We've talked about the accuracy of Wikipedia before. It has been in the news recently because of its mistaken assertion that John Seigenthaler plotted to kill the Kennedys. It turns out that the author was a prankster who was trying to tweak a member of a prominent family to amuse a co-worker. This is just one of many inaccuracies that have been found on Wikipedia.

I realize that it can be a good source, but it's bothersome that the authors don't have to identify themselves. While many people use it as a primary source, even the founder of Wikipedia, Jimmy Wales, says, "People definitely should not be using Wikipedia as a primary source."

How much do you use Wikipedia? How do you check its accuracy?

[On another front, this green background in Potpourri is just too dark. If my laptop screen is a certain way, I can't even see the writing at all.]

This message has been edited. Last edited by: Kalleh,

I use wikipedia fairly often, but usually just to browse through reading interesting articles. Often, I just need to check a date or some such and use it, although it always for personal satisfaction, I never actually need that information to be cited for something important.

People who use an online encyclopedia for a primary source get what they're paying for.

I mainly use it as a quick reference, usually for information I am familiar with but to check some minor / forgotten fact or spelling, etc. I have found one inaccuracy in an article, and I corrected it. For what it is, quick, free, and online, I like it and use it, but I usually cross-check facts in reference books or with friends. This is just common sense. It's also rather good as a compendium of up-to-date news and facts. It has features that real print encyclopedia do not have, discussion about controversial articles from people with different points of view. The list of external links at the bottom of an article also helps with cross-checking of facts.

It gets worse. Other sites copy Wikipedia's articles, but fail to update when Wikipedia itself has made corrections.

This was brought forcibly to me when looking up definitions of 'waterboarding'. The article in Answers.com purports to be taken from Wikipedia, but in fact the two differ significantly, even in the very first sentence!

• Where Wikipedia says, "There are several different varieties of interrogation techniques referred to as waterboarding,"
  
• Answers.com says, "There appear to be two different varieties of torture referred to as waterboarding." (emph. added)

Answers.com has that same discrepency five times, calling it 'torture'. The Wikipedia article, however, clearly states that some people claim that this constitutes 'torture'.

This message has been edited. Last edited by: shufitz,

For those of us in academia, the Web can be terrible when grading papers. It is so hard to tell if a paper has been copied from some site, though it is possible. However, when you have fifty 20-page papers to grade in a couple of weeks (I once had 108!), along with all your other work, it is nearly impossible to check all of this out. Recently I heard a speaker citing statistics on high school and college cheating (student reports, so if anything the statistics are conservative), and approximately 50% of students cheat. Worse, some of them don't even consider certain activities cheating, such as copying papers from Internet sites. The person giving the report said the answer is not to try to catch all forms of cheating (because you never will), but instead it is to talk about the importance of integrity in society, clarify what constitutes cheating, and ask students to sign honor codes. I don't know, but it is all very disconcerting. I can honestly say that not only have I never cheated in any of my academic programs, but I have never even thought about it or considered it. I think this rampant cheating is somewhat new and encouraged by all the data that is available on the Internet.

I suppose my subject (ESOL/EFL) is a little different but I find that you do get a feel for the type of language a student is likely to produce and anything that student copies stands out like a sore thumb. Once you have noticed that something looks wrong it's relatively easy to verify it on the internet. Choose a selection of phrases that are fairly unusual and do a google search on each with quotes to search for the exact phrase. If you manage to go through all of them without a match then give the student the benefit of the doubt. In practice you'll usually find a match (I've done this several times with suspicious work and always found the copied source in under ten minutes!).

I remember my high school teachers complaining about students plagiarizing the old fashioned print encyclopedias. Most places I've taught at have definitions of plagiarizing and copying. I simply include the URL in my syllabuses, and tell the students that I use the institution's definitions of the terms. They should familiarize themselves with these difinitions because it could affect their grade. Also, I tend to tell the students during the first class that if they are going to cheat, they had better be good enough at it that I can't tell that they're cheating. I've just resigned myself to the fact that students have been cheating at least since Sumerian times.

Kalleh, there are computer programs which analyze students writing and then compare to various sources to check for plagiarism. Perhaps you should look into those.

quote:

Also, I tend to tell the students during the first class that if they are going to cheat, they had better be good enough at it that I can't tell that they're cheating. I've just resigned myself to the fact that students have been cheating at least since Sumerian times.

I absolutely would hate to cheat in Zmj's course; he'd find it immediately! I know that cheating has been going on since forever. However, I do think that it has gotten worse with the Internet. I could be wrong, though. Sean, as for those programs, yes, I have used them. However, it takes time, and I'd so much rather be helping the student with his/her paper than to be checking for cheating.

My teachers in college started insisting we turn our papers in on disc, so they could just load the documents right into the program. This requires almost no work for you, and is very fast. Plus, the students know you are checking for cheating, and are much less likely to try it. Obviously, email works better than discs, if you like large emails with attachments.

I suppose that's an answer. Yet...isn't it sad that it has come to that? After all, I am educating young men and women who will have the lives of patients in their hands. If they give a wrong medication, instead of reporting it immediately so that steps can be taken, will they hide that fact while the patient gets sicker? It's that integrity piece that bothers me. I tend to agree with the speaker I heard in that it's better to prevent the cheating, rather than to go crazy trying to catch it.

BTW, out of curiosity, what do the professors do when they find that a student has plagiarized part of a paper?

There has been a move away from examination to coursework assessment for many years, culminating with the ludicrous NVQ debacle in the UK, whose apologists isisted it would get rid of eaminations for ever. Of course, it has belatedly been realised that examinations, in spite of their admitted deficiencies, are very cheap to run and can be 100% secure against cheating.

Quetly, and without much commentary, the whole NVQ rubbish is quietly disappearing and companies are starting to insist on examined qualifications, validated by a proper awarding body.

Now that coursework cheating is so easy, I forsee an even greater move back to "old-fashioned" examinations, for all their faults.

quote:

ludicrous NVQ debacle

What's that? How can examinations be 100% secure?

Going back to the Wikipedia subject, this from the BBC report of five days ago:

quote:

The free online resource Wikipedia is about as accurate on science as the Encyclopedia Britannica, a study shows. The British journal Nature examined a range of scientific entries on both works of reference and found few differences in accuracy.

Since then, other things have led to a concentrated effort by a number of wikipedia posters to post inaccurate information, and damage the credibility of the site. There is a lot of bad blood over there right now. Today, for instance, it was reported by Wired that founder Jimmy Wales has edited his own bio, a definite no-no.

NVQs are National Vocational Qualifications. They were started as work-related, competence-based qualifications in a whole series of jobs. They were intended to reflect the skills and knowledge needed to do a job effectively, and show that a candidate is competent in the area of work the NVQ represents. They are entirely based on course-work or on-the-job observation and questioning.

Assessors ‘sign-off’ units when they think the candidates are ready, and the main complaint is the lack of any rigorous checking of candidate's work and cross-country equality of standards. Many people have thought that, in the lower levels at least (there are five in all), they were too easy.

They are being phased out now in favour of qualifications validated by a proper supervisory body, as RE says.

I'd mentioned before that answers.com quotes wikipedia but ignores wikipedia's updates.

Just as a spot check to see how bad it is, I checked Answers.com's schedenfreude entry, quoting Wikipedia, against Wikipedia's own's. Notice that only the latter has a section on "popular culture references; a check shows that it added that section as far back as its June 24, 2005 entry. So on this entry, Answers.com is quoting six-month old wiki data, lacking any error-corrections that may have been made. Not very reasurring.

This is an answers.com problem and not a Wikipedia one, surely.

Surely so, but also a reliance-on-the-net problem. Error runs swiftly, while correction merely ambles.

'Twas ever thus. It's rather like a newspaper that slashes a scandalous story boldly across page 1 but, when it gets around to printing a retraction several days later, buries that retraction in small type at page 76C.

There is definitely a future in the web in things similar to RSS. The answers.com page should literally reference the wikipedia page, grabbing content from it in real time, or possibly in a daily cache. This is not trivial to do right now, but it is the future of the internet.

quote:

What's that? How can examinations be 100% secure?

They can be close to 100% secure - which is far more secure than can be any form of workplace assessment undertaken by untrained and often biased observers.

Richard, I'm not understanding the linkage you made. "Secure" and "unbiased" are not the same thing. An essay test, no matter how secure, is only as unbiased as the grader.

Anticipating your response, I can see you mentioning that "secure" would include blind grading, where the grader is not told whose paper he or she is grading. That of course would tend to eliminate personal bias as to the test-taker bias, but would have no effect on bias for content, style, appearance, etc.

PS: How does one, economically, make a test 100% secure?

And do include the possibility that the test-administrator will cheat, as by altering the answer sheets before turning them in.

I have been an examiner for City and Guilds since 1978 and I reckon I know a bit about the job! Nothing can be 100% secure but a properly invigilated written test can be made close to it.

To take your points:

1. Security is not the same as bias and I didn't mean to suggest it was. However, assessors who know the candidates (and may even be their managers or employers in workplace assessment) have a vested interest in getting their staff through the qualification and will be more likely to cheat.

2. Test administrators might cheat but there is little reason to do so when they do not know the candidates. And in any case there are checks in place to stop this sort of thing, although their exact format is not something that awarding bodies necessarily make all that public.

3. Bias regarding content, style and appearance is largely eliminated by a proper marking scheme and, indeed, most awarding bodies specifically forbid examiners from deducting marks for such things as poor spelling and grammar (unless the test is specifically to do with such skills). Furthermore there are also checks and balances in place to help iron out examiner bias prior to grades' being issued.

4. Bias and subjectivity are only a problem in the higher levels of test, where answers are of the narrative or structured answer type, and then only if a proper marking scheme has not been devised. Simple knowledge tests, frequently tested by multiple choice, are right or wrong and the only way that examiner bias could affest the result would be if the examiner were to deliberately alter a paper. This could certainly happen but, as I said, why should he or she when the identify of the candidates is not usually known? And, as I said, there are certain procedures in place that guard against this sort of thing.

Examinations, for all their faults, are more secure than workplace assessment, except when such assessment is done by trained assessors who do not know the candidates. And this was what the NVQ system tried to impose but it failed lamentably for one simple reason. It costs far more money to send an assessor to a workplace, there to observe a candidate's performance for half a day (which was what was expected) than it costs for an invigilator to go to an examination centre for 3 hours, there to watch over maybe thirty candidates, and then to have their papers marked by an assessor. By my estimate about thirty times as much per person.

British companies who tried the NVQ scheme dropped it very quickly when they found it was costing them so much more to get their staff qualified - and to no real purpose.

The scheme is a typical concoction created by academics who have never in their entire lives had to work in a commercial environment and it would never even have got off the ground had it not been that the Government (another bunch of spendthrifts who would never make it in commerce) made NVQs a condition of the funding that educational establishents receive.

Fortunately good sense is now prevailing and the whole ghastly NVQ mess will soon be a minor glitch in educational history (if anything costing so many millions could ever be minor).

Richard, I would say that you are overly sanguine. Rather than respond to your points, let me just note that "insecurities" and anomolous grades arise even in what may be the most thoroughly invigilated test in the world

I refer to the Olympic games. Almost any other test or contest has less protection, if only for reasons of cost-efficiency. Yet even in the Olympics disputed 'grading' arises: figure skating last year; the US "loss" to the USSR in the basketball championship some years ago; the Chinese gymnast. And only with the massive cost of drug-testing can one be secure against chemical cheating.

Your reasoning has its merit. Of course, it is possible to deter, discourage or reduce the scope of cheating or subjectivity -- and most of your points go to that -- but of course that is not the same as near-elimination of it.

Interestingly, the goal is not to eliminate or reduce cheating, for if you know that cheating has taken place, you can re-test. The holy-grail goal is to eliminate undetected cheating. And unfortunately, that very goal presents a conceptual problem. How can you measure how successful you are in reducing the problems which you did not detect?

quote:

How can you measure how successful you are in reducing the problems which you did not detect?

Clearly it is impossible to reduce, or even measure, that which is undetected. Having said which, the various systems in place to detect examination cheating have been honed over the years and are pretty good these days.

To give just one example, the City and Guilds examination series I am presently invigilating requires the use of calculators and the candidates are provided with calculators by the invigilators. Candidates are not permitted to use their own, as some sophisticated calculators allow text input and it has been discovered that a few candidates have been loading subject cribs into such calculators.

So calculators have now been added to a long list (including pieces of paper, mobile telephones and dictionaries) that are not permitted on candidates desks during examinations.

So examinations 100& secure? Maybe not. More secure than classwork assessment? Almost certainly.

It is relatively easy to bar against any particular method of cheating that has occurred to you. The difficulty is two-fold.

First, you have to be cost-efficient. To take your example of calculators: how does one prevent the testee from bringing in a crib-sheet subtly coded, as by use of a microdot hidden in his clothing? In theory you could require each testee to strip and to take the test in clothing provided by the invigilators. Hardly practical, however!

The second limitation is that the industrious cheater will find methods you may not have thought of. For example ...

No, I've deleted the example, in favor of a challenge. PM to me the complete set of precautions (or so much thereof as you may, subject to confidentiality), including the nature and purpose of the test and the population to which given. And let's see if I can come up with a way to beat it. This could be a fun problem for each of us!

This message has been edited. Last edited by: shufitz,

PS: You might want to take a browse at a partial chapter on cheating in the Chicago exam given to measure overall student progress. It is in Freakonomics, a recent best-seller co-authored by a prominent young economist. Not to suggest that your test would be subject to such obvious vulnerabilities, of course.

I can tell you from the lecture that I attended, students can find ways to cheat that you'd never predict. It may be similar to computer hacking in that advice on how to cheat maybe is best given by those who cheat!

quote:

how does one prevent the testee from bringing in a crib-sheet subtly coded, as by use of a microdot hidden in his clothing?

I might take up your challenge but I can tell you how the two examples mentioned here are dealt with.

1. NO papers, other than those given out by the invigilator, are allowed on candidates' desks so no crib, no matter how cleverly coded, could be used.

2. A microdot needs a microdot reader. No equipment (and this includes pens and pencils), other than that provided by the invigilator, is allowed on candidates' desks. Candidates who are short-sighted or have other visual problems are provided with large print examination papers (so they can't use a coded magnifying glass) and the invigilator has large screen calculators for the same reason.

Clearly cheating, like other crimes, will evolve to try to overcome any precautions taken against it - but I still stick to my base contention - an examination can be made close to 100% secure. Classwork assessment cannot.

quote:

Originally posted by Richard English: I might take up your challenge.

I look forward to it, to the common good of testing bodies!

The rest of our disagreement is now largely eliminated. You agree that a testing cannot be 100% secure, and I agree that security can be made pretty darn good. And we certainly agree that some testings are far more secure than others!

Reviving a thread
This was a good article, talking about how the Wikipedia authors are collaborative and even a bit modest and can't lay claim to their articles.

quote:

Wikipedia has become the world’s master catalogue raisonnée for new clumps of data. Its legion nameless authors are the Audubons, the Magellans, the Berensons of our time. This was made clear to me recently when I unknowingly quoted the work of Randy Dewberry, an anonymous contributor to Wikipedia, in a column on the video game Angry Birds. Dewberry’s prose hit a note rare in exposition anywhere: both efficient and impassioned. (“Players take control of a flock of birds that are attempting to retrieve their eggs from a group of evil pigs that have stolen them.”)

The passage described Angry Birds so perfectly that I assumed it came from the game’s developers. Who else could know the game so well? But as Dewberry subsequently explained to me in an e-mail, that’s not what happened. In fact, according to the entry’s history, the original description of Angry Birds was such egregious corporate shilling that Wikipedia planned to drop it. That’s when Dewberry, a Wikipedian and devoted gamer, introduced paragraphs so lively they made the pleasure of the game palpable. The entry remained.

Like many Wikipedians, Dewberry is modest to the point of self-effacement about his contributions to the site. Because entries are anonymous and collaborative, no author is tempted to showboat and, in the pursuit of literary glory, swerve from the aim of clarity and utility. “No one editor can lay absolute claim to any articles,” Dewberry told me. “While editors will acknowledge when a user puts a substantial amount of work into an article, it is not ‘their’ article.”

I wonder, though, if it really will "never be surpassed," as the author says. Everything related to technology seems to be replaced from time to time.