Whilst I can understand why many of us would like to see the results of this poll as the work of a single crackpot and therefore ignored out of hand, I'm not yet convinced that that's the case.
Looking at the actual report from the link posted above by Mats, it appears that the results are based upon five previous surveys (2000-2005) plus the current one (January, 2006). Two organisations were involved: the Indonesian Survey Institute (LSI) (2006) and the Freedom Institute and Center for the Study of Islam and Society (PPIM) of the State Islamic University (UIN) in Jakarta. (A total of 3,200 people surveyed, between Aceh and Irian Jaya.)
In order to attempt to look at the state of radicalism throughout Indonesia, what's called "multistage random sampling" was used to produce a statistically representative sample of Indonesia from which to make inferences and generalisations about the national population. In my opinion, this is a worthwhile thing to attempt. Somebody had to do it sooner or later. However, as in all undertakings of this ("scientific") nature, the quality of the methodology has to be extremely high and I'm not sure that it was.
I have only seen the document mentioned above and have no access to the complete technical version so my next comments may be unfair to the "researchers":
1. All the interviews were conducted by "students trained to do interviews". Without passing judgment upon them, that might be an oxymoron (a contradiction of terms). Still, everybody has to learn somehow.
2. Looking at the graphs in the report, I see that there are points for March, 2006. If the latest surveys were done in January, 2006, what are these? Extrapolations, and if so, on what basis? Again, if so, how come some that look like they should go down go up, and vice-versa?
3. There seem to be far too many reported intercorrelations supposedly statistically significant at the 1% probability level (p<0.01). Virtually all experiments in behavioural/social sciences operate at the 5% level (p<0.05). A probability level in this context is simply the chance of getting the same result using the same method. At 5%, it means if the survey were to be repeated in the same way a hundred times, the results would be the same 95/100 times. That's pretty reasonable for the inexact social sciences. But 99/100 is rather an ambitious undertaking and might just reflect upon the level of expertise of the academics teaching in those institutions.
4. Combining results of various surveys undertaken by different people at different times is always problematic, at the very least.
My "conclusion" about all of this is that whilst the study may not be perfect, it was also not simply a random telephone poll. I believe it was a serious attempt to obtain a clearer picture of the growth/decline of radicalism in Indonesia. And that's admirable. Deep down, I hope that 48-58% of Indonesians don't turn out to really believe adulterers should be stoned to death. And maybe it's nice to know that the number of people who believe in amputating the arms of thieves might be decreasing.
The following is a quote from part of the conclusions of the report:
I hope so.
:shock:
Looking at the actual report from the link posted above by Mats, it appears that the results are based upon five previous surveys (2000-2005) plus the current one (January, 2006). Two organisations were involved: the Indonesian Survey Institute (LSI) (2006) and the Freedom Institute and Center for the Study of Islam and Society (PPIM) of the State Islamic University (UIN) in Jakarta. (A total of 3,200 people surveyed, between Aceh and Irian Jaya.)
In order to attempt to look at the state of radicalism throughout Indonesia, what's called "multistage random sampling" was used to produce a statistically representative sample of Indonesia from which to make inferences and generalisations about the national population. In my opinion, this is a worthwhile thing to attempt. Somebody had to do it sooner or later. However, as in all undertakings of this ("scientific") nature, the quality of the methodology has to be extremely high and I'm not sure that it was.
I have only seen the document mentioned above and have no access to the complete technical version so my next comments may be unfair to the "researchers":
1. All the interviews were conducted by "students trained to do interviews". Without passing judgment upon them, that might be an oxymoron (a contradiction of terms). Still, everybody has to learn somehow.
2. Looking at the graphs in the report, I see that there are points for March, 2006. If the latest surveys were done in January, 2006, what are these? Extrapolations, and if so, on what basis? Again, if so, how come some that look like they should go down go up, and vice-versa?
3. There seem to be far too many reported intercorrelations supposedly statistically significant at the 1% probability level (p<0.01). Virtually all experiments in behavioural/social sciences operate at the 5% level (p<0.05). A probability level in this context is simply the chance of getting the same result using the same method. At 5%, it means if the survey were to be repeated in the same way a hundred times, the results would be the same 95/100 times. That's pretty reasonable for the inexact social sciences. But 99/100 is rather an ambitious undertaking and might just reflect upon the level of expertise of the academics teaching in those institutions.
4. Combining results of various surveys undertaken by different people at different times is always problematic, at the very least.
My "conclusion" about all of this is that whilst the study may not be perfect, it was also not simply a random telephone poll. I believe it was a serious attempt to obtain a clearer picture of the growth/decline of radicalism in Indonesia. And that's admirable. Deep down, I hope that 48-58% of Indonesians don't turn out to really believe adulterers should be stoned to death. And maybe it's nice to know that the number of people who believe in amputating the arms of thieves might be decreasing.
The following is a quote from part of the conclusions of the report:
The strengthening of support for radical agendas and means will change the national political map. Parties that have a tendency to struggle for radical agendas will become stronger. Conversely, mass rejection of radical agendas and means will strengthen political forces that do not have an orientation toward radical agendas. Politically, the strengthening of the large parties (PDIP, Golkar, PPP, PKB, PD) can help reduce the tendency for support of radicalism; socially, the spread of anti-radical conceptions of religion will strengthen a national political map that is more open.
I hope so.
:shock: