1. Detailed results of a large survey of attitudes to a proposed cable car on Mt Wellington have been claimed to have settled the question of in-principle support for the development once and for all.
2. They do not do this, because the survey was conducted using opt-in survey methods, which are not statistically reliable irrespective of sample size.
3. Additionally, the use of a preamble stressing a (probably unrealistically) favourable view of the project is likely to have affected the results.
4. Lower support rates in certain inner-city suburbs are probably not just legacy effects from previous cable-car proposals but probably also reflect innate aspects of the cable car debate.
UPDATE: Adrian Bold has responded to this piece. My response to his response appears at the bottom!
(See also the later piece from September 2014: Polling On The Mt Wellington Cable Car Proposal)
|The mountain from not far from my place.|
Proposals to build a cable car on Mount Wellington (or as it may soon be known, Kunanyi-Mount Wellington) have been a stock in trade of Tasmanian conservation vs development debates over the years. Such proposals most recently surfaced in 1988 and 1993-4, when they were significant Hobart City Council election issues. A revised push to gain approval for a cable car project has been running for the last few years. Its spearhead is young entrepreneur Adrian Bold. In 2011 Bold was widely expected to win a seat on Hobart City Council after a very accomplished online pre-campaign, until he blew it by failing to cause his nomination to be received in time.
Public opinion on the concept of a cable car has seldom been measured and I am not aware of it recently having been accurately polled. Results of "surveys" of public opinion have been released from time to time, but these are usually the dreaded politician-conducted opt-in "elector surveys", which are utterly useless. Measures of opinion like numbers of submissions or petition signatures prove little given that only the most motivated few percent of the population are likely to bother expressing their views in such ways.
One sometimes encounters references to a 2009 EMRS poll claimed to have shown very strong support for a cable car. (Adrian Bold on Twitter claims 78%). I have never been able to find the original results and questions of this poll, only second-hand references to it. According to Mr Bold it was commissioned by a previous proponent.
Now, Adrian Bold has released partial results of a very detailed survey indeed. The results in brief are outlined on the MWCC website here and there is a very detailed PDF report (192 pages long!) here. Many results were considered "commercial in confidence" and not released.
The survey was an opt-in survey filled out by visitors (including me!) to Bold's website. The survey was open for nine weeks from late January this year, and was filled out 2,219 times.
Bold has made many claims about the merits of this survey, and some of them are extremely dubious.
Does The Sample Size Establish Bold's Survey Is Accurate?
No. Bold has claimed "without this robust sample size achieved, locals would otherwise likely continue to speculate over public sentiment", but in fact the sample size provides no reason for anyone to stop speculating. The reason for this is that small sample size is but one of many factors that can impact upon the reliability of an opinion survey, and while too small a sample size means that a survey is unreliable, a large sample size does not demonstrate that a survey can be trusted.
The main limitation of Bold's survey, one that cannot be overcome by pointing to the sample size, is that it is an opt-in poll. The use of opt-in methods means that the survey develops biases based on who is aware of the survey, who has the time to complete it, and who is motivated to consider the survey to be worth completing. Any form of opt-in polling skews received opinion in favour of the views of those who are interested in an issue and motivated concerning it. The same is also true of submissions processes, meaning that the numbers of submissions for or against an issue prove nothing.
That is not to say that the use of an opt-in survey method proves the results are significantly wrong. For instance, the survey greatly under-represents voters over 70 years old, because such voters are not heavy internet users on average, and would be much less likely to have been aware of the survey or to have felt comfortable filling it out. But if the views of voters in that age group are similar to those of voters overall, that doesn't necessarily matter.
A further concern with this mode of online surveying is that preventing multiple responses, if someone happened to be determined to submit them, is quite difficult. The MWCC website does state that the poll is "Limited to 1 response per computer and mobile device - multiple responses from an identical IP address are disregarded" but there are many ways a person who wants to could still cast multiple votes.
There are other issues with the accuracy of Bold's survey design, on which I'll comment further below.
Do The Political Leaning Questions Establish Bold's Survey Is Accurate?
The results of Bold's question "What is your typical political persuasion? (optional)" were as follows:
Swinging voter 30.5%
Bold writes "The graphed result suggests a trending pattern similar to current State Election Polls (*EMRS, survey sample size of 1,000) available here.
This data simply provides us with further confidence our survey has been answered by the broadest representation of society, and not 'stacked' by any particular affiliation or belief."
Bold makes much of the 71% "fair and strong support" level among Greens voters, suggesting that his proposal enjoys clear tripartisan support. (Indeed, Independent voters, at 55%, recorded the lowest "fair and strong support" level according to this twitpic.)
Firstly, the results don't suggest a "trending pattern" of anything, because a trending pattern is determined from a time series, not the results of a single survey. But that's just a minor quibble. Another minor quibble is linking to a Wikipedia page that lists results of EMRS polls using figures in a format that EMRS still publishes but no longer uses as its headline result.
Considering the most recent EMRS headline results (55 Lib, 23 Labor, 18 Green, 4 Ind) then if it is assumed the bulk of so-called "swinging voters" swing between Labor and Liberal (and currently support the Liberals) then it is possible to argue that the claimed political persuasion results are similar to EMRS statewide figures. On the other hand, given that the "swinging voters" are less supportive of the project than both Labor and Liberal voters, there may be some Green-leaning voters in the swinging voter mix anyway. But so what? Whether or not the poll's breakdown closely resembles the EMRS state vote is irrelevant, given that EMRS polls statewide, while survey responses would have been concentrated in Hobart and surrounds. (My suspicion is that Green voters are underrepresented in the survey given that, but perhaps not massively.)
If those filling out the survey were politically honest, then the results do show that the survey was answered by people with a rather broad range of political views, including substantial numbers from all parties' support bases. No more than that.
However, even that would still prove only that no party's support base was universally opposed to the cable car. That 71% of those claiming to be Greens supporters who responded said that they supported the concept in no way means that this is true of Greens supporters overall. It may just mean that those Greens supporters who liked the proposal were more likely to feel motivated to fill in the survey (especially to send a message to their party!), while those who were opposed to it were more likely to think that the survey was a waste of time, a push-poll or a stunt.
Another Problem: The Preamble
Suppose I cut Mr Bold's survey a lot of slack and assume all of the following:
- that there was no or negligible multiple voting
- that there was no or negligible dishonest voting
- that the survey respondents accurately represent the way those in the general community would answer if asked the same questions
...does it then follow that the public in Hobart are very strongly in support of a cable car?
Again, no. The reason is that the the survey included an introductory statement that was extremely capable of loading the response. The introductory statement made the following contestable claims:
* "The eco-tourism development would be entirely funded by the private sector [..]"
While this may well be Mr Bold's intention (and he does tend to push political views compatible with a purely free-market approach) experience with similar development proposals is that there is invariably public expenditure during the assessment process. Whether the proposal would, if approved and built, then run without further public funds injections is a matter of conjecture.
* "This proposal would increase visitor demand to the pinnacle and therefore includes an alternative, (not a replacement) form of transport to the pinnacle to avoid expansion of carparking at the pinnacle and reduce traffic volume on Pinnacle Road."
Whether the Pinnacle Road would be closed or partially closed to the public at some future time as a cost-saving measure in the event of the successful development of a cable car is a matter that is outside the developer's control.
Various other statements about what would be included were made, including a visitor centre at the Pinnacle that would aim to be invisible from the summit carpark (a very difficult thing to achieve, in my view, without substantial disturbance or placing it well away from the carpark and hence limiting its access to the vehicle traffic market.)
The general problem with an introductory statement of this kind is that all that is being measured on the basis of it is public response to the most favorable possible (or perhaps the most favorable impossible) conception of the development. A similar survey that started with an introductory statement outlining all the potential problems with the development would receive a much more negative result, but would it then prove that a massive percentage of voters had reservations about the idea?
While it might be argued that the whole idea is about finding the levels of "in principle" support, I think there's a slight difference between an in-principle positive response (that might be subject to a few reservations) and a response that is essentially against the concept, but that can be turned around provided all underlying concerns are overcome. The wording used might well tweak a supportive response out of the latter category and seems designed to do exactly that.
Asking people for a response that is predicated on the assumption that all the commonly held fears are wrong is likely to result in "opposed" responses being dominated by those who are utterly opposed to a cable car on Mt Wellington, absolutely no matter what the project's impacts, economics or logistics. Even if the finding that only 20% or so were utterly opposed to the concept no matter what was valid for the community at large, I'm not sure it would be all that interesting.
Is the MWCC survey a push-poll?
Not as such, but it's a related genre of suspect opinion surveying which seems to have no established name, but which I call a skew-poll.
The term push-poll is widely overused. A push-poll is a method of lobbying the public, in the guise of a poll, which actually seeks to convince the public of a claim of fact (usually a negative and false one about an opponent). The results are generally of no interest to the push-poller.
What I call skew-polling, on the other hand, doesn't necessarily aim to alter the respondent's views about a proposal. Rather, it uses wording to prod the respondent into responding in a favorable manner so that results can then be reported as favorable in the press and in lobbying.
Skew-polling need not be intentional and often results simply from the person conducting a survey being unaware of how deficient their survey methods are and how distorted the results may become.
When it is intentional, there can be three reasons for doing it:
1. Bandwagon effect: some people may be convinced to support an idea if they think most people already support it.
2. Influencing politicians: since some politicians are clueless enough to believe their own "elector surveys", they may believe other badly designed polling efforts too, and shift their views in the belief they will be rewarded with votes.
3. Getting media publicity: a startling poll result can serve as a hook for more general publicity about a group's views.
Ironically, in Tasmania this kind of "polling" is most often seen from environmentalists (many "polls" on the Bell Bay pulp mill were of this ilk.)
It should be noted that not everything in Bold's survey design is bad. For instance:
"Mr Bold said the survey was deliberately commissioned during summer to avoid any spike in local emotion that historically peaks with the frustration of unreliable access and demand during the snowfall season."
Legacies of 1993
An interesting claim in Bold's media release is this:
“The unfortunate truth here is that some inner suburbs which carry legacy concerns from the ghastly 1993 proposal were more likely to oppose our project."
He's referring to South Hobart recording the lowest support level, with Mt Nelson, West Hobart, Mt Stuart, Glebe, Dynnyrne and Fern Tree all listed as also having low support levels.
I'd suggest there's more to it than just memories of past campaigns. These are the suburbs where people feel most closely "tied" to the mountain in a psychological sense and are hence cautious about development impacting on their immediate and constant view of it. They are also the suburbs with high Green and Wilkie votes in recent elections. I connect this to the low support level from Independent voters via the assumption that many ultra-greens no longer call themselves Greens supporters, since they think the Greens have sold out. Although Wilkie is a lighter shade of green than The Greens, he is seen by such voters as more authentic. Conversely, the only suburbs with close ties to the mountain and especially high levels of support, Battery Point and Tolmans Hill, are wealthy suburbs with high Liberal votes.
In the case of South Hobart, my own suburb of residence, there is more, because South Hobart is seen as highly likely to suffer adverse traffic impacts and disruption from a successful cable car construction. For instance, Bold says "The proposal's route is yet to be determined however options exist in South Hobart, West Hobart or Lenah Valley." A cable car leaving from near McRobies Gully would be very likely to create traffic issues along Davey and/or Macquarie Streets, which are already streets with serious traffic issues created by the layout of the city.
It would be remiss of me not to mention I have some ancient history with this concept! In the late 1980s I was strongly opposed to the version of the cable car being proposed at the time, to the extent of letterboxing against it and also, in the height of rebellion, wearing a No Cable Car badge to school on one of the last days of the year until I was asked to remove it.
My stance then was probably symptomatic of a general conservationist mindset at that time (eg I was also a strong supporter of the then Green Independents in the 1989-92 term of parliament). I suspect there would have been many errors in my early views of the likely impact on the natural (as opposed to visual) environment of a cable car project, and also agree that what little direct environmental impact such cable car schemes would have had will be even less in the current proposal - provided its base site manages to avoid the numerous and in cases very genuine threatened species issues in the Hobart foothills.
There are many views that I held then that I don't hold now, but for all that, I'm still a bit suspicious of the concept, and the main reason for my suspicion is that development proposals in Tasmania seem to often fall into two broad classes.
The first class is the class that is the subject of divisive political debate and that is championed prominently by one or both major parties. Developments in this category tend to have in common these three things: (i) a long period of acrimonious public campaigning and polarisation (ii) a lot of public money being spent on assessment processes and legal and other appeal processes (iii) the development doesn't happen. The Bell Bay pulp mill and Ralphs Bay canal estate are the most recent examples of the genre.
The second class is the class of tourism success stories that happen quietly, without public division and that are usually roaring successes before anyone has really noticed. Any proposal for a cable car on Mt Wellington will be an example of the first class. I'm particularly keen that we avoid a repeat of the two above-mentioned fiascos, which between them cost millions of taxpayer dollars for what was, in the end, nothing. Even cable cars have past history here, with the Groom Liberal government wasting public money on a viability study. Nothing Adrian Bold has so far said has provided me with much confidence that the approvals process for a cable car will be anything but more public money down the drain.
What we really need, before major contentious projects like the Gunns mill, Ralphs Bay and the cable car are evaluated in Tasmania again, is major reform to the approach for evaluating them, to eliminate the current approach in which both developers and objectors benefit from the public purse, but often both end up dissatisfied, and in the long run all sides lose. In principle at least, I'd like to see something like the following combination of aspects for major and divisive developments:
1. Streamlining to reduce unnecessary costs and delays in approval processes and reduce potential for activist obstruction. This would include a review of existing barriers to development to identify and remove those that are not well justified.
2. User-pays assessment. Developer pays cost of all public assessment processes related to the assessment of their proposal. (I'd be open perhaps to some of these being returned if the development is approved, after it has been running successfully for a while and contributed to the state's economy.)
3. Objector-pays appeals. Community groups taking out appeals or court cases against development approvals foot the whole cost if their appeal fails, as in many other legal processes. If their appeal succeeds, developer foots the cost. If groups lodging appeals cannot demonstrate they can afford the costs of defeat, they can't appeal.
The Survey Overall
As usual with this sort of surveying, the problem isn't so much the data as the excessive claims made on the survey's behalf. It might be that a correctly designed opinion poll would return vaguely similar results, but we don't know that until it happens.
The survey doesn't end the debate about public support levels, it doesn't prove overwhelming support or "public licence", and it doesn't prove Green politicians are on the wrong track on the issue in the eyes of their constituents.
Above all, it doesn't "indisputably" show that the "the question of whether or not the public support our concept in principle has been answered", and in proof that it does not, I've written this article disputing it. To claim that an opt-in online poll of any nature can indisputably demonstrate anything about a broader population is either remarkably headstrong or clueless, if not both, and I can do no more than recognise the puffery of this claim by awarding Mr Bold this site's inaugural Porcupine Fish.
|Porcupine Fish Award for Ultra-Fishy Polling (image credit)|
Reply to MWCC response
Adrian Bold of the Mt Wellington Cable Car Company has now responded to this piece in an article posted here. This follows an in-person chat with Mr Bold which I agreed to attend on the basis that he was going to show me the 2009 commissioned EMRS polling, which he forgot. He did however indicate that it had a similar preamble to his own survey, which means it was fatally flawed as a measure of overall genuine opinion regarding the cable car.
At no point did Mr Bold indicate that this meeting was an on-the-record interview, and some of his accounts of what I said in this meeting are either not the full story or else taken out of context. More on those later. Bold seems to have a bit of form in this regard, recently trying to embarrass Rob Valentine MLC (also on this subject) on Twitter with claims about an in-person conversation that Valentine rejected and Bold has failed to substantiate.
At the top of Bold's response is the following:
"Dr Bonham’s focus on the survey however is clearly misguided with [sic] what the primary purpose of the survey was, or has chosen to ignore this. Unfortunately, this lack of acknowledge [sic] has led Dr Bonham’s critique on a journey of assumptions, which, if true would have provided a perfectly valid reason to have led him to ridicule the survey almost outright."
All this is nonsense. The basis for my comments in the article was that even if the MWCC survey was not intended as a representative opinion poll, Bold's public comments following the release of the findings treated it as if it was. Whether the intention of Bold's survey was purely to gather data for qualitative market research, or a combination of market research and attempting to demonstrate support for his concept, is not my concern. My concern is that he then used his survey to attempt to make claims about the state of general public opinion, that were demonstrably unsound. If your survey isn't really a poll, but you make poll-sounding claims that it has proven things about public sentiment, then you are representing it as a poll, and I will treat it accordingly. Don't like this? Then don't make the claims in the first place.
My responses below are grouped according to Bold's own headings.
"Opt-in Methodology" [sic]
(Pedantic note: lots of people, especially many scientists, use the word "methodology" needlessly when what they actually mean is "method"; it doesn't assist in communicating with the public and is one of my pet gripes. "Method" is how you do something; "methodology" is the broad study of approaches to method, or a broad philosophical position about method.)
Mr Bold claims that I told him that a form of surveying akin to "those tele-marketing calls we seem to attract around tea-time?" is "the only avenue available for a true ‘random-sample’ survey". Tele-marketing and phone polling are not the same things, and our discussion did not actually mention tele-marketing at all. It's true that he asked me if there is anything else besides telephone surveying as a way of truly scientific random polling and that I initially said "that's basically it", but later I mentioned alternative approaches such as the online panel methods used by Morgan and Essential. Of course, someone meaning to survey an area as small as Hobart via such methods would probably need to assemble a specific panel first to get a reasonable sample size, and gather appropriate data in order to scale it.
It is true that phone surveying is not ideally suited to very long qualitative surveys. However if the point of Question 1 was to hook people in and drive publicity, then there is no need to make silly noises about what the results to Question 1 show after the survey is completed.
I reject Bold's claim that his survey would have captured all "who cared enough and utilized the mountain substantially enough, to endure around 15 minutes" and that "the people that matter in this debate" filled it out, as I believe that some hardline objectors would have declined to fill out the survey out of cynicism, as already stated in my article. Obviously, there were hardline objectors who also did fill it out, but for hardline supporters the decision to fill it out is much more of a no-brainer than for objectors.
Bold argues that based on my argument that his opt-in is unrepresentative of public opinion, I should also consider the number of signatures on petitions and the number of responses to submissions unrepresentative because they are also opt-ins. He's got that right, I do reject them for exactly that reason, and already indicated this in my article!
There is also some strawmanning (and plenty more later on): "Yet according to Dr Bonham the size of sample collected is irrelevant [..]" - anyone reading my article can see I argued no such thing - merely that large sample size is not a magic bullet that cancels out all other methodical defects.
"Without becoming too academic, I disagree that an opt-in survey can’t accurately gauge public opinion regardless of sample size. For example, if we commissioned a voluntary survey that gathered a response from 100% of the public concerned, is that not truly representative? Are voluntary local government elections, garnering just 50% of the voting public in places, not deemed representative?"
Concerning the first, get back to me when you've done it. The use of the term "sample size" implies that there is sampling, which means that some people answer and others don't. A 100% response is by definition not a sample of the whole - it is the whole! Concerning the second, there's a massive body of debate arguing that point back and forth out there for anyone who wants to read it.
"Dr Bonham has acknowledged the survey did gather demographical data from respondents, and having since shared this detail with him, he appears to accept that the findings are generally what he would expect"
Here he is taking things I said at our meeting out of context. For instance, I said that I would expect that males would be slightly more supportive than females and hence am not surprised that that was the case in the MWCC survey. However that does not mean that I consider the overall support levels to be representative for either gender - only that the comparative difference makes sense. Ditto for comparisons between different suburbs.
Online IP Control
Bold basically accepts my comment about the possibility of routing around multiple vote controls, and validly notes that there are other forms of opt-in processes that are also prone to fraudulent multiple voting, including petitions. That is correct but doesn't concern me. In terms of a comparison between opt-in surveying and scientific polling, when he has found a way for a person to be included nineteen times in the same Newspoll, he should get back to me!
Bold attempts to paraphrase the "Do The Political Leaning Questions Establish Bold's Survey Is Accurate?" section above and makes a mess of it. I suggest readers check the original.
Here Bold rather puts his foot in it, by writing "Without this preamble inclusion, respondents would have no idea on what caliber or quality of cableway solution MWCC was intending to propose, and so I defend its inclusion entirely."
and following that up with
"When was the last time a political party went to an election without announcing any policy? Of course it is legitimate to provide respondents with some key information, pertinent to our project particulars, for us to gain their feedback from."
When a political party announces policy, its aim isn't to find out who you intend voting for. Its aim is to influence your vote. "Selling" your product by presenting a rosy view of what it will be and then asking people if they like what they hear works much the same way - whether you intend it to or not.
Bold is guilty of severe misquoting when he writes "Of course, I accept Dr Bonham’s observation and welcome his opinion that our adopted principles are “the most favourable possible conception of the development.”. It's pretty stupid to misquote someone when the correct text can be seen at the link you've provided, but perhaps he hopes to hoodwink some silly cablesheep who reads his article but does not even bother following the link to mine. My actual comment was "the most favorable possible (or perhaps the most favorable impossible) conception of the development." (my bold (pun unintended).) Misquoting of this form is pretty dodgy practice and not a way to make a good impression!
"My View of Dr Bonham's View"
"Dr Bonham correctly acknowledges that our survey is not an under-handed push-poll [..]" Well, I stated that it isn't a push-poll, but I made no comment on whether or not it was under-handed!
Bold attempts to prove that his intentions were thoroughly sound via this question:
"So if his initial cynicism was spot on and we did intentionally wish to skew the survey result in our favour, would we have not run the survey at a time of year to gain maximum exposure and feed off local sentiment?"
The first part is strawmanning since I was very careful not to take a position on whether the survey method was deliberately designed to present a skewed pro-cable-car response for public consumption. But in any case, there is nothing implausible about the designer of a deliberately dodgy skew-poll including some sound features in their methods in order to ward off the most obvious criticisms. So I don't think this defence against a charge I didn't actually make really works.
This section generally repeats a lot of ground I dealt with at the start of my reply.
This section also largely repeats ground covered at the top. It's a paradigm case of something I've had a zillion times in online debates. Someone says that A proves B, C and D, I point out that they're wrong and that A is unfit for proving any of B, C and D, and instead of just accepting they should never have made such a claim in the first place, they say, ah, but A was only designed to prove E and F and I should acknowledge its value on that basis.
As far as I'm concerned, the purpose of someone's survey is the purpose that they choose to use it for. If it's not a poll, don't pretend that it can "resolve the long-debated public topic", that it ends speculation about public sentiment, that it proves things about Green views on the cable car, and above all that it "indisputably" "the question of whether or not the public support our concept in principle".
I note that some of the comments I quoted and criticised on Bold's website have now deservedly disappeared. Nothing has disappeared from my original article following Bold's critique of it, because he didn't actually lay a finger on anything I said.
Finally, Bold argues that "Given that our recent opt-in results closely matched a 2009 random sample poll on the topic, and [context-allergic strawmanning snipped] , I remain confident to claim that the survey commissioned by the Mt Wellington Cableway Co. is robust enough to be considered broadly representative of local sentiment"
One skew-poll getting the same result as another is not representative of anything!
Quid Pro Quo
This section isn't in response to any one of Bold's sections. I don't normally refer to in-person conversations in my articles, and respect granted confidences utterly, but since Adrian Bold considers it legitimate to write a piece in which he refers to an in-person conversation, I will point out that Adrian Bold, in a publicly viewable tweet, wrote "Tks for opinion piece, quite fair assessment":
.. a view he repeated to me via LinkedIn message, and to a more qualified degree in person. He also admitted both by phone and at our meeting that he was asking for trouble by using rash words like "indisputably". He did courteously tell me that he was going to reply to my piece, and I welcomed this because it is good to get some debate, but I do find his attitudes towards my work in the lead-up to, and largely during, our conversation, to be quite incompatible with the tone of his critique!
Tidying Up Some Loose Ends (May 10)
Thanks to it being mentioned on Tasmanian Times, I have now come across some discussion of this article that occurred on the MWCC Facebook page. It started off with Mr Bold complaining:
"Bonham's response reads more like a personal jibe, lacking substance on the subject matter. Anyone can poke holes at anyone, at least we can back our positions with a decent sample of the population."
He's clutching at straws in pretending that because my reply above contains exactly two "personal jibes" that therefore the whole thing reads like one. In any case, the core problem here is that a decent sample and a large sample are not always the same concept - adequate sample size is a necessary but not a sufficient condition of getting it right.
There is a response from Downhill Dan Donaldson which commences:
"Please avoid associating his views with UTAS, even if he does work there, his views have nothing to do with the Uni."
I don't work there, although I am an Honorary Research Associate of a department there. I do not claim this site to be departmental research, though if the department wishes to claim it as such I have no objection.
"As I said before, very few surveys are ever conducted using all theories of survey design and yet most are widely accepted through peer reviewed research journals as valid surveys provided a few attempts for quality survey design are made such as target sample size, non misleading questions, internal construct validation."
The key question here is what purposes they are accepted as valid surveys for. Opt-in surveys are indeed widely used in a range of research, but not to attempt to establish the percentages of people in the general community holding particular opinions. The original article here was sparked by Mr Bold making claims on the basis of his data that his methods and data were unsuited to support. That those methods and data are probably suitable for answering many other questions is irrelevant to that critique. Donaldson expresses confidence that the survey was adequately designed for its intention; however, it was not adequately designed for the kinds of statements made on its behalf when it was released. I already pointed out that in these circumstances, the actual intention is not relevant.
Anthony Rochester writes "Of course Kevin's academic analysis is motivated by his personal opinions on the whole thing. He isn't objective and doesn't claim to be. Consider it to be his one vote against."
I don't think so. I did disclaim that I am cautious about the concept of a cable car, but recently you could have also seen me here criticising comments about the Legislative Council results made by pro-same-sex marriage campaigners, although I am also strongly pro-same-sex marriage.
The experience of Mr Bold's attempts at sneaky tactics during this debate has made me more cautious!