1
$\begingroup$

I am looking for a word to describe a certain type of bias, something between Framing and Selection Bias.

I am analyizing a dataset with measurements and their location in Berlin. The dataset includes the information whether these measurements were taken in the former eastern or western part of the city. Including this information in the dataset was not necessary in the sense that the information where the wall used to be is publicly available and easy to find. The fact that this information is included makes it easy and thus likely to analyze it, e.g. by a test for difference of means.

I am looking for a term to describe how the design of a dataset encourages/discourages certain types of analyses and therefore conclusions. I am not looking for selection bias since this refers to the data itself, not just its presentation.
I also believe I am not looking for Framing either, because I understand this to work differently, i.e. through emotionality, e.g. by contrasting a low value with an even lower one to make the first one seem higher. In my case it's more about how the design of the dataset includes certain distinctions which are then in turn confirmed by the statistical analysis, while other distinctions, which might be equally significant never get researched, because they are not included in the initial data.

I am thus looking for a term to describe that researchers will often happily go for the low-hanging fruit early, such as the distinctions already present in the dataset, without the need to further aggregate information, something along the lines of the joke about the person who looks for their keys under a street lantern, even though they hadn't lost them there, but under the lantern is the only place with the necessary light to find keys.

Do you think the case from my example constitutes a type of bias? If not, why not? If yes, would you include it in the term of "Framing"? If you think this is a type of bias, but "Framing" is the wrong term, what would the correct one be?

Also, I'm fairly new to statistics in general and this board in particular, so I'm not sure if this is the right place for this question.

$\endgroup$
1
  • $\begingroup$ Not statistical bias, but Tversky and Kahneman talk about availability bias. Just because you measure something does not mean it is informative. $\endgroup$
    – Andy W
    Commented Jan 30, 2022 at 18:53

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.