T
tjm190
Guest
A sample size of 1000-2000 or so is good no matter how large the population. Let’s say you want to take a sample of a well mixed pot of soup. You decide a certain spoon is adequate for you to get a good idea of what the soup tastes like. Guess what? If the pot is 1 million times as big, the same spoon size still works. From an actuarial perspective, ‘500’ is far too small.Give me a break. What is your source as to what is a properly conducted survey? Gallup consistently gauges the pulse of the whole nation using only a few thousand responses. Given the tiny population which consistently practices NFP, 505 couples is quite persuasive.
They can discuss the survey however they like, it is completely inadequate.Let me adjust the underlying assertion slightly. Among those who receive training in NFP methods (and training is often/usually necessary for successful practice), a very small percentage later obtain divorces. See, additional discussion of the survey at physiciansforlife.org/content/view/193/36/.
A- It was drawn from the wrong population B- Mail surveys are invalid anyway.In that study, even if EVERY couple who failed to mail back in a questionnaire had obtained a divorce (an obviously unlikely scenario), the rate of divorce among those who had received NFP training is something like 25%, still remarkably lower than average (even among Catholics).
The results are presented in a completely biased manner- it makes me suspect that the survey they send out may also be inadequate.And while one should certainly consider an organization’s mission and agenda in evaluating studies it conducts, this is certainly not dispositive. It is unlikely that the organization is fabricating their results outright. This is not a situation where responses could be exaggerated, where subjects could be cued by the survey conductors, etc. This is not an experiment, its a survey, not about opinions, but about facts.
Those results are generally presented in the format “when asked XXX, Y% answered this”And where in the world do you get this idea that the survey questionnaire needs to be presented? I once conducted surveys for Gallup. Even a simple presidential poll was at least 15 questions long. Have you ever heard each question given when evaluating the poll? And surely “Are you divorced or not” is a question much less subject to bias than asking which president a person prefers, intends to vote for, identifies with, etc.
The survey talks about things such as “they are more happy with their marriage” and other things like that.
Speculation is not a form of statistical evidenceFurther still, the findings of the above study are entirely plausible and expected. NFP increases cooperation and communication between spouse. Secular psychologists and sex therapists would overwhelmingly agree that this would make divorces less likely. Is there iron-clad, statistical proof that NFP reduces divorces? Perhaps not, but I would assert that there are neither sufficient people practicing NFP nor institutions sufficiently disinterested to yield the kind of proof you demand.
It isn’t just imperfect, it’s completely invalid. It is drawn from the wrong population, suffers from nonresponse bias, we can not see if the survey itself was biased, a margin of error is not presented, it is a mail-in survey, fails the ‘success failure condition’ (a survey of this type must have atleast 10 'yes’es and 'no’s), and on top of that is far too small.This study may not be perfect, but you cannot deny that it provides evidence for the assertion that couples who practice, and receive training in, NFP are less likely to be divorced. A rational person believes things most based upon evidence, no? If you wish to dispute that above assertion, the real question is, where is your evidence to the contrary?
It’s a complete train wreck that nobody with even the most basic statistic knowledge could help but laugh at. There is no evidence for anything present in that survey, aside from the fact that it proves the individuals conducting the survey had no idea what they were doing.