R-squared should not necessarily be used to justify the choice of a model. You should choose your model based on underlying theory or your sense of what the correct specification is, not based on r-squared. See for example What’s a good value for R-squared? https://people.duke.edu/~rnau/rsquared.htm