HSK paper marking question

The April 2006 HSK results can be checked online since yesterday afternoon. I was pleasantly surprised to see that I got 100 for 阅读. I thought that I had made a few mistakes in that paper . For me, it was the most difficult and challenging paper of the four. Does anyone know about how they mark the papers?

Congratulations!

The multiple choice sections are marked by optical scanner. For the intermediate exam, that's pretty much everything apart from the writing of characters at the end of the 综合 section. I presume they're done by hand.

For the writing and speaking sections of the advanced exam I know BLCU gets groups of CSL postgrads together, trains them and then takes them to a hotel outside of Beijing for some intensive marking action. I'm not sure if they do all papers there, or if there are regional marking centers.

thanks!

Does getting 100 for 阅读 mean that I got all the answers right for that paper? Or maybe they adjusted the marks accordingly? Example, if the highest mark gotten for that paper was 90, do they make it become 100 instead? I don't know how to explain this clearly. Hopefully, someone understood what I meant.

I know what you mean - I just don't know the answer. I think I might have known it once, but it's been a couple of years since I last took the HSK. . .

I think there is some curving, or at least some wierdness going on. I got 100 too, but wasn't that confident about that section. Not to mention in my total score, they gave me three more points than the average of all four sections....which is wierd.

Oh well, an 'A's and 'A', I guess.

You're telling me there is some curving going on. Last year I got an 8 (filling in at random would have got 25 (assuming no curving)).

I got 84 for 阅读, answering randomly the last third of the questions

so I'm pretty sure they do some curving

This is from the website:

"1. About the scores of HSK (Elementary-Intermediate)

A section score of HSK (Elementary-Intermediate) is a scale score with 50 as the score mean and 15 as the standard deviation, while the total score is a scale score with 200 as the score mean and 60 the standard deviation. These scores show the relative position of the candidate in a standard sample group. The table below helps to locate the relative position of a candidate in the whole group. The right column indicates the percentage occupied by those whose scores are lower than the corresponding scores in the standard HSK reference group. Therefore, as it shows, the total score of HSK (Elementary-Intermediate) is not simply an add-up of the section scores. "

http://www.hsk.org.cn/english/Intro_hsk2.aspx

In other words, I don't think it is possible to know if a "100" really necessarily meant that you got all the questions right, although I'd assume you would be very close to that.

In any case, congrats weiming, youpii, and aline! I am away from home, and so I can't check my score. I have a horrible feeling that I bombed the grammar section. Bombed it enough to ruin any other positive things I might have done. That's another feature of the HSK I am sure many people curse!

That's weird, I got an 84 for 综合whatever that means. I was so ashamed. But then I did write 健康的康 as 唐， that probably pissed off some test reviewer somewhere.

Oh, now don't none of YOU make that sam mistake either! 康　used to be a clever ideograph: a hand holding rice [food: written to look like 水] 隶 under a shield 广 [which used to look more like a shield] access to food, a good way to 'protect ones health'.

Of course since simplification it's just a dumb top/bottom phono graph 广[guang] rhymes with 康 [kang], 隶[li4] in it's simplified form is very reminiscent of the original form, but means servitude in it's modern meaning 奴隶, slave [and meant a kind of scribe or subordinate in it's traditional meaning].

So that's why 康 is now a silly picture of 隶 under 广. No wonder I forgot it.

I love the 'grammar' section. It's so funny to be tested on 'grammar' in a language where dictionaries don't even list the part of speech. *nyuk*

aarrgh!

I can't get my results from the website! Does anyone know when we'll be able to go to BLCU and pick them up in person?

Thanks!

I posted about this a while back.

From what I can understand of the website, it's much more serious than curving as I traditionally understand it (making slight adjustments to marks to allow for an exam being slightly harder/easier than other years). Your entire mark is based entirely on the performance of the other people on the test, rather than an objective standard. As I read it on the Basic test for example, the lowest scoring 20% or so will always fail the test for being in the bottom 20%.

I think it's a pretty shoddy system as it makes no attempt to hold the exam board responsible for maintaining consistancy each year. If, for example, you assume that the quality of Chinese teaching is getting better (the textbooks certainly are), then every years it's going to get harder and harder to achieve the same grade. Or suppose the Korean or Japanese government issues it's own Chinese test and suddenly there's a sudden shift in the demographic of the people taking the test. That would cause serious problems.

Worse still, I can find any statistics from previous years, so I've no idea at the moment whether I'm in a reasonable state shoot for level 3 in the summer. Not impressed at all with the HSK. .sulk.

I think 100 must mean that you got the most questions right out of anybody. That probably means you did get 100 percent, but there's no way to be sure.

P.S The systems pretty complicated and I may have misunderstood it completely. Please tell me I'm wrong!

That makes sense - so if you get '100', it means you got (joint) top mark - ie, nobody else got a higher mark than you.

Or suppose the Korean or Japanese government issues it's own Chinese test and suddenly there's a sudden shift in the demographic of the people taking the test.

I've wondered about this posibility because it seems that the number of people taking the HSK is skyrocketing, especially in Korea. Here is a bit of an article talking about the surge in studying Chinese in Korea:

"Spurred on by the China boom, interest in Chinese language learning has surged in recent years. In South Korea, annual increases in the number of students taking Chinese have hit an astonishing 38 percent, while corresponding figures for those taking French, English, Japanese or Spanish have hovered around 2--4 percent."

http://www.chinatoday.com.cn/English/e2005/e200505/p64.htm

One might expect that since the test is graded on a curve, it might be hard for native English-speakers to do well in the areas that a majority of Koreans do well in, say, the grammar or 综合. While it might be easier for English-speaking laowais to do well on the curve in areas like listening. Also, as a stereotype, it seems that due to cultural patterns of language learning, Westerners might be a bit better at speaking compared to some Asian nationalities, and of course this is not a component of the 初中.

Also, I couldn't help but think about what a massive influx of test takers might do during the actual test. The testing center I took it at in Shanghai seemed to be 90% Korean, with a few Japanese, and myself the only noticable non-Asian. One of the Korean guys in front of me kept filling in answers after the time had run out. It looked like he left most slots blank, and he was just trying to frantically fill in some random guesses. The examiner then asked him to stop. He wouldn't. So, finally he gave his booklet and ripped up the test paper. He had then yelled at the examiner and threatened to fight the examiner. In the end, after threatening and yelling at five examiners, he was removed from the test taking center. (I couldn't help but feel a bit sad that he ripped up his test paper!) In any case, if this is the kind of non-serious test taker is at all representative of even a significant percentage of Korean students, then this might mean that the test is a bit easier then it was, say, 10 years ago, even if the quality of textbooks has become better. Who knows? I guess it is is only possible to speculate.

I don't mean to sound like I'm picking on Korea. But it does seem to be one of the few countries in which the HSK might be a known test to many employers, which I don't think can be said of other countries. If so, this would give an added incetive for many people to take the test. See mrtoga's comment, for example:

http://www.chinese-forums.com/showpost.php?p=69508&postcount=17

Anyway, does it seem that there website has crashed? A bit frustrating.

When you finally get the hard copy version of your results, there used to be a slip of paper that explains how the number of questions you got right corresponded to what percentage of questions you got right and what grade boundary it put you in. When you finally see that, it might clear things up a bit.

With reading, I think it's designed to be challenging, its a comprehension test not a speed reading test. So there should be enough potential marks to score over 100%.

Your entire mark is based entirely on the performance of the other people on the test, rather than an objective standard.

What if too many people who should have taken the advanced test chicken out and take the elementary-intermediate test instead? There's nothing to prevent that from skewing the curve?

There are rumours that certain month's tests are skewed by the ethnic minorities in China who take the HSK as part of their university entrance exam, can't remember which months tests though. Although it should only affect the advanced level.

What if too many people who should have taken the advanced test chicken out and take the elementary-intermediate test instead? There's nothing to prevent that from skewing the curve?

Hehe. I assume that the top 10 or 20% of test takers are really fairly good. I can imagine, based on the curving system and a limited number of questions, a few wrong guesses might really hurt you on the curve.

Anyway, so, the test score goes from 0 to 400. 200 is the mean. Standard deviation is 60. What I what to know is, what percent of people get scores of 337 and above? It seems to me that this is a crucial question in understanding whether or not to take the 初中 (trying for an 8 ), or whether to take the 高 (trying for a 9 or above). My math skills have really gone down hill, so I would love if someone could help me out!

(337 and above is the level for an 8, the highest level. )

In any case, here’s how I did: 听力 79 (band 7), 语法 72 (6), 阅读 78 (7), 综合 82 (8 ), Total 311中B (Band 7).

My reading and listening weren’t as good as I thought they’d be, especially since I got an 88 on reading on the other time I took the test. So that’s a bit disappointing. My grammar is better than I thought. I wrote characters and practice essays for three months in preparation for the advanced level test, so that paid off in the 综合 section. The other time I took the test, I had a 64 in the 综合, and I only wrote two or three characters correctly. This time, I’m pretty sure, I only missed two or three characters in that part.

In any case, the scores for the advanced level test don’t seem to be out yet.

See these pages on how to calculate the area under a normal distribution curve:

http://mathworld.wolfram.com/StandardDeviation.html

and

http://cnx.org/content/m11166/latest/

http://www.physics.csbsju.edu/stats/descriptive2.html

Fraction Number of Standard

of Data Deviations from Mean

50.0% .674

68.3 1.000

90.0 1.645

95.0 1.960

95.4 2.000

98.0 2.326

99.0 2.576

99.7 3.000

the test score goes from 0 to 400. 200 is the mean. Standard deviation is 60... what percent of people get scores of 337 and above?

(337-200) / 60 = 2.28 standard deviation (sd) away from the mean

Extrapolating from the table above, that equal to about 97% of the curve, which means that there's 1.5% on either side. In other words, about 1.5% of test takers score above 337 (i.e. 98 percentile or so).

(311 - 200) / 60 = 1.85 sd or about 93%

7% / 2 = 3.5%

Your score of 311 is at about the 96-97 percentile.

Extrapolating from the table above, that equal to about 97% of the curve, which means that there's 1.5% on either side. In other words, about 1.5% of test takers score above 337 (i.e. 98 percentile or so).

Thanks gato. I'm bad at math, but I had guestimated that the percentage of people who could get an 8 was very small (I had estimated it at about 5% or so, but it seems less than that).

So, I certainly imagine that being within the top 2% isn't easy, especially if there are any significant amounts of fluent ethnic minorities that need the test for college, or whatever. Even if there are no ethnic minorities taking the test, there are numerous clever, kick-ass test takers and intelligent people with great Chinese out there that might occupy that territory. It almost looks like getting a 9 might be easier than getting an 8, even though the Gao is a much harder test.

Thoughts? Any other math-inclined people who can shed some light on this?

thanks for the explanation.

Congrats to all the others.