In his excellent book A Question of Intelligence: The IQ Debate in America, former Fortune magazine editor Daniel Seligman describes what it’s like to take the WAIS-R IQ test. In part 4 we learned he did poorly on the Picture Completion subtest, and on page 4, he describes his performance on the next subtest:
I did much better on the next exercise: the Digit Span. This is a test of short-term memory in which the examiner reads off a string of digits and asks you to repeat them. He begins with three digits and proceeds incrementally to nine. If you miss twice at any level, you are judged to have scored at the next lower level. I made it to nine with no special difficulty.
But there is a fiendish round two to Digit Span. Beginning with two digits and proceeding on to eight, the examiner asks you to repeat the digits backward. This was a lot tougher. I found that once we were up to six and seven digits, it helped me to “chunk” them into groups of three or four. Of course, you still have to hold the first chunk in your head while focusing on the second, and twice during the exercise I lost the first chunk. On the other hand, I batted a solid .500 on the eight-digit span backward-accomplishing it once in two attempts-and found my morale somewhat restored.
From Seligman’s description, it’s clear that he made it up all the way to the hardest item on both rounds, but got only partial credit twice. This implies a raw score of 26/28. This equated to a scaled score of 17 in the peak age group, but in his own age group (55-64) it equated to an astonishing scaled score of 19 (IQ 145, U.S. norms; IQ 144 U.S. white norms).
[Update (Nob 20/2015): because WAIS-R norms were a decade old when Seligman was tested, and the Flynn effect increased WAIS Digit Span scaled scores by 0.059 points a decade from 1978 to 1995 (Flynn, 2012), his scaled scores must be reduced to 16.94 and 18.94 respectively. An age adjusted scaled score of 18.94 remains equivalent to a spectacular IQ of 145 (U.S. norms; 144 U.S. white norms) on that subtest].
But the great David Wechsler would not have been impressed, writing that “as a test of general intelligence”, Digit Span “is among the poorest.”
In The Measurement of Adult Intelligence (third edition, 1944) Wechsler wrote (pg 83-84):
The ability involved contains little of ‘g’, and as Spearman has shown is more or less independent of the general factor. Our own results confirm these observations. For a long time we considered the desirability of eliminating the test from our battery altogether, but finally decided to retain it for the following reasons: (1) While memory span for digits backwards and forwards is on the whole a poor measure of intelligence, it is nevertheless an extremely good one at the lower levels. Except in cases of special defects or organic disease, adults who cannot retain 5 digits forward and 3 backwards will be found, in nine cases out of ten, to be feebleminded. (2) Special difficulty with the repetition of digits forwards or backwards is often of diagnostic significance. Obvious examples are the memory defects which constitute clinical symptoms in certain organic and other types of cases. A marked falling off in memory span is often one of the earliest indications of them.
According to scholar Arthur Jensen, Digit Span forward has a g loading of 0.3 while Digit Span backward has a g loading of 0.6. The school board I attended as a kid labeled Digit Span a measure of “Span of Attention”.
[Update: Nov 20/ 2015: Sadly, it has been brought to my attention by commenter “jeanbedelbokassa” that Seligman is actually saying he got 25/28, not 26 out of 28. This means his scaled score (relative to the peak age group) is 16, and his scaled score relative to his own age is 18. Because of old norms, these get reduced slightly to 15.94 and 17.94 respectively. An age adjusted scaled score of 17.94 equals an IQ of 140 (U.S. norms; 139 U.S. whites norms). Still an incredibly high score. For making this error, points also need to be deducted from my reading comprehension IQ :-)]
Interesting, I thought digit span was a good, simple measure of g. And it’s too bad it doesn’t count for more in Seligman’s case.
Nevertheless, I still expect Mr. Seligman full scale to be around 130. His Jewiness will help him find success on Vocabulary (he did take that subtest, right?)
Yes, he did take all 11 subtests of the WAIS-R which includes vocabulary. Fortunately, that was one of the subtests for which I was able to determine a score
Forgot to deduct points for old norms on this subtest too, not that it makes much difference on digit span which had very little Flynn effect. But update added in bold.
Tested myself. 8 digit backwards, can’t do 9 (maybe 1/4 the time correct at 9). LSAT 180.
Sounds like you have a BDS IQ of about 130, assuming you haven’t practiced the test.
By contrast your LSAT IQ is around 150:
https://pumpkinperson.com//?s=lsat&search=Go
I didn’t practice beforehand. Memory has always been one of my weaknesses and I was surprised I tested above average.
Very impressive!
What would a 20/28 correspond to (%ile wise)?
Two questions:
1.) What counts as practice? I do one of these once every say 11-12 days. These days I get 8 digits, although when I started it was more like 6-7.
2.) Isn’t ‘chunking’ considered cheating?
1) if you’ve done the task more than a few times in your life, and especially within the last 6 months, you’re likely to show a practice effect which invalidates your score
2) Chunking is a mental process and the psychologist has no way of getting inside your head. It’s technically not cheating but it is an example of how test sophistication can invalidate IQ tests
Please use another username.
Interesting. Out of curiosity, what does a 7 on forward and a 5 on backwards equate to? 50th percentile? How about 7 on forward and 6 on backwards?
For young American adults the mean forward is 6.8 (SD = 1.27) and mean backwards is 5.1 (SD = 1.51). In theory 68% of the people will fall within 1 SD from the mean (above or below), and 95% will fall within 2 SD from the mean, though the distribution’s not perfectly Gaussian, especially at the extremes.
Well that’s disappointing…=(
Can’t you chunk most everything even in the real world?
I saw something in one of the publicly available WAIS-IV clinical use books that the subtests measure how well you can rehearse or chunk. It seems like the test is assuming that you will chunk or rehearse, and that’s probably what’s happening too.
I wonder though, do most people rehearse or chunk?
Also, if you improved your digit span, doesn’t that mean you are improving your memory too? I mean, you are actively using your working memory, and it’s not like you can just ‘know’ the answers, unless you memorized those numbers which would make absolutely no sense whatsoever.
If you practice digit span, you only improve your memory for English numbers, not your auditory memory in general. Essentially all you’ve done is invalidated the test as a sample of your STM
I think the low digit span helps explain mug of pee’s reading disability. Reading comprehension requires tons of working memory, especially reading complex articles
yeah. the digit span is the most easily “hacked” of the wechsler subtests. maybe that’s why it has the lowest g loading.
I see, so it won’t translate into improvement with letters or words. Only numbers.
Correct
Isn’t auditory memory and read memory somewhat different?
I also have another question, which applies more so for the Verbal sub tests. Say you were thinking about the questions, and then you get new bells in your head, does that shout ‘practice effect’, or does that mean that you didn’t show your potential on test day?
Practice effect
How is that a practice effect? In the real world, this happens a lot.
I’m not trying to argue, I just want the reasoning.
Because the whole point of an IQ test is to test everyone’s ability under a fixed set of conditions that are the same for everyone. If we count the insights you have after the test, then we have to do the same for everyone in the norming sample which would be like a rising tide that lifts all boats without changing their relative heights.
Just for curiosity, how common are these new insights in your experience?
not that common, but then most people don’t think much about the test after they take it. If you have unique insights after, it’s hard to know whether they’re because of superior intellect or just interest in the test.
If you want to take a test that allows more time to think and reflect, I suggest the Ron Hoeflin tests, which are unsupervised and taken over days, sometimes years.
Some people take IQ tests within 3 months, practice effects on these verbal sub tests aren’t high. Wouldn’t that mean that the insights are due to higher intellect?
I must be honest, I’m really really insecure about my intelligence. That’s why I’m asking these questions. I also have OCD which helps.
I’m not necessarily an IQ believer either, but I do believe that it measures something, it certainly does.
Also, out of people who DO think about the tests, how common are these new insights?
Also, if people could theoretically get the right answers, wouldn’t most do so on the test itself?
Also, if I practiced digit span with 8 numbers, but not letters, but if I get the letters as easily as the numbers, would that mean my memory improved since I got the letters?
It’s probably just narrow transfer to a very similar type of memory or learned strategies for taking that type of test. I doubt your auditory short-term memory improved in real life situations
What kind of other memory tests would be there to test my real auditory working memory, independent of any practice on the digit span? I do remember after a 6 month period, taking a digit span and getting 34-39, but I don’t remember the exact number, so that’s what I’m trying to figure out.
Also, is letters that similar to numbers? I do remember you saying that digit span stuff won’t carry into letters. I’m just wondering.
In one study people who practiced digit span to the point of repeating as many as 100 digits, were still only able to repeat 7 letters.
But that’s just one study. The general finding is cognitive training only slightly improves very similar abilities and has no impact on more different abilities
Wait, so my letter ability would be doing a mixture of 7s and 8s? My real ability to do 7s easily, and 8s with a little bit of an effort?
How many digits could they initially remember? Can I have a link to the study?
It’s described in Jensen’s book. I don’t recall if he provided a citation but I’ll check late tonight
If I practiced remembering a list of words in order and went from a 7 to a 8 forward digit span with numbers shown one at a time, and a spatial span of 5 to 7 with getting 8 once, did I improve my working memory? To be clear I didn’t practice the spatial span or the forward digit span, but just thought of it as a good measure to see if my WM improved. Also I am able to remember more of what people say, but that’s a bit less measured and more anecdotal. These abilities seem different so the idea that cognitive training can’t improve overall abilities seems false. It’s worth noting I improved from memorizing 3 words in order to 5 of the list of animals given 2 seconds an item, along with an improvement of 3 to 4 words in order of people of various professions which is unpracticed and after like 4 tries I got 5 with 18 items I made sure to pick from a different pile so I couldn’t have been familiar with the items.
Any thoughts?
pumpkin, if you practiced digit span to the point of getting the same score on letter span with the same strategy, wouldn’t using the strategy improve performance in real life?
In real life you don’t get time to rehearse so no
Pumpkin, if you take a digit span test once every four days, and in between you don’t employ a strategy, but after 12 days your score goes up- the items are completely different to. I must say that the first time I took it I wasn’t really focused- the next time I got a 6 and then I got a 7 twice. Can I assume that on a good day when I’m focused, I’d have a digit span of 7?