I am using survey panel data and a linear fixed effects model with an event dummy variable to examine the effect of Biden visiting Ukraine on the concern of the US citizens about US getting more deeply involved in the Ukraine War(Dependent variable). The same survey participants answer the same survey questions every week.(The waves of the survey are weekly). I want to use two weeks before and two weeks after the event in order to check if there is a significant difference in the concern of survey participants about US getting involved in the war before vs after Biden visiting Ukraine. I'm using this equation with individual fixed effects :

concern about country getting involved in war = eventDummy + depression + financial_difficulties +trust on military + trust_on_mainstream_media_de + political position.

'eventdummy = 0 for survey answers before biden visiting ukraine and 1 for survey answers after'

The other variables (depression, financial difficulties, trust on mainstream media) are used as time variant controls. My concern is since fixed effects variation accounts only for within-individual variation and not between-individual variation: Depression or financial difficulties may not vary at all in a period of one month (since im only using data from 2 weeks before and 2 weeks after the event) and therefore they will be useless in the model. (we dont exepect people with financial difficulties to improve their financial situation within one month therefore there would be no variation in the variable)

Note: Im using linearmodels.panel.model.PanelOLS in python for fixed effects regression

Do you have thoughts about this?