Note: up-to-date information is maintained here: A study in Noise and Health shows that wind farms cause people to lose sleep.  How reliable is this study?



Summary:  Its reliability is low.  This is a flawed and misleadingly titled study by long-time anti-wind lobbyists.

• It mistakes correlation for causation, and overstates correlation
• It downplays or ignores long-understood impacts of both bias and impacts of change in creating annoyance
• There are significant unstated conflicts-of-interest, biases and allegiances to an anti-wind lobbyist group among the six authors and reviewers
• One of the authors has been actively involved in creating anti-wind bias and annoyance in these sites for years
• It should be considered against the 17 major world-wide reviews to date which have found no health impacts from wind generation.

1. The study overstates causation and correlation, and understates the impact of bias of the studied groups.
   
   a.  Nissenbaum et al are overstating the strength of the correlation that their data shows. In contrast to the conclusions, figures 1 and 2 show a very weak dose-response, if there is one at all. The near horizontal 'curve fits' and large amount of data scatter are indications of the weak relationship between sleep quality and turbine distance. The authors seem to use a low p-value as support for the hypothesis that sleep disturbance is related to turbine distance. A better interpretation of the p-value related to a near horizontal line fit would be that it suggests a high probability of a weak dose response. Correlation coefficients are not given but should have been to indicate the quality of the curve fits. Intrinsik points out an additional failing of the report:
   

   Although there was a statistically significant difference between the mean PSQI scores in the near (7.8) and far group (6.0), it is important to remember that both of these average scores are greater than 5, which would qualify both groups as “poor sleepers”. When one examines the reported “% of PSQI score >5” no statistical difference between the near and far groups was found (p=0.0745).

   
   b. As the Intrinsik assessment points out:
   

   Given that the relationship between noise from wind turbines and health concerns is the fundamental premise of the study by Nissenbaum et al., it is surprising that the authors gave such little consideration to collection of actual sound data measurements at the study participant homes. The use of post-hoc sound data, visually obtained from figures in reports, is not scientifically defensible and should not have been used to draw conclusions about the findings of the questionnaires with distance from turbine locations. [16]

   
   c. Intrinsik also points out the misleading title of the study, another case of overstating conclusions available from the data:
   

   We also believe that the title of the paper “Effects of industrial wind turbine noise on sleep and health” is not supported given the nature of the data presented. No evidence with respect to sound level (noise) and its effect on sleep and health has been presented in this paper  [16]

   
   Given that the authors themselves admit that they can't construct a dose-response curve, their conclusion that wind farms affect sleep is surprising:
   

   In their paper Nissenbaum et al. state that noise emitted by IWTs can affect sleep. However, their results do not support this statement. In fact, the authors state that “The data on measured and estimated noise levels were not adequate to construct a dose-response curve...” and no statistical analyses were conducted to assess this supposed relationship. Therefore, we do not believe that Nissenbaum et al. (2012) show any statistical difference in overall “poor” sleep quality or sleepiness between the groups. [16]

   
   d.  The studied communities, via agitators such as Nissenbaum, have developed strong negative attitudes to wind farms.  As this study shows, this is a much better predictor of the effects Nissenbaum is claiming than any actual noise from wind farms. [4] 
   
   Intrinsik goes further and asserts the conclusion that the authors, if unbiased, would have found from the data, that the study groups were annoyed by changes in their environment and self-reported health impacts arose from annoyance with the change:
   
   

   The authors pointed out that visual cue and attitude towards wind turbines “are known to affect the psychological response to environmental noise”. While this may be true, visual cue and attitude by themselves have been shown to be stronger drivers of psychological responses than a wind-turbine specific variable like sound itself (e.g., Pedersen 2004). Therefore, a conclusion that can be drawn from this study is that the self-reported health effects of people living near wind turbines can be likely attributable to physical manifestations from an annoyed state, rather than a wind-turbine specific factor like noise. Indeed, the weight of evidence in the wind turbine and human health literature points to a causal relationship between self-reported health effects and annoyance, which is to say annoyance brought on by the change in the local environment(i.e., a decrease in amenity) that wind turbines represent (Knopper and Ollson 2011). [16]

   
   e.  The authors' treatment of bias is poor, stating without evidence that accounting for selection and reporting bias would make their conclusions stronger, not weaker.
   
   f.  The sample size is small as is the control group (many studies have small sample sizes; this makes this less authoritative, not worthless in and of itself.)
   
   
2. The study group has undisclosed biases influenced by one of the authors.
   
   This is not Nissenbaum's first study of the Mars Hill or the Vinalhaven wind farms.  He ran studies there in 2010 and 2011 as well, using a format pioneered by anti-wind folks in the UK and then by Nina Pierpont, creator of Wind Turbine Syndrome. [3], [9] The a-scientific studies are so poorly constructed that they are guaranteed to make people ascribe new symptoms to wind turbines, and to take completely ignored minor symptoms and turn them into major complaints. These studies have pre-loaded the biases of these study groups, making it difficult to accept the the conclusions of this better structured study.  The data this study is based upon is not new data, but data that was obtained several years ago that has been presented at the ICBEN conference in London (UK) over a year ago and has been presented as part of anti-wind farm submissions, which is not unusual, but the data and the conclusions Nissenbaum et al have been putting forward has been strongly criticized in the past. [10] For example, a 2012 Massachusetts expert panel had this to say about the conclusions:
   

   details of how homes were identified, how many homes/people were approached, and differences between those who did and did not participate are important to know. Without this, attributing any of the observed associations to the wind turbines (either noise from them or the sight of them) is premature. [16]

   
   
3. The authors' pre-existing bias is not disclosed or accounted for.
   
   Jeffery Aramini is the person most often interviewed in newspaper reports but the study is co-authored with Michael Nissenbaum and Christopher Hanning. 
   
   a. Aramini has maintained a lower profile than Nissenbaum and Hanning, but is on the Advisory Group of an anti-wind lobbyist group, Wind Vigilance. [11] 
   b. Nissenbaum is a long-time anti-wind activist and also a member of the Advisory Group of Wind Vigilance. [1]
   c. Hanning is a long-time anti-wind activist as well, who has been writing anti-wind papers that have not been able to get into even low-impact, peer-reviewed journals and is also on the Advisory Group of Wind Vigilance.[2] [12]
   
   Fronting with the lower profile Aramini in newspaper interviews appears to be a convenient way to disguise the deep and long-standing bias of the authors.
   
   The authorial bias can easily be detected by the inappropriate use of the emotionally laden "industrial wind turbine", a term which was selected and focussed grouped by anti-wind lobbying organizations associated with the Koch Brothers.[14]  Neutral language includes "wind turbine" and "wind turbine generator".
   
   
4. The thanked reviewers have unstated biases and conflicts-of-interest as they are paid anti-wind experts who have a long history of directly testifying against wind energy.
   
   a. Carl Phillips is relatively new to this group, having been asked to leave his post at an Alberta university for taking tobacco industry money and remarkably finding that tobacco products were much less harmful than people thought.[6]  He has found a new source of funds in anti-wind lobbying.  As he says on his blog, Ep-Ology:
   

   I knew what answer I was going to present from the start. So when I wrote my COI [conflict of interest] statement, I did not hesitate to describe, matter-of-fact, that I do work as a testifying expert on behalf of communities fighting the siting of local wind turbines.

   b. Rand not only testifies, his firm gains revenue from measuring sound near wind farms for complainants and to assist litigation. [7]  Rand, in any event,
   c. James has been testifying for fee against wind farms since 2006.[8] 
   
   Phillips and James are both members of the Advisory Group of Wind Vigilance as well.  The three thanked are bolstering their court room pitches, and cannot be considered credible unbiased assessors. If they are the only peer reviewers, this is grounds for retraction. As their conflicts and pre-existing biases are unstated, this too is grounds for serious concern.
   
   
5. Noise and Health is a rarely referenced journal of low impact.
   
   The journal, Noise and Health has a very low impact index of 1.2, meaning that few researchers reference their studies; there can be a variety of reasons for this including poor quality or trivial papers. [5]  The journal may not have sought independent reviewers who would have pointed out the flaws in the article, but may have accepted the reviewers that came with the article as suggestions.  If the journal did not gain separate review, this is a reason for retraction in and of itself.
   
   
6. Wind Vigilance, on whose Advisory Group five of six authors and reviewers sit, has been promoting wind health issues in the absence of peer-reviewed evidence and against 17 major studies' findings for years.
   
   As Wind Vigilance is a central theme to this, it would be useful to understand their positions and the degree of evidence behind them:
   
   

   Based on a review of the evidence, the Society for Wind Vigilance is satisfied that there is a significant probability of adverse health effects for human subjects living within 2.0 km of land based industrial wind turbines. The Society for Wind Vigilance recognizes the urgent need for further human health research to finalize guidelines for siting and noise levels that will protect human health. In the interim the Society for Wind Vigilance recommends that land based industrial wind turbines be sited a minimum of 2 km from the property line of non participating residents. Distances greater than 2 km will typically be required for special terrain such as turbines on ridges and offshore turbines. [13]

   
   Bolding is mine to indicate statements which require elaboration:
   1. This statement was released April 2012, six months before the first peer-reviewed paper that found any issues with wind energy and health was published.  In the meantime, 17 major reviews by independent and credible organizations worldwide of thousands of pieces of peer-reviewed research found no issue with wind energy and human health.  On what grounds did Wind Vigilance make the assessment that all of the other studies and the vast majority of medical, engineering and acoustic professionals were wrong?  On what grounds did they decide on 2 kilometers?
   2. "Industrial wind turbines" is the preferred emotive phrasing of anti-wind lobbyists.  It is not neutral language in this discussion, just as "wind farms" is the preferred emotive language of pro-wind advocates (including me).  This language was created by lobbying organizations associated with the Koch Brothers and other astroturf funding undiversified fossil fuel organizations.[14]
   3. The reasoning behind greater setbacks on ridges and lakes is not explained, but given the weakness of the rest of their position, it can only be to ensure that wind turbines will never be seen or heard.  Obviously this is an extreme and foolish position.
   
   That five of six authors and reviewers of this paper are so tightly associated with an organization with such a strong and strident opposition to evidence-based siting guidelines and wind energy in general is indicative.  That they do not make clear their association, their long-standing bias and their conflicts-of-interest is also indicative.
   
   
7. 17 major reviews have found no health impacts from wind energy
   
   This single paper must be contrasted to the 17 (to-date) major reviews world-wide of hundreds or thousands of peer-reviewed articles related to wind energy, health and noise which have found no health impacts.  All studies agree that a small subset of people very close to wind farms find the noise annoying. [15]  The best consensus is that the vast majority of health complaints attributed to wind energy are the result of a psychogenic or communicated psychosomatic illness; people are making themselves sick when they are told that they will get sick. [15]

Full disclosure. This assessment was developed with the assistance of:

• Dr. David Perry -- Dr. Perry holds degrees in electrical engineering and neuroscience from The University of Melbourne and a PhD from the Bionics Institute examining how sound stimulation from a cochlear implant is represented in the brain. Dr. Perry is also on the Board of Directors of community-owned wind project Hepburn Wind in Australia.
• Richard Mackie - Mr. Mackie is an Engineer with degrees from the University of Auckland and the Australian Graduate School of Management. He is Managing Director of Advanced Energy Consulting (Australia) which does work related to wind energy projects.
• The Intrinsik professional assessment upon which some of the comments are based was funded by CanWEA (Canadian Wind Energy Association).

References:
[1] Dr. Michael A. Nissenbaum - The Society for Wind Vigilance
[2] http://docs.wind-watch.org/Hanni...
[3] http://www.google.com/url?sa=t&r...
[4] http://www.sciencedirect.com/sci...
[5] Journal and Academic Rankings
[6] Tobacco researcher leaves U of A
[7] Wind Turbines: Published Articles
[8] "Wind turbine syndrome" is more wind than syndrome
[10] http://www.ert.gov.on.ca/files/2...
[11] Jeff Aramini, DVM, MSC, PHD - The Society for Wind Vigilance
[12] Dr Chris Hanning - The Society for Wind Vigilance
[13] News - The Society for Wind Vigilance
[14] Turbine foes try to forge national opposition movement
[15] Wind farms don't make people sick, so why the complaints?
[16] http://www.canwea.ca/pdf/Intrins...

Additional references:
[1] Additional background on the authors and the nature of the preceding studies performed by Mr. Nissenbaum here:  A Vet, A Radiologist, And An Anaesthetist Walk Into A Scientific Controversy...
[3] Nissenbaum paper recycles claims on wind energy and health already found inadequate by courts and expert panel
[4]BigCityLib Strikes Back - The Full Nissenbaum

(this answer addresses some of Mr. Barnard's objections, and also presents my perception of the relative validity of the paper in question--see also one comment from me under Mike's answer above, regarding the new Intrinsik review/critique)

Mr. Barnard appears to be overstating his objections, much as the media coverage of this paper may be overstating the significance of its findings.  He selectively quoted statements in the paper re: causation.  In fact, the authors note that further study, for replication and to prove causation is needed.  While sound levels were not measured during the period that surveys were being filled out, the survey questions were looking for longer-term effects (not simply of that moment). In keeping with this temporal focus, previously collected sound level data WAS gathered from both sites and is presented along the bottom of the graphs that show the survey response data. (That said, the sound data was still, as the authors note, "not adequate to construct a dose-response curve.")  Likewise, their section on bias included possible bias factors in both directions (including the big one, re: no masking of the intent of the questionnaires), and did not, as Barnard implies, presume all possible bias would only strengthen their conclusions.

I agree to a point with Barnard's suggestion that the results are less dramatic than press coverage claims, and that the paper in places may also imply. What they found was modest shift in average sleep quality and mental health scores among those living closer to turbines, more than dramatic differences. It is worth noting that while they found modest (and statistically significant) changes in average sleep quality and mental health scores, they did NOT find any difference in overall physical health scores.  My more measured analysis (for the non-advocacy Acoustic Ecology Institute) of the findings can be seen here: New paper quantifies sleep, mental health effects near wind farms

Barnard does not put this study into a proper context.  Its purpose is to use established public health research metrics to see whether the scattered (and at times concentrated) reports of sleep and health issues around wind farms can indeed be quantified, and if so, whether there are trends indicating more problems closer to wind farms.  In research language, we've had "case reports" of sleep and health issue, but few real studies of those closest.  By including control groups from the same communities, but out of earshot of turbines, and using established research protocols and metrics, they moved this body of research forward in an important way.  Barnard implies it may have been an ad-hoc survey (as Nissenbaum's initial inquiries around Mars Hill were); in fact, the study employed three widely-used public health questionnaires. Yes, the results were presented at a conference a year ago; this is standard practice in the research community.  Published papers routinely appear a year or two after results are ready to share at conferences. Throughout the paper, the authors follow standard procedures in presenting their data and are transparent as to the statistical methods used to quantify their results. 

To label these researchers and reviewers "anti-wind activists" is playing loose with language.  Nissenbaum, Hanning, James, and Rand were all well-established professionals in their field, who responded to a rising tide of noise complaints around wind farms by moving beyond standard practice of assuming that turbine noise levels can't bother people, and going to these towns to try to learn what was behind the problems.  They did not go into it with preconceptions, besides a professional interest in understanding what was going on.  It's true that they have come to feel that current siting standards are not offering as much protection from community noise impacts (primarily sleep, secondarily stress) as planners generally aim for when siting new sources of community noise, and that they've become well known for such views, in contrast to more sanguine views on similar issues that are presented by other acousticians who are often contracted to do sound studies by wind developers. Their focus in public statements have been on the results of their research, not on broad-scale anti-wind rhetoric.  (However, some community activists with a broader anti-wind agenda certainly glom onto their work and overstate the level of proof therein)

(An aside: I'm not sure why being routinely contracted by developers does not create a bias while sometimes getting paid by community groups or individuals supposedly confirms a bias--I prefer to read the studies produced by each on their own terms, and to trust that the professional ethics of all these acousticians means that none are biased--they're just presenting noise numbers, after all; what we find is different approaches to collecting and assessing the data, rather than untrustworthy data.  And while both groups accentuate the interpretation that reinforces their opinion about what is a reasonable noise level for wind farm neighbors, the data itself in studies by all these folks speaks for itself.) 

James and Rand have each been acousticians for decades, with a focus on community noise and noise control; both do far more pro bono work than paid work on wind farm issues, though both have at times been paid to testify at various proceedings (just as acousticians working for wind companies are, which doesn't discount the validity of their measurements, either).  In particular, it's laughable to suggest Rand's wind work is "lucrative".  Hanning is an elder in the field of sleep studies (his university in the UK named their research center after him), whose primary contribution on wind issues has been to repeat that which has been long known in sleep studies: that non-waking "arousals" caused by noise sources can degrade sleep quality as much as actually being woken up.  Other than his supporting co-author work with Nissenbaum, I'm not aware of him attempting to publish on wind turbine noise; he is basically retired and I'm sure content with his reputation in his field.  I am less familiar with Aramini or Shepherd's background.

While this paper is a useful step forward, it is only one step.  Similar studies elsewhere, perhaps in communities with more people living within a mile of turbines, would be the obvious next step. I agree with Barnard that it would be constructive (and more convincing) to see these sorts of studies published in journals that have more impact and respect/reputation. (However, it should be noted that the cost of preparing papers for publication is a significant hurdle for researchers not employed by universities; also, many of the literature reviews that find limited evidence for health effects are themselves not formally published or peer reviewed).

It's also likely that different types of communities may lead to different results; there are indications that ranchers and owners of large farms (eg in Texas, Oklahoma) are not as bothered by turbine noise as are people in more pastoral rural communities. Most of the communities where noise has become a major issue are in places where a significant proportion of the population is seeking peace and quiet, rather than a working landscape.  It may be relevant that the two sites in this study are among the more contentious in recent years: Mars Hill appears to be a Altamont-of-noise worst case situation, with a ridgeline turbine array above sheltered homesites, that further obtained a state waiver to run 5dB louder than normal state standards, triggering high rates of annoyance and sleep/headache/depression effects.  Vinalhaven is a situation in which there is a difficult divide between those living near the turbines, and the rest of the town, which could contribute to some of the stress-related mental health effects.  However, in both places, the fact of noise-induced sleep loss and severe impact on daytime enjoyment of property is clear and incontrovertible. 

Overall, this paper is worth reading carefully, and should not be dismissed on the grounds that Barnard claimed.  It's not proof of health effects, but it is well-designed and valid, within the modest parameters of an early attempt to quantify these effects.