Pre-test measures

Image and question selection was twofold, primarily, the author identified possible images against three criteria -- political, general and racial, selected from a general cross-section of images located on internet sites identified by the search 'race hate groups' 'white supremacy' and 'political symbols'. Various Discussion groups were then facilitated, one prisoners and one civilian group[1]. The dual grouping was to negate any prisoner only bias. Twenty selected images were displayed in each of the three categories to each group separately. The means of display was in a room which had reduced lighting and the individual images being projected onto screen. A semi-structured discussion then followed with the facilitator noting each image, which appeared to raise awareness or interest within the group. The images selected for treatment use were seven racial, five political (with a balance between parties) and two general (see appendix A).Further discussion groups with different prisoners and civilians; incorporating a variety of closed and open-ended questions being tabled around a different selection of images, which were similar in design to those already selected for use in the experiment. The responses from within these groups were noted as to which questions appeared to raise awareness or interest from each pilot group. Thus the questions selected were:

Three scales were used as outcome measures in this experiment:

British Prejudice Scale: The British Prejudice Scale (BPS) (Lepore & Brown, 1997) is the amalgamation of several existing instruments: the Modern Racism Scale, the New Racism Scale and the Subtle and Blatant Prejudice Scale. The authors subtly altered these scales post-amalgamation to make them appropriate for white British respondents. The scale is designed as a general measure of anti-black prejudice. This scale was chosen because it contains a substantial component of anti-immigration, anti-foreigner sentiment. This sentiment closely approximates out-group hostility, distance and perceived worldview threats that are central to the theoretical argument being developed in this thesis. The internal consistency for this scale is high, Cronbach a=.92 (Lepore & Brown, 1997).The scale consists of 15 questions answered on a scale from 1 - 7 (strongly agree to strongly disagree) and are summed indicating a range from 15 (high prejudice) to 105 (low prejudice). Thus a respondent scoring lower on this scale will be deemed more racist than a person with a high score.

Social Dominance Orientation Scale: The Social Dominance Orientation Scale (SDO) (Pratto, Sidanius, Stallworth & Malle, 1994), measures individual differences in the extent to which respondents prefer inequality among social groups. There are four separate versions of this scale; the 16-item scale has been selected for use in these experiments to allow other measures to be included. According to the scale's authors, SDO is not a direct measure of racial attitudes, but, like authoritarianism, is a focal part of social ideology that predicts a wide range of political and racial attitudes. The measure has strong internal consistency, Cronbach a= .84 (Pratto et al., 1994). SDO conceptualises the importance of racial and political attitudes and therefore will be a direct measure of racial attitudes. It is purported that a person high in SDO will quite probably develop a negative attitude towards some group that is low in status or prestige. Further indications are that for those respondents high in SDO display a tendency to favour hierarchical relationships within groups and suggest an alignment of 'superior-inferior' dimensions. According to the authors, individuals high in SDO accept 'legitimising myths' that include racial and ethnic prejudice, nationalism, patriotism, separation between 'high' and 'low' culture, sexism, meritocracy and political conservatism (Pratto et all., 1994). Scale items are scored on a 1-7 (very positive to very negative) scale and scores are averaged across items. Thus a respondent scoring higher on this scale would indicate a stronger social dominance orientation or more prone to racist thinking than a person with lower scores. Selection of this scale was important to the study because those high in SDO display out-group hostility or denigration and would therefore likely seek to ensure a greater social distance from members of those out-groups. In addition the scale has been designed to independently indicate racism as apart from traditional political ideology.

Social Distance Scale: The Social Distance Scale (SDS) (Bogardus, 1933) was designed to measure the extent to which people want to keep a distance and avoid intimate contact between themselves and people from different racial, ethnic, national or social groups. The scale has a high internal consistency, with Cronbach a greater than .90 (Kleg & Yamamoto, 1995, Mielenz, 1997, Osei-Kwame & Achola, 1981). The scale consists of measuring 'first feeling' reactions to a list of social, racial, ethnic and national groups ranging from regarding distant social contact (e.g., as visitors to one's country) to the most intimate (e.g., as a family member by marriage). In Bogardus's original scoring method (1933) a low score on this scale indicates the person is less inclusive or welcoming of out-group members than those respondents with a high score. Unlike the British Prejudice Scale and the Social Dominance Orientation Scales, which utilise closed-ended questions and Likert scales, the Social Distance Scale design is based on the uni dimensional Guttman-type scale. Clear written instructions on how to respond to the questionnaire and each scale were provided to all respondent prior to each scale.

The British Prejudice Scale and the Social Dominance Orientation Scales being of the closed question style were pre-coded and the response sets were in a Likert scale, this allowing for ease of analysis. The Social Distance Scale design is based on the uni-dimensional Guttman-type scale, with increasing levels of intimacy. Clear written instructions on how to respond to the questionnaire and each scale were provided to all respondent prior to each scale. The main body of the questionnaire, as previously noted, was pre-determined by the use of existing scales.

Consideration at this stage was given to a methodology of image presentation, due to the social undesirability of racial comments, and it was further considered unlikely that respondents would answers openly and honestly if they were fully aware that they were being questioned about racist attitudes. Furthermore, fully informed consent was not realistic prior to the experiment. The selected scales were already of a political nature by being measurements of political attitudes. The scales used dictated the information required, therefore consideration only had to be given to the construction, format ordering and 'filler' questions required to ensure the deception was successful.

The filler questions were selected by using a pre-general election governmental questionnaire[2] . It has been acknowledged that the style of question can influence the reported response, thus this method, by which elimination of any bias from the researcher is effected was deemed the most appropriate methodology. The questions were subsequently piloted,[3] with random introduction, the facilitator noting the level to which questions appeared to lead or invoke discussion or a better response from the group members. These questions were then used to fill the questionnaire and disguise the real intention of the measure.

The design of the questionnaire was against two specific criteria, simplicity of administration and level of cognition of intended recipients - prisoners. Instructions and layout were deliberately simple with additional concentration on ensuring the design was short, thus ensuring a minimisation of respondent fatigue and high response rate. The demographics were selected to identify length in prison, first or further prison sentence, number of previous prison sentences, age and level of education, thus allowing for statistical tests for possible influence or bias.

Post-test measures

The study investigated whether exposure to racist signs and symbols impacted participants' scores on measures associated with hate and prejudice. The hypothesis was tested in seven different ways, firstly against the British Prejudice Scale, which was designed as a measure of anti-black prejudice; Secondly, against the Social Dominance Orientation Scale, the Social Dominance Orientation Scale conceptualising the importance of racial and political attitudes, therefore being a direct measure of racial attitudes. The next five tests of the hypothesis were sub-components of the Social Distance Scale which measures the extent by which people want to keep a distance and avoid increasing amounts of intimate contact between themselves and people from different racial, ethnic, national or social groups.

A Man-Whitney U[4] test was conducted to determine whether the experimental and the control groups were comparable, on the demographic information from the sample. The demographic variables used in this experiment were length in prison (time spent in present establishment), first time in prison (whether or not this was a first custodial sentence), previous times in prison (number of previous custodial sentences), length of sentence (length of present custodial sentence), age and education. Independent t-tests were also conducted using the British Prejudice Scale, and the various elements of the Social Distance Scale scores. In order to take into account multiple testing, a Bonferroni adjustment[5] was also calculated. Analyses of covariance[6] were considered but no variances in dependant variables were apparent.

Hypothesis

The primary hypothesis of this research is that viewing racist signs and symbols increases prejudice against out-groups and factors associated with racism and hate.

FINDINGS

British Prejudice Scale

On average, the control group scored significantly higher on the British Prejudice Scale (M= 64.67, SD= 16.41) than the experiment group (M= 36.68, SD= 18.46). The results of the test were significant (t(126) = 9.08, p=<0.001). A low score on this scale means high prejudice. Participants in the experiment group showed significantly more prejudice as measured by the British Prejudice Scale than the participants in the control group.

Social Dominance Orientation Scale

The control group also scored significantly lower on the Social Dominance Orientation Scale (M= 37.86, SD= 16.92) than the experiment group (M=75.14, SD= 26.26). This distance between the group means was also significant (t(131) =-9.95, p<0.001). A high score on this scale means high prejudice. This means, those participants in the experiment group showed significantly more prejudice than the participants in the control group.

Social Distance Scale 'Black'

The third dependant variable in this study was the Social Distance Scale, which provides a measure of the degree to which people want to maintain social detachment from people from dissimilar social, racial, ethnic or national groups. The five minority groups used in this study were identified through the pilot group discussions. These appeared to be the most salient groups mentioned as targets of prejudice or hate. Separate analysis was conducted to see if any group stood out more than another. For all subscales independent t-tests were conducted to investigate if participants in the experimental group differed from the control group and the results are described in detail below.

On average the control group scored significantly higher on the Social Distance Scale 'Black' (M=5.90, SD= 1.89) than the experimental group (M= 3.15, SD= 2.37), (t(120) = 7.117, p<0.001). A low score on this scale means high segregation. This means, those participants in the experimental group showed significantly more segregation than the participants in the control group.

Social Distance Scale 'Chinese'

On the Social Distance Scale 'Chinese', the control group also scored significantly higher (M=5.71, SD= 2.11) than the experimental group (M= 3.30, SD= 1.99), (t(117) = 2.216, p<0.029. As before, a low score on this scale means high segregation. This means, those participants in the experimental group showed significantly more segregation than the participants in the control group.

Social Distance Scale 'Mixed'

The t-test on the on the Social Distance Scale 'Mixed' revealed further that on average the control group scored significantly higher on this scale, too, (M=5.66, SD= 2.08) than the experimental group (M= 3.15, SD= 2.62),t(123) = 7.562, p<0.001. 122) = 6.163. As low scores on this scale mean greater desire for social distance, those participants in the experimental group showed a higher proclivity toward segregation than the participants in the control group.

Social Distance Scale 'Immigrant'

The control group also scored significantly higher on the Social Distance Scale 'Immigrant' (M=5.71, SD= 2.11) than the experimental group (M= 3.30, SD= 1.93), t(127) = 6.608, p<0.001. This means that participants in the experimental group showed significantly more segregation than the participants in the control group.

Social Distance Scale 'Asian'

And finally, the control group also scored significantly higher on the Social Distance Scale 'Asian' (M=5.66, SD= 2.08) than the experimental group (M= 3.15, SD= 2.45), t(122) = 6.163, p<0.001, meaning that here too, those participants in the experimental group showed significantly more segregation than the participants in the control group.

These analyses have shown that for all seven dependent measures those participants that had been exposed to racist signs and symbol earlier showed greater prejudice.

In order to take into account multiple testing, a Bonferroni adjustment was calculated. After this correction, the p-value that the t-tests needed to show in order and count as significant was p<.0071 (alpha of .05 divided by 7 t-tests). In all but one of the scales ('Chinese') the p-values were smaller than this adjusted alpha level (see results above). Thus, all but one difference between the experimental and the control group remain significant even after Bonferroni adjustment. It is therefore possible to assume that these differences are not the result of chance.

Analysis of Findings

The information concerning the demographics of the sample will be presented first. Then a discussion is presented separately for each of the different prejudice scales that were used as dependent variables.

In order to determine whether the experimental and the control groups were comparable, Man-Whitney U tests were conducted on the demographic information on the sample. This test was used rather then the t-test because the demographic variables had been collected on ordinal level, not continuous level. An independent t-test was therefore not considered. The demographic variables used in this experiment were length in prison (time spent in present establishment), first time in prison (whether or not this was a first custodial sentence), previous times in prison (number of previous custodial sentences), length of sentence (length of present custodial sentence), age and education. No significant differences were found for first time in prison, previous times in prison, age and education. This means that on these variables the groups were comparable. However, there were significant differences found in length in prison and for length of sentence

The results of the test were significant, z= -3.10, p<.05. The control group had an average rank of 79.28, the experiment group had an average rank of 58.45, and thus participants in the control group had on average longer time in prison than the experimental group. (See table 1). The biggest group for the experimental group had been 7 to 12 months in prison and the biggest group of the control group had been 13 to 24 months in prisons.

Participants in the control group had on average also significantly longer sentences than the experimental group. E.g.: Up to 63% of participants in the control group had sentences between 4 and 10 years, while the lower 40.7% of participants in the experimental group had only sentences between 2 and 4 years. In the Man-Whitney test, the control group had an average rank of 83.62, the experimental group had an average rank of 50.98, and thus the control group had on average longer sentences than the experimental group z= -5.087, p=<.05.

In order to find out where the differences for length of the time spent in prison already were coming from the frequencies were examined. On closer examination of the frequencies of the different categories in each group, it appears that more people are having a longer length of sentence in the control group than in the experimental group.

For length of sentence, further examination was undertaken to try and establish where the significant differences were coming from. In order to do this, the frequencies were examined. On closer examination of the frequencies of the different categories in each group, it appears that more people are serving a sentence of 1 to 4 years in the experimental group than in the control group. It also appears there are more people with sentences of over 4 years but less than 10 years in the control group than in the experimental group.

Analyses of covariance were conducted to investigate if the differences remained once the differences in the length of sentence and time in prison were taken into account. For the British Prejudice Scale as the dependent variable, length of sentence and time in prison were entered as covariates, manipulation as the independent variable. After controlling for these covariates, which significantly predicted the prejudice scores (F(1,123)=6.01, p=.016 for length in prison, F(1,123)=10.77, p=.001 for length of sentence), group membership still significantly predicted the prejudice score, F(1,123)=63.64, MSE= 272.45 p=.000. This means that after controlling for differences in length of sentence and time in prison, the differences in the British Prejudice Scale between control group and experimental group were still significant.

For the Social Dominance Orientation Scale, an analysis of covariance was conducted to investigate if the differences remained once the differences in the length of sentence and time in prison were taken into account. Length of sentence and time in prison were entered as covariates, manipulation as the independent variable. After controlling for these covariates, length in prison did not significantly predict the prejudice score, (F(1,128)=.99, p=.33 but for the length of sentence variable it did significantly predicted the prejudice score, F(1,128)=7.26, p=.008), group membership still significantly predicted the prejudice score, F(1,128)=75.18, MSE= 433.82 p=.000. This means that after controlling for differences in length of sentence and time in prison, the differences in the Social Dominance Orientation between control group and experimental group were still significant.

For the Social Distance Scale 'Black', an analysis of covariance was conducted to investigate if the differences remained once the differences in the length of sentence and time in prison were taken into account. Length of sentence and time in prison were entered as covariates, manipulation as the independent variable. After controlling for these covariates, length in prison did not significantly predicted the prejudice score, (F(1,118)=1.28, p=.26, for length of sentence variable, however, it did significantly predict the prejudice score, F(1,118)=4.14, p=.04), group membership still significantly predicted the prejudice score, F(1,118)=36.69, MSE= 4.36 p=.000. This means that after controlling for differences in length of sentence and time in prison, the differences in the Social Distance Scale 'Black' between control group and experimental group were still significant.

An Analysis of covariance was conducted for the Social Distance Scale 'Chinese' to investigate if the differences remained once the differences in the length of sentence and time in prison were taken into account. Length of sentence and time in prison were entered as covariates, manipulation as the independent variable. After controlling for these covariates, neither significantly predicted the prejudice scores (F(1,115)=.55, p=.46 for length in prison, F(1,115)=1.82, p=.18 for length of sentence), group membership did not significantly predicted the prejudice score, F(1,115)=2.67, MSE= 3.77 p=.11. This means that after controlling for differences in length of sentence and time in prison, the differences in the Social Distance Scale 'Chinese' between control group and experimental group were not found to be significant.

For the Social Distance Scale 'Mixed' an analysis of covariance was conducted to investigate if the differences remained once the differences in the length of sentence and time in prison were taken into account. Length of sentence and time in prison were entered as covariates, manipulation as the independent variable. After controlling for these covariates, length in prison did not significantly predicted the prejudice score (F(1,121)=2.40, p=.12, but the length of sentence variable was significant in predicting the prejudice score, F(1,121)=8.69, p=.004), group membership still significantly predicted the prejudice score, F(1,121)=36.70, MSE= 4.11 p=.000. This means that after controlling for differences in length of sentence and time in prison, the differences in the Social Distance Scale 'Mixed' between control group and experimental group were still significant.

Using the Social Distance Scale 'Immigrant' an analysis of covariance was conducted to investigate if the differences remained once the differences in the length of sentence and time in prison were taken into account. Length of sentence and time in prison were entered as covariates, manipulation as the independent variable. After controlling for these covariates, which significantly predicted the prejudice scores (F(1,125)=7.67, p=.006 for length in prison, F(1,125)=8.87, p=.003 for length of sentence), group membership still significantly predicted the prejudice score, F(1,125)=32.94, MSE= 3.84 p=.000. This means that after controlling for differences in length of sentence and time in prison, the differences in the Social Distance Scale 'Immigrant' between control group and experimental group were still significant.

For the Social Distance Scale 'Asian' an analysis of covariance was conducted to investigate if the differences remained once the differences in the length of sentence and time in prison were taken into account. Length of sentence and time in prison were entered as covariates, manipulation as the independent variable. After controlling for these covariates, which significantly predicted the prejudice scores (F(1,120)=3.33, p=.07 for length in prison, F(1,120)=7.06, p=.009 for length of sentence), group membership still significantly predicted the prejudice score, F(1,120)=26.65, MSE= 4.80 p=.000. This means that after controlling for differences in length of sentence and time in prison, the differences in the Social Distance Scale 'Asian' between control group and experimental group were still significant.

In summary, this means indications are that although the groups are not perfectly similar, it is likely the results are due to the manipulation.

Discussion

The study investigated whether racist signs and symbols increase the salience of factors associated with hate crimes in self-reported levels of prejudice. The hypothesis was tested in seven different ways, firstly against the British Prejudice Scale, which was designed as a measure of anti-black prejudice; Secondly, against the Social Dominance Orientation Scale, the Social Dominance Orientation Scale conceptualising the importance of racial and political attitudes, therefore being a direct measure of racial attitudes. The next five tests of the hypothesis were sub-components of the Social Distance Scale which measures the extent by which people want to keep a distance and avoid increasing amounts of intimate contact between themselves and people from different racial, ethnic, national or social groups.

The experiment entailed two randomly assigned groups, one experimental and one control. The independent variable in the experiment being the viewing of certain visual cues and the dependant variable being the changes in salience of factors associated with hate crime, specifically scored on measures of racial prejudice. The experimental group was exposed to the racist signs and symbols, the control group was not. Afterwards both groups received the same questionnaire containing questions measuring their racial prejudice.

The sample was selected from a category C male training prison in England, two variables were used for selection, self-reported White-British and of single cell occupancy, control (n=100) and experiment (n=100).

The results of this study clearly indicate that those prisoners who received the additional material in the package, the symbols, self-reported higher levels of prejudice and segregation than those in the control group who did not. The analysis of the demographic variables revealed that although there were no significant differences between control and experimental group found for first time in prison, previous times in prison, age and education. Significant differences were found in length in prison and length of sentence. Participants in the control group had on average spent longer time in the prison and have significantly longer sentences than those in the experimental group. The findings for the demographic variable length of sentence do not to explain the differences in the reported scores for the prejudice scales, as it could normally be assumed that the longer a sentence the more racist a person would become within the prison environment. The findings for the demographic variable length in prison are conversely interesting. It can be seen that for the control group the higher grouping is thirteen to twenty four months, with the experiment group being seven to twelve months. Again this indicating that the control group have spent a longer time at the prison. These analysis revealed for the British Prejudice Scale that the difference between the experimental and the control group remained significant even after length in prison and of sentence had been included as covariates and were statistically controlled.

With the Social Dominance Orientation, the manipulation also remained significant after these two covariates were included, as was the case for all social distance scale variables except the Chinese subscale, for which none of the variables became significant once the covariates were included. This suggests that although there are limitations due to the imperfect randomisation, the manipulation rather than the demographic differences drove the differences in the prejudice scales.

It is argued therefore, that the non-conscious influence of the symbols has had effect on the experimental group at least on some measures.

The findings from the British Prejudice Scale indicate that the experimental group experimental group showed significantly more prejudice than the control group, with a low score indicating more prejudice. From this result, it is argued that those respondents in the experimental group not only demonstrated a higher degree of anti-black prejudice, but also demonstrated a higher degree of anti-immigration and anti-foreigner sentiment. These additional measures being an integral component of the British Prejudice scale, and incorporated when Lepore and Brown changes to original scale to suite white British respondents. According to Biernat and Crandall (1999) this scale offers an interesting conceptualisation of prejudice in cognitive terms. They support the argument that prejudice may be reflected in the strength of associations between social categories, in this case 'Blacks' and the mental linking of the category to the attributes of those within the specific category. Thus, the low score on this scale reflecting the negative attributes for highly prejudiced respondents. According to Devine (1989) 6, the fundamental concept is that everyone, whether blatantly bigoted or not, mechanically activates racial stereotypes. This she postulates is the automatic constituent of response. She argues that low-prejudice people consciously suppress or inhibit the stereotype once it has been activated, the controlled component. In this experiment the activation being the increase in salience of the signs and symbols, thus those in the experiment group were unable/ or unwilling to control or suppress their prejudice. This was captured in their self-reporting.

The findings from the Social Dominance Orientation Scale indicate that the control group scored significantly lower on the scale than those in the experiment group. High scoring on this scale indicates more prejudice indicating those in the experiment group were significantly more prejudice than those in the control group. Social Dominance Orientation can be understood as a point to which a person's preference is measured concerning disparity inside social groups. According to Pratto et al., (1994) those who indicate towards the higher end of the Social Dominance Orientation Scale have a preference for hierarchical relations among groups, as apposed to equal relations among groups. They support a variety of 'legitimising myths' which alienate social groups on a superior-inferior dimension. Pratto et al., suggest these myths include racial and ethnic prejudice, nationalism, patriotism and separation between high and low culture.

The results from the Social Distance Scale are subdivided into five sections, Black, Chinese, Mixed, Immigrant and Asian. Lower scores on this scale indicate higher segregation. The experimental group scored significantly lower on all subsets than the control group, thus indicating a higher level of segregation. This scale being designed as a measure of how people want to maintain social distance and avoid increasing levels of intimate contact between themselves and members of different social, ethnic, racial or national groups. This measurement of out-group attitudes, it is argued, measures group likings and acceptance rather than group beliefs or cognitive representations. Stanger et al., (1991). According to Hartley (1946) who used the Social Distance Scale on groups to measure prejudice and narrow-mindedness, he found that a want for a high quantity of social distance from any one out-group was strongly positively correlated with preference for distance with virtually any other out-group.

According to Weisbuch-Remington et al, (2005) anthropological, sociological and psychological theories suggest that religious symbols should influence motivational processes during performance of goal-relevant tasks. Their study evolved around subjects viewing religious symbols, which were presented outside of the participants' conscious awareness. Those who viewed the symbols were more affected than those in the control group. Jung (1964) argues that conscious awareness was not required for the symbol to have an influence. Similarly according to Blascovich and Mendes (2000) in Weisbuch-Remington et al, (2005) that affective symbols influence motivational states both consciously and unconsciously. As Geertz (1973, p. 90), stated, "Religion could be defined as a "system of symbols" which act to establish powerful, persuasive, and long lasting moods and motivations in people". It is argued that this can also be said for racist signs and symbols thus establishing powerful and persuasive moods in people. It can also be further argued that the symbols are summarising symbols that has the purpose of making the defining point of an ideology salient, as supported by Ortner (1973). Thus, the neo-Nazi swastika or the burning cross is used by certain groups to raise the salience of their ideology, as found in the experiment. Jung (1964) postulates that the viewing of symbols considering a persons limitations conscious perceptual ability that people often view symbols without any awareness that they have in fact seen them. He presents that this action is likely to have profound consequences for a persons experiences in the world. The respondents exposed to the manipulation in this papers experiment may or may not have realised what they were viewing, they did see the symbols, and possibly might not have known what they represented. However, with only four respondents from the experimental group documenting refusal on the questionnaire because, they either (n=3) did not agree that political surveys should be undertaken in prisons, or (n=1) saw through the deception and refused to participate because he felt the questionnaire was not about political views but was a racist questionnaire, and one he did not wish to participate in.

To further support this study, according to Weisbuch-Remington et al, (2005) the argument proffered is one of symbols evoking responses even when the symbols are outside of the recipient's conscious awareness as with the symbols embedded within the political questionnaire element of this study. According to Ortner (1973) the image should have an influence even when the viewing is brief. According to Weisbuch-Remington et al, (2005, p. 1206) "... within social psychology, a large empirical literature demonstrates that unreportable stimuli in general (i.e. not necessarily religious) can be influential. Aggressive behaviour, interpretations of others behaviour, attitude, memory ability, and even physiological responses have been influenced by stimuli presented outside of the participants awareness."

This study set out to investigate whether racist signs and symbols increase the salience of factors associated with hate crimes in self-reported levels of prejudice. The hypothesis was tested in seven different ways, firstly against the British Prejudice Scale, with the experimental group showing significantly more prejudice than the control group. Secondly, against the Social Dominance Orientation Scale, with the experimental group displaying significantly more prejudice than the control group. The next five tests of the hypothesis were sub-components of the Social Distance Scale with again the experimental group showing more segregation than the control group.

There have been considerable limitations to this study, from original design, administration and the manner in which the demographic variables were collected. I will deal with each in turn for clarity. Selection of the sample population, although one would think it easy in a captive audience situation, it was not. As was discovered in the return questionnaires, self-perceptional codings do not always concur with the records held, in this transitional data recording period for the Prison Service. Thus, for the chosen population, four had their ethnicity incorrectly recorded. This proved problematic as it led to sample error, albeit a minor error. The intended sample of this experiment being self-identified White-British and who must occupy a single cell to aid confidentiality and singularity of response. This again was difficult, the chosen prison was designated all single cell occupancy. However, following recent dramatic rises in prison population there had been a number of 'doubling up' of cellular accommodation, thus resulting in a cell designed for one, being occupied by two prisoners. Some times from the same ethnic group and sometimes from different ethnic groups. Therefore, a decision was made to eliminate all double cell occupancy respondees from the sample. These limitations, it was felt could cause a concern about generalisability, but the resultant sample size (n=200) negated these concerns.

The main limitation of this study is that there is no before or after test. The before test was not possible because of two major reasons, firstly to provide a sample size which assisted in the elimination of selection bias and which should also be generalisable to the establishment, about 45% were chosen, this being the resultant size after those not deemed suitable for reasons of ethnicity or cell occupancy were dismissed. Any pre testing would have diminished this group further and it was not deemed suitable considering the sample size that would have resulted. Secondly, and attempt, it is argued, to pre test the respondents would have possibly given the real reason for the survey questionnaire away. Thus negating the considerable deception methods utilised. It was assumed that the large sample size would even out any differentials between respondent's degrees of prejudice and hatred. There have been to date no post testing carried out. It would be interesting to further this experiment in this area. However, the reason for the experiment was made aware to each respondent, and due to the transient population of prisons, respondents would not be available for post testing, in some cases. It was attempted to address this lack of pre-test through random assignments to the condition, but inspection of the demographic information showed that there were group differences in length of sentence and time in prison. It could be argued that participants who serve longer sentences have higher prejudices in general and that it is not the increase in salience of the signs and symbols that has made the difference in scores. As summarised before, in order to address this analyses of covariance were run with length of sentence and time in prison added as covariates to group membership (control and experimental) as the independent variable. Group membership still remained a significant predictor of prejudice scores. This is not a substitute for pre-tests but helps to explore whether the differences were due to the experimental manipulation or the demographic differences between the groups.

The results of this study support the hypothesis that racist signs and symbols will increase the salience of factors associated with hate crimes in self-reported levels of prejudice. This study provides the first empirical evidence consistent with the idea that these symbols contribute to the expression of attitudes and emotions in relation to hate crime. This is indeed preliminary work in this area. There have been studies on the non-conscious influence of symbols, referred to earlier in this paper. However, to the authors knowledge this is the first study of this type, and the resultant findings have been found to be significant. It has to be acknowledged that this experiment would have benefited from better design. It is with hindsight that the author strongly urges and replication of this experiment to include a pre-test and continuous level demographic data. The original design and piloting of the experiment was adequate. However, for replication purposes much improvement and benefit would be gained from utilising the disregarded focus group method of administering the questionnaire. It was for resource issues alone that this did not take place, and it is felt that this had a negative effect on the experiment. Ideally, now a follow up experiment should be undertaken using a similar generalisable sample, but with the concerns regarding the demographic variables and delivery of the questionnaire satisfied. This would, it is argued, support and strengthen the findings of the original experiment.

Hate crime, born from prejudice, fuelled by neo-Nazi, anti-Semitic and White Supremacy groups, it can be argued, is kept alive and nourished by propaganda distributed throughout society, throughout prisons, supported by the purported racist views and action of group members or Prison Service staff, these actions, being self fulfilling, a vicious cycle that needs to be broken. Society in general, but more importantly, the Prison Service must find more, far-reaching ways to combat hate crime and prejudice, it must find more ways to minimise prejudices within our prisons. Policies and procedures go some way to achieving this, however, only a short way. The recent improvements in race equality following the Commission for Racial Equalities investigation and subsequent joint action plan with the Prison Service and the then Director General Martin Narey's foreword in the Annual Report and Accounts 2000-2001, 7. Which shook the Prison Service and those employed within it. He stated in this foreword, "... the Prison Service is an institutionally racist organisation, which reflects an institutionally racist white society. We have to add to this, our knowledge that there are pockets of blatant and malicious racism within the Service. It is time to face up to these things". It is time and the minimal policy changes and new key performance target measurement will go some way towards achieving this. This section is closed with a quote from Gerstenfeld (2004, p. 244). "Hate crimes do not occur in a vacuum. Research from the United States and other countries strongly suggests that hate crimes are the result of complex combinations of individual and social factors, and general societal prejudice certainly plays an important role". More research in British prisons is needed to try to understand this complexity of social factors. Therefore, further research must be directed towards gaining a deeper understanding of the impact of these racist signs and symbols. Further research should be undertaken to gain a deeper understanding of the meaning of these signs and symbols and a better insight into the processes involved with them.

However, it is argued, for this study and for the new learning for both HMPS and wider society, the issues have been missed. It is not the localised actions, the reworking of policy and procedure, the implementation of new and fresh data monitoring systems that will bring about the radical change that is needed. It is the attention and action to the recruitment, retention and promotional processes as outlined within this paper that need to be reviewed to address the 'missed' issues of testing and reducing for SDO, SD, hatred and prejudice, thus ensuring that the approaches taken in the future are based upon the theory and discovery of this study and paper.

Main Prisoner Study

Introduction

The second study to be conducted for this research will seek to explore these issues further and also better interrogate the psychology behind these findings.

Sample

Approximately 200 participants will be drawn from a prisoner population of a category B life sentence training prison. No inducement will be offered for participation. Only the scores of white-British prisoners will be included in the findings, although all prisoners will be allowed to participate. Random assignment to the four groups will be facilitated.

Procedure

Method of Administration

The participants will be divided into two experimental (n=50 each) and two Control (n=50 each) conditions as per above. Each group, as with Study 1 will be informed that they are participating in a study of political attitude and views. Each will be provided with a survey with clear instructions and a consent form and will be instructed to complete the survey in one sitting thus mirroring the methodology of study 1.

Furthermore there will be an addition to the experimental design by adding a mortality salience component to the research. Control group A will receive neither the mortality salience conditions nor the hate symbology. Control group B will receive the morality salience conditions, but no symbology. Experimental group C will view the hate symbology, but will not take part in the mortality salience intervention. Finally experimental group D will view the hate symbology and take part in the mortality salience intervention.

Pre-test measures

For Study 2 (and all subsequent studies), all participants' surveys will begin with a series of demographic questions (for self-reported race, class, ethnic background, etc) as well as close-ended questions designed to tap into individual's political and worldviews. This section will include the Social Distance Scale "Black" (one of the measures that was found to be a significant outcome variable in the previous study). The inclusion of this scale at the beginning of all the surveys will allow for a measure of baseline views on a key social prejudice prior to the different interventions.

The next part of the survey will involve a short series of more open-ended questions about social and personal issues and about social attitudes. For groups B and D, these will include two mortality salience questions commonly utilised in TMT research. It is common practise when administering this type of intervention that for the control group, these two questions will be replaced by two neutral questions exploring their thoughts around television viewing. The TMT questions would be as follows:

  • Please briefly describe the emotions that the thought of your own death arouses in you.
  • Jot down, as specifically as you can, what you think will happen to you as you physically die and once you are physically dead.

For the control group, the two replacement questions would be:

  • Please briefly describe the emotions that the watching television arouses in you. (Television being a major recreational factor for prisoners)
  • Jot down, as specifically as you can, what you think will happen to you if you were not able to watch television any more.

However, because this study involves a prisoner sample where thoughts about mortality and the loss of television may possibly be more traumatic, the two mortality questions will be embedded with the short questionnaire, thus alleviating the initial impact. The purpose of this intervention is to influence the mortality salience of groups B and D, to see if such thoughts influence the reaction to the symbol priming exercise.

Next, the survey will include the questions (from Study 1) about awareness of and reaction to common symbols. As in Study 1, the surveys for the experimental groups C and D will also include a series of hate symbols that participants will be asked to respond to. For control groups A and B, this section of the survey will only feature national symbols (flags of various countries) and corporate symbols (e.g., a Mercedes symbol). Following the learning from the pilot study and further review, the choice of symbols have been improved - See Appendix B.

On all surveys, this section will be followed by another, longer section of closed-ended questions that include all of the different scales used as dependant variables in Study 1 (e.g., the British Prejudice Scale, etc). Upon completion of the survey, respondents will be debriefed by a psychologist. The reason for the experiments will be revealed to them along with details of support services available to them should they experience any issues with the subject matter raised.

Experimental Priming

Results from the pilot study support the hypothesis that viewing racist signs and symbols can increase the salience of factors associated with hate and prejudice and increase hostility towards out-groups. The next question, then, is why this is so and how this effect works. The purpose of Test 2 will be to attempt to replicate these findings, but also to explore the dynamics of these symbol effects theoretically. In particular, this further study will build on terror management theory (TMT) research and test whether there are differential impacts of viewing hate symbols based on whether individuals were reminded of their own mortality. Research, reviewed in the previous section, suggests that raising the salience of participants' mortality increases hostility to out-group members who threaten their cultural values. This study will explore whether such reminders interact with exposure to hate symbols.

Post-test measures

Study 2, will be analysed independently and in the same fashion as the pilot study. Each of the four groups will be compared to one another on the dependent variables designed to measure aspects of prejudice and discrimination. It is hypothesised that the most prejudiced group in each of the four samples will be Group D (exposed to both the mortality salience intervention as well as the hate symbols). It is hypothesised that this relationship will hold up across all three samples, despite the fact that each (prisoners, students and prison officers) start from a very different perspective. Man-Whitney U tests will be conducted to determine compatibility of group(s) against the demographic information from the sample(s). Independent t-tests will be conducted using the British Prejudice Scale, and the various elements of the Social Distance Scale scores. Furthermore, in order to take into account multiple testing, Bonferroni adjustments will also be calculated. Additionally for study 2, 3, and 4 analysis of variance (ANOVA) will be undertaken also for study 4 on the levels of punitive awards provided. It is anticipated that those participants who are exposed to the mortality salience questions will report higher levels of punitivness. Analyses of covariance will also be considered to test for variances in dependant variables.

Finally, the different samples will also allow for cross-sample comparisons and within sample analyses to explore whether those starting with very different worldviews (e.g., those people who are strongly in support of multiculturalism or have strongly liberal pluralist values as measured by the initial questionnaires) may be less influenced by exposure to hate symbology than those who are already inclined toward prejudice. Additionally, the unique social circumstances of life prisoners might make them uniquely susceptible to the influence of hate symbology. These and other questions can be explored in the analysis of the data once all four studies are complete.

Hypothesis

It is hypothesised that those participating under control condition A will have scores that are similar to those of the control group in the pilot study, and those in experimental group C will have scores similar to those of the experimental group in the previous study (i.e. more prejudiced than control group A). Of interest in this second study, however, will be the impact of the addition of the morality salience intervention. It is hypothesised that those subjects made to contemplate their own death (control group B) will be more prejudiced than control group A, but less than experimental group C, who were exposed to the hate symbols. On the other hand, experimental group D (who receive both the mortality salience and the hate symbol priming) should be the most prejudiced of all the groups in their scores. These results will help to provide a better understanding of what is going on with the symbol priming results from a psychological perspective.

Study 3: The Staff Study

Sample

The purpose of Test 3 will be to replicate Tests 2 but with a respondent group of HM Prison Service staff. This sample will provide an older (in the case of study 1), more working class sample with a direct power base over the prisoners, but also occupy a very different social status than the prisoner samples. These differences in backgrounds, attitudes and worldviews will be measured in the first part of the survey to allow for comparisons across the samples.

Prison Officers, Managers, civilian administration workers, educational teachers and instructional officers currently employed in a British Category B gaol. No inducement will be offered for participation. Only the scores of white-British participants will be included in the analysis. Random assignment to the groups will be facilitated.

Procedure

Method of Administration

Identical administrative procedures and documentation will be undertaken as with study 2. However, the packet will be provided to all prison staff attached in a sealed envelope to their month ending pay advice slip. This will ensure that every staff member will be afforded the ability to participate. There are at present four staff who do not report a White British background within the current staff group. Identification from the respondents is not possible due to the anonymity of response and for the researcher to ensure that it is only those four staff that did not have the document package attached to their pay advice slip, may have the propensity to cause additional enquiry about the study. Therefore it has been decided to allow for the four non-applicable responses and report this in the study summary. This is recognised as not ideal but with this small number having a percentage impact of 2% it is not deemed a major concern.

An additional front sheet will be provided instructing all staff respondents to complete the questionnaire in private and to place it once completed in the sealed envelope provided and put it in a dedicated locked post box set up for the study. It will be further requested that the questionnaires are returned within seven days - after that they will be counted as a non-return. The post box will be emptied by the researcher's representative only.

Pre-test measures

Additionally in Test 4, the last section of the survey will include a final self-report measure of "punitivness" involving items regarding everyday violations of normal structure and expectations regarding 13 specific disciplinary scenarios involved in prison life (see Appendix C).

Experimental Priming

Post-test measures

See study two

Hypothesis

An additional hypothesis of this study is that exposure to the hate symbols and the mortality salience interventions will increase the punitivness scores of these prison staff members.

Ethical Issues

Research Summary and Ethical Considerations

Full ethical approval to undertake this study was granted on 10 th February 2010.

This study will set out to investigate whether racist signs and symbols increase the salience of factors associated with hate crimes in self-reported levels of prejudice. The primary hypothesis of this research is that viewing racist signs and symbols increases prejudice against out-groups and factors associated with racism and hate. This study will involve interaction with human subjects, thus it is imperative that full and appropriate ethical considerations and understanding are formulated to support the research and those involved in the study. This study will be guided by a collection of recognised and appropriate ethical guidelines, including Queen's University Belfast School of Law Code of Practice April 2007, the Queen's University Policy on the Ethical Approval of Research and Queen's University Code of Good Conduct. Additional external instruments have also been studied to provide the researcher a broader understanding of research of this type including British Society of Criminology, Code of Ethics for Researchers in the Field of Criminology, the RESPECT for research ethics guidelines (2006), Code of Ethics and Conduct, Ethics and Criminal Justice: An Introduction (Kleinig, 2006), the guidelines of the British Sociological Association, the Social Research Association and the Economic and Social Research Council. The research proposal incorporates these principles and outlines the ethical framework that underpins the study and the research design.

It is understood that a primary principle of ethical research is responsibility. The researcher fully understands that the onus of this responsibility must be satisfied by him towards the subjects, resultant data and importantly any post-study support to the subjects should it be required (more so due to the slight deceptive element adopted in the delivery of the instruments), that he has a responsibility to the study subjects, those persons in and around the penal establishments, the staff and managers of the establishments, those who have allowed the study to proceed and those who have provided funding for the research. Furthermore it is imperative that the reputation of the funders, the researcher's supervisors and any associated persons are upheld and not jeopardised in any manner. It is additionally recognised that all associated research and resultant study must be able to uphold and be tested against the rigours of established ethical guidelines and requirements and not fall short of the required standards.

The grounding for and a primary aim of this study is to bring focus upon the impact of racial graffiti in prisons, and (following the recent worldwide terrorist attacks) provide further understanding on the effects of visual nationalistic symbology such as flags and markings on behaviour and attitudes. Although hate symbols have been studied outside of social psychology (e.g. in historical research), the researcher has not been able to identify any previous empirical research that rigorously tests the outcomes of viewing race hate symbols on variables associated with racial prejudice. Thus impact evidence is sparse. The symbols used to adorn the bodies of members of hate groups, prison walls, cell confines, pseudo uniforms or clothing markers and tags can enforce in-group collectivity. The displaying of these symbols may provide a powerful sense of group identity and a method of either immediate visual recognition or more clandestine forms of recognition in the manner of coded wording and numerical presentations. However, this study will test whether the symbols associated with these racist and white supremacy groups are more than just a form of club regalia or some fashion statement. I will test whether the viewing of the signs and symbols increases the social distance, social dominance and prejudice of majority group viewers.

With this hypothesis in mind, the potential ethical considerations of utilising hate symbols in a research context are fully recognised by the researcher. Full balance has been brought to this study against the possible effect that participation in the study may have as a trigger to possible resultant anguish or discomfort. It is anticipated that such distress will be minimal, however, as the symbols being used are common to the prison environment and presented in a non-threatening manner in the research process. Moreover, full psychological support services are in place and on hand should they be required either during the study or at any time post study. Participation in the study is wholly voluntary and no inducements will be offered. Participation will also be anonymous. Because of my position in HMPS, I will have no personal access to the list of who did or did not participate in the research. As such, participants who choose to complete the questionnaire will at no time be subject to any coercion or those who withdraw or do not participate will not suffer any resultant detriment.

An initial pilot study (n=400) has already been undertaken with full ethical approval obtained from HM Prison Service (before the re-establishment of the School of Law's ethics committee). Further replications and additional testing will be undertaken to explore whether increasing a participants' mortality salience has an impact on racial and prejudicial views across a range of samples. The initial pilot study will be replicated across a series of different sample groups, chosen on theoretical grounds, including: prisoners, university students, and the full range of prison staff members. The preliminary study gave no indication either during or after the study of any harmful effects of the research. However, psychological support will still be offered to participants after the survey as it is suggested that that symbols are a powerful form of non-verbal communication and participants may want to talk about their experiences after the research. Symbols can prime emotions and feelings in both in-group and out-group, and influence both attitudes and behaviour.

Furthermore as this research is intended to observe the effects of symbolic priming on out-group hostility[7], it has to be recognised that the study of HMPS staff in particular may provide data which indicates a deeper cause for concern beyond the hypotheses of this study. As such, a crucial aspect of this research will involve the preservation of participants' anonymity throughout. As principle researcher I will not be able to link the responses to particular respondents, nor will I know who has and has not refused to participate in the research.

The main methodology of the study will involve the administration of a questionnaire to participants. Each participant will be informed that they are being asked to participate in a study of political attitudes and views. Each participant will be provided with a pack containing typed instructions and a closed-ended survey (with a small number of open-ended items). For both the control and experimental groups, the first part of the survey will involve a political questionnaire containing images taken from either the current political theatre (e.g., a Union Jack symbol) or every day life (e.g., motor vehicle emblem symbol). For the experimental group only, this section will also included additional "hate" imagery (e.g., swastikas) interspersed among the other symbols. Both groups will be lead into the symbols with the statement:

  • 'You are about to view a collection of pictures, please take the time to study each one'.

Directly following this questionnaire each participant will be presented with questions from the following three scales:

  • British Prejudice Scale
  • Social Dominance Orientation.
  • Social Distance Scale

Additional tests will involve minor variations to this design. For instance, for certain experimental groups, the study will involve a short series of more open-ended questions about social and personal issues and about social attitudes including two mortality salience questions commonly utilised in TMT research. For the control group, these two questions will be replaced by two neutral questions exploring their thoughts around television viewing.

Additionally in Test 4, the last section of the survey will include a final self-report measure of "punitivness" involving items regarding everyday violations of normal structure and expectations regarding 13 specific disciplinary scenarios involved in prison life. An additional hypothesis of this study is that exposure to the hate symbols and the mortality salience interventions will increase the punitivness scores of these prison staff members.

In accordance with the recommendations of various ethical guidelines, each participant will be provided with details of the research aims (e.g., to learn about prejudice, political views, etc), although the true research hypotheses will be kept from participants until after their participation. This element of deceptive delivery is necessary for the testing of the hypothesis as will be explained along with other aspects of the study upon completion of the research. All those individuals who were asked to participate in the study will receive a written explanation of the study after the research is completed. This will include an explanation for how to ask any questions they may have about the research. A staff psychologist will be available for all participants with concerns about the research. As the researcher, I will also make myself available to participants wanting further explanations of the study; however, as I am not a trained psychologist I will refer any issues of distress or concern to the psychology team in the first instance. Again, no such distress was apparent in the pilot test, and none is anticipated.

Further information will be contained which outlines and satisfies the requirements of Schedule 2 of the Data Protection Act (1998). Furthermore, details outlining the ongoing right of withdrawal, conditions of sanitisation and confidentiality and the full psychological support mechanisms in place and available to them, also the process of application for access.

Collection of the sealed packets will take place the following morning from the prisoners and from a discreet mailbox for the staff participants. Subsequent opening, sight and handling of the questionnaires and any other associated documentation or logs will only be carried out by the researcher and consequent sanitised sharing of data - only to the researcher's direct supervisor(s). All identification of participants will be eradicated; there should be no request for voluntary identification as it is deemed not relevant to this study at this time. All original questionnaires and associated documentation will be stored within a security safe inside a secure prison building and office. Once full analysis has been made these documents will be placed in a sealed security container and placed is a secure emoluments storage area, subsequent electronic analysis, notes and records and any unused material will be encrypted and stored electronically in a manner which satisfies the stringent requirements of governmental information handling. However, procedures will be put into place to ensure that all data and documents will be destroyed as confidential waste if requested by a participant or in accordance with the five year policy of data retention promulgated by Queens University, thus complying with the indefinite retention requirements currently contained in the Data Protection Act (1998). There may be occasion in the future to publish in whole or part something consequential from this study. However, as the questionnaire participant or comment made on the questionnaire is unidentifiable, there is not seen any requirement to establish any post study contact with any participant as this is not achievable. Furthermore there would be no information or data contained within any future publication that could lead to the identity of any participant or compromise of any position.

The researcher considers a closed establishment to be a community; indeed it encompasses all aspects of one. The ethical framework for this study has been informed by this understanding -- including the researcher's responsibility and obligations towards those individuals and groups that this study involves. The study will be informed by the ethical guiding principles of an assortment of research literature and governing bodies However due to the considerations given in understanding a closed prison environment and a micro culture or community; the researcher gave particular focus to consideration of the Social Research Association. Hence by applying these further considerations the study will be conducted with emphasis upon professional integrity, openness and honesty in dealing with participants and with a view to continual improvement and reflection upon the study and those it affects. The researcher will as matter of course seek out advice, guidance and critique from his supervisor(s) whilst at the same time drawing from previous learning or study. Full record and documentation will be established and maintained throughout the study period also encompassing and recording any subsequent actions post study. Thus, it is felt that this project satisfies the requirements of 'academic citizenship', leading to a full, frank and worthwhile outcome.

Summary of Methodology

The pilot study has demonstrated a relationship between the viewing of racist signs and symbols and increased measures of prejudice, social dominance and social distance. The pilot study has provided significant indication that those who received the additional symbols of hate in their packets self-reported higher levels of prejudice and segregation than those who did not. The main study will replicate this but against a different prisoner population profile, thereby further interrogating the relationship between the symbols and measures of prejudice, thus attempting to establish a better and fuller understanding of the dynamics of these symbols of hate. The staff study will replicate again the previous study but with the addition of a mortality salience element and a measure of punitivness. The staff study will be using a very different sample from studies one and two, HMPS employees. This will bring an external validity to the pilot study and study two findings.

  1. Non-HMPS staff
  2. Obtained following internet search for governmental political questionnaires. This questionnaire was selected because of its recent use and up to date questions around British current political views and positions.
  3. Randomly selected prisoners (n=8) by cell location number.
  4. Non-parametric test used to test the null hypothesis that two samples come from the same population or, alternatively, whether observations in one sample tend to be larger than observations in the other.
  5. Method used to address the problem of multiple comparisons.
  6. ANCOVA - all-purpose linear representation which utilises a quantitative continuous outcome variable with one or supplementary qualitative factor variables
  7. Measured with the British Prejudice Scale, designed as a measure of anti-black prejudice; a Social Dominance Orientation Scale, measuring the importance of racial and political attitudes; and various sub-components from a Social Distance Scale, measuring the extent by which people want to avoid social contact with those from different racial, ethnic, national or social groups.