1

I have a dataset that includes in situ measurements of:

Dependent variable- growth rates (continuous) Independent variable - temperature (continuous) Independent variable - food concentration (continuous)

and I want to test the effect of food concentration on growth rates, while accounting for the different temperatures (study organism is known to grow faster at higher temperatures). Can I do an ANCOVA analysis with two continuous (i.e. not categorical) independent variables? If not, what is an alternative analysis/method to statistically account for temperature?

Note: It's been suggested that I run an ANOVA with an interaction term between temperature and food concentration, but I'm not convinced that this sufficiently standardizes growth rates to temperature rather than testing any relationships between temperature and food concentration. Is ANOVA actually the right choice?

Thanks!

user182007
  • 13
  • 2

1 Answers1

1

You can definitely do this, but if there are no categorical predictors then it's just called multiple regression.

Jake Westfall
  • 11,539
  • 2
  • 48
  • 96
  • Good to know, thanks! So, just to make sure I understand, you're confirming that running a multiple regression with food concentration and temperature will statistically 'normalize' growth rates to temperature and then test the effect of food concentration? I'm essentially looking for a growth/degree that I can test against food concentration. – user182007 Oct 24 '17 at 15:05
  • @user182007 Yes, the estimated slope for food concentration in the multiple regression will represent the change in temperature per unit change in concentration, while "[controlling for](https://stats.stackexchange.com/questions/17336/how-exactly-does-one-control-for-other-variables?noredirect=1&lq=1)" growth rate. – Jake Westfall Oct 24 '17 at 15:33
  • Thanks again, that's a good way to explain it. Sorry for being a needy OP, but it seems like I would actually want a slope that represents the change in **growth** per unit change in concentration, while "controlling for" **temperature**, slight reversal of what you described for the multiple regression. Is this still the analysis I should run? Thanks again! – user182007 Oct 25 '17 at 13:24
  • @user182007 Sorry, I misspoke! What you said (not what I said) is what the slope will tell you. I accidentally reversed the DV and the covariate. – Jake Westfall Oct 25 '17 at 14:49