The comic is highlighting the absurdity of taking something that is technically undefined, and thinking that you’ve got a counter-factual (with is, like, exactly what is happening for most people in this thread).
If no felons have ever previously run for president, you have no data on how felons perform. You have an N of 0 because the event hasn’t occurred. Its a null result. NA. Undefined. You have no information. Its untested.
Even further, it highlights the very exact point of the comic, which is that when you rely on currently has an N of zero as a counter factual, you are going beyond the scope of what your data is capable of speaking to.
To assess the impact of a candidate with a felony on their chances of winning a presidential election, we need to know how many felons have run and how many have won. However, if no felon has ever run for president, we have zero data points for both felons running and winning. This means our calculation for the probability of a felon winning would involve dividing by zero, which is mathematically undefined and impossible. Without any previous instances to examine, we simply cannot make a statistically grounded prediction about the impact of a felony on a candidate’s electoral prospects; we lack any empirical evidence to base such an assessment on.
Except that we actually have approval ratings and polls for about 90 years of elections. From which we can build the appropriate counter-factuals to actually create a statistic because an approval rating is a continuous variable, not a discrete variable. An approval rating of 51% is directly comparable to an approval rating of 31%, and all Presidents ‘have’ this condition, even if it went unmeasured. I also have a sufficient range of variation to build the negative case example because I have presidents and candidates across the range of variation observed in the condition, and variation in the outcome: winning an election.
Being a felon is also a condition, but 100% of the data we have is “not a felon”. And we have no variation in the observed outcome. Some non-felons won, some non-felons lost. We’re not testing if they are a felon or not, we’re testing if they win the election or not.
Look I get that this is beyond you, but you really aren’t making the point you think you are here. Also, you are on the wrong side of the fallacy the comic is presenting. I’m not trying to interpret being a felon has on becoming president, you are. I’m interested in what the polling data has to say about the probability of winning, which is a statistically and scientifically grounded thing to do.
You mostly seem like you have an axe to grind because Biden is losing the election for you. I’m sorry for that.
The comic is highlighting the absurdity of taking something that is technically undefined, and thinking that you’ve got a counter-factual (with is, like, exactly what is happening for most people in this thread).
If no felons have ever previously run for president, you have no data on how felons perform. You have an N of 0 because the event hasn’t occurred. Its a null result. NA. Undefined. You have no information. Its untested.
Even further, it highlights the very exact point of the comic, which is that when you rely on currently has an N of zero as a counter factual, you are going beyond the scope of what your data is capable of speaking to.
To assess the impact of a candidate with a felony on their chances of winning a presidential election, we need to know how many felons have run and how many have won. However, if no felon has ever run for president, we have zero data points for both felons running and winning. This means our calculation for the probability of a felon winning would involve dividing by zero, which is mathematically undefined and impossible. Without any previous instances to examine, we simply cannot make a statistically grounded prediction about the impact of a felony on a candidate’s electoral prospects; we lack any empirical evidence to base such an assessment on.
Refer to the title panel of the comic, which says the problematic statement is…
What you said was,
Or to summarize…
So is it sounding familiar?
Except that we actually have approval ratings and polls for about 90 years of elections. From which we can build the appropriate counter-factuals to actually create a statistic because an approval rating is a continuous variable, not a discrete variable. An approval rating of 51% is directly comparable to an approval rating of 31%, and all Presidents ‘have’ this condition, even if it went unmeasured. I also have a sufficient range of variation to build the negative case example because I have presidents and candidates across the range of variation observed in the condition, and variation in the outcome: winning an election.
Being a felon is also a condition, but 100% of the data we have is “not a felon”. And we have no variation in the observed outcome. Some non-felons won, some non-felons lost. We’re not testing if they are a felon or not, we’re testing if they win the election or not.
Look I get that this is beyond you, but you really aren’t making the point you think you are here. Also, you are on the wrong side of the fallacy the comic is presenting. I’m not trying to interpret being a felon has on becoming president, you are. I’m interested in what the polling data has to say about the probability of winning, which is a statistically and scientifically grounded thing to do.
You mostly seem like you have an axe to grind because Biden is losing the election for you. I’m sorry for that.