Many improvements have been proposed for the basic gravity model specification, most of which are confirmed by standard statistical tests due to the large number of observations often used to estimate such models. We use Monte Carlo experiments to examine situations in which features of models may be found statistically significant (or insignificant) when it is known ex ante that they are absent (or present) in the underlying data process. Erroneous assumptions about the presence or absence of lagged dependent variables, fixed effects, free-trade associations and custom unions are shown to introduce an economically important bias in estimates of the coefficients of interest, and in some cases to be confirmed spuriously. Policy effects for such initiatives as free trade associations and currency unions can also be confirmed spuriously when they do not exist in the data-generating process. 17 Here is another, more positive interpretation: since gravity models routinely have outstanding fits, fixed-effects models among others, perhaps the under-4-percent R-squared finding in column 4 can be taken as evidence that researchers are seldom in the position of estimating an LDV world with an FE model.