2 May, 2009 - 15:15 — admin

From Frank Le Fever.

Because numerous momentary events influence sample data, summary data points from month to month or season to season zigzag up and down in a way that makes "eyeballing" an unreliable way to discern trends. Each little upward or downward move invites ad hoc "explanations" which may or may not be valid, and can be over-rated by people not sophisticated about the role of non-systematic ("random") sampling error can contribute to departures from an overall, systematic, lawful trend in the data.

The graphs uploaded here show data points, season by season across the years, but without the usual connecting lines which provide a fallacious appearance of continuity. Instead, straight lines are drawn, not fitted by eye but calculated to balance momentary upward and downward departures from the general trend.

Both statistics (AQH and cume) have a statistically reliable downward trend from Fall 2003 through Spring 2008. (The decline in AQH seems steeper than that for cume, but I have not tested the difference statistically.)

**COMMENTS ON METHOD**:

(1) The slope shown for cume is with the unusually high Summer 2004 score omitted, so the decline is not an artifact of starting from an unusually high data point. The deliberate omission of an "outlier" is fairly common. I have calculated both ways, with and without the Summer 2004 cume data, and the difference between the results is trivial). I retained the all the 2004 AQH samples because none departed markedly from the general trend line.

(2) I managed to obtain the missing data for Winter 2007 (which Nathan did not have for his report to the PNB), and used it for the AQH graph, but too late to use in the cume graph (I used average cume as an estimate for the missing value), so I wil re-do it including the Winter 2007 cume value.

- Login to post comments

## Comments

## Dishonest & Disingenuous Presentation of WBAI Arbitron Data

Conveniently for the plotter of the data, all the known data is not used to generate this graph. For example the 8 years prior to 2003 are omitted. When you factor that data in, information that has previously been made public, the overall trend line is up, not down. The data for summer 2004 probably was so high the graph would have looked ridiculous revealing that 2003-4 was an anomalous period, indeed it was the height of the invasion of Iraq. Wars have usually resulted in an increase in listeners for Pacifica stations.

This kind of data crimping is unacceptable. It is an example of how easily graphical presentation of information can be manipulated to lay the ground work for unsuspecting viewers to accept what would otherwise be considered unreasonable conclusions and assertions.

There is no wonder why many people in the WBAI community feel that WBAI is being hijacked. Astute listeners should wonder, why they lie if truth is on their side? I don't know who generated the graph, though Mr. LaFever posted it.

## JUC's combination of incompetence and malevolence

I have belatedly found this crude hatchet job posted by JUC's Minister of Disinformation over two months ago.

It illustrates well the combination of incompetence and malevolence that has characterized WBAI's performance during the reign of Program Director Bernard White and his mutual-protection society, "Justice & Unity".

For example, this wind-up toy (who is obviously on auto-pilot transcribing Bob Lederer's agitprop piece) apparently does not realize that his lie is found out within 10 or 20 seconds by anybody with wit to LOOK AT THE GRAPH(S) and the will to do so. He foolishly says "...The data for summer 2004 probably was so high the graph would have looked ridiculous revealing that 2003-4 was an anomalous period..."

I invite more diligent readers to take a moment now and LOOK at the graphs (both for CUME and for AQH): obviously I have included the summer 2004 data point, and because Dr. LeFever [sic: Ph.D., Experimental Psychology, NYU], having taught statistics, is wise enough to choose appropriate parameters for his graphs, so neither graph looks "ridiculous".

Indeed, I even COMMENT on the problem of dealing with the 2004 data: "...The slope shown for cume is with the unusually high Summer 2004 score omitted, so the decline is not an artifact of starting from an unusually high data point. The deliberate omission of an "outlier" is fairly common. I have calculated both ways, with and without the Summer 2004 cume data, and the difference between the results is trivial). I retained all the 2004 AQH samples because none departed markedly from the general trend line..."

Riley piously intones: "...This kind of data crimping is unacceptable."

Rather ironic coming from a "videographer" who routinely "crimps" videos of LSB meetings to show JUC in the best possible light and to misrepresent what JUC does to obstruct transaction of business!

But even more relevant to the specific case of misrepresentation of trends in Arbitron statistics, he asserts: "...Conveniently for the plotter of the data, all the known data is not used to generate this graph. For example the 8 years prior to 2003 are omitted. When you factor that data in, information that has previously been made public, the overall trend line is up, not down..."

Presumably he alludes to the unfortunate graphic exercise promulgated by ex-GM Tony Riddle on Oct. 2, 2008.

Talk about "crimping" data! To begin with, he omits summer and winter data, constructing his graph entirely on the basis of spring and fall data.

Moreover, he omits more than two years of data (missing between spring 2001 and fall 2003) with NO indication of a break on the X-axis. If there IS an upward trend between 1998 and 2008, the effect will be to exagerate the visual impression.

I say "IF" there is an upward trend, because during the years in which our graphs overlap, his omitting half the data manages to create an impression of an upward trend (from 2005 to 2008) WHICH IS CONTRADICTED WHEN COMPLETE DATA ARE USED.

What else? Well, instead of CALCULATING trend lines [I explain the need for this in the notes to my graphs], he "eyeballs" the data and draws a line CONVENIENTLY between two isolated data points that catch his eye.

I have taken a preliminary look at the trend from 1998 onwards (I would not make a serious formal analysis without those missing data points), and if one accepts the data that Tony shows, there was a small upward trend from 1998 to 2001 which IF CONTINUED would have projected a CUME in 2008 far above what the CUME actually was.

Indeed, not only did Bernard's operation fail to retain even a portion of those new listeners during the war years, it failed to continue the less spectacular but fairly consistent growth in listeners WBAI had before he took over.

One can discuss the relevance of looking at a 10-year trend and accept or reject such an approach on various grounds. Given the many changes in our signal area and society as a whole since 1998, I would argue that its utility is limited. It makes more sense, I think to focus on trends during the "Bernard era". Let me emphasize, however, that even for THIS period, during which our graphs overlap, Tony's data representation and "analysis" are absurdly flawed.

Moreover, there are other ways to look at the downward trend that COMPLETE Arbitron data shows (yes, a downward trend, NOT a rebound or upward trend) since 2005. Besides the evidence of COMPLETE listening data [take another look at my two graphs and see how they differ from Tony's EVEN for those years], there is the evidence of data which can reasonably be expected to have at least SOME positive correlation with numbers of listeners:

[1] number of members (losing nearly 1000 per year during this period, although at least two JUC candidates for the LSB in 2007 got away with saying membership was growing and we had "28,000" members when in fact we had scarcely more than 1/2 that many).

[2] number of pledges [look at decline in AQH and decline in number of pledges side by side at http://www.takeforwardwbai.org/downwardtrend.html (note especially DECLINE between 2006 and 2008)]

[3] actual dollars DONATED (not just "pledged") per day during these years [Financial Report, PNB meeting July 2009 (if no one heard it, that's not surprising considering how hard Bob Lederer and other JUC mobsters tried to disrupt the meeting and how LOUD their efforts were -- as well as how physically obstructive and disruptive) -- I'll come back and post a link to the relevant graph.]

Talk about hijacking! JUC tried to hijack the meeting of people given the responsibility for keeping WBAI and Pacifica alive, given that responsibility by elected representatives of listeners and staff. An important meeting HIJACKED by people who prefer mob rule and disinformation. Disinformation on the same level as Riley's knowing or unknowing distortions, or even worse.