The Cardboard Citizens Glasshouse Tour, 2014. Data and Analysis. WIP. Technical,issues with editing.
Sound Point Notation
It is by definition impossible to observe directly what the spectators are thinking as they react to a play. An idea can be gained of how engaged they are in a performance by the noises they make during it, although, here too, only an incomplete picture is possible. Incomplete or not, if spectators laugh, chuckle, snort, gasp, cheer, boo or hiss, there can little doubt that they are involved, even if involvement is defined as refusing to play along! Thus, a count of noises made seems to be a reasonable proxy for engagement. Similarly, in the forum, the extent to which spectators shout out, shout stop, come up to the stage, join in the discussion, and show their appreciation (or otherwise) through noise seems to be a reasonable proxy for engagement. If it is assumed, furthermore, that the extent to which the spectators engage in the play may be one of the factors that drives their active participation in the forum, then the question is whether such a relationship is observable. This was the question we attempted to answer during the course of the Glasshouse tour. Throughout the tour, the writer attended as many performances as possible, and counted sound points, for both the play and the forum, on predesigned forms. This raw data was transferred to a spreadsheet. By the end of the tour, the observer had accumulated data for 42 performances out of 56.
Comparing Like with Like – Large and Small Audiences, Longer or Shorter Forum Sessions
An audience of ten spectators watching a play that is ten minutes long enjoys, in aggregate, one hundred spectator minutes. (This idea from analysis of the airlines industry, in which the common unit is the passenger mile – this being the number of passengers in the plane multiplied by the number of miles travelled). Let it be assumed that there is one collective moment of laughter, five people are heard to gasp separately, and everyone claps at the end of the performance. In the system of sound-point notation devised for this tour, this amounts to five single noises, and two group noises. The total number of noises made is calculated as 5 plus two times 10, giving a total of 25 noises. The number of noises per spectator minute is therefore 0.25 (25/100). A further practical matter is how to record group noises when the entire auditorium is not involved in the group noise. Here, the approach was to guesstimate the number of audience members engaged in the clapping, cheer, laugh, hiss or boo. A moderate group chuckle might be recorded as 0.30 denoting the involvement of roughly a third of the audience (equivalent to three single spectator vocalizations in this small audience of ten). A very loud prolonged applause might be counted as 1.5 or 2 group reactions, denoting noise levels above average for the whole group, of an extremely loud noise from part of the group. The danger of this approach is the difficulty of maintaining consistency. In this case the sound point recorder was the same person throughout, which might partially offset the risk. It should be noted that, sitting in the audience, the observer may not always hear what is going on in seats any distance away from them, thus, the sounds recorded may be an underestimate (or, if the people in the nearby seats are very noisy) an over-estimate. The relationship depicted in the following analysis should thus be taken with a good pinch of salt. This is art, not science.
For each performance, the number of sound-points per spectator minute was calculated for the play and the forum, for as many performances as possible on the tour. An experiment with linear regression on an excel spreadsheet suggested the expected relationship had been found: the more sound-points there were in the play, the more there seemed to be in the forum. The next question was whether this was different from once audience to the next. The data was therefore grouped into the following categories: Adult hostel and associated services; Young People’s hostel; Theatre for Young People; Prisons; London Theatres ad Regional Theatres. A control for audience size was also added. Eviews was used to run an OLS regression. The results are shown in Table 1.
This suggests the following: engagement in the forum is directly related to engagement in the play, as reflected in audience sound points. The number of sound points per spectator minute is significantly greater for London theatres. For very small audiences, the relationship does not seem to hold. This should not be interpreted to indicate a lack of engagement – after all silence can also be a measure of engagement. Observations collected in the narrative thread of this research suggest a very intense level of emotional engagement in association with a small number of highly focused interventions, in small audiences.
For many people in the audience (65%) Forum Theatre was a new experience and all who answered this question indicated their willingness to do it again. 59% of those who responded said that the forum had changed their view of the characters and their problems, about half changed their view of the play, and half had a favourite character (and only 17% a least favourite character) – all of which suggests significant levels of engagement in the play as well as the forum. Although only 22% of this group said they had shouted “stop” and 19% said they had gone up to the stage (not necessarily the same people), about half of the respondents gave some textual explanation as to why or why not. It is probably reasonable to interpret this to suggest that the audience was highly engaged in the forum process, and thought about the possibility of going up to the stage, thus was likely to be engaged in the fate of those who had the courage to step up, as suggested above.
Table 1 – Summary of Responses to Questions About the Forum.
The Word Cloud
The third key element of the feedback form was the ‘word cloud’ shown in Figure 3.
Figure 3. The Word Cloud
The respondent was asked to put a circle round any word they agreed with, and to cross out any word they disagreed with. The words are a combination of words used in the play by one person of another, or words that describe the behaviour at work in the play. The content was designed to have a negative skew to provoke disagreement, in the expectation that, if the audience tended not to have a view, they would not push back on the negative view that the protagonists got themselves into this mess because of their behaviour. The responses suggested that, overall, the audience seemed to believe that the situation the protagonists found themselves in was not entirely their fault – that the ‘system’ in the form of economics and housing regulation were circumstances too powerful for them to control. This in no way implies that the audience espoused the passive response to circumstances the theatre practitioners fought against. The point embedded in this trend in the data - that the socio-political order is tilted against the marginalized and dispossessed - is indisputable. The frequency with which people were motivated to come up to the stage to present their ‘solution’ suggests anything but a passive acceptance of circumstances. Moreover, there were some interesting differences in some contexts. For instance, the chart below suggests that the prison respondents were hard on Paul, blaming him for the mess the family was getting into.
Source: Cardboard Citizens Glass House Tour 2014 Database
Exploring further, the Word Cloud responses, in aggregate, could be used to address two questions. Did the audience differentiate between the three main characters in the play? And did the view taken on the protagonists in differ with the type of audience?
The above chart already suggests differentiation on both counts. The questions were explored further through a series of correlation matrices, applied to the net total number of times each of the words was encircled (reduced by the number of crossings-out) in the relevant sample of forms: hostels, specialist theatres, young people's hostels and their theatres, prisons, London theatres and regional theatres. Each of these sub-samples was normalized on the total number of words encircled or crossed out in the sample, in order to avoid a comparison of apples and oranges. Taking the first question first, the protagonists had weakly correlated profiles (correlation coefficients ranging between 0.5 and 0.7). In other words, the audience saw them as behaving in similar ways some of the time (which corresponds to what happens in the play). However there were some interesting exceptions to this general rule. Hostel and prison audiences seemed to see Paul and Rhea completely differently, as reflected in very low (0.06) to slightly negative correlations between their profiles. London theatre audiences on average saw more in common between Rhea and Jess than between Rhea and Paul, but for regional theatres it was the other way round: Rhea and Jess were less closely correlated, and Paul correlated slightly more strongly with both. This suggests that regional audiences were somewhat more critical of all of the characters, and less forgiving of Rhea and Jess than London theatre audiences.
Taking the second question, and examining three correlation matrices, one for Jess, one for Paul and one for Rhea, audiences seem to be in broad agreement with each other on Rhea and Jess. Hostels, London and Regional audiences had similar views on Paul (correlation coefficients around 0.8), but young people and prisons held different views (correlation coefficients around 0.3). Although there is a limit to what can be read into these numbers, the key point to draw from them is the indication of a variance in view between different audience groupings. These numbers are thus supportive of intuition and common sense.
With caution, an excel F.Test was applied to the same groupings of data. This statistic tests for similar variance in different series of data. In this case, if respondents are tending to choose the same words from one type of venue to the next, a number close to 1 would be expected. If they are choosing very different words, the result would be close to zero. In this context, numbers in the middle of the range - for instance, 0.5 or 0.6 - are also of interest. Here, there is no certainty that the audiences are doing the same thing, and no certainty that they are doing completely different things. In this context - concerning the behaviour of different audiences watching a play - certainty would be a strange result.
Considering the protagonists relative to each other (within venue types), that is to say, Jess Versus Rhea, or Rhea versus Paul, the expectation would be numbers closer to zero than one, reflecting differences between the characters. This is broadly what was found, however on average the audiences in hostels, hostel theatres and London theatres seemed to see similarities between Jess and Rhea (F test results over 0.9 in some cases). Moving on to consider the reactions by venue for each protagonist in turn, the audiences would be expected to agree at times and to disagree at others, and this is exactly what the F-tests suggest: for Jess and Rhea, about half of the F test numbers were 0.5 or higher, and a few were around 0.95. For Paul, however, results above 0.5 happened only 15 percent of the time. In the play, Paul can be described as the most contentious, misunderstood character. Once again, the statistics chime with common sense and intuition.
The Word Cloud game also asked whether the respondent had changed their view on the protagonist after the forum. The majority had not. Most people thus seemed to stick to the view they had formed of the character in the play through the forum, too. Changes worthy of note: two-thirds of the audiences in young people's hostels changed their view of Rhea. Thirty percent of London theatre audiences changed their view of Paul, and about a quarter of the hostel audiences did so. Just under forty percent of prison spectators said they had changed their view of Jess, and one quarter of young people's audiences also took this position. Overall, then, there were some significant swings in perception (or at least, perceptions of such swings on the part of the respondents), and considerable variance in those changes.