Community Forum : General Discussion Forum
Welcome Guest   
 
 Subject : Membership Data for the System Dynamics Society (over time).. 04/29/2019 05:29:05 AM 
Guido W. Reichert
Posts: 30
Location
Hello Everyone,

Currently Jean-Jacques has opened a discussion about the "future of the field" and John Sterman has recently published a much noted article about the "path forward".

That had me look up some data for the development of membership in the System Dynamics Society and as a practitioner of the art I would of course like to look at time series data.

An unfiltered search in the directory of members currently reveals 599 entries which is worldwide. I do have some difficulty in coming up with time series data - are there any provided by the SDS -- a pointer would be greatly appreciated? If not, why are time series data not provided?

Best regards,
Guido

PS: For a field that is utterly dedicated to show and explain the development of quantitative entities of interest over time it strikes me as rather odd how fed time series charts I can find on our pages. As an iconic observation I note, that on the infographic published by the SDS to highlight "60 years of progress" one cannot find a single time series as a reference mode of behavior indicating progress.
Last Edited On: 04/29/2019 05:48:38 AM By Guido W. Reichert
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 04/29/2019 01:25:02 PM 
Martin Schaffernicht
Posts: 41
Location
Hi Guido,

great question! As far as I understand, the answer to the question "how many members do we have" depends on "when exactly?".

During 2019, individuals who have been members during 2018 can renew at any moment in time,a s well as new members can subscribe at any time. This means that some sign up or renew in January, others in April, others at the conference... you get the point.

When the PC reports state that "in the year X, we had Y members", this refers to how many members were on board at the end of the year X. Now we are 1/3 into the year 2019, so mane people have not subscribed (or renewed) yet.

Of course, this also implies that we do not know how many people are leaving until we detect that the did not come back!

On top of this, there is a subgroup of individuals who have been members until 2 or 3 or more years ago, did not renew for some time, and now they "come back". If the criteria for "new member" is "was not a member last year", then these "returning former members" are marked as "new members", which they are not really.

So: we are currently revising our way to look at members, so we can better monitor what is going on in the future. This also means that the data from the past will become difficult to compare because of this modification.

As for making membership development openly visible, I believe that for the past, you can find that information on a yearly base in the reports from the PC meetings.

If the question is to support on-line real-time access, I think I'd like to think about a couple of aspects first. We would need to decide the granularity of time. Would it be by year, by quarter, by month, by day? Then, how would we make sure to avoid incorrect inferences (due to the progressive filling up of the membership stock)? And: would such information have a self-reinforcing effect (think of the "likes" in social media)?

Saludos,
Martin
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/01/2019 08:06:26 AM 
Guido W. Reichert
Posts: 30
Location
Hello Martin,

Thank you for your answer which is confusing: How can counting being an active member be difficult? It is after all a stock and we teach that the number in the stock is clearly defined for any point in time even if it is not observed or reported.

From what you describe we can come up with a simple structure:

oJAFX.png

What you are saying is that the number of members is oscillating around some "fair reported value" (e.g. it is now 599 while in fact reporting >1000 may be more appropriate). SDS membership needs to be actively renewed and telecom providers (and many other organizations) are avoiding that kind of uncertain fluctuation by nudging their members into automatic renewal, e.g. many membership contracts or subscriptions will automatically be prolonged for another year unless terminated. Couldn't the SDS do so too?

As far as I understand it the membership follows the calendar year, i.e. if I renew my membership in May it will still terminate on December 31 that year?

So while the number of members could be reported any second (or at least day) that number would fluctuate too much. So for reporting the members we would have to use a filter (a moving average or smooth) and then indeed we would have to decide upon the averaging time.

Note, how such simple processes already indicate the difficulties of System Dynamics to model such rather operational realities:
  • The members who return will likely not have the same character as those who enter as newcomers (so we need a different stock maybe).
  • The stock "Former Members - may reenter" is likely a delay of some higher order and it is not always trivial to model a leak or bi-flow from such a delay in a consistent manner.
  • The probability of reentering will probably decrease as residence time in the stock of former members increases. Since all members in that stock have the same average residence time we may start to introduce age cohorts for former members.
  • Some of these condsiderations also apply to modeling the members themselves (e.g. residence time considerations).

There is a reason why modeling individual behavior instead of group behavior has some appeal. ;-)

Best regards,
Guido
Last Edited On: 05/01/2019 08:11:22 AM By Guido W. Reichert
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/01/2019 09:55:19 AM 
Jean-Jacques Lauble
Posts: 42
Location
Hi Guido and Martin
I do not make things more complex than they are.
Finally, I need to know how many members were active each year. It is the number of members at the end of the year. Eventually the same kind of curve that is built by demographers that indicates the proportion of members having a certain age and what would be interesting is to see how that curve changed over the years and a survey of the reasons people become members for the first time and why they quit. That would be already enough and is not so complicated to realize.
Best JJ
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/01/2019 10:26:05 AM 
Guido W. Reichert
Posts: 30
Location
Hi JJ,

You do not really find the structure I posted "complicated", do you? ;-) You are of course right, that the end of year value should reveal the maximum count for members being active within the year.

The complication still exists: If you are interested in using the member directory (after all societies are for socializing) you will be some hundreds of names short for some time each year, which I find unfortunate.

You yourself have discovered, that not being an active member "right now" will not allow to post in this forum "right now". Maybe that explains the very, very low participation here?

Best wait for socializing and posting until December?

Best Guido
Last Edited On: 05/01/2019 10:28:17 AM By Guido W. Reichert
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/01/2019 12:33:27 PM 
Guido W. Reichert
Posts: 30
Location
Thanks to Martin's adivce to look at the Policy Council Reports we find historic data reaching as far back as 1984 in the Annual Report on Home Office Operations for Fiscal Year 2016.

Adding the most recent value for 2017 from another report I came up with this plot:

dZiQk.png

Running this by Mathematica's FindFormula we find three distinct periods according to average annual growth rates:

1985 - 1997 ~30 members per year
1998 - 2007 ~25 members per year
2008 - 2017 ~ 9 members per year

Another interesting fact to note is the development of the ratio regular/students-subsidized this has come from a value around 5 to 1 down to 2.5 to 1.

Best regards,
Guido
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/01/2019 01:16:58 PM 
Jean-Jacques Lauble
Posts: 42
Location
Hi Guido

I can see that you have not lost your sense of humour, which is a blessing in such a serious place; of course your model is simple enough, but I am interested first by basic information and this is what I need and i wold like to have the demographic curve i was talking about, this all for the moment and eventually a non directive survey of the reasons of becoming a member and leaving membership. This is all and I do not have it; Remember that i am an end user and i do not mind if the information comes from a model or from a Crystal ball as long as I get it.
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/02/2019 06:42:40 AM 
Guido W. Reichert
Posts: 30
Location
Hi JJ,

Certainly information can come from lots of sources, but a model does not need be be a source (my model above does not run and simulate anything...). Instead, a model - like the one given - helps to tell why we need information and what the information means.

For example, why having the member count from December 31 in each year is the correct number to use to model membership dues (probably), for the reasons given (e.g. participation in the forum) the number of "active members" might better be given by a moving average or smooth. Again, this imo becomes much clearer from looking at the model?

BTW: I find this example a good case in point for using SFD first and CLD only later (if ever :) ).
Last Edited On: 05/02/2019 06:45:45 AM By Guido W. Reichert
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/06/2019 06:44:16 AM 
Guido W. Reichert
Posts: 30
Location
While I am also curious for the additional data that Jean-Jacques has asked for, we may in the mean time complement the membership data with a look at publications.

As a naive proxy I have simply gone to Google Scholar and searched for publications given an exact search string ("System Dynamics") and a filter restricting results to a given year.

Below you will find the data for the years from 1984 to 2018.

BMbxf.png
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/15/2019 04:28:01 PM 
Dr. Mark Nelson
Posts: 5
Location
Hi Guido,

I have a draft report with 2018 numbers. I have added some additional data points and series. If there is a particular series you would like to see in there, I can attempt to determine if we have the historical data. I am primarily waiting for our annual audit information to be returned to finalize the current version.

On the membership numbers, what MemberClicks shows externally is somewhat misleading on numbers. For example, some members do not want their information public to the directory, so an unfiltered search will not return the whole set.

That aside, both you and Martin are correct -- in my experience working in associations, the method of calculating and reporting membership here is quite complicated. It took me quite a bit of time to be able to replicate prior year numbers so that I could calculate 2018 with confidence in a comparable way. Part of that results from what I would view as a complex dues structure. Another part is more unique to the past practices and internal policies for the Society and its renewal cycle. I am not opining on whether the prior approach was right or wrong, only that it is complicated and non-standard. The approach was built with some logic and rationale that made sense within that context. Complications aside, what I have concluded at this stage is that historically our "official" membership number for the year is set around Sept 30, when the next year's renewal cycle starts and around that point in time we have a more solid membership number for a given year.

While we are discussing time series, prior to 2008, the Society demonstrated fairly consistent membership growth. In 2007 the membership was 1106 members. Since 2007 the membership has roughly been within plus or minus 50 members of that figure with the exception of two years (2008 and 2017) where the numbers fell below and above that range respectively. In 2008 the new dues framework was released, producing what one might have expected as a drop in that year. 2017 was also an outlier because of the 60th anniversary of the field and other events. What is interesting to me is that a dues structure that was changed specifically to increase "inclusivity and affordability" within the Society coincides with a leveling and stagnation of membership growth that has lasted more than a decade. Looking at the subdata, there is also no significant evidence of increased membership among the originally targeted segment. Certainly other factors and forces are in play, but the timing of the plateau and stagnation, and its duration are notable.
Mark R. Nelson, Ph.D., MBA, CAE
Executive Director
System Dynamics Society
 Subject : Re:Re:Membership Data for the System Dynamics Society (over time).. 05/15/2019 04:50:03 PM 
Dr. Mark Nelson
Posts: 5
Location
Jean-Jacques Lauble Wrote on 05/01/2019 08:55:19 AM:
I need to know how many members were active each year. It is the number of members at the end of the year. Eventually the same kind of curve that is built by demographers that indicates the proportion of members having a certain age and what would be interesting is to see how that curve changed over the years and a survey of the reasons people become members for the first time and why they quit. That would be already enough and is not so complicated to realize.
Best JJ


I can help you with the annual numbers. I believe Guido provided a graph with the figures, but I have those as well. As he noted, the data has been reported annually by Roberta for years in her end of year report to the Policy Council. That report typically comes out around mid-year for the following year as we wait for the results of the annual audit.

I have not seen data from prior member surveys and I do not see that information in the files I inherited. We did conduct a member survey in the fall, shortly after I arrived. I am happy to share some of the results from that survey. We are about to conduct a survey among lapsed members (i.e., those who quit) from the last two years. That survey should go out soon. I personally have also conducted nearly 500 member interviews at this point with current and past members across a spectrum of demographics. I could make some educated guesses as to why some of the trends exist.

As you might expect with any small nonprofit, there are many areas within the Society that need attention. Providing the types of data you mentioned and making it easier for members to find is just one of those. We do not have the staff or resources of larger Societies, and some items have, by necessity, taken priority. I also find the learning curve for the Society as an organization is also quite a bit steeper than a number of other associations. I offer those points not as intended excuses, but in an effort to be honest and transparent when I say that from my perspective the information you seek is not being held back intentionally, but that as a very small organization in the middle of several large transitions there is limited bandwidth to do everything we could and should.

I am happy to try to schedule a time to talk with you more directly and discuss what data might be available. With the annual conference approaching in less than 100 days, my calendar is filling quite fast. If we plan ahead, then I am happy to try to accommodate your availability.
Mark R. Nelson, Ph.D., MBA, CAE
Executive Director
System Dynamics Society
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/16/2019 12:17:29 PM 
Martin Schaffernicht
Posts: 41
Location
Hi,
after looking at Guido's "overshoot & collapse" reminding graph of publications (I think there is an article about that theme in the SDR, but I've not read it yet), a quick comment on "time".

As Mark has written, usually the SDS report will state membership figures on a year by year base. That may be sufficient for what the usual audience of that report needs. Mark also writes about the membership campaign, which - of course - could not operate based on year-by-year data. It may not be "by second", but the granularity of time may be month or even week. The intended purpose defines what is useful, right?

BTW, I thought that Guido's membership S&F-Diagram nicely resumes some important aspects (even though I'd add an inflow to the "Potential members" stock as well as an outflow towards "alternative associations"). S

till I believe we could use a slightly less aggregated representation of the situation. Even without a formal model, we are confident that the membership is not as homogeneous as "one stock" suggests.

There is a subset of members who are "experts" (I'm not a fan of "high quality research", but if "quality" means "free of beginner mistakes and open to critique & improvement", I can live with that), while others try to become more "proficient" or to grow beyond the "novice" stage of competence development. Still others are more "general interest", rather potential users or clients than potential modelers. The individuals in these sub-groups have important differences, and it probably affects their decisions to stay or leave.

At the same time, it is known that association members can be more or less engaged, meaning willing to give (volunteers), wanting to take ("value-seekers") or just "rank and file".

Add to this that sometimes we talk about the "field", and at other times we talk about the "SDS". Members of the SDS are members of the "field", but not all members of the "field" are members of the SDS. As the SDS, we have given ourselves the mission to support the development of the "field", and as the PC of the SDS, we must take care of the SDS, knowing that the SDS is not the only entity in that "field".

Nothing makes it impossible that a SDS member who is a SD expert feels "engaged" with the field but is only a "value seeker" in the SDS, for instance.

So I do think we need to decompose the single "Members" stock, and this is currently being done by means of an ongoing dialog (rather than formal model development) - but I am confident that the personal experience of the participants allows to make substantial progress without paying the price of developing a detailed formal model.

Greetings,
Martin

So if we assume four competence stages and three levels of engagement, then we will want to know how and why individuals move across the resulting states. This is a current and ongoing discussion
 Subject : Re:Membership Data for the System Dynamics Society (over time).. 05/18/2019 02:49:55 AM 
Jean-Jacques Lauble
Posts: 42
Location
Hi Martin
Could you precise the four stages of competence and the objective and preferably factual criteria they are based on?
Best
JJ
 
# of Topics per Page