the materials, controlling for the demographic variables.
We see a similar pattern of effects on the probability of correctly identifying the themes of the materials, conditional on having seen them (Panel B). Consistent with the bivariate findings in table 1, Rostov city residents who saw the campaign were more likely to identify the theme of the campaign as related to human rights. Those under 30 were also more likely to correctly identify the theme, but only if they had graduated from high school (because the -2.827 effect for less than high school offsets the 3.074 effect for under 30). In essence, young adults who had finished high school at the time of the survey were more likely to correctly identify the theme, while youths still in high school were neither more nor less likely than those over 30 to do so.
The message was also more likely to evoke the desired reactions for young adults who had graduated from high school: the significant positive coefficients of 2.507 for thinking about abuses in schools (Panel C) and 1.716 for thinking about the right to high quality education (Panel D) imply, respectively, 12.3 and 5.6 times higher odds of the desired reactions for those under 30 with high school than for those over 30. Here too, it is noteworthy that the "under 30" effects are offset for those still in high school. (These results also imply that those over 30 with less than high school are especially unmoved by the campaign and unlikely to correctly identify its theme.) Female respondents who saw the materials were less likely to react with concern about violence in schools.
Very few respondents visited the centers so we cannot place too much stock in the model predicting visits (Panel E). All who visited were under 30, so for technical reasons we cannot enter an "under 30" dummy in the model. Instead, we enter a dummy variable for "under 20." Although it is not quite significant, it is very close. Given the small sample size and low variance on the dependent variable, we can reasonably conclude that youths under 20 were indeed more likely to visit the receptions than those over 20. No other effects approach significance. Also, we could not enter the Rostov city dummy variable, but since no respondents in Rostov city said they visited the centers, we can conclude that the probability of doing so was much lower in Rostov city.
Dynamics of questions from the Rostov group specifically relating to children (Table 3)
Finding: In Novocherkassk, for those who understood the campaign, we find the desired effect of increasing concern about laws on children's rights, and to a lesser extent, concern about abuse of children.
Given the small number of Novocherkassk respondents in the 2003 round of the survey (22), we must interpret the aggregate in that town with great caution. Nonetheless, it is encouraging to see that concerns about observation of a law protecting children's rights increased substantially from 2003 to 2004: while 68% believed that a law exists and that it is observed in 2003, that number fell to 14% in the 2004 survey (Panel A). This trend is consistent with a scenario whereby the "Sovmesti" and the "My s pravami" campaigns raised the general level of concerns about children's rights in the Novocherkassk community. In light of the small 2003 sample size, however, we must refrain from making a strong claim on this count. But it is also consistent with such a scenario that in 2004 (where the sample size is larger) skepticism about the observation of a law protecting children's rights is higher in Novocherkassk than in any of the other three locations (combining those who say the law is not observed and those who do not know if it is observed.)
In contrast, the picture is muddier on this score in Rostov city: while fewer respondents think a law exists, more think it does and is not observed. The results from Rostov city are quite similar with respect to a law protecting the rights of students (Panel B): fewer think it exists, but more think it does and is not observed.
In Novocherkassk, we see a sharp growth in the number who say that no such law exists and a steep drop in the number who say that such a law is observed. We suspect these dramatic swings have more to do with the small 2003 sample size in Novocherkassk than with the campaign.
Although the campaign in Rostov oblast is not likely to have increased the frequency with which respondents hear about abuses in schools - either in general or from their children - we nonetheless examine the dynamics of these variables (Panels C-E). It is not impossible, if unlikely, that the campaign could make people more sensitive to such reports, more likely to "tune in" to them when they hear them. In fact, we do see an increase in reports of hearing about abuses in schools in Novocherkassk, though not hearing about them directly from one's children. Again, we hesitate to reach any conclusions based on such a small 2003 sample, but this pattern is consistent with a type of "percolating" effect of the campaign: it may have subtly increased awareness of and concern about abuses of students in Novocherkassk. No such pattern at all is evident in Rostov city: if anything, fewer respondents hear about abuses in schools there in 2004.
Assessing campaign impact at the individual level (Table 4)
Finding: Those that were exposed to the campaign and understood its meaning in Novochkassk express increased concern about children's rights. These findings are strongest for those under 30.
We estimate the same type of multivariate models for outcomes that might plausibly be affected by the campaign as we do for the other campaigns. We begin with models containing only measures of demographic traits, location, and year. We then enter a single variable denoting both exposure to and correct identification of the theme of the "Sovmesti" campaign materials. In addition, we test the possibility that the impact of campaign varied by cohort using the appropriate interaction term ("sovmesti"*under 30). We only report the results for the interaction term when it is statistically significant though; otherwise, such results are redundant. The same caveats regarding the relationship of the campaign to the "concern about causalities" from the war in Chechnya discussed in the Ryazan report apply here as well: we cannot claim definitively that exposure to the campaign "caused" people to have a particular view even when we identify a statistically significant association at the individual level. However, such associations bolster the case that the campaign had some impact, because they are consistent with such a scenario.
We begin by examining three variables from the "fear" battery that might plausibly be influenced by the campaign. The content of the campaign could be expected to increase the salience of concerns about the future of children in Russia (Panel A), the loss of human rights (Panel B), and access to education (Panel C). In fact, our logistic regression analysis confirms that those exposed to the campaign who correctly identified its theme are significantly more likely to express concern about the future of children: the coefficient of 1.081 in Model 2, Panel A, implies their odds of doing so are nearly 3 times higher than the odds for those who did not see the campaign materials, controlling for age, gender, education, and place of residence. Also, once we enter campaign exposure in the model, the increase in concern about children's problems in 2004 in Rostov oblast (denoted by the "2004 survey" coefficient) falls from .272 (Model 1) to .203 (Model 2) and is no longer statistically significant. This is one of the most encouraging results regarding the impact of the campaign at the individual level: the data suggest that those who saw the campaign and grasped its subject are more concerned about the future of children, and campaign exposure helps account for the increase in this concern in Rostov oblast from 2003 to 2004.
Model 3 sheds additional light on the nature of the association between campaign exposure and concern about the future of children in Russia. The significant, negative coefficient on the interaction between "Sovmesti" exposure and "under 30" implies that the effect of the campaign only pertains to those over 30: if anything, those under 30 who saw and correctly interpreted the materials are somewhat less likely to voice concern about children's future than those under 30 who never saw the campaign (since adding the main and interaction effects produces a sum of -.367). Perhaps contrary to the aims of the activists, the campaign appears to have been most effective at reaching those over 30.
We do not observe any associations at all between campaign exposure and concerns about the loss of human rights in Russia or access to education: in neither case is the relevant coefficient statistically significant, and adding the interaction with under 30 does nothing to change the conclusion.
Next, we test for associations between campaign exposure and views on whether there exists a law protecting children and whether it is observed. The dependent variable is complex, so we break it into two separate variables: the probability of saying that a law exists at all (Panel D) and the probability of saying that it exists and is not enforced (Panel E). We presume the goal of the campaign was to increase both probabilities - i.e., create awareness of legal protections of children's rights and concern that existing protections are not enforced.
In fact, we find that, if anything, exposure to the campaign is associated with decreased probability of saying that a law protecting children's rights exists (Model 2, Panel D). That is what the negative coefficient on "Saw/understood 'sovmesti'" implies. In hindsight, this may actually be a positive result of the campaign: perhaps the campaign created the impression that there are no legal protections of children's rights. There is no evidence that the campaign increased the probability of saying that an existing protection of children's rights is not enforced: that coefficient, while positive, does not even approach statistical significance (Panel E.)
We conducted parallel analyses to those in Panels D and E regarding a law protecting the rights of students. We found the same results. First, exposure to the campaign is associated with decreased probability of saying that such a law exists (Panel F, Model 2). Second, exposure has no association at all with the probability of saying that an existing law is not enforced (Panel G).
Next, we examined for associations between campaign exposure and awareness of abuses in schools - either in general or via one's children using ordinal logit models (since reports of abuses were measured using an ordinal scale). We find no effect at all of campaign exposure on hearing about abuses in general; adding the exposure variable in Model 2, Panel H does nothing to improve the model fit and yields a highly non-significant (and near-zero) parameter estimate.
We do find some effects when we look at reports of abuses from the respondents' own children. First, respondents over 30 who saw the campaign and have children indeed report hearing more about abuses of their children in school. This is implied by the positive and significant main effect of "sovmesti" in Model 3, Panel I. Perhaps the campaign inspired some parents to ask their children about what takes place in their schools. If so, that would be a very positive result. On the other hand, the negative interaction effect in Model 3 implies that this association holds only for parents who are over 30: those under 30 who saw the campaign are, if anything, slightly less likely to hear of abuses of their own children. Also, we found a negative association between campaign exposure and hearing about abuses witnessed (not experienced) by the respondents children: the -1.79 coefficient in Model 2, Table I is statistically significant. Thus, while there is some indication of the hoped-for individual-level effect, that must be weighed against other null or counter-intuitive findings.
Finally, we examined whether the individual level campaign exposure is associated with views on the priority of specific rights. In fact, we saw a similar pattern with respect to three rights (and no effect with respect to the others - we do not show these null results.) In the case of freedom of expression, freedom from torture, and freedom from arbitrary arrest, we find that campaign exposure is positively associated with support for rights being a top priority, but only among those over 30. In Panels J, K, and L, the main effect of campaign exposure is positive and significant, but it is offset for those under 30 by negative and significant interaction effects. (In fact, neither effect is quite significant in Panel K - freedom from torture - but they are close enough, given the small variance in the key independent variables, as to merit some credibility). This is further evidence that the campaign's impact may have been strongest among those over 30, despite the hopes of the organizers to influence those under 30 in particular.
Thus, overall we get a mixed picture regarding the impact of the campaign at the individual level. We do not see overwhelming evidence of a strong impact of the type hoped for, but given the measures available and the limitations of the data, we are encouraged to have found some evidence consistent with a degree of impact. Most surprising, we have several cases where the impact of the campaign appears to be most pronounced for those over 30. We are surprised by this, given the campaign's emphasis on children in its form, substance, and target audience.
Aggregate dynamics of other variables (Tables 5, 6, 7)
Finding: While the impact on general views of human rights is not great, we do find an increase in support of human rights NGOs in Novocherkassk.
Turning now to the other measures on the survey for which the campaign might have been relevant, we do not find any evidence of trends consistent with the type of impact hoped for. Here again, though, we must be cautious (particularly in regard to null findings) due to the small size of the 2003 sample from Novocherkassk.
The one positive finding is an increase in positive feelings toward abstract groups that defend human rights in Novocherkassk (Table 5, Panel A): the percentage viewing such groups in highly or somewhat positive terms grew from 68% to 83% between 2003 and 2004. This may well reflect the appeal of the campaign in Novocherkassk. In Rostov city, the only noteworthy change is in the opposite direction: an increase in the proportion who have never heard of such groups. On the other, positive views toward Memorial increased in Rostov city (Panel B). Knowledge of Memorial appears to have decreased in Novocherkassk, but again this could reflect the problematic sample size in 2003 (and in any case, we would not expect an improvement in views of Memorial in Novocherkassk, since the name (or brand) "Memorial" does not appear on campaign materials.)
It is difficult to discern any coherent pattern that could be related to the campaign in what people in Rostov and Stavropol identify as the main problems facing the country (Table 6). The only patterns common to the campaign sites are declining concerns about corruption and bribery and about weak government. We cannot think of a reason why these decreases would be connected to the campaign. Other trends are opposite in Novocherkassk and Rostov city (HIV, drug abuse, police abuse). We do not the sense that the campaign shaped peoples perceptions of the pressing issues in the country from these data.
At the aggregate level we discern at best weak evidence that the campaign affected views toward the priority of the nine human rights we asked about (Table 7). We do see some positive movement in Novocherkassk regarding freedom from arbitrary arrest, association, slavery, and, though not quite significant, slavery. But we also see a decrease in support for freedom from torture, and all these contrasts are based on comparisons with the very small 2003 sample. We see little change at all in Rostov city. The most dramatic changes are observed in Stavropol (increased support for all three civil liberties, coupled with decreased support for freedom from torture and arbitrary arrest).
 Taganrog activists were also originally part of this project but failed to campaign. Accordingly, we do not report results here.
 The Russian survey and focus group company Validata did test a sample of 197 young people in Novocherkassk who had been exposed to the campaign on recognition of this phrase with highly positive results. These are available upon request from Memorial.