Continuous DV (amount sold in $)- what model to use

Hi everyone,

I have been working with a company to determine the effect of an event on their product sales per day. As we want to study the effect per day for the whole month (31 days) for many different brands, I used a negative binomial model with random effect (to be able to take into account brand variation) with unit sales (quantity sold) as DV. However, they really want to see the effect by quantity sold ($) per brand per day. I've tried, but as there are many brands with zero sales on some days, I find that normal regression has a lot of variation and I'm not sure what statistical model could account for this by looking at daily sales by brand when DV is the amount sold (in dollars). My understanding is that unit sold is generally preferred because counting models are better able to take this type of analysis into account. Does anyone have any recommendations?

Thank you in advance!!

  • Savionf Savionf
    +2

    Questions at this level should come with a bounty.

    • I am a student, someone suggested to me this website to help me with my problem. I do not have any funding to give

  • That's understandable. Hopefully a good samaritan will be able to help. You may also enroll in the affiliate program: https://matchmaticians.com/affiliates and use the money you earn for asking questions.

    • thank you, that is a good idea I will try it out!

1 Answer

I would combine brands to make sure there are fewer or no brands with zero sales per day. Or combine days into weeks. Or both

Mathe Mathe
3.3K
  • Thank you for your answer! Yes, that is what I have done, but I was wondering if there is any model allowing for the same longitudinal analysis per brand, as the count model when looked at unit sold. I think will just keep it aggregated then.

Join Matchmaticians Affiliate Marketing Program to earn up to a 50% commission on every question that your affiliated users ask or answer.