Drug Abuse
THE INTERNATIONAL JOURNAL OF DRUG POLICY 1993 4 3
MESSAGE IN A TOILET
Stephen Mugford reports on evaluations of a novel narrowcasting strategy to communicate with hard to reach populations which is being used for conveying harm-reduction messages in the drugs and the AIDS fields
INTRODUCTION
For any social policy to succeed, one essential element is communication. Whatever messages one wishes to convey and whoever the target audience may be, it is vital to choose a strategy that gets appropriate and effective messages across to the audience. Furthermore, unless the social policy issue is very simple and its application extremely general, it is very likely that a wide variety of messages will he needed, each one targeted to a different subsection of the total population. Indeed, in a complex modern society, where there are many 'lifestyles', different messages will not merely be'appropriate' to different groups of people; rather, some messages that will be appropriate to one group will be very poorly received by others. The problem is increased if the topic concerns activities - such as drug use - that are subject to politico-moral controversy. It takes little imagination, for example, to see that a very explicit message about needle sharing which may be entirely' appropriate and useful to an injecting heroin user who lives in 'theCross' (King's Cross, Sydney's 'red light' district) may be distinctly offensive to a conservative, elderly spinster living in a small country town. By the same token, a message about limiting alcohol use which could have a positive impact on young women might be treated with derision by middle-aged, blue collar workers. Moreover, although some groups can be identifiedby geographical location, it is characteristic of modern societies that the bulk of the population lives in cities and that social diversity within those cities is high. This highlights the difficulty of relying upon broadcasting of messages within those areas. Any strategy that delivers a message to the full range of the population, be it electronic broadcasting such as TV adverts or passive advertising on billboards, is caught on the horns of a dilemma. Either the messages are reduced to a bland common denominator which offends nobody or messages are used which are useful to some and offensive to others. In a parliamentary democracy, if those citizens/voters who are offended by a message are more numerous and/or of higher status than those for whorn the message is (potentially) useful we may he fairly conident that the default option will he the bland ritessages. Thus conveying well-designed andeffective messages to numerically smaller or lower status groups can be a very challenging enterprise.
Summarising this, we might suggest that where policy communication is concerned, it is commonly the case that:
1 The overall population (conceived in broad demographic terms) exhibits diversity (by gender, by regionflocation, by age and by employment status). Thus a message suitable for one group may he unsuitable for another.
2. Target populations for key messages are subsets of that demographic band, and that, without careful design, many people might receive messages that are redundant or unwelcome in order to reach the few that need those messages.
3. Yet it is important to convey messales to people in a fashion and a language that is appealing to them, even if is not so obviously appealing to other groups.
It follows that good communication is precise communication - precise both in fitting the message to the audience and in getting the message to the audience. This article explores a technique for addressing the latter question, that of getting the message to specific audiences. It does not deal with the question of fitting the message to those audiences, which is a separate question, both analytically and practically. Rather, it deals with the following practical question:
... assuming that adequate messages based on appropriate research and testing have been designed, how might those messages be precisely delivered to the relevant audiences?
In order to explore this question, we need to look at two basic issues - what types of communication are available and what is the relationship between lifestyle and the use of space? 1 examine these in turn, suggesti rig first a brief typology of communication types and second a set of lifestyle/territory relations.
TYPOLOGY OF COMMUNICATION
Communication types can be arrayed along several dimensions. The first dimension is the degree to which communication is individual or general. A communication addressed (in one medium or another) to'Bill'is more individual than one addressed to Mr W Smith. That, in turn, is more personal than one addressed, over a loudspeaker, to'Customers', which itself is still slightly more personal than a TV ad.
A related, but analytically separable, dimension is whether the communication comes from a friend, an acquaintance or a stranger. Of course, these two intersect in practice, so that friends are highly likely to use individual communication (speech, telephones personal letters) whereas, at the other extreme, strangers are likely to rely more on general means such as loudhailers, press or TV - in broad terms, mass media corn- ~ munication. Indeed, exceptions to this latter association are 'out of place' in various ways. The young man who hires a plane to tow an aerial banner above a sporting event to propose to his loved one in the crowd is amusingly romantic, whereas the political party that uses the electoral role to send letters that start 'Dear Christopher...'is risking a very negative reaction (this example is based upon a wave of complaints following ALP letters in the recent ACT election campaign
Combining these factors, we can come up with three broad types of communication - which are points along a continuum - that we might label inter-personal communication, narrowcasting and broadcasting.
Another important dimension of communication concerns directness, that is, is the manifest content of the message directly concerned with the purpose of communication? For example, one may advertise a product directly, by placing a TV advert or sending material through mail. Orone may advertise it indirectly through sponsorship. Dividing this dimension fairly crudely (for it too is a continuum) we can distinguish direct and indirect communication as the basic division.
These two major aspects can be used to create a typology- illustrated inTable 1. Each of the six types of communication strategy can be described, and each has particular features.
TABLE 1: Six types of communication
|
Interpersonal |
Narrowcasting |
Broadcasting |
| Indirect |
Type 1 |
Type 3 |
Type 5 |
| Direct |
Type 2 |
Type 4 |
Type 6 |
It is important to stress at the outset that these six types cannot be arranged in any single hierarchy of suitability or desirability as communication strategies. Which one is'best'and which is 'worst' depends on the type of message to be conveyed, the nature of the sender, of the receiver and of the relation between the two. Furthermore, although some strategies might seem ideal in principle, there may be various practical barriers to using them in a given context, not least of which may be cost. This caveat must be stressed in order to avA simplistic thinking. For example, type 1 communication might be the best strategy for message X in context Y, and its marginal unit cost in terms of dollars per desired outcome might be lower than type 6. On the other hand, the fixed costs of establishing a widespread method for using this strategy might be prohibitive.
In the case of harm-reduction messages, a variety of channels may be used, but the most important type for our consideration here is type 4 - direct narrowcasting. This is the case because official communication can rarely make systematic use of interpersonal communication channels (although one may hope to create flow-on along these channels) and broadcasting is limited because of the reasons outlined in the previous section. Thus narrowcasting is the main option, and direct communication is to be preferred to indirect communication for getting information messages across.
Assuming that one prefers direct narrowcasting, the question arises as to the nature of the channel that will be used and the location of message transmission. this raises thequestion of the relations between lifestyle and territory.
LIFESTYLE AND TERRITORY
What is the basis of a shared lifestyle? Five fairly obvious (and analytically separable) bases of lifestyle (or, at minimum, shared views of the world) exist in modern society -shared work, shared leisure, shared sexual preference, shared gender and shared ethnicity/religion. Although there may be others that we can imagine, these five axes account for most of the differences in the way that people are differentiated from one another in terms of their everyday world views, beliefs and so forth.
In different ways, each of these intersects with the way that geographical and social space is used. For example, people with common work share their workplaces and, not uncommonly, share some common residential region and/or leisure time activities. Thus we can, forexample, point toplaces where'mostof thepeople who live round here' are from industry X, or where;~ 'most of the people who drink in this pub' are from industry Y. In the same way, people who share leisure activities centre at least some of their time around particular clubs, pubs and other locations. People with common sexual preferences - especially gay groups -' tend to do the same. Men and women use different spaces -if only bathrooms- whereas members of ethnic and religious minorities tend to share both residential locations as well as some urban areas where one may find clusters of 'ethnic' shops, restaurants and so forth.
The most useful concept here is that of a 'territory'. In very many cases, we can consider that there are places that constitute a territory, in the sense that those who use it and view it as 'their space' are disproportionately likely to share important elements of their lifestyle. Gay bars, ethnic stores, neighbourhood pubs and so forth are all examples of lifestyle-related territories.
It follows that the first stage in direct narrowcasting is to identify locations that are territories in this sense. Then one may seek to choose a message known to be relevant to people who share the lifestyle typically found in that territory. One may then transmit that message in that territory with some confidence that the message will be seen as relevant to those who use it.
The next question that arises is to find modes of transmission which will convey the message accuratey and effectively within the territory. A wide variety of options present themselves. There may be places where one can play videos containing some kinds of messages waiting roomsand foyers where people queue might lend themselves to this. Or it may be that sound transmission over PA systems is relevant. And finally it is obviously the case that display of printed materials such as posters is likely to be a major option.
With the last named, however, it is important that poster material be displayed in locations where they will be paid some attention. A darkened dance floor with loud music, flashing lights and gyrating dancers, for example, is hardly a place to locate materiAl which needs any detailed attention, other than something which displays the simplest of messages (such as brand name advertising).
One possible display location, however, which is found embedded within most territories, is the toilet areas. The important feature of toilets, especially the area inside toilet stalls or on walls above urinals is that individuals typically spend a short period of time there carrying out primary biological functions but with little to catch their conscious attention. Moreover, where urinals are concerned, social mores about the permissible parameters of the gaze tend to focus it on otherwise blank walls. These characteristics suggest that toilets may well be ideal places to display posters with simple messages. If the message displayed is chosen to be relevant to the people who typically may be expected to use that toilet location because of known intersections between territory and lifestyle, we have an ideal context for direct narrowcasting.
This is precisely the underlying logic of the technique known as convenience advertising. Modelled on commercial applications of direct marketing techniques (which may best be considered as a subset of narowcasting, applied to selling goods by matching lifestyle to consumption patterns), convenience advertising has been applied to convey harm-reduction messages. In the following sections, 1 outline three studies which 1 have carried out in Australia (Mugford, 1990) and Ireland (Mugford, 1991) in which AIDS-related messages, targeted at young people, were displayed in toilets on various university campuses. As we shall see the data from these studies suggest that the technique is extremely effective at conveying information in the desired manner.
THE THREE STUDIES
In each of the three studies reported here messages relating to AIDS and both sexual activity and drug using were transmitted to a group of young people (students) for which previous research indicated those messages to be germane and helpful. They were displayed on campus locations. Therefore, the results provide a good test of the convenience advertising technique.
General method
In each of the three locations, the same basic method was used. An on-campus location or locations were chosen and a variety of poster displays was placed in toilets in that location(s). In each case, it was possible to work out which sets of students were likely to use d toilets. Inonecase, the toiletswere in halls of residence. In the others they were either in common areas (such as the student union) or in particular buildings where classes were held for specified subjects.
Following the display, survey samples were chosen, either by contacting students in the residences--- ,, ~ sampling whole classes of students who were known to use the relevant buildings. Participation was always voluntary and stressed to be so, but in practice we experienced no refusals with the in-class surveys and very few within the halls. Each respondent was asked to complete a questionnaire which, in addition to a few brief socio-demographic and attitudinal variables, contained a series of questions concerning the posters. Every poster was reproduced (in photo-reduced form) and a series of questions asked about the recall of the poster and evaluation of its message.
In addition, the first study included a control group surveyed in a hall where posters were not displayed as well as an experimental group where they were dis-' played. In the other two studies, questions were asked about use of buildings and regular use of toilet facilities therein, which allowed the groups to be divided in low exposure (never or rarely used the facilities involved) and high exposure (regular use of facilities).
The three studies
The first study (Mugford, 1989) - which amounted to a pilot for the later studies - was carried out in Canberra in halls of residence at the Australian National University. This involved a relatively small sample of students: 62 (of a target of 100) responded in the experimental hall and 78 in the control hall. Of the 62 in the experimental hall, 52 indicated that they had seen one or more adverts in the previous week and in the following analysis 1 concentrate mainly on this subsample.
The second study (Mugford, 1990) was carried out on the campuses of universities in Melbourne and Brisbane. Here the sample was much larger - 845 in all, and 779 in the age range 17-30 -which, being the main target group, is the group 1 concentrate upon in the following analysis.
The third study (Mugford, 199 1) was carried out on two tertiary campuses in Dublin. It was identical to the second study, except for the changes in detail required for wording specific questions - such as the names of buildings, the photos of adverts, etc. The sample size here was 367.
OVERVIEW OF RESULTS
The Canberra study
In the first study, questions asked in the questionnaire evaluated the degree to which students might be considered at risk of contracting AIDS through heterosexual intercourse, by calculating whether they had sexual intercourse, if so whether with multiple partners and whether condoms were used. There was also a set of questions (Figure 1) that dealt with some general attitudes towards risk indaily life. Results on risk taking and on related data are reported here - for the Canberra study only - because the overall study showed an interesting connection between risk and message evaluation.
On sexual behaviour, this was an overwhelmingly self-identified heterosexual group, with 132 claiming that preference, 2 homosexuality and 4 bisexuality. About two-thirds (88) had had heterosexual intercourse and one-third (49) had not. The females were somewhat more likely to have had sexual intercourse (50 of 7 1) than the males (38 of 66), a difference that is not quite statistically significant (X' = 2.5, p = 0. 11 ). Given their wider experience, it is not surprising that the young women tended to have earlier sexual experiences, 16 of the 49 having first intercourse between 13 and 16 years of age compared with 8 of 3 7 young men.
FIGURE 1: Risk Question in Study 1
13. These questions deal with personal choices. (Circle a number.)
(a) Do you prefer
Leisure activities that are just exciting or 1
Leisure activities that have a purpose or 2
Are you unsure? 3
(h) Do you prefer
Work that earns promotion 1
or Work that you enjoy doing 2
or Are you unsure ? 3
(c) Do you prefer
Taking life seriously or 1
Taking life light heartedly or 2
Are you unsure? 3
(d) Do you prefer
Having continuity in the place you live 1
or Having frequent moves of house 2
or Are you unsure? 3
(c) Do you prefer
Fixing long term life ambitions or 1
Taking life as it comes or 2
Are you unsure ? 3
(f) Do you prefer
Leisure activities 1
or Work activities 2
or Are you unsure? 3
(g) Do you prefer
Taking risks or 1
Going through life safely or 2
Are you unsure? 3
The majority of those who were sexually active with more than one partner claimed to use condoms'always' in casual relationships (40) whereas 13 said they used condoms 'most times'I 9 'sometimes' and 3 'never'. In regular relationships the numbers were 23, 12, 27 and 27 respectively. The dataon condom use incasual relationships permitted the construction of a simple threepoint scale of 'riskiness in condom use', where one equalled 'not applicable' (no sex or at least no casual sex), two equalled consistent use of condoms in such relationships and three equalled only partial use of condoms. Interestingly, the scale showed a marked associa, tion with the measures of risk-taking covered earlier in the questionnaire - question 13; 13a, c, e and g in particular relate to risk taking and lack of conscious emphasis upon planning (Figure 1). These were rescored, combined into an index and again split into three groups of high (10-12), medium (8-9) and low (4-7) scores. As Table 2 shows, there is a very statistically significant association between general risk taking scores and condom risk (X-' = 13.7,P<0.01). Indeed, there is also an association between low-risk scores and not having had sex at all.
Although some variability existed on condom use and condom risk-taking, only one person admitted to injecting drugs, so no analysis can be attempted relating to drug use risk.
Overall, the respondents were a moderately sexually active group exhibiting variability in their risk levels for condom use but not in their needle-related behaviour which was negligible. Their knowledge of AIDS itself was quite good.
From here I concentrate upon the reaction to the adverts of the 62 students who responded in the experimental group. The overall result of greatest significance is that, of the 62 usable responses, 52 or 84% identified having seen at least one advert in the coWege in the previous week. Moreover, of those at higher risk ot AIDS (such as those more likely to have sex and/or less likely to use condoms) reported exposure was, if anything, higher rather than lower. The 10 who claimed not to have seen any adverts included 8 who were at virtually no risk (no sex or no casual relationships) and 2 who claimed always to use condoms in casual relationships, whereas those who scored high on condom-related risk were likely to report having seen several ads. Precisely what this means is not certain, although a 'salience ffect'seems likely (that is, those who are more sexually active are more likely to register AIDS messages). The combination of the high level of recall plus the trend towards the more at-risk registering the adverts is encouraging and does suggest that, at least within a population of this nature, there is good reason to believe that the strategy is generally successful at the most basic level, that is, those who most need to see the messages see and register them when delivered in this format.
TABLE 2: Association between condom risk taking and general risk-taking (n= 14 1)
|
Condom use inapplicable |
Always use condoms |
Sometimes or never use condoms |
| Low risk |
34 |
14 |
2 |
| Medium risk |
25 |
16 |
10 |
| High risk |
17 |
10 |
13 |
Turning to look at the adverts in more detail, for each of the six adverts used we can report on whether they were recalled and what people thought of them, the latter arrayed along several dimensions. It is not possible to infer that different rates of recall reflect different levels of effectiveness for two reasons. First, although all adverts were placed in the hall, not all adverts would be in every toilet section, so not every student would have seen every advert. Second, some, like the second advert in particular, had been verywidely disseminated elsewhere and were thus more quickly recognised and (perhaps) recalled at the stage of questionnaire completion. Table 3 presents basic recall data.
No differences in terms of exposure to the adverts were observed in the data by year of enrolment or gender. A number of other variables were examined, such as family background and religion but no trends emerged. Exposure seems to be across the board in terms of general social features.
The adverts were also assessed on six other aspects: namely case of understanding, offensiveness, usefulness, novelty, relevance and ease of recall. These data are not reported in detail here, because they relate more to the question of the message rather than to transmission method. Suffice it to say that results were highly positive, although there was variability in the degree of positive response by advert and by dimension measured.
An observed variation in perceived usefulness and relevance also raised an interesting question as to whether there was any association between variables such as risk taking or sexual activity and the perception of utility or relevance. Both dimensions were examined for all six adverts in relation to condom risk scores and risk scores generally, as well as other plausible variables such as number of friends, religion and so on. The pattern that consistently emerged, sometimes at statistically significant levels, other times at levels that are close to significance, was that the utility and relevance of the adverts were rated higher than expected (on assumptions of random association) by those who are at real risk (multiple partners) but do not always use condoms. For those who are not at risk or are at risk (multiple sexual partners) but consistently use condoms, the rating of utility and relevance of the adverts are lower than would be expected on the assumption of random association, that is to say, these responses indicate that to a reasonable extent the posters were specifically reaching the at-risk target group within the general target population.
TABLE 3: Recall of adverts, including location seen (n=62)
| Advert no. |
Advert recalled |
Seen in college |
Correct on colour (%) |
| 1 |
39 |
24 |
49 |
| 2 |
59 |
41 |
68 |
| 3 |
52 |
17 |
46 |
| 4 |
54 |
38 |
74 |
| 5 |
36 |
19 |
86 |
| 6 |
37 |
22 |
65 |
Overall, the data reported here strongly suggest that the 'convenience advertising strategy' was a valuable and a viable option to be included in harm-reduction strategies related to AIDS. Furthermore, impru'kssionistic data concerning the desire expressed by students to researchers for follow-up material suggest that, where possible (e.g. in clubs), the strategy should be supplemented by a notice saying that copies of the poster of AIDS leaflets are available from the front office. This makes the strategy much more interactive than passive, and although it is likely that the most receptive'viewers'will be the more likely to avail themselves of the facility it cannot hurt to have the message both reinforced and more widely spread.
The Melbourne/Brisbane study
The results here are analysed for the 779 students inthe target group of 17-30 years of age. Although the two campuses differed in magnitude in respect of recognition of the adverts, all the important relationships are in the same direction, that is, a pattern found at one campus is found at the other and vice versa. In the main part of the analysis, therefore, the two campuses are melded for analysis purposes.
The key independent variable examined measured exposure to the adverts. Such exposure presumably results when students frequently use the places where the adverts were displayed - the relevant toilet blocks on the two campuses. The data were examined to see what the distribution was on these variables. Two findings were important: first the answers to the questions concerning building use and toilet use were very highly correlated - the numbers frequently using the building but not the toilet were small, as were those who reported frequently using the toilet on those infrequent occasions they went to the buildings. Second, when complex measures derived for building and toilet use were compared with simple ones - toilet use alone - the latter showed an equally strong fit with the dependent variables as the former. For simplicity's sake, therefore, the simple measure was used.
Thus 'exposure' was calculated as a binary variable, such that those reported as 'low' on the variable were those who reported that they rarely or never used the toilet facilities in any of the relevant buildings where adverts were displayed and those 'high' usually or fairly often used them
For most variables, a very small number of cases had missing data, rarely more than 1 %. In these cases, the missing value was substituted with the modal value to ensure the maintenance of the maximum number of cases for analysis.
Starting with the questions of advert recognition, the proportion who recognised each advert is given in Table 4.
It can be seen that recognition levels vary widely, but, as noted already, the inter-campus ordering is highly consistent. What is of note is the low level of correlation between the adverts. One might have expected that those who recognised (say) advert 1, would be much more likely to be numbered among those who recognised adverts 3 or 5 (the other high recognition adverts) than among those who did not. In fact, although that expectation is correct in the strict sense (that is, the correlations are positive) the magnitude of the association is very small. For example, the correlation between advert 1 and advert 3 is 0. 14 and between advert 1 and advert 5, 0.16. For this reason, it is important to treateach advert as a separate matter, notassume that what will be true of one will be true of another.
Turning to the main independent variable, in the reduced sample of 779cases the exposure variable split 1 approximately 42: 58 with 327 (39.5%) being, coded as high exposure and 428 (60.5%) as low exposure.
For each advert, six factors were examined ~ whether the respondent recognised the advert and, for those that did, how recently they had seen it and in what location (the four locations listed for each advert).
In all cases the data were examined to answer the following questions:
Was there a difference in recognition level depending upon exposure?
Was there any evidence that those with high exposure recalled seeing the adverts more recently?
Did those with high exposure report seeing the adverts in different locations to others?
Recall
As we noted above, there were clear differences in the extent of recognition of each advert, ~~ith the first and sixth reporting extremes offligh and low levels ofrecogrution and the other four in between. When we turn to Table 5, we find that this pattern interacts with exposure. For the two extreme cases, the relationship between exposure and recognition is in the predicted direction (more exposure, more recognition) but the magnitudes are small and the relationships not significant. For the other four, however, which are better tests because the distribution is less extreme, the results are extremely clear cut. Differences are large in magnitude, in the expected direction and highly statistically significant.
Recency
If those exposed to adverts via convenience advertising were recalling the messages from the toilet posters, they should say that they saw the adverts more recently than others. Data on this are shown in Table 6. Once more, the pattern of results is striking and in the expected direction. All but one column pair shows very significant differences. The exception- the lastcolumnpairis as striking in percentage terms as the other but, mainly because n is small, it does not reach conventional levels of significance (p is about 0. 1). What this shows is that those exposed to the adverts clearly recall seeing them more recently.
Locations
If those exposed to adverts via convenience advertising were recalling the messages from the toilet posters, they should say that they saw the adverts as'a poster on campus' (the nearest neutral phrase we could derive) than others. On the other hand, they should not be more likely to say that they saw them in other locations.
TABLE 4: Level of recognition of each of the six adverts, percentage of sample (n=779)
|
Ad no. 1 |
Ad no. 2 |
Ad no. 3 |
Adno.4 |
Ad no. 5 |
Ad no. 6 |
| Recognised |
83 |
41 |
57 |
30 |
57 |
16 |
| Not recognised |
17 |
59 |
43 |
70 |
43 |
84 |
TABLE 5: Recognition of adverts by level of exposure to adverts
|
1 |
2 |
3 |
4 |
5 |
6 |
|
|
|
|
|
|
|
| Recognised (%) |
|
|
|
|
|
|
| Didn't recognise (%) |
|
|
|
|
|
|
| Number |
|
|
|
|
|
|
| Significance level |
|
|
|
|
|
|
Significance levels are: * signifies difference at the 0.001 level; + differ in the expected direction, but at >0.05. High and Low indicate level of exposure to adverts.
TABLE 6: Recency eing the adverts by level of exposure to adverts
|
1 |
2 |
3 |
4 |
5 |
6 |
|
|
|
|
|
|
|
| in the last week (%) |
|
|
|
|
|
|
| in the last month (%) |
|
|
|
|
|
|
| one month plus (%) |
|
|
|
|
|
|
| number |
|
|
|
|
|
|
| significance level |
|
|
|
|
|
|
Significance levels are: * signifies difference at the 0.001 level, ** signifies difference at the 0.01 level; + differ in the expected direction, but at 0.05. ' High and Low indicate level of exposure to adverts.
TABLE 7: Summary of significance tests for the relationship between level of exposure to adverts and location in which they were reported seen
Advert no.
|
1
|
2
|
3
|
4
|
5
|
6
|
| Magazine |
NS |
NS |
NS |
NS |
0.05 |
NS |
| Pub poster |
0.001 |
NS |
NS |
NS |
NS |
NS |
| Campus poster |
0.001 |
0.001 |
0.001 |
0.001 |
0.001 |
0.05 |
| Wall poster |
NS |
NS |
NS |
NS |
NS |
NS |
Respondents could claim that they had seen the adverts in four locations, including magazines, club/pub posters and wall posters, as well as campus posters. Looking at the four locations, the results can be summarised as shown in Table 7 in which 1 show not the raw data for these calculations (which would be daunting in its detail) but rather a summary of the significance tests in the relationships. Under an ideal outcome, the row for 'campus poster' should show a series of highly significant relations, in which those exposed to the adverts much more frequently report seeing them on campus than not. In general, other rows should show no significant relationship.
The results shown in Table 7 are almost startling. Of the 18 relationships where no statistical relation would be expected, 16 are not significant. Of the six that should be significant, all six are, with five of them, where the numbers are larger, at extremely high levels. Table 7 demonstrates that, without any reasonable objection, one can definitely conclude that the convenience advertising strategy is a very clear success.
The data reviewed in respect of study 2 indicate that the convenience advertising strategy is extremely effective in reaching the target audience and leading to high recall of the messages. The level of recognition of the adverts, the recency with which they were recalled as being seen and the accuracy of the recall of the location in which they were displayed speak of an extremely effective strategy.
The Dublin study
As with the previous study, recognition of the adverts was a key dependent variable. In this study, the number of people who recognised each advert was highly variable. For example, advert no. 1 was recognised by 207 respondents (58%) whereas advert no. 3 was recognised by only 29%.
The key independent variable measured was exposure to the adverts. Aswith study 2, the simplermeasure of exposure proved more than adequate as the independent variable. Overall, 227 (64%) were ranked as high in exposure to the adverts, whereas 126 (36%) were ranked as low in exposure (14 missing data).
Starting, then, with the questions of advert recognition, the proportion who recognised each advert is summarised in Table 8.
Unlike earlier findings with the Australian study, there is a substantial level of correlations between the adverts, that is, those who recognised (say) advert 1 are more likely to be numbered among those who recognised adverts 2 or 6 (the other high recognition adverts) than among those who did not. All the correlations are positive and the magnitude of the association is moderate (around 0.4 on average). Nevertheless, it is important to treat each advert as a separate matter, not assume that what will be true of one will be true of another.
TABLE 8. Level of recognition of each of the six adverts, percentage of sample (n=367).
|
Ad no. 1 |
Ad no. 2 |
Ad no. 3 |
Ad no. 4 |
Ad no. 5 |
Ad no. 6 |
| Recognised % |
58 |
43 |
29 |
31 |
36 |
41 |
| Not recognised % |
42 |
57 |
71 |
69 |
64 |
59 |
TABLE 9: Recognition of adverts by level of exposure to adverts
|
1 |
2 |
3 |
4 |
5 |
6 |
|
|
|
|
|
|
|
| Recognised % |
|
|
|
|
|
|
| Didn't recognise % |
|
|
|
|
|
|
| Number |
|
|
|
|
|
|
| Sign level 1 |
|
|
|
|
|
|
Significance levels {ire: * signifies difference at the 0.001 level; **signifies difference at the 0.05 level; + differ in the expected direction, but at >0.05. ' High and Low indicate level of exposure to adverts.
TABLE 10: Recency of seeing the adverts by level of exposure to adverts
|
1 |
2 |
3 |
4 |
5 |
6 |
|
|
|
|
|
|
|
| In the last week % |
|
|
|
|
|
|
| In the last month % |
|
|
|
|
|
|
| >1 month (%) |
|
|
|
|
|
|
| Number |
|
|
|
|
|
|
| Significance level 1 |
|
|
|
|
|
|
Significance levels are: * signifies difference at the 0.001 level; + differ in the expected direction, but at >0.05. ' High and Low indicate level of exposure to adverts.
For each advert, two principal factors were examined - whether the respondent recognised the advert and, for those that did, how recently they had seen it. The data were examined to answer the following questions:
- Was there a difference in recognition level depending upon exposure?
- Was there any evidence that those with high exposure recalled seeing the adverts more recently?
Recall
As we noted above, there were clear differences in the extent of recognition of each advert, with the first and third reporting extremes of high and low levels of recognition and the other four in between. When we turn to Table 9, we find that this pattern interacts with exposure. As with study 2, for two extreme cases where recognition was low (adverts nos 3, 5) the relationship between exposure and recognition was in the predicted direction (more exposure, more recognition) but the magnitudes are smaller and in one case the relationship is not significant. For the other four, however, which are better tests because the distribution was less extreme, the results are extremely clear cut. Differences are large in magnitude, in the expected direction and highly statistically significant.
Recency
If those exposed to adverts via convenience advertising were recalling the messages from the toilet posters, they should say that they saw the adverts more recently than others. Data on this are shown in Table 10. Once more, the pattern of results was clear and in the expected direction. All column pairs shows differences in the first row, with those exposed saying that they had seen the adverts in the last week, whereas those not exposed said they had seen them longer ago than that (which may mean they saw them in the toilet areas but not recently because they don't use them often). Nevertheless, only the first column is statistically significant, although the last column pair is close to conventional levels of significance (p is about 0. 1 ).
Overall, however, the pattern is identical in direction to earlier studies, and it is reasonable to infer that the data in this show that those exposed to the adverts recall seeing them more recently.
IMPLICATIONS OF THE RESULTS
The data reviewed here indicate that the convenience advertising strategy is extremely effective in reaching the target audience in a variety of contexts and leads to high recall of the messages. In all cases, the level of recognition of the adverts and the recency with which they were recalled as being seen, along with other. data on risk, on location accuracy and so forth, speak of an extremely effective strategy.
The results clearly demonstrate the capacity to'narrowcast' specific messages tailored for a given client~le. This is better than broadcasting to a wide range of potential audiences, risking either offence or irrelevance to many although being too bland for those particularly in need of the message.
This is a very clear demonstration of the efficacy of the strategy in three contexts.
OTHER HARM-REDUCTION APPLICATIONS
Although the specific data reviewed here refer more to AIDS messages than to drug-related material (albeit, some of these adverts referred to injecting drug use), the extent to which the idea and technique can be generalised seems unaffected by that fact.
In so far as we can identify particular patterns of drug use as being connected to particular lifestyles and insofar as we can design messages appropriate to specific drug-using groups, direct narrowcasting to those groups is obviously essential if harm-reduction goals are to be pursued.
Convenience advertising is by no means the only way to pursue such direct narrowcasting - but it is one that has been systematically evaluated and found useful, reliable and highly effective. Moreover, it seems highly likely that in employing it for harm-reduction and education messages in the drugs field, at least two extensions are warranted.
First, attention should be paid to making the strategy more interactive. This can be done by backing up posters with 'further information' to be obtained elsewhere in the site used - for example, the'front office'or equivalent. Material gathered informally as part of the first study strongly supports this idea.
Second, as toilets are almost exclusively unisex, opportunity exists for targeting gender-specific messages. So far, although this has certainly been used in cases where the advertising has been a commercial variant of direct marketing, little has been developped where education and harm reduction has been concern d 1 part, this is because although the strategy has now been fairly comprehensively evaluated, most of the messages have been taken from programmes designed on the assumption that readers may be of either gender. But, congruent with its capacity for direct narrowcasting, potential exists for focusing messages more precisely than this.
Dr Stephen Mugford, Department of Sociology Australian National University, Canberra, Australia
REFERENCES
Mugford, S.K. (1989) An evaluation of the convenience advertising strategy: report on the the Hall of Residence study. A Report to the Director, AIDS Education and Information Service, Department of Community Services and Health.
Mugford, S.K. (1990) Evaluation of the convenience advertising strategy for AIDS messages on two university campuses. A Report to the Director, AIDS Education and Information Service, Department of Community Services and Health.
Mugford, S.K. (1991) Evaluation of the convenience advertising strategy for AIDS messages on Irish university campuses. A Report to Convenience Advertising for the Dept of Health, Republic of Ireland.
Wednesday, 15 December 1993 00:00 | 
Written by Stephen Mugford | 
| 
| 