4 ms·

> I think it's more likely a measurement error, or some subtle statistical effect caused by a combination of probability skewing and demographic changes. Can y

> I think it's more likely a measurement error, or some subtle statistical effect caused by a combination of probability skewing and demographic changes. Can you elaborate?

stupidcar 9y ago
A possible statistical contribution I can think of is the skewness of the age distribution within the population of readers. Here is the distribution charted: https://docs.google.com/spreadsheets/u/1/d/e/2PACX-1vRrops3JL4ZDBo4CAygpEz_J8imEGB8GPekC1XAgZPM2s4mQT-ey1iefV3T2bjPKxkamFYifTYBBk74/pubchart?oid=312146988&format=interactive https://docs.google.com/spreadsheets/u/1/d/e/2PACX-1vRrops3J... Consider just the bins 20 (ages 10-20), 30 (ages 20-30) and 40 (ages 30-40). We can see that, of the three groups, teens are by far the least likely of these groups to read the blog, and people in their 20s are the most likely. Now consider two groups of two-sibling families that lie within this range, where one sibling is in one bin, and the other the next. E.g. the first group where one is a teenager, the first-born is in their twenties, and the second where one is in their 20s, and the first-born is their 30s. Case 1: If you take a random family from the [teen, 20s] group, how likely is it that the reader from this family is younger or first-born sibling? Since those in their 20s are much more likely to read the blog, it is much more likely that the reader is in their 20s, and is the first-born. Case 2: Now take a random family from the [20s, 30s] group. How likely is it that the reader is the younger sibling or the first-born? In this case, because readers in their 20s are more common than readers in their 30s, it is more likely that the reader is the younger sibling. However, the ratio of readers in their 30s to readers in their 20s is much closer than that between the 20s and teens. This makes it more likely in that in case 2 that the reader is still the first-born. Therefore, if the number of families in each group is equal, we would expect to see, in aggregate, more first-born readers than younger siblings. I'm not saying that this is happening in the actual data, you'd need to do more analysis, and probably need more data about the ages of siblings, or that it is a significant enough effect to cause the result found. But it demonstrates that statistical analysis can involve subtle effects that are not obvious without careful consideration. As such, we should be very wary of making any conclusions about causes, either birth order or anything else.
- AstralStorm 9y ago
  This is a good possible explanation, presuming something akin to Flynn effect is in operation on Openness to Experience value?
- sevenfive 9y ago
  But by that logic younger siblings would be overrepresented next bin, [30,40]. What goes up must come down. Personally, I mulled over it for a while and I don't see any mere "statistical" effect that's plausible. I think we are forced to conclude there's a psychological difference with firstborns.

> I think it's more likely a measurement error, or some subtle statistical effect caused by a combination of probability skewing and demographic changes. Can you elaborate?

stupidcar 9y ago

A possible statistical contribution I can think of is the skewness of the age distribution within the population of readers. Here is the distribution charted: https://docs.google.com/spreadsheets/u/1/d/e/2PACX-1vRrops3JL4ZDBo4CAygpEz_J8imEGB8GPekC1XAgZPM2s4mQT-ey1iefV3T2bjPKxkamFYifTYBBk74/pubchart?oid=312146988&format=interactive https://docs.google.com/spreadsheets/u/1/d/e/2PACX-1vRrops3J... Consider just the bins 20 (ages 10-20), 30 (ages 20-30) and 40 (ages 30-40). We can see that, of the three groups, teens are by far the least likely of these groups to read the blog, and people in their 20s are the most likely. Now consider two groups of two-sibling families that lie within this range, where one sibling is in one bin, and the other the next. E.g. the first group where one is a teenager, the first-born is in their twenties, and the second where one is in their 20s, and the first-born is their 30s. Case 1: If you take a random family from the [teen, 20s] group, how likely is it that the reader from this family is younger or first-born sibling? Since those in their 20s are much more likely to read the blog, it is much more likely that the reader is in their 20s, and is the first-born. Case 2: Now take a random family from the [20s, 30s] group. How likely is it that the reader is the younger sibling or the first-born? In this case, because readers in their 20s are more common than readers in their 30s, it is more likely that the reader is the younger sibling. However, the ratio of readers in their 30s to readers in their 20s is much closer than that between the 20s and teens. This makes it more likely in that in case 2 that the reader is still the first-born. Therefore, if the number of families in each group is equal, we would expect to see, in aggregate, more first-born readers than younger siblings. I'm not saying that this is happening in the actual data, you'd need to do more analysis, and probably need more data about the ages of siblings, or that it is a significant enough effect to cause the result found. But it demonstrates that statistical analysis can involve subtle effects that are not obvious without careful consideration. As such, we should be very wary of making any conclusions about causes, either birth order or anything else.

AstralStorm 9y ago
This is a good possible explanation, presuming something akin to Flynn effect is in operation on Openness to Experience value?
sevenfive 9y ago
But by that logic younger siblings would be overrepresented next bin, [30,40]. What goes up must come down. Personally, I mulled over it for a while and I don't see any mere "statistical" effect that's plausible. I think we are forced to conclude there's a psychological difference with firstborns.