> Calculating the first difference of predicted probabilities
from the logit regression shows an increase in the prob-
ability of a successful coup of about 0.8% (Figure 7).
The absolute change is smaller than that for attempts
(but successful coups are rarer than attempts). The per-
centage change in the predicted probability is around
150%. _Successful coups are strongly associated with
IMET training and spending._ (Emphasis mine)
The fact that successful coup attempts do not go up as much as total coup attempts after US military training, does not alone to speak to whether US military training increases the success of coups.
After all, people only attempt coups if they think there is some likelihood of success. IMET training could very well increase the _perception_ of likelihood of success more than it increases _actual_ likelihood of success. Given this, even if IMET training makes a given coup attempt more likely to succeed, if more risky coup attempts are made after IMET training, the overall success rate of coups could go down after IMET training.
While my previous paragraph of texts explains that the data that you cite does not support your conclusions, I must also ask, where the data you cite comes from. I read the article that you link [0] and it says the opposite of what you say it says. Total military coups in the dataset are 110 for non-trained soldiers and 165 for trained soldiers (a 50% increase). However, _successful_ coups are 37 for non-trained soldiers and 72 for trained soldiers (a 100% increase). This information comes from Figure 2 and Figure 3 in the linked article.
> Effects of military training on coup success
> Calculating the first difference of predicted probabilities from the logit regression shows an increase in the prob- ability of a successful coup of about 0.8% (Figure 7). The absolute change is smaller than that for attempts (but successful coups are rarer than attempts). The per- centage change in the predicted probability is around 150%. _Successful coups are strongly associated with IMET training and spending._ (Emphasis mine)
The fact that successful coup attempts do not go up as much as total coup attempts after US military training, does not alone to speak to whether US military training increases the success of coups.
After all, people only attempt coups if they think there is some likelihood of success. IMET training could very well increase the _perception_ of likelihood of success more than it increases _actual_ likelihood of success. Given this, even if IMET training makes a given coup attempt more likely to succeed, if more risky coup attempts are made after IMET training, the overall success rate of coups could go down after IMET training.
While my previous paragraph of texts explains that the data that you cite does not support your conclusions, I must also ask, where the data you cite comes from. I read the article that you link [0] and it says the opposite of what you say it says. Total military coups in the dataset are 110 for non-trained soldiers and 165 for trained soldiers (a 50% increase). However, _successful_ coups are 37 for non-trained soldiers and 72 for trained soldiers (a 100% increase). This information comes from Figure 2 and Figure 3 in the linked article.
[0] https://journals.sagepub.com/doi/abs/10.1177/002234331771355...