Monday, April 26, 2010

Peak Fundamentalism

Percent Maximum Fundamentalism & TRB Posts

A recent pilot study attempted to quantify a fundamentalism construct in Universal House of Justice (UHJ) membership over time (1963-2010). The present chart adds the peak in talk.relgion.bahai (TRB) posts in 2004, about twelve years after peak fundamentalism.

Gosh. That is one half a generation among humans; hundreds of generations in my species (house mouse). What does it all mean?

1. As previously reported, the blue chart dots show a composite variable (UHJ Member Index) used as a preliminary estimate of fundamentalism among the membership. That is, it has no necessary relation to UHJ statements, rulings or policies as such. Indeed, I concluded that the Index -- a purely conceptual variable derived from third-party data (TRB), was actually showing NSA voter sentiment changes over time. As such, the surprise was not the Index rise (1987-1993), but rather its dramatic crash (2000-2010). In short, I concluded that fundamentalism was rapidly loosing favor among NSA voters, who elect the UHJ.

2. Being an incurable optimist, I was encouraged to find something unexpected and perhaps not fully noticed and applauded by many liberal commentators on Baha'i matters -- the collapse of fundamentalism sentiment (2000-2010).

3. While the source data (TRB) is admittedly noisy, it did provide an operational definition independent of any bias on my part or that of a group of experts I might have selected to rate UHJ members, past and present. An obvious critique of the results charted is that they are totally trivial. Namely, it goes up when perceived controversial members are elected and down when they retire.

While professors often dwell on trivia in search of pearls of wisdom, please consider what might have happened in the five most recent UHJ elections (2000-2010). NSA voters could have elected equally or more fundamentalist members; but they did not.

4. In retrospect, the observed change in NSA voter sentiment makes sense. Having participated in Baha'i communities myself since 1963 in several countries, as resident or visitor, I have not seen any particular interest in, or desire for, increased authoritarian, theocratic, anti-freedom persons or actions in the Baha'i community. In fact, among Baha'is, the UHJ, for example, was a non-issue -- the highest administrative body in the Baha'i Faith and as such, its rulings would obviously be final and binding (what else?). Nothing mysterious there. Everybody knows the buck stops somewhere.

In short, if it is true that the UHJ members do vary significantly over a fundamentalism-freedom dimension, as simple observation and the TRB result count for each member both suggest, then it seems clear that those toward the fundamentalist end of the spectrum have no market, no audience, at least in the Baha'i community I've known. Mingling among Baha'is at all sorts of meetings, I do not recall overhearing "Maybe a Hollywood celebrity should be the final and binding authority in the Baha'i community." Until there are 20, 40 or 80 thousand Baha'is saying something like that, there are definitely more pressing issues.

In fact, one has to consider whether some of the theocratic "do not think; just do as I say" rhetoric itself may have been damaging, given that such issues of authority and obedience regarding the administration are near the bottom of the priority list for the average Baha'i, struggling just to be a better person vis a vis the very high standards in the Baha'i teachings, and that most likely, many decades more will pass before the underlying ideas are more widely and better understood.

Why damaging? Well, every community has its fringe persons, with iconoclastic ideas which might seem strange or even be disruptive. A heavy authoritarian hand might actually increase disunity, while a moderate "we are all in the same boat" approach might be more appropriate if an outside person doing an intervention were to use a more 'Abdu'l-Baha model.

I fondly recall a presumed pilgrim's note cited by my teacher Dr. Daniel Jordan in the first days I was investigating the Baha'i Faith at the University of Chicago, as a newly arrived freshman. A pilgrim from the U.S. was said to have complained to 'Abdu'l-Baha in Palestine that the American Baha'is were doing too much back-biting.
Ending Number 1:
'Abdu'l-Baha pounded his fist on the table, the loud bang shocking the visiting pilgrim, saying, "I told them not to back-bite; don't they know my authority?"
Ending Number 2:
''Abdu'l-Baha smiled saying softly, "I went there, I taught them, but they just don't listen", shrugging his shoulders to suggest, "What can I do?"
The real story was Ending Number 2, of course. The same day I declared -- here was a loving, patient Center of the Covenant, knowing that perfection among humans takes time to attain.

5. Concerning my pilot study, Baquia writes in a comment:
The study you have conducted contains a fatal flaw. The quantity of messages and the amount of discussion in online discussion groups (talk.religion.bahai) tends to ebb and flow so when you see an in/decrease in the mention of something, you could very well be simply measuring the volume of the discussion in the group as a whole, by proxy.
Thanks for the review. With all due respect, I beg to differ. For each past and present UHJ member, the total results count for search member's name was used and this value was a constant, an attribute of the person, over the entire period of TRB activity. Thus, when the mentions occurred would not change the computed result for each election -- the sum of the nine member values. The only thing changing, is who served when, not the dates of TRB posts.

Indeed, let us say I handed you a list of all the past and present UHJ members and you alone made a fundamentalism rating, say, 0 to 10, for each. Those numbers then are the only input into the computed sums plotted -- UHJ Member Index. It would not matter when you did this rating, because once you did, those values would be treated as constants, an attribute (according to you) of each person.

Therefore, my use of the source database (TRB) was independent of any ebb and flow of the post volume.

Quite rightly, however, your TRB volume variable may be of interest and is plotted in pink/red in the chart above. There may be a rather dramatic story to tell:
The cards are dealt and Dr. Peter Khan is elected in 1987. TRB not even born yet. Meanwhile, alarmed chatter starts in other forums, as more and more intellectuals find their way onto the internet. Next it's Douglas Martin, and so forth. Approaching the end of the century, Baha'i authorities notice the rapidly rising pink line of TRB post volume, get upset stomachs when some of the posts are read, and scramble to full alert status. Then it's code red when fundamentalism had peaked (1993-1999), but horrors, the pink TRB line keeps rising into the new century.

Somebody says, "The Baha'i presence on the internet is getting slaughtered here", and at the peak of TRB post volume in 2004, the Baha'i Internet Agency is formed (purportedly by the UHJ, although I'm still waiting to get confirmation on that) and presto, wave the magic wand, free speech by individuals and groups of Baha'is is suddenly the new watch-word.
Baquia continues:
What you've done in essence is statistical sampling. And your sample is showing a decrease. But you need to normalize for a potential decrease in the total population to rule out the probability that this trend is reflecting that instead.
My take here (show me I'm wrong) is that the populations that I used are totally constant. First is the total number of TRB posts. Second is all past and present UHJ members.

Technically, "statistical sampling" was not used. Instead, I used the total population of TRB posts matching search result with the total population of UHJ members. There was no need to sample in either population, and there was no sampling, so such could not have shown a "decrease". The observed decrease in the UHJ Member Index is due only to changes in UHJ membership over time. When an investigator can get measurements from the whole population and thus does not need to sample, all the better since sampling error is absent from the results.

In sum, the most interesting result is perhaps the fall in fundamentalism starting in 2000, about five years before TRB posts peaked. This key fact almost forces attention beyond the two groups, TRB posters and UHJ members, toward a third group -- the NSA voters.

Indeed, the Pearson product-moment correlation between the percent maximum values for Member Index and TRB post volume from 1999 to 2010 (see Appendix), is 0.44, and its square, 0.19, indicates that only 19 percent of the variance of one accounts for the other. That is, over 80 percent of the variation of the UHJ Member Index is not related to TRB post volume in those years (and this, without considering the statistical significance, if any, of the correlation above). Baquia continues:
I'm not sure if you were active or lurking on trb at the 'peak' years (1999-2002) but the chart you've produced looks to be an accurate depiction of the total volume of posts.
For a time, I participated under a handle. Contrary to "accurate depiction", we see that the rise in Member Index precedes the TRB rise by many years. If anything, the direction of cause-and-effect is TRB activity reacting to the fundamentalism surge shown, rather than the other way around. Baquia continues:
Many people left the heated debates to start their own blogs or to other venues such as more moderated and focused online groups such as Talisman or Tarikh. And many others simply moved on to other interests.
...which is a good description of the post-2004 drop in TRB post volume. Baquia continues:
One way you could work around this flaw is to show the relative mention of a UHJ member. So for example, for each mention in a month or year, divide it by the total number of posts in that year/month. In this way you are factoring out the effect of the total volume of discussion.

This would be a lot of work and even after that, we're only left with a tenuous link to 'fundamentalism'. After all, these are public figures within the Baha'i community and it is natural for them to be mentioned in discussions. The nature of the discussion need not be a condemnation of 'fundamentalism' of course. It is more akin to a popularity index, if you will.
Yes, but I am not sure you have established any "flaw" as such, beyond it was a pilot study that could have included many other internet sources to perhaps reduce noise in the data, etc.

I do not understand the value of your "relative mention" idea, since the essence of the study design already controls for that, since the only variance of any interest was within-group, that is, among the past and present UHJ members. Each one has exactly the same "relative mention" since the same total population of posts was used for each. In short, there is nothing to normalize there. The "total volume of discussion" is a constant in the pilot study, much as the "total volume of thoughts in your head" might be, if you had done the ratings (instead of TRB posters), as described above.

On the other hand, Baquia, you may be perhaps pointing to a more refined or weighted composite variable than my Member Index, which was just a simple sum of numbers. Or even different methods which yield more precise findings and/or address different research questions.

As someone who has been there posting in TRB, believe it, the individual member result count was not even remotely a "popularity" rating in my study, which is precisely why the TRB database was chosen. My report recognized that many of the mentions would be favorable, but that this would probably be a negligible noise factor. Of course, anybody can go post by post and subtract out the favorable mentions to reduce that noise.

On the other hand, if one's favorite UHJ members seem to have a higher or lower count in the pilot study than one might want to see, not to worry. The study does not and could not evaluate these members as people -- they all are part of Baha'i history. They all have toiled for the Cause of Baha'u'llah. So the study simply tried, perhaps with some success, to quantify certain trends using an objectively defined method, so that others can repeat the work, if desired, to confirm or not the findings.
Regarding author disclosure, I have personal experiences with at least the following members: David Ruhe, Peter Khan, Douglas Martin and Glenford Mitchell. All are most friendly, charming persons. Some highlights. Peter and I were in the same community for years in Ann Arbor, MI, and you know the routine -- feasts in each others places, etc. I got to know Glenford quite well when we were both serving on the National Teaching Committee (NTC) in the U.S., I being the first "youth member", though treated equally by all. Of course, that leads to Dr. Ruhe, including all the joint NSA/NTC meetings, and in particular those times we traveled together for conferences calling for one member of each (NSA and NTC), not to mention the blanket which I still have he and wife Meg gave us at my first wedding at the House of Worship. I came to know Douglas at the Davidson Baha'i school and later, at BIC as a courtesy call, while I was in New York promoting "The Dawn-Breakers" film in the 1980's.
© 2010 James J Keene
Appendix:
Year, % Max UHJ Member Index & TRB Posts

1663  12  0
1964 12 0
1965 12 0
1966 12 0
1967 12 0
1968 12 0
1969 12 0
1970 12 0
1971 12 0
1972 12 0
1973 12 0
1974 12 0
1975 12 0
1976 12 0
1977 12 0
1978 12 0
1979 12 0
1980 12 0
1981 12 0
1982 15 0
1983 15 0
1984 15 0
1985 15 0
1986 15 0
1987 59 0
1988 74 0
1989 74 0
1990 74 0
1991 74 0
1992 74 0
1993 100 0
1994 100 0
1995 100 0
1996 100 0
1997 100 0
1998 100 0
1999 100 33 ;major TRB starts here
2000 90 25
2001 90 37
2002 90 72
2003 87 93
2004 87 100
2005 65 71
2006 65 35
2007 65 33
2008 62 24
2009 62 20
2010 27 15

6 comments:

  1. Allah'u'Abha, I'm glad that my comment was taken with the constructive intention it was delivered.

    However, I'm a bit confused. This is what I understand you have done, please let me know if any of it is wrong:

    You searched for the keyword X in TRB where X is the name of a UHJ member. You then summed X for each member to come up with Y and you then plotted Y along a time axis.

    You then say that since the trend of Y is falling, this proves there is less "fundamentalism".

    The reason I believe the above is because you said:

    "For each UHJ member from 1963 to present, the name as shown in the Appendix: Raw Data below, was entered for a "Search this group" operation and the number of results was recorded as an "Index" associated with the person throughout."

    "For each election where UHJ membership changed, the individual indexes for the nine elected members were summed to obtain the UHJ Member Index plotted in the chart."

    "a higher index value does correlate with a greater amount of general mayhem associated with the person in TRB. That is, the signal-to-noise ratio to quantify fundamentalism was believed to be more than sufficient for this pilot study."

    As I wrote before, the pitfall of the study is that you are measuring popularity (incorrectly at that) and there is a tenuous to non-existent relationship with popularity and 'fundamentalism' - which you haven't even defined btw.

    There is no question that the volume of posts ebbs and flows on TRB as with any other discussion forum. You've simply dipped a thimble in a stream and taken out a small sample (those that mention a UHJ member). This is called statistical sampling.

    Since having more discussion would mean a higher probability of the mention of high profile persons such as UHJ members, you need to normalize for the ebb and flow of the volume in the discussion.

    That is to say, if we were measuring how many people wear blue at feasts, we would have to normalize for the total number of people at those feasts. Other wise, we would come to the erroneous conclusion that NY feasts feature the highest number of blue wearers. Where in fact just because there is a higher Baha'i population in NY, there are more people, and therefore, a higher probability of some of those people wearing blue.

    Get it? So we need to normalize for the total population. One simple way is to divide the total blue wearers at any feast by the total number of participants and come up with a blue percentage.

    This way we would see for example, that it is in fact Irish Baha'is who wear more blue. So while there are 144 Baha'i blue wearers in NY, that only represents 6.4% of the total Baha'is. While in Ireland, we only have 40 blue wearers but that represents 56% of the total Baha'is at feasts.

    So more Baha'is in Ireland wear blue at feasts proportionally than do Baha'is in NY.

    But as I mentioned, even if you normalize for the volume of the discussion, you are then faced with proving that a mention of a person is somehow connected with 'fundamentalism'. I have yet to read a cogent reason for this.

    I hope that I've explained things clearly. If not, let me know.

    ReplyDelete
  2. Baquia. We all love you and now even more, due to your interest -- thank you -- in my little effort to understand recent Baha'i history.

    Your comment begins OK, I think, but I wonder if the independence of two components of the pilot study is sufficiently clear (if not, it's my fault): (1) the rating of all UHJ members and (2) plotting the rating sum for elected members from 1963 to 2010.

    If there is any merit at all here, one could substitute (1) above with any other rating method and proceed to plot the sums in step (2). Let's say we are rating UHJ member height in inches for step (1). Then in (2), we would see if total height is trending over time.

    My approach was that I know very little -- especially compared to the abundance of highly intelligent commentators, such as yourself, on Baha'i matters. Thus, I did not do the ratings in (1) although I suggested that perhaps you could and then plot the results in (2).

    However, I have defined "fundamentalism", very specifically. First, in the old-fashioned sociology way via an operational definition -- the number of mentions of a UHJ member in TRB. This definition assumed that all past and present members had an equal chance to be mentioned in TRB. In our favor is that the TRB database is recent (mostly in the last ten years), where even recently elected members would have a chance to be mentioned, particularly given that they all have a history going back a long time (no members in the 20's age bracket). A flaw would be that members back in the 60's and 70's might be somewhat out of sight, out of mind, among TRB posters, and receive fewer mentions.

    All this is sort of obvious, but it was a pilot study.

    There are other sources, such as Talisman and others, but TRB was selected in part due to its huge volume of data and its more recent history.

    Second, "fundamentalism" was defined conceptually as conservative, authoritarian, theocratic, anti-free-speech notoriety, which may not satisfy the kind of rigor a true religion scholar might hope for, but for my purposes, was sufficient.

    Summary: once we separate the rating method (1) from the plotting step (2), any real debate might attack the former. For example, does my operationsl definition of fundamentalism actually measure my bare-bones, "notoriety" conceptual definition?

    I think it does, at least until further research can convincingly show otherwise.
    --more--

    ReplyDelete
  3. --contiued--
    Scientists consider two aspects of operational definitions of a construct such as fundamentalism.

    First, is the measurement (TRB mentions) reliable? Here we are 100% safe. If we go back and redo the measurement (number of search results), the numbers are going to be the same and nearly so, given new posts are added to TRB over time.

    Since there was no "statistical sampling", namely, the entire population of TRB posts was used for each member, there is nothing to normalize there. Each member has equal chance to attract attention of the TRB posters.

    Second, is the measurement valid? Does it really mean what we say it does? If a personality profile asks "Do you remember your dreams?" does a "Yes" answer contribute to measuring the concept of creativity? In the end, readers and other investigators will decide on this validity question for my pilot study. How are we doing so far? Well, no major issues have surfaced ... yet.

    In general, validity is addressed by using other operational definitions and seeing how they correlate. Cattell and other psychometricians developed personality tests by seeing if item A correlated with item B. If A was considered valid, then a correlation would enhance B's validity.

    So with validity, we focus on things like is Peter Khan two times more fundamentalist than Doug Martin and is Mr. Martin twice as fundamentalist (as I defined it conceptually) as any of the other past or present members?

    If this is really the question, then the ball is in the court of the investigative community to address it. A scholar may not use search results at all, and simply write a tome on the history of each of these individuals, and through sheer effort and insight, provide a rating of these persons based on historical data (their statements and actions), which could then be plugged into step (2) above of my research design. Or a more computer data-processing type might come up with a different, better operational definition and plot that.

    Re "pitfall" and "popularity", there is no such pitfall unless you define popularity as the extent to which TRB posters were outraged by statements and actions of particular persons. If you have another popularity definition, prove it. Show us that the totality of text about person X goes to how admired and beloved they were in TRB.

    Re "number of blue wearers", that would apply only if I had multiple sources, like, say, both TRB and alt.religion.bahai, in which case, yes, weighting by the total posts in each forum would clearly be required.

    Finally, the present validity question may arise only because I reported a result -- the ongoing collapse of fundamentalism in recent years, which may contradict the idea many may hold that findamentalists still have and will continue to have overwhelming control in the Baha'i community. Well, again, prove me wrong.

    ReplyDelete
  4. Dear Baquia. More feedback. You write: "You've simply dipped a thimble in a stream and taken out a small sample (those that mention a UHJ member). This is called statistical sampling."

    You mean the UHJ members are only a "sample" of all people mentioned in the posts? Well, not in my design. For step 1 (the ratings), I did a one-to-one mapping from one space to another. Namely, each UHJ member has a certain position in a multi-dimensional TRB space. Mathematically, a position is a point, which I reduced to a single number (the mention count) and mapped that into UHJ member space. No sampling anywhere here.

    Example, for a number of objects of all shapes, my research is only interested in the cubes (UHJ members in the space of all the objects). I determine to map each cube into another space -- a vector of all cubes and do this by noting the volume of each cube. I have mapped a point in one abstract vector space to another abstract vector space. This how a math guy would describe it. A common example is a 2D drawing of a 3D object, where points are uniquely mapped from the 3D to the 2D space. If the vertical deflection on a TV goes bad, you get a horizontal line where the original 3D objects are mapped into a 1D space.

    In my procedure, a single point in TRB space is mapped to a single point in UHJ member space. There is no sampling anywhere here. In the latter space, the axes are the several dozen members past and present, the coordinates of the single point are their ratings.

    Baquia writes: "Since having more discussion would mean a higher probability of the mention of high profile persons such as UHJ members, you need to normalize for the ebb and flow of the volume in the discussion."

    Since I used the total volume (population) as described, I've dismissed this point. But let's go with it and see what it might mean.

    It is asserting that mention of member A at time T1 has a different meaning with respect to the study's operational definition than a mention at time T2. And the meaning is said to be different in nature due to how many other posts there are in the respective time intervals around T1 and T2. Gosh. off hand, I don't see how that is pertinent to the operational definition used.

    Whether at T1 or T2, if James Keene had somehow thrown darts at Baha'i intellectuals, James Keene would be condemned in TRB. It would not matter who else is also being dragged over the coals. See? If some sort of scaling is proposed, it needs some justification, else such scaling itself could be questioned as an attempt to manipulate the raw data.

    BTW, James Keene is not mentioned even once in TRB; sorry, I just don't have a history of causing mayhem in the Baha'i community.

    Nobody has mentioned it yet, but please consider also that many posts will praise and defend person A and those are counted also. But I don't consider that to be an artifact, since almost all of those posts were in reaction to the original commotion arising from something person A did/said/wrote.

    To confirm, then, note that there is no praise of James Keene either; nobody needed to defend me because I didn't initiate anything anti-freedom/anti-intellectural, to provoke the freedom fighters in the bloody pages of TRB. All this goes to why the TRB post population may have been a very good choice indeed for my pilot study.

    ReplyDelete
  5. My apologies, I believe that I misunderstood your initial message. You didn't sum the number of mentions of each UHJ member along a time line (that is to say, X number of mentions of Douglas Martin in 1999, etc.) but instead have used a total sum of the number of their mention. I still don't fully understand exactly how/what you're calculating.

    Perhaps it would be useful for tiny brained readers like myself to be shown a very simple step by step guide as to what exactly you did and what you counted, etc.

    To return to the discussion at hand, which I believe is the question of whether the UHJ is 'fundamentalist' or not, may I present to you what happened at the 2007 US national convention as 'Exhibit A'?

    If you missed the proceedings, the US NSA presented their annual report as they normally do but did so in 2 parts. The second part was a shockingly honest portrayal of the state of the US Baha'i community.

    The UHJ sent a representative from the ITC - basically an enforcer - to force the NSA to nix this report and instead to supplement a letter from the UHJ in which they basically tell the US Baha'is to stay the (Ruhi) course.

    You can read the details of the report/letter here at BahaiRants.com

    If that doesn't strike you as an extremely 'fundamentalist' reaction, then I don't know what will.

    ReplyDelete
  6. Baquia asked for a "very simple step by step guide as to what exactly you did and what you counted, etc." Your wish is my command:
    1. I set up the historical election data shown in the Appendix of "Fundamentalism Bubble Burst". Please notice that each time point is a year of an UHJ election containing a list of nine names of new or continuing members.

    2. There are 25 such persons listed here in the order they first appear in the Appendix:

    Hugh Chance,1
    Hushmand Fatheazam,0
    Amoz Gibson,0
    Lutfu'llah Hakim,0
    David Hofman,49
    Borrah Kavelin,0
    Ali Nakhjavani,81
    Ian Semple,172
    Charles Wolcott,13
    David Ruhe,4
    Glenford Mitchell,74
    Peter Khan,1180
    Hooper Dunbar,170
    Adib Taherzadeh,268
    Farzam Arbab,187
    Douglas Martin,484
    Kiser Barnes,7
    Hartmut Grossmann,7
    Firaydoun Javaheri,0
    Paul Lample,85
    Payman Mohajer,0
    Gustavo Correa,0
    Shahriar Razavi,1
    Stephen Birkland,214
    Stephen Hall,213

    3. The fundamentalism rating shown above was the TRB result count for the member. Thus, the entire plot is based only on the election years and results and the 25 ratings shown above, plugged into the spreadsheet in the Appendix.

    4. Notice that anybody can, as I suggested, make their own ratings by any credible method desired. E.g., a scholar might consider the history of each of these members -- writings, speeches, activities, etc, and make an "informed" rating for each, and then plug those numbers into the Appendix spreadsheet.

    That's it. I mentioned the distribution is skewed. Take a look:

    range, number of members
    0-99, 17
    100-199, 3
    200-299, 3
    300-399, 0
    400-499, 1
    >500, 1
    Total, 25

    17 of 25, a sound majority are in the lowest interval; only 8 rate greater than 100.

    ReplyDelete