3 ms·

> the first step is acquiring sufficient high-quality data and reproducing findings I agree that good data is a pre-requisite. An online survey of blog reader

> the first step is acquiring sufficient high-quality data and reproducing findings I agree that good data is a pre-requisite. An online survey of blog readers isn't going to upend psychology as we know it, but Scott's findings of a heavily skewed numbers of first-borns in the survey probably can't be explained as just "online survey crap." The only "online crap" explanation I can think of is if the "first born" box was pre-filled and people changed the response for their number of siblings but didn't change the pre-filled response for birth order. This unbalanced first-born ratio is something that could be looked at in a bigger and more rigorous survey. > some want to talk about it prematurely because it speculatively supports conclusions that fit their agenda We have to talk about it otherwise we'll never get any data- I'm not advocating policy decisions based on incomplete research, I'm advocating discussion and follow on research. You can't do research without discussing ideas ahead of time. > yet here we are, hijacking an interesting if incomplete study for your own purpose I didn't hijack anything. I was making a comment that stayed on topic and is related to the post. BTW, I'm a statistician, and have a pretty good understanding behind the science of surveys. I'm admittedly not an expert in psychology. From the HN guidelines: "Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize. Assume good faith."

richmarr 9y ago
> The only "online crap" explanation I can think of is if the "first born" box was pre-filled and people changed the response for their number of siblings but didn't change the pre-filled response for birth order. In the survey, birth order is inferred from the field that records the number of elder siblings. If the form design nudged respondants to not completing that field more than other fields (via unclear or biasing wording, ordering effects, etc) that could produce a similar outcome. I haven't looked at the survey so no idea, but there are more potential causes than defaults.
- jm_l 9y ago
  I took the survey, am a younger sibling, and have very strong memory of the questions being very fair. If I think I my hardest I can't imagine any reason another responder would have been pulled to unconsciously report wrong data on the birth order questions.
  - richmarr 9y ago
    With respect, your opinion is anecdotal. These things aren't always obvious, or even visible to respondants at all. I'm not saying there's no effect, just that right now this claimed finding seems contrary to the existing body of evidence. This finding may well be accurate (and the existing studies in error). Alternatively it may be a mistake. Alternatively both could be true and this survey represents a special case, some kind of selection bias. Given that this survey is yet to be peer reviewed or replicated I think the most likely explanation is a mistake, but that's just my perception of the balance of probabilities.

richmarr 9y ago

> The only "online crap" explanation I can think of is if the "first born" box was pre-filled and people changed the response for their number of siblings but didn't change the pre-filled response for birth order. In the survey, birth order is inferred from the field that records the number of elder siblings. If the form design nudged respondants to not completing that field more than other fields (via unclear or biasing wording, ordering effects, etc) that could produce a similar outcome. I haven't looked at the survey so no idea, but there are more potential causes than defaults.

jm_l 9y ago
I took the survey, am a younger sibling, and have very strong memory of the questions being very fair. If I think I my hardest I can't imagine any reason another responder would have been pulled to unconsciously report wrong data on the birth order questions.
- richmarr 9y ago
  With respect, your opinion is anecdotal. These things aren't always obvious, or even visible to respondants at all. I'm not saying there's no effect, just that right now this claimed finding seems contrary to the existing body of evidence. This finding may well be accurate (and the existing studies in error). Alternatively it may be a mistake. Alternatively both could be true and this survey represents a special case, some kind of selection bias. Given that this survey is yet to be peer reviewed or replicated I think the most likely explanation is a mistake, but that's just my perception of the balance of probabilities.