I suppose it depends upon what we mean by believe. If belief implies accepting the truth-value of something, then I we need only observe the subject’s stated beliefs. But such a definition doesn’t allow for deception. Ok, then modify the definition: belief implies accepting the truth-value of a proposition without deception. Evidence for deception would exist if at a minimum there is an incentive to deceive others. But without that incentive (which non-professionals won’t have by and large) we can take a person’s word at face value, by this definition. Of course there are social reasons to profess faith - it’s reasonable to set aside the claims of those dependent upon an insistent religious member. But there also lots of people who could leave the church and do not. Their continued worship and payment to the church is evidence for their faith.
As Roderick_Femm noted in post 3, we can’t observe people’s hearts directly. But surely indirect means shouldn’t be limited to whether people’s actions are in accordance with the the religious claims that they accept. There is other evidence that can be gathered.
So what definition of belief is in use? Is is attested-belief? Attested-without-deception belief? Or act-belief? A definitive answer to this is available: we consult with the dictionary: lexicographers after all are acknowledged experts in everything. The 2nd definition listed in Webster’s New World College dictionary defines believe as, “to have religious faith”. Baptists have this and the OP is thereby refuted. QED.
Space considerations prevent more detailed investigation of bad faith, self deception, or the act of lying to oneself. Discussion here will be limited to sophestry and red herrings.