Abstract
Using the data of the EURISLAM project, this article investigates the impact of dramatic events on the public debate with regard to Islam in Belgium, the Netherlands, France, Germany, Switzerland and the UK. It assesses the extent to which dramatic events such as 9/11, the bombings in Madrid and London and the murder of Theo Van Gogh in Amsterdam impacted on the debate on multiculturalism, notably the accommodation of Western-European institutions to Islam. In this contribution, we analyse the overall evolution of public debate in the written press on Muslims, and more particularly Muslim rights, for the period 1999–2009. Our aim is to empirically analyse the transitions that the public debate on Islam in Europe has undergone in the wake of the most dramatic terrorist acts perpetrated by Muslim extremists during the last decade. We hypothesize that dramatic violent events involving Muslim extremists had an impact on the number of claims about Muslims and Islam in general, but not on the debate about religious rights for Muslims in Europe. Descriptive analysis and time series plus intervention analysis were undertaken to test these hypotheses.1
Introduction
Communication studies have shown an increase in the number of articles about Muslims and Islam in the Western press during the past 10 years (see Deltombe, 2008; D’haenens and Bink, 2007; Kabir, 2006; Poole, 2002; Poole and Richardson, 2006; Saeed, 2007). Dramatic events such as the terrorist attacks of 11 September 2001 have put extremist Muslim actors in the headlines of newspapers all over the world. Terrorist attacks undertaken in the name of Islam in New York, Bali, Mumbai, Madrid and London have fuelled concerns about Islamic extremism. Worries about radicalization among the Muslim youth in the West are omnipresent and Islamophobia is on the rise, not least because some of the most radical Islamist leaders have been given a public platform in the Western media (Aldridge, 2007). Populists all over Europe use these dramatic events and radical discourses to state that all Muslims should be treated with suspicion and they routinely warn against the alleged ‘Islamization’ of Europe. In their wake, in the mainstream political debate several authoritative voices have been claiming that multiculturalism has failed, pointing mainly to the non-integration of citizens of Muslim background.
Consequently, the influence of such dramatic events on the general debate regarding the accommodation of Islam is worth studying because of the deep societal impacts they underline. In the decades preceding 9/11, several social scientists had already documented the very negative and stereotypical way in which Muslims and Islam are often portrayed in political and media discourse (Blommaert and Verschueren, 1998; Saïd, 1981; Van Dijk, 1993). It is not much of a surprise to read that most pundits seem to agree that it has only become worse since 2001 (Brown, 2006). Jaspal and Cinnirella (2010), for instance, talk about a new negative ‘hypervisibility’ of Muslims in the UK. At the other side of the globe, Kabir (2006) notes that there is an unbalanced coverage of Islam in the Australian media, focusing on Islamic militants and demonizing all Muslims. Focusing on the USA, Kumar (2010) even claims that there are now five key hegemonic frames for debate on Islam in the post 9/11 world: ‘(1) Islam is a monolithic religion, (2) Islam is a uniquely sexist religion, (3) the “Muslim mind” is incapable of rationality and science, (4) Islam is inherently violent and (5) the West spreads democracy, while Islam spawns terrorism’ (Kumar, 2010: 254). Let us note that while Kumar invests considerable energy in debunking all of these stereotypes, the article does not, however, convincingly prove that these frames are indeed as omnipresent and hegemonic in the public debate as is postulated. This actually seems to be a weakness of quite some academic work on stereotypes about Islam and Muslims in the media: they either take their large-scale existence for granted – in line with pre-established ideas about the dominance of elite racism (Van Dijk, 1993) and hegemonic Islamophobia (Kumar, 2010) – or provide only anecdotal proof. Moreover, they often lack systematic large-scale empirical demonstration. Other scholars sketch a less gloomy picture. Trevino et al (2010), for instance, revealed that American newspapers tend to use more unfavourable than favourable or neutral terms when portraying Islam, but they also note that, overall, the media coverage on Muslims has not become significantly more negative since 9/11.
The European Monitoring Centre on Racism and Xenophobia 2 has nevertheless documented that the terrorist attacks of 9/11 led to an increase in Islamophobic acts in Europe (Allen and Nielsen, 2002). There is, furthermore, abundant proof of negative stereotyping of Muslims in the European media, especially the tabloids and the blogosphere. Does this lead to the conclusion that the entire public debate about the presence and rights of Muslims in Europe has gone only downhill during the last decade? We will try to answer this question in this contribution by analysing the evolution of public debate in the written press on Muslims, and more particularly Muslim rights, in six European countries for the period 1999–2009. We will do so by making use of descriptive analysis, time-series analysis and event analysis.
We hypothesize that dramatic events such as the 9/11 terrorist attacks, the Madrid and London
bombings, the cartoon affair and the assassination of Theo Van Gogh in Amsterdam did have an
influence in terms of the number of claims and tone of the general public debate
surrounding the presence of Islam and Muslims in Europe (H1
Our aim is not to prove that Islamophobia is on the rise in the public domain, nor to sketch a rosy picture of interreligious and intercultural dialogue. Our aim is simply to empirically analyse what transitions the public debate on Islam in Europe has undergone in the wake of the most dramatic terrorist acts perpetrated by Muslim extremists during the last decade, without a priori assuming it turned to the negative. For the sake of argumentation and in line with the scientific practice of falsification, we take as a point of departure the hypothesis that the dramatic events did not push calls for accommodation of Islam off the public agenda. In order to better understand which type of political claims suffer from the influence of the dramatic events analysed, we will divide the claims into four groups according to their issue. This allows us to more precisely define the specificities of the public debate reported by print media regarding the accommodation of Islam in Europe. We are aware that reports in the written media do not show the exact reality of the extent to which political claims are made in the public sphere (Barranco & Wisler, 1999; Koopmans & Statham, 1999; Koopmans et al, 2005). Media coverage inherently suffers from selection bias: the coverage of particular events, for instance, depends on the ‘news worthiness’ of other events occurring on the same day. One could, however, claim that, because they lack representation in the media, unreported political claims will have less impact in the public debate. As a dataset to test this working hypothesis, we make use of a database of political claims with regard to Muslims and Islam in Europe, collected within the framework of the FP7 project EURISLAM, funded by the European Commission. The database analysed in this contribution is composed of 5199 political claims about Muslims or Islam in Europe, reported on in the written media of the six participating countries.
Political claims about Islam in Europe
As the sampling procedure is random, it allows us to highlight those periods in which the
political debate about Islam and Muslims in Europe was more intense than others. The graph in Figure 1, pooling all European data together over
a 10-year period, clearly shows that there are peaks in the frequency of political claims at
particular points in time. Peaks are to be observed in the wake of dramatic events such as the
terrorist attacks of 9/11, the murder of Theo Van Gogh and the Madrid and London bombings.
Furthermore, there is a spectacular increase in February 2006, at the moment of further escalation
of the Danish cartoon controversy (which started at the end of 2005). One should not, however, jump
to conclusions that the public debate on Islam in Europe is entirely led by violent incidents, as
other peaks in the number of claims, such as that observed in October 2006, seem to be unrelated to
dramatic international events. Number of political claims with regard to Muslims or Islam in Europe for the pooled data (six
countries) per month in the period 1999–2009.

Number of political claims with regard to Muslims or Islam in the UK per month in the period 1999–2009.

Number of political claims with regard to Muslims or Islam in the Netherlands per month in the period 1999–2009.

Number of political claims with regard to Muslims or Islam in Switzerland per month in the period 1999–2009.

Number of political claims with regard to Muslims or Islam in Germany per month in the period 1999–2009.

Number of political claims with regard to Muslims or Islam in France per month in the period 1999–2009.
To better evaluate the potential significance of these peaks, we first need to evaluate whether
this pattern is present in all six countries in the same way or whether there might be a
methodological artefact created by the merging of the national data. Indeed, it might be the case
that particular events such as the London bombings or the murder of Theo Van Gogh have only a
pronounced impact in particular countries, in these cases the UK and the Netherlands, respectively.
Hence, in Figures 2 to 7 we present the frequencies per investigated country. Number of political claims with regard to Muslims or Islam in Belgium per month in the period
1999–2009.
We begin with the UK in Figure 2. As can be observed, the first peak coincides with 9/11, the largest peak coincides with the London bombings (July 2005) and a third peak is observed in October 2006, when the UK debates the veil. If we compare the UK results with the Dutch results in Figure 3, we notice that the pattern is quite different, with the exception of the impact of 9/11. As can be clearly observed, there are two outspoken peaks in the Dutch debate: the period after 9/11 and the period following the murder of Theo Van Gogh in November 2004. Figure 4, representing the Swiss data, also shows a pronounced peak in November 2004, attesting to the fact that the murder of Theo Van Gogh did also have some international resonance.
We can clearly see in Figure 5 that 9/11 is also a crucial marker for the German debate, with another peak in February 2006 at the time of the international escalation of the Danish cartoon controversy. Interestingly, the Theo Van Gogh murder in neighbouring country the Netherlands in November 2004 has more resonance than the London bombings in July 2005.
In France, the highest peaks are to be observed in February 2004 (preceding the new French law on religious symbols, in essence leading to a ban on the veil in schools) and October 2006 (the period of the so-called ‘affaire Redeker’, a debate triggered by a vehement critique of Islam by a French philosophy professor). We are not aware of any factor that might explain that, during the same month, there is also a peak in the debate in the UK to be observed. Interestingly, the riots in the banlieues (November 2005) – during which some public figures made a link between the riots and the Muslim background of the banlieues’ young population – do not appear as a peak in claims on Islam in France.
Finally, Figure 7 presents the evolution of claims making in Belgium. Peaks are to be observed after 9/11, in November 2002 (assertive position-taking by immigrant youngsters in Antwerp through the Arab European League 3 ), in November 2004 (murder of Theo Van Gogh) and in February 2006 (Danish cartoon affair).
Summing up the results of this descriptive analysis, we can state that 9/11 had an impact on the debate all over Europe, that the escalation of the Danish cartoon controversy also had a wide resonance; the murder of Theo Van Gogh had an impact in several countries (Switzerland, Germany, Belgium) beyond the Netherlands; the London bombings at first sight had only a significant impact in the UK; and that in every country local events played a key role in triggering more political claims about Islam at particular moments in time. It does therefore seem to be the case that political claims making on Islam in Europe is influenced by dramatic events, but they are clearly not the main driving force in a univocal way all over Europe.
The direction of the debate: More positive or more negative?
Another way of assessing the impact of these dramatic events is by looking at the tone of the debate: are Muslims over time more negatively addressed? In our database of political claims we assessed the ‘position’ of claims, based on a simple scoring system (see Introduction). All claims whose realization implies deterioration – making abstraction of the relative importance of the change – in the rights or position of Muslims received a −1 score. All claims whose realization implied an improvement in the rights and position of Muslims (minor or major) received a + 1 score. Neutral or ambivalent claims were given a code of 0.
In Figure 8 we present the results of the
mean position score by year per country. A first observation that can be made is that the most
negative tone of the debate (in terms of direction of claims) is to be found in Germany from 2001
onwards. More generally, in 2002 (the year following 9/11) there is a backdrop in all countries,
with the exception of Belgium. The debate is the most stable in the UK (with the worst result to be
found in 2005, the year of the London bombings). Lower results are to be observed in 2004 in
Germany, the Netherlands and Belgium. The discursive context seems to be quite similar for all
countries in the year 2006. Let us refrain from trying to comment on all changes, given the risk of
overinterpretation. It does seem to be the case that 9/11 has pushed down the number of positive
claims for Muslims and pushed up the number of negative claims in the year 2002 (with the exception
of Belgium), but the effect did not linger on during subsequent years. Assessing the entire 10-year
period, one cannot say that the tone of the debate on Muslims in Europe has become dramatically
worse. Overall, the debates are still slightly positively orientated towards improving the rights
and the position of Muslims in Europe. Mean ‘position’ of claims by year per country.
Times series for different types of claims
In order to try to better understand what types of claims show an impact of the dramatic events,
we classified them into four substantive categories (Figure 9). A first group, under the heading ‘Minority integration
politics’, contains claims related to Muslim minority rights, to integration of Muslims and to
discrimination towards Muslims. A second group, labelled ‘Racism/Islamophobia and
anti-racism/Islam’, contains issues around racism and Islamophobia in Europe as Islamophobic claims
but also anti-racism and anti-Islamophobia issues (moral appeals, etc.). A third group, ‘Political
and Islamic extremism and violence’, refers to claims with regard to Islamic extremism. It contains
claims directly or indirectly related to violent terrorist attacks (either the attacks themselves or
responses to them). A last group contains all ‘Other issues’: asylum and aliens politics,
interethnic relations, crimes (committed by or towards Muslim people), transnational politics (which
were exclusively coded when they had a direct link with Muslims in Europe), and so on. Number of political claims per specific issue, per month in six countries.
In the graphs of Figure 9 we have pooled the data of all six countries together for the four types of claim. 4 Different patterns can be observed in the time series. The time series of the Minority integration politics does not show the peak of September 2001, nor for that matter any other peaks linked to violent and dramatic events appearing in the previous graphs. This seems to suggest that the debate(s) on Muslim minority rights remain fairly unaffected by the dramatic violent events such as terrorist attacks. The exception is the spike that occurs in November 2004 at the time of the murder of Theo Van Gogh. It suggests that this event could have affected the claims concerning Muslim minority rights whereas the other dramatic events did not. The other kinds of claims do, however, seem to be affected by these violent and dramatic events. The 9/11 events stand out every time. Furthermore, the Danish cartoon controversy appears in the time series for Racism/Islamophobia and anti-racism/Islam, and in the time series for Political and Islamic extremism and violence we see a flare after the London bombings. In the Other issues time series we can observe the 9/11 attacks and the murder of Theo Van Gogh. Other smaller peaks are, however, unrelated to the dramatic and violent events on which we focus.
In sum, a glance at the four figures shows that the four different groups of issues are not influenced in the same way by the dramatic events. Minority integration politics seem to be less influenced by dramatic external events than other kinds of political claims related to Islam. In particular, 9/11 seems to have had an impact on all kinds of claims, except those related to minority integration politics. In order to examine this more closely, we will now undertake a time-series and intervention analysis.
Time-series and intervention analyses
Time-series analysis was developed in economic sciences and has three goals: first, to identify a model representing the evolution of the time series; second, to estimate the impact of one or various events on a time series; and, finally, to make forecasts. This last tool will not be used in our research because it specifically applies to econometric research goals. We will, however, statistically test whether or not there was an impact of particular dramatic events on the evolution of our time series. In doing so, a first step is the identification of the ARIMA-model, which is most appropriate to capture the type of evolution of the time series (see Box and Jenkins, 1976; Desbois, 2005). The AR-I-MA model is governed by the assumption that any time series can be represented by the sum of two elements: a deterministic component (AR) and a stochastic (random) component (MA) (Montgomery et al, 2008: 231). A next step is to undertake an ‘intervention analysis’ to determine the impact of particular exogenous events (see Jacobs et al, 2011).
Three elements compose the ARIMA model; they are usually called p, d and q for ‘(p, d, q) model’. The first element p is the autoregression element. It represents the lingering effects of preceding scores and tries to predict future values from the deviations observed in the previous values. The trends term d, for differencing, corresponds to the number of times the time series needs to be differentiated to be stationary. This process, called identification (the “I ” in AR-I-MA), is the first step of the definition of an ARIMA model. The moving average term q is ‘the number of terms in the model that describe the persistence of a random shock from one observation to the next’ (Tabachnick and Fidell, 2007: 18–4).
Intervention analysis implies cutting the series into two segments: the pre- and the post-intervention series. In the conventional approach of intervention analysis ‘the preintervention series is modeled first, and the impact is modeled afterwards’ (Yaffee and McGee, 2000: 282). To perform the estimation of the model of the pre-intervention series, the number of observations before the intervention should be sufficient otherwise the model could be incorrectly estimated and the number of observations would not give sufficient information to detect the problem with the verification. In our data, the number of observations before the first intervention is relatively small (the 9/11 attacks occur at time 33). Even though some authors say that 30 observations could be sufficient to estimate models, the majority of authors prefer using samples of 60 or more observations. An alternative approach gives a solution of dealing with short pre-intervention series by using the entire series to estimate the model and not only the pre-intervention series. This alternative method is used in our analysis.
We will shortly explain the analysis procedure that has been used, focusing first on the identification of the ARIMA models and subsequently presenting our intervention analyses.
Identification of the ARIMA models
First, the identification procedure requires an observation of the plots of the time series and of the plots of the autocorrelation functions (ACFs) of the series to verify if the series is stationary. If the series is not stationary, one or more transformations have to be performed in order to obtain constant mean and variance. The observation of the four time plots in Appendix 1 does not conclude that the series is not stationary. No visible trend, seasonality, outliers, cycles or other elements that can affect the stationarity are noticeable. Figure 9 shows a rapid decay of the ACF for the four issues, which indicates that the time series are stationary. Therefore, no transformation of the data is necessary: the d of the ‘(p, d, q) model’ is equal to zero.
The determination of the values for p and q requires the observation of the ACF and the partial autocorrelation function (PACF) of the time series. First, several models were selected by the observation of the ACF, PACF and the extended sample autocorrelation function (ESACF) giving further information on the order of the model which should be selected. Then, the models selected were tested, but also the models with a lower and a higher order to check overfitting and underfitting. If the model gives a good fit of the data, the parameter estimates should be significantly different from zero and within the bounds of stationarity and invertibility and the residuals should be random. The models that gave a good fit of the data were compared. The model with the lowest goodness-of-fit statistics (Akaike’s information criterion [AIC], Schwarz Bayesian criterion [SBC] and variance estimate) was chosen.
The determination procedure of the ARIMA models concluded that the best ARIMA model for the time series Minority and integration politics was a model (1,0,1), a model (1,0,1) for the time series Racism/Islamophobia and anti-racism/anti-Islamophobia, a model (1,0,0) for the time series Political and Islamic extremism and violence and a model (1,0,1) for the time series Other issues.
Intervention analyses
Intervention analysis is used to evaluate the impact of a particular event on a time series,
comparing the pattern of the time series before and after the event. Different types of intervention
can be performed; they can be abrupt or gradual, permanent or temporary (Tabachnick and Fidell, 2007). As we have seen with the
observation of the time series of the four groups of claims (see Figure 8), important peaks appear but they decline immediately.
No permanent or gradual effects can be observed. However, each time series shows the occurrence of
some events having abrupt and short-term effects. As a consequence, only intervention analyses for
abrupt and temporary effects will be undertaken. The impact of five events, already mentioned in our
descriptive analysis, will be analysed. They are the attacks of 9/11 (11 September 2001); bombings in Madrid (11 March 2004); bombings in London (7 July 2005); murder of Theo Van Gogh (2 November 2004); and escalation of the affair of the Danish cartoon of Mahomet (February 2006).
For each of these events, a dummy variable was created with a unique code 1 in the time unit at which the event occurred and a code 0 for all other units. These variables were used as independent variables to run in Statistical Analysis Software (SAS) intervention analyses for each of the groups of claims with the corresponding ARIMA model. 5
Results
Parameters of the intervention analyses for the four issues.
Note: ω: the magnitude of the asymptotic change in level after intervention; SE: Standard Error; t: t value; Sig. = Significance of the t-value.
The intervention analyses for the issue of Political and Islamic extremism and violence revealed significant values for the attacks of 9/11 and London, the murder of Theo Van Gogh and the cartoon affair. The magnitude is slightly higher for the attacks of London (with a ω of 59.49; p < 0.001) than for the 9/11 attacks (with a ω of 43.73; p < 0.001). The other events have smaller magnitudes: 27.34 (p < 0.05) for the cartoon controversy and 24.78 (p < 0.05) for the murder of Theo Van Gogh.
The intervention analyses for the issue of Racism/Islamophobia and anti-racism/anti-Islam confirm that three exogenous events had an impact on the time series. The attacks of 9/11, the murder of Theo Van Gogh and the Danish cartoon affair impacted the public debate by an increase of (anti-)racist and (anti-) Islamophobic claims. We can conclude that the debate about this issue was tense after these three events, with a higher intensity after the cartoon affair (with a ω of 58.58; p < 0.001) than after 9/11 (ω = 41.21, p < 0.001) and the murder of Theo Van Gogh (ω = 24.95, p < 0.05).
The intervention analyses for the Other issues show a significant effect only for the murder of Theo Van Gogh, with an ω of 16.2 (p < 0.05). However, we can note that 9/11 impacted the claims with a delay of 1 month (the top of the spike occurs in October and not in September 2001); therefore, to be evaluated, this impact should be modelled differently than the other interventions.
Summing up, 9/11 had a consistent impact on the public debate with regard to Islam but left the debate on Minority integration politics unaffected. One might conclude that this debate is of an ‘endemic’ nature – is constantly present – whereas debates on other issues are more affected by exogenous events. The murder of Theo Van Gogh is the only event that affected each type of issue, even the issue concerning Muslim minority rights, which was not affected by the other events. However, we can note that the impacts occurring at the time of the murder of Theo Van Gogh have a lower magnitude than the impacts of the other events.
Conclusion and debate
In this article we set out to investigate the impact of dramatic events on the public debate with regard to Islam in Belgium, the Netherlands, France, Germany, Switzerland and the UK. Making use of a large dataset covering political claims making in six countries over a 10-year period, we empirically investigated the impact on the debate of dramatic events such as 9/11, the bombings in Madrid and London, the murder of Theo Van Gogh in Amsterdam and the cartoon affair. Going against common academic knowledge, we hypothesized that dramatic violent events involving Muslim extremists did not influence the number of claims discussing accommodation of Islam in Europe. Descriptive analysis and time series plus intervention analyses were undertaken to test this hypothesis. A confrontation with the empirical data does not allow us to reject our hypothesis in a clear-cut way: the debate on religious rights for Muslims and accommodation of Islam in Europe is indeed of an endemic nature and has remained relatively unaffected in amplitude by the dramatic terrorist events of the start of the new millennium, with the exception, however, of the murder of Theo Van Gogh.
Therefore, the article has shown the importance of distinguishing public debates concerning religious rights from other issues when the impact of dramatic and violent events is analysed. It suggests taking a critical look at previous studies showing an undifferentiated increase in the number of (negative) articles about Muslims after dramatic and violent events without distinguishing the issues of the articles. Indeed, the number of articles about Al-Qaida and other Islamic terrorist actors obviously increased after 9/11. Numerous articles reported on the events and the reactions they have triggered. However, this did not push the debate about accommodation of Islam and granting of religious rights to Muslims off the public agenda. Indeed, we have seen that debates about group rights for the Muslim population living in Europe remained largely unaffected by the sudden apparition of dramatic and violent events.
Most experts routinely state that the public debate on Muslims in Europe has gone downhill the last decade as a result of stigmatization and generalization. In line with this interpretation, one would expect the debate on the rights of Muslims and accommodation of Islam to have developed in negative terms or to have dwindled. However, assessing the entire 10-year period, one cannot say that the tone of the debate on Muslims in Europe has become dramatically worse. This somewhat surprising finding should make us wonder whether studies that use content analysis or qualitative discourse analysis should not pay more attention to the dangers of selective sampling and partisan theoretical framing. When one analyses in an in-depth way only the most ‘interesting’ interventions in the public debate, one might be tempted to focus on only the most shocking or negative ones. Furthermore, if one takes as a point of departure a solid belief in the existence of a dominant or hegemonic discourse, there is the intrinsic – probably unconscious – tendency to find only what one is looking for. Researchers who do not make a conscious effort to avoid these methodological flaws undermine the credibility of a potentially powerful analytical approach, which has proven its worth many times in the past.
In our opinion a lot of scholarly work has been overemphasizing the deplorable stigmatization and demonization of Islam and Muslims in Europe, failing to see that there was also a counter discursive movement. Several actors have invested in showing the ‘other’ faces of Islam, not in the least mainstream Muslims who have been resisting to be ‘represented’ by extremist radical voices. Furthermore, citizens and policy makers have grasped the challenge of a debate on the future of state–religion relations and the place of Islam in Europe, resisting the challenges posed by populism and radicalization.
Footnotes
Funding
This research was undertaken as part of the EURISLAM-project, funded under the Framework Programme Seven (FP7) scheme of the European Commission.
