allenm@ittvax.UUCP (01/03/85)
I think many siblings correlates with low income level, which correlates with low test scores. > Recent studies suggest that kids with many siblings get LOWER SAT scores > than do kids with fewer brothers and sisters. The proposed "theory" is that > less time is spent with or available from "adults". > > [Let's avoid discussing the "studies" which are clearly lacking adequate > control groups for quality of interactions, spacing of siblings, > environmental factors, etc.] > Often a reaction to a study is to disagree with the conclusion and to propose a different one based on personal obversation or "common knowledge". This is surely not science. Why not discuss the statistical or sociological methodology? Why not propose alternative conclusions based on similar studies? > Can Scolastic Aptitude Tests that measure `common sense' and what was once > thought to be inate ability or lack of it really be affected by degree of > exposure to `adults' or adversely affected by substituting exposure to > siblings? You are arguing here with the reasoning for what you claim is the hypothesis ("The proposed 'theory' is that less time is spent with or available from 'adults'.") for the statistic ("Recent studies suggest that kids with many siblings get LOWER SAT scores than do kids with fewer brothers and sisters."). I agree that such reasoning should be supported by additional data, were it to be seriously advanced. Who actually proposed this theory? > ... My observation is that siblings 4 to 5 years older often supply > better inputs to their younger siblings than do adults! Often because they > share or can relate to similar interests (like drawing, playing with dolls, > playing with cars) than adults can supply. > > Personally I blame the last 20 year drop in SAT's on TV, which is a better > sitter for several kids than for one. With smaller families perhaps kids are > spending less time with the TV and more time with their parents? Whatever! Your hypothesis is at least as unsupported as the one you are attacking. Do you believe that people will agree with you on the basis of personal "observation" or what you want to "blame"? How about a plausible alternative hypothesis, even with missing citations? I recall reading about high correlation between large family size and low income level (plausible, but no reference). From there, a connection between low income level and low standardized test scores also seems plausible. Since I'm not a Sociologist, I'm satisfied with such reasoning. Since I believe in the scientific method, I think such correlations are capable of being checked. I don't think it would be easy (possible?) to check whether siblings give "better input" to each other than adults do. And the effect TV may well be independent of family size (that may be possible to determine). -- Allen Matsumoto ITT Adv. Tech. Center, Stratford, CT 06497 203-385-7218 (decvax!ittvax!allenm)
elt@astrovax.UUCP (Ed Turner) (01/04/85)
I did not see the original article in this discussion, but it might be worth mentioning that there was a very elaborate study done in either Denmark or Holland (I forget which) in the 1960's which concluded that the determining factor (= single most important factor) in IQ scores of children is related to the number of older siblings a child has at birth. In particular, the mean age of all persons in a household at a child's birth was the best correlating factor with IQ score at some relatively mature age like 10 or 12. This study was based on one of the largest (2nd largest I think) populations ever analyzed with respect to IQ and was carried out with impressive (to my fairly expert eye) statistical sophistication. Moreover, it was shown that this correlation either largely or entirely explained many other observed correlations such as those with birth order, economic and social class, race, and so forth. In other words, children born into households with the same mean age showed little or no difference in later IQ scores regardless of many other factors such as those just mentioned, and observed correlations of IQ with these other factors could be explained by differences in the distributions of mean household ages at birth. Although I read about this work in some detail when it came out, I do not really follow the subject (since I have a low opinion of the whole concept of IQ) and never heard how it fared as far as verification in other samples, acceptance by sociologists, etc. Does anyone know? Ed Turner astrovax!elt
lat@stcvax.UUCP (Larry Tepper) (01/04/85)
> Since I'm not a Sociologist, I'm satisfied with such reasoning. Since I > believe in the scientific method, I think such correlations are capable > of being checked... > > Allen Matsumoto > (decvax!ittvax!allenm) Allen was responding to another article discussing the CAUSES of low SAT scores. The thing to remember about correlations in general is that there is no implication of cause involved. The classical example, as I have heard it, goes something like "There is a strong correlation between sun spots and droughts. Now, do sun spots cause droughts, or do droughts cause sun spots?" The point is, all that one can say is "a correlation exists". The cause, however intuitively obvious, could well be something else. -- One of the survivors... {ihnp4 hao ucbvax!nbires}!stcvax!lat Larry Tepper Storage Technology, MD-3T, Louisville, CO 80028 303-673-5435