First of all, I don't think the article states whether it was a longitudinal or cross sectional study. Most national youth studies these days tend to be longitudinal. Second, even if it's cross sectional, it's not meaningless. Let's say you found that 8% of 15 year olds in 2000 reported being "gay," but in the same year only 2% of the 25 year olds did likewise. Such evidence by itself would suggest that "gayness" is not an intrinsic or pre-determined characteristic of individuals.
Your remarks about margin of error are simply speculation.
I have read the published results of the entire study (Laumann, et al: The Social organization of sexuality in the United States.) since I've written a text to which they were relevant; they were not longitudinal. Moreover, this study was not conducted "these days" but rather in the late 80s/early 90s. The sexual behavior questions were all between 1988 and 1992 iirc. The questions did not even appear in the GSS before 1988 and the cited results were published in 1994. That alone eliminates the idea of a 9 year longitudinal study (between 16-25 years).
BTW, the funding was yanked and that's why similar studies have not continued on a rolling, longitudinal basis.
Moreover, youth results in general are extremely problematic for two more reasons: (1) they require parental permission, and more conservative households are far less likely to permit adolescents to participate in sexual surveys; (2) most or all respondents are drawn from a restricted demographic, oftentimes related to a clinical setting.
My remarks on the margin of error are based 100% on the actual results which I analyzed. Your remarks are assuredly pure conjecture.