5 ms·

> One new study of 6,300 heterosexual couples Why are major sociology studies done with such small samples? I understand that if everything is done correctly

by htag 4y ago

> One new study of 6,300 heterosexual couples Why are major sociology studies done with such small samples? I understand that if everything is done correctly a relatively small sample can strongly reflect a population. However wither someone has a full time job or not (1099 or W2) is reported to the IRS, and marriage/divorce records are public. Why are government departments not collaborating with academic institutions publish these types of insights?

sidlls 4y ago
Assuming proper control for biases in the population, a sample size of 6300 is not tiny. The estimated error for such a population is going to be around 1% (give or take).
- htag 4y ago
  It's just that 1. Sampling and self reporting introduces many concerns to the validity of the study. Let's not downplay the negative effect these can have. 2. It's expensive to collect consent and data. 3. Federal grants often fund these studies. Tax payers are essentially paying to collect the same data twice.
- projektfu 4y ago
  That's true assuming the sample is not biased. I'd be interested to see a kaplan-meyer curve for the two groups.
ellisv 4y ago
Most government departments don’t have research in their mandate and are underfunded/staffed.
- htag 4y ago
  This type of record collecting and academic collaboration might be a good responsibility for The Library of Congress.
  - tomrod 4y ago
    It's a completely reasonable question -- reach out to your Congressional representatives to kick off the initiative!
- woodruffw 4y ago
  Not research, but most government departments do have dedicated statistical bureaus! This would probably fall under the BLS[1] or BEA[2], which are bureaus of the DOL and DOC, respectively. Another potential fit would be the IRS's SOI[3] division. [1]: https://en.wikipedia.org/wiki/Bureau_of_Labor_Statistics https://en.wikipedia.org/wiki/Bureau_of_Labor_Statistics [2]: https://en.wikipedia.org/wiki/Bureau_of_Economic_Analysis https://en.wikipedia.org/wiki/Bureau_of_Economic_Analysis [3]: https://www.irs.gov/statistics/soi-tax-stats-statistics-of-income https://www.irs.gov/statistics/soi-tax-stats-statistics-of-i...

by htag 4y ago

sidlls 4y ago

Assuming proper control for biases in the population, a sample size of 6300 is not tiny. The estimated error for such a population is going to be around 1% (give or take).

htag 4y ago
It's just that 1. Sampling and self reporting introduces many concerns to the validity of the study. Let's not downplay the negative effect these can have. 2. It's expensive to collect consent and data. 3. Federal grants often fund these studies. Tax payers are essentially paying to collect the same data twice.
projektfu 4y ago
That's true assuming the sample is not biased. I'd be interested to see a kaplan-meyer curve for the two groups.

ellisv 4y ago

Most government departments don’t have research in their mandate and are underfunded/staffed.

htag 4y ago
This type of record collecting and academic collaboration might be a good responsibility for The Library of Congress.
- tomrod 4y ago
  It's a completely reasonable question -- reach out to your Congressional representatives to kick off the initiative!
woodruffw 4y ago
Not research, but most government departments do have dedicated statistical bureaus! This would probably fall under the BLS[1] or BEA[2], which are bureaus of the DOL and DOC, respectively. Another potential fit would be the IRS's SOI[3] division. [1]: https://en.wikipedia.org/wiki/Bureau_of_Labor_Statistics https://en.wikipedia.org/wiki/Bureau_of_Labor_Statistics [2]: https://en.wikipedia.org/wiki/Bureau_of_Economic_Analysis https://en.wikipedia.org/wiki/Bureau_of_Economic_Analysis [3]: https://www.irs.gov/statistics/soi-tax-stats-statistics-of-income https://www.irs.gov/statistics/soi-tax-stats-statistics-of-i...