Hot answers tagged

128

Science aims at revelation and insight. Before you can even consider dropping these samples, you need to understand why they exist. The reason is that unexpectedly "good" data can be just as much a sign of problems with your theory as unexpectedly bad data. Are these data points telling you that you've got a bug in your algorithm? Are they saying that ...


93

Revoking a degree is rarely done and, then, only for the most extreme reasons such as explicit dishonesty and such. Any results in any thesis are subject to revision as new information becomes available that was not present in the past. It doesn't mean that the work was wrongly done, but only that what is known has advanced. Since you re-did the experiment ...


69

Maybe Zenodo or other "Academic Data Repository". Googling this would give you a list. Zenodo have some advantages. Gives you a DOI, Digial Object Identifier, a unique link and a academic standard for citations. You don't need acceptance to publish your data. Is a official EU project, used for giving research grants in Open AIRE project. Is hosted by CERN....


68

Good research practice (which I would argue is not country specific), requires an independent ethical review of any research that involves human subjects. A university can make blanket determinations that certain types of research do not require individual review (e.g., interviews). It is generally considered academic/research misconduct to not have proper ...


60

Is it unscientific? Yes. The purpose of these plots is not to show that the algorithm is fast, but rather to give an accurate picture of its speed. Intentionally removing outliers without explanation gives a distorted picture. It's OK if you are clear about it (for example, explaining in the caption or text about the outliers that were removed and why), ...


50

If you also have some code associated with this data that you might like to share, another option might be GitHub. You wouldn't host the 12GB dataset in a GitHub repository itself; instead you would host your code, and create a readme.md file (GitHub will do this virtually automatically for you) where you write out instructions or other narrative. This is ...


48

There is a trove of documents from Microsoft with advice on GDPR compliance, such as "Windows and the GDPR: Information for IT Administrators and Decision Makers" and has a pretty thorough explanation of what data moves where. According to the document itself, it takes 17 minutes to read. I think you'll feel better after you've done so. There's a lot of ...


47

There is only one reliable way to do it, which is to try to replicate their results. The unreliable, but not completely useless way, is to see if the numbers fit Benford's Law. Benford's Law describes the distribution of the first digit of many very diverse data sets. This is the distribution: (public domain chart from Wikipedia) Andreas Diekmann ...


44

You should ask the company where you do your internship. Perhaps they are fine with whatever you want to do. Perhaps they will allow you to use their data but only release summaries, not the raw data. Perhaps they will require you to anonymize it. In addition, how will you do the analysis? Will you take the company's data and transfer it to your own ...


43

Yes. I would have a separate section of your paper entitled something like "Further exploratory analysis", report what you did and what you found, and note that until a study has been design to specifically test your hypothesis, it remains a hypothesis, but suggest that it might be an attractive target for further study.


38

Edit After thinking about some of the points raised in the comments, I would like to expand on my answer, but also defend its form against the criticism that it is so vague as to be unhelpful. [In case you are wondering what the original answer was, it is roughly the sections 'Looking for mistakes' and 'Trusting your feelings'.] Benford's law This was ...


37

I really don't get this. This is a little like the Movies Classification Board or the Classification & Ratings Administration feeling awkward about asking to view the movies they are going to rate. If you are developing an algorithm for detecting pornography, why would it be awkward to request funding to acquire test resources? This is an entirely ...


36

Edit: I think I should add a little more background. I am in a PhD position actually hired for doing research. Yet due to my background in computer science, I am 'officially' responsible for everything related in our lab to electronic data processing. First, I think there is a serious management issue in your lab: leaving the responsibility of data ...


34

I agree with both previous commenters: publish your algorithm first and then point them to the YouTube video. If you cannot or don't want to go the traditional journal publishing route, at least, consider self-publishing a paper, describing your algorithm, at a preprint or e-print server, such as arXiv. At the same time, I don't see a problem with this ...


33

Don't do this. Here is some random data with 21 experiments A-U, each one repeated 3 times. In both cases, the experimentwise means (indicated using red crosses) are identical, but the within-experiment standard deviations are very different (1.0 in the top graph and 0.2 in the bottom one). R code is below. Just seeing the experimentwise means is very ...


31

Yes. For a practical scenario, imagine someone trying to replicate your research, going only by your paper, and beating themselves over the head thinking they have a bug, because their plot shows these weird outliers. Basically, you have to give people all the information, because you don't know how they're going to use your paper. Probably, most readers ...


30

There are services that provide enough to support 12 GB of data. For example, Figshare provides 20 GB of free space (file size limit 5 GB) for private storage and apparently unlimited public space. They state they can support larger files but not through user upload. When you publish data you can assign a doi to the data set (this can actually be done much ...


26

This will, as it seems nearly all questions on this site, vary based on field. My answer applies to Epidemiology and medical research only. Your mileage may vary. It is very common for this to happen in my field. There has been an increasing emphasis on using meta-analysis and systematic reviews to summarize bodies of work, and with those there is almost ...


26

You need to run this by your Institutional Review Board (IRB), as you would any time you use data from human subjects for research. From what you describe, it has a good chance of qualifying for an IRB exemption under one of these categories (if identifying information is removed before giving the data to the researcher): It is using existing data the data ...


26

Your university should have some sort of data privacy compliance office. You absolutely need to talk to them. Well-meaning advice from strangers on the internet is great for giving you an idea of what the issues are but there is potential legal liability for the university here and you must talk to the people whose job it is to manage these issues.


25

Drawing the lines implies a continuous relationship between the parameters. So if you can expect continuity, then connecting the points is fine. A second point to make is to avoid colours that are as similar as the green and blue you have chosen. One reason the difference is hard to see in the first scatter plot is due to colour. Try to experiment with ...


24

Is there a way to express this (not as a citation format), that maybe this data is accurate, but not sure, since it is the only available source about that data. How about something like: "X, measured by Jones in 1950, is commonly given as the value for Y (as cited in Smith, 1989 and Cutler, 1995). However, the original manuscript by Jones describing this ...


24

As you know (or may not know!), basically all the pornographic movies are not free. I find this not quite realistic: A fair chunk of the total internet bandwidth goes to porn. There are a lot of "free" websites. And a lot of "studios" offer samples which would suit your research, in my opinion. You don't need a two-hour porn movie to test your algorithms. ...


24

The image manipulations reported on Retraction Watch are most of the time naive collages of gel photographs or spectrograms. They get caught, among other things, because repeating patterns in the noise appear on closer inspection, or linear disruption of the noise are visible, see this. For 1D data, the case you mention, there is the Benford's law and other ...


24

There are a number of ways in which a DOI could be linked to a dataset: Figshare will provide a DOI for any deposited work, which includes data. Zenodo also provides DOIs for any kind of research output, including datasets Dryad provides DOIs for data submissions linked to papers (for a fee, which includes storage, curation, archival, & checks for best-...


24

The most common and sustainable thing to do is to deposit the data in a research data repository. Depending on which one you choose, the data will get a persistent identifier, e.g. a DOI, can be cited properly in publications, might be reused by other researchers, ... You can find a list of available data repositories at re3data, the registry of research ...


23

My main confusion was not realizing that The Open Archives Initiative Protocol for Metadata Harvesting is a separate protocol, not a subset of arXiv API. In this case, the relevant queries are ListIdentifiers (10k items per query) and ListRecords (1k items per query). To get just identifiers we need to write: http://export.arxiv.org/oai2?verb=...


23

One major pitfall is going into analyzing the data without an a priori protocol to deal with outliers. Time to time, tension rises between analysts and investigators on whether a point should be removed or not. Investigator may wish to keep it because it drives the significant result; analyst may be anxious to keep it because, well, it drives the significant ...


22

Academic misconduct is the last thing you have to consider, not the first. First, you have to consider that there might be something you are overlooking (e.g. an updated version of the code, or the data set is not really the same, or something related to the machine or etc.). Then, you have to consider that mistakes happen, and as I wrote in this answer, ...


21

There area number of routes. They do cost money to create, so you want to get it via one of many providers: Figshare gives DOIs to items you put in your account with them - this is the most DIY route Zenodo gives DOIs to objects Many data repositories assign DOIs, e.g., Dryad


Only top voted, non community-wiki answers of a minimum length are eligible