Martin James writes,
I came across this article in NYTimes and it raised a few doubts. They write,
For example, in a study that Professor Dweck and her researchers did with 400 fifth graders, half were randomly praised as being “really smart” for doing well on a test; the others were praised for their effort. Then they were given two tasks to choose from: an easy one that they would learn little from but do well, or a more challenging one that might be more interesting but induce more mistakes. The majority of those praised for being smart chose the simple task, while 90 percent of those commended for trying hard selected the more difficult one.
What I have reservation on this example is that, these fifth graders have already given many tests before giving this test. So how is that we can attribute their choice of task on just the basis of this test. Are we assuming that these children have no prior conditioning before? Unlike say a Milgram experiment, these children are not subjected to anew behaviour, it is a plain test.
Also on the basis of this test I can conclude that for an organization to motivate its employees to take up difficult challenges is to give them a test and then commending for trying hard.
My reply: I dunno, but maybe this sort of study is relevant for kids in schools nowadays who really do get tested a lot.
First, I think it is kind of difficult to criticise research on what somebody else (NY Times) reports about that research.
But the main problem is that Martin seems to take issue with the research design that is described, i.e. that the kids already are very used to taking tests. But actually this does not look like a strong case against the findings: Even though they are already used to taking tests, the randomly assigned "treatment" seems to have had some effect – by usual standards of social experimentation, that even seems like a strong argument in favor of the findings.
Finally, the whole point isn't about taking tests, rather it is about telling people that they can change their behavior (i.e., it's not too bad to make mistakes as long as you learn from them) instead of telling someone there "smart" or "intelligent" or "stupid" which implies that they can not change their performance by trying harder. The consequence for organizations is obviously not to give out tests but rather to praise their employees.
I agree with the comment above that the randomization of assignment should take care of the doubts of prior testing versus this particular test. This is not an uncommon experiment in education, and it is generally found that praise and encouragement leads to better results and that the feedback from adults (generally teachers, obviously) matters. The biggest issue that I always had with such experiments was that they are often held in one classroom with one teacher, which leads to all kinds of entangled with personalities, how the classroom was run previously and is run now, etc.
I was always interested to see if this experiment has been duplicated in other countries. I feel like this result is inextricably linked to the American culture, where kids are, generally speaking, used to being encouraged and sensitive to criticism. What happens when this is done in a culture that is not accustomed to such praises but more used to criticism? If anyone knows of any good papers on this, I'd be interested to read them.
My guess would be that it's not a change in their beliefs about how smart they are that's having the effect, it's the short term boost in self esteem that arises because of the belief. If someone says to me "Wow, that paper you wrote was really great!" my opinion of the paper's greatness isn't likely to change, but I'll still feel pleased, get a self esteem boost, and maybe therefore take the less safe of two choices in the next few minutes.
The report of the report doesn't talk about the hypothesised mechanism. I don't think we can argue over whether the effect exists (assuming that they randomized appropriately, etc) but we can argue over the mechanism.
I also would like to read about similar studies performed in other countries. On http://www.thecyberprofessor.com/ I was reading that researchers can design a study that will yield the results that are desired. Statistics is fairly new to me and I think it would be interesting to see what the results are for this study outside the US.
We tend to follow others' expectations of ourself. The pygmalion effect. If a teacher believes little Johnny is not going to do well, then little Johnny could possibly not do well because his teacher had no confidence in him.
If I am complimented on being a nice person, I will be nice person, even though it may only be for a short while until someone makes me angry.
The importance of mindset seems very real based on Dr. Dweck's research. I agree that there is some question about how much effect an individual teacher has vs. another.