Implications for analysis of survey data when recall time frames differ bt questions

#1
Hi, I am advising some folks on setting up a brief survey containing questions that measure some well-known health indicators.

My curiosity was piqued because amongst these questions the recall time frame differs between certain indicators.

For instance, there's this question:

During the past 30 days, how often did you smoke tobacco?

Daily
Less than daily
Not at all
Refused to answer

Then, later, there this question:

In a typical 7-day period, on how many days do you eat fruit?

(Range 0 - 7)

There are other questions that ask for recall of other time frames, as well.

In thinking of analyses, I wonder what, if any, the differential recall time frames have an impact on the type of model, variance estimation, and even on conclusions.

Simply put, what does it even mean to examine how daily tobacco use in the past 30 days relates to the number of days a person ate fruit?

I have analyzed survey data before but generally the recall time frame is consistent throughout. But this is a survey where the PI is cobbling together questions from different international surveys. I might be overthinking things but I thought I'd ask the community.
 

hlsmith

Not a robit
#4
Re: Implications for analysis of survey data when recall time frames differ bt questi

Surveys are beasts.


Well you are probably fine moving forward, you just need to be very specific when reporting results. State the qualifiers and time frame for each component.


Respondents reporting smoking Daily during the past month had two times greater odds of reporting eating 0-7 Fruits in a typical week than respondents reporting smoking Less than Daily during the past month.
 
#5
Re: Implications for analysis of survey data when recall time frames differ bt questi

Hi hlsmith, thanks for responding. I definitely agree with being very specific in reporting. I worry, though, that folks will run away with interpretation.
 

hlsmith

Not a robit
#6
Re: Implications for analysis of survey data when recall time frames differ bt questi

Well you have three options, so you will have to select the lesser of the three evils:


-Change questions (then questions can't be contrasted to other survey results but better internal comparisons).
-Keep questions (then you have to be explicit in interpretations)
-Quit your job (then you have to get sunblock for the beach and unemployment insurance)


Good luck!
 
#7
Re: Implications for analysis of survey data when recall time frames differ bt questi

Well you have three options, so you will have to select the lesser of the three evils:


-Change questions (then questions can't be contrasted to other survey results but better internal comparisons).
-Keep questions (then you have to be explicit in interpretations)
-Quit your job (then you have to get sunblock for the beach and unemployment insurance)


Good luck!
Can I retire now? :D