The authors should decide what paper they want to write. The initial history of SSDs with references is good, but much too brief for a review paper and omits one important topic (censored data) and barely mentions another (Bayesian methods).
9 Responses
Comments are closed.
RESOLUTION
No change. This is all too hard and as noted by a number of others, we are already over the word count; the MS has been substantially modified anyway; and we stated upfront that it wasn’t a detailed review of all aspects of SSD modelling.
Here’s a suggestion…and yes it does require a bit of re-organization: what if the purpose of the paper is to investigate solutions (i.e. model averaging, multimodality, left censoring) to the problem of fitting single distributions to small datasets and then to discuss the different software tools currently in use and how well each of them allows the user to deal with this problem of small data sets. Then finally, we can make the point that all jurisdictions struggle with these same issues and that collaboration between jurisdictions will help with efficiency. And that R is designed to collaborative and transparent and could support collaboration across jurisdictions. The information is all there and it could help to streamline the paper some so that there will be room to provide some more detail about the potential solutions.
This is a solution to the reviewer’s comment, but does require significant effort to restructure the paper, including removing most if not all of the upfront SSD methodologies current status content (which, I think, all reviewers provided positive comments about) and adding a fair bit more text evaluating/comparing the different tools. I’m not sure we have the time or appetite to do this.
I certainly don’t!
As indicated in our introduction: “The current paper seeks to summarise the current status and elaborate on several recent developments for SSD methods, specifically, model averaging, bimodality and also software development.” It was never our intention to provide a full review of the history of SSDs, but rather outline the commonly used methodologies (and their associated supporting software) and highlight some recent developments.
I think by the time we respond to many of the other comments, we can easily add here the the manuscript has been substantially modified, with greater clarity and detail regarding the current methods.
The only problem with this, is that I think we submitted it under the “Critical review” category, because we were over the word limit for a regular research article. We could potentially seek to change the category to a regular research article, and negotiate with the editor over word limit.
Dealing with censored data – see Nysen paper.
Bayesian methods — this is a whole paper in itself!
I would point out that the focus of this paper is not comprehensive review of everything about SSD but focusing on two specific aspects -software and bimodality.
Rick wrote:
From: Rick van Dam <rick.vandam@wqadvice.com.au>
Sent: Monday, 20 July 2020 10:49 AM
To: David Fox <david.fox@environmetrics.net.au>
Cc: Rebecca Fisher <r.fisher@aims.gov.au>; Batley, Graeme (L&W, Lucas Heights) <Graeme.Batley@csiro.au>; Tillmanns, Angeline ENV:EX <Angeline.Tillmanns@gov.bc.ca>; Joseph Thorley <joe@poissonconsulting.ca>; Carl Schwarz <cschwarzstatsfuca@gmail.com>; Spry, Doug (EC) <doug.spry@canada.ca>; McTavish, Kathleen (EC) <kathleen.mctavish@canada.ca>
Subject: Re: Environmental Toxicology and Chemistry – Decision on Manuscript ID ETCJ-Jun-20-00450 [email ref: DL-SW-3-a]
Hi All
In amongst the comments on the ms, reviewer 4 noted “The authors should decide what paper they want to write.” The reviewer has picked up on an issue that we had already identified when drafting the ms, in that our paper is a bit of a mix of two types of paper – a review and a detailed statistical analysis – and, thus, runs a risk of not being either. I’m loathe to make major changes to the content and structure, because I think it works well enough as it is to be a very useful contribution. I would suggest we accept the reviewer’s suggestions where valid, but do not attempt to re-structure or completely remove some parts of the paper. We are word limited so can’t be expected to make major additions. However, in response to the above reviewer comment (and other comments) we may need to justify why we have not removed some sections of text in order to make room for more detailed text elsewhere. As David has pointed out, we can make reference to planned subsequent papers that look at issues and compare approaches/tools in more detail, which should address numerous comments. And, hopefully, the editor is a little relaxed about the word limit given the reviewer comments.
Cheers
Rick