Replicate or repeat?

[Geoff]

Does anyone else find that replicate is often used instead of repeat? Here are a couple of examples where I believe replicate has been misused.

A letter in the daily paper a day or so ago was talking about a survey a local school had conducted on the incidence of cyber bullying. In it was this:

The student research team presented their findings in Mexico City last month sparking interest among the international audience to replicate the study.

On the radio, there was discussion about the earthquake in China:

Speaker 1: There are [quantity X] dead in the region of [wherever].

Speaker 2: That's right. and that figure is replicated over and over in other regions.

What do you think?

[Dave M]

I think "replicate" is correctly used in the first example, and I'm not sure about the second.

If the same student research team carried out the same survey again, on the same target population (though not necessarily on the same individuals), they would repeat the survey. However, the newspaper is referring to the interest shown by the audience in carrying out the same sort of survey, elsewhere around the world - which is replicating it.

In the second example, the use is less justifiable - numbers aren't "replicated" in the same way. We don't say I have ONE house, and gosh, there is ONE school nearby and ONE name of the street, so the mumber one is being replicated! It's plainer speaking just to say "repeated".

[Paul Doherty]

I agree about example one. If another team repeat the experiment they might get different results. But if they replicate it they will get the same results. They might write to the original team and say "we repeated your experiment but we could not replicate your results".

I'm tempted to allow it in the second example also. To replicate is to produce the same thing in (broadly) the same way: if there are 10,000 dead in Szechuan and that figure is replicated elsewhere, the implication is that both disasters have the same cause. In the context, however, that probably goes without saying, so repeated would have been just as good and would have been simpler and less pretentious.

[Pete]

Sorry to disagree with eminent posters but I really dislike the second example. The SOED does suggest that 'replicate' means, inter alia, 'repeat', but I do not think they should be used as synonyms.

I agree that the first example is correct. As an ex-scientist, I take it as axiomatic that no experiment is valid until it has been repeated and the results have been replicated, preferably by other labs or researchers. This is the correct use of 'replicate', meaning to make a replica, copy, reproduce exactly.

I suggest, however, that the second example is wrong on several levels:
- it is almost certainly incorrect as a matter of fact, i.e. it is extremely unlikely that the number of casualties is replicated, even within an order of magnitude, in several different regions;
- I don't think'replicated' would be the correct word in this case as there is no reproduction of the same number;
- I am uncomfortable with 'repeated', too, and I would prefer to recast the sentence to say something like, " and similar numbers were reported in other regions";
- It seems to me that "replicated over and over" is a tautology.

[Paul Doherty]

Does one replicate the experiment or replicate the result?

[Pete]

May 16, 2008 18:49:20 GMT Paul Doherty said:

Does one replicate the experiment or replicate the result?

You repeat the experiment to see if the result can be replicated.

[Dr Mildr]

I also would repeat a study to try to replicate the results.
Strange how different disciplines have a different take on the use of the word 'experiment'.

[Paul Doherty]

So to replicate the study (Geoff's example) is wrong?

[Pete]

In the context of the previous few posts, I think so, yes. But it is possible to replicate a study or experiment, by doing everything in exactly the same way. As a scientist, I would suggest that if you replicate the study, you must also get identical results, so that they, too, are replicated.

Somehow I am not sure this is helping, sorry.

[Sue M-V]

Wow, that made me nervous! Having just used "replicate" in my last post, I had to have a quick look back to see whether I was guilty of misusing it! I think it's all right, though.

I agree entirely with what Dr Mildr has to say, and what most of you seem to be agreeing about, and I share Pete's misgivings about the use of "repeat" or "replicate" with numbers of dead - perhaps because the numbers are not actually exactly the same. I also have an undefined feeling that either of the words applied to this subject is somehow inappropriate. I'd want to say something that expressed the horror of all the thousands of dead people, and not just talk about repeated numbers.

Sue

[Tone]

Pete,
>As an ex-scientist<

Me going off at a tangent, as usual --

How do you manage that? Surely your whole "attitude-to-reality" hasn't been wiped out? You surely view issues with the same "questioning" or investigative outlook and apply the same mental procedures.

I asked a similar question of another poster some years ago (the stroppy one who departed after calling us all "sheep") when he claimed to be "an ex-philosopher". He didn't answer, and no-one else picked up on it.

But the statement intrigues me.

Tone

[Paul Doherty]

I agree with Pete's and Sue's misgivings about the second example; I can see why the reporter used replicate rather than repeat though. No doubt it's hard to hone every on-the-spot report to choose exactly the right word.

In the first example, I think we all agree that one might replicate an experiment by repeating it and getting the same results. Indeed, that one should be able to do so is a necessary part of the scientific method.

[Geoff]

May 16, 2008 18:51:40 GMT Pete said:

You repeat the experiment to see if the result can be replicated.

Which is exactly my point about the first example. In the context of an international audience listening to the findings of the students' survey, I think it's reasonable to assume the letter writer was saying that audience members were keen to conduct similar surveys to see the extent of cyber bullying in their own countries? That doesn't sound like replication to me.

[Pete]

May 16, 2008 21:06:23 GMT Tone said:

Pete,
>As an ex-scientist<

Me going off at a tangent, as usual --

How do you manage that? Surely your whole "attitude-to-reality" hasn't been wiped out? You surely view issues with the same "questioning" or investigative outlook and apply the same mental procedures.

Tone

Good point, Tone. I spent my whole school and university career wanting to be a scientist, then spent 5 years discovering that I wasn't very good at it! So I switched to something completely different (a la Monty Python) and became a tax inspector! So when I wrote "as an ex-scientist", I meant "as someone who used to be a professional scientist but who now makes a living another way", (as a tax consutant - I left the Revenue 11 years ago).

But you are right to suggest that I "view issues with the same questioning or investigative outlook and apply the same mental procedures." This is of course, quite frustrating at times, since interpretation of statute or judicial decisions is necessarily a lot less "scientific" than interpretation of facts in a scientific discipline.

Not that it matters but, since I am baring all, when I was "being" a scientist I was a biochemist.

[Dave M]

> I think we all agree that one might replicate an experiment by repeating it and getting the same results <

I don't think you even need to get the same results, Paul.

If you've replicated the results, then of course the results will be the same, but suppose you just replicated the experiment: you took the original group's method, and applied it to your own population. You haven't now repeated the original experiment, but you have replicated it. (You've run a "copy" of the original, not re-run the original.)

If, having replicated the experiment, you get different results, then you have some new findings: you can hypothesize that the original results are incorrect, or that the results depend on the population.

As with "mathematics", we're dependent here on definition: what makes "an experiment" that particular experiment?