Suppose you want to spread the prejudice that Xs are Ys - e.g. Jews are greedy and money-obsessed; women are terrible drivers, etc. What's a good method?
STEP ONE: Find some Xs that are Ys. That probably won't be hard. Inevitably, some women are bad drivers and some Jews are money-obsessed. So you will easily be able to come up with a handful of true anecdotes.
STEP TWO: Pad out those core anecdotes with more examples that could easily be interpreted as examples of Xs being Ys, particularly if you edit the context a bit. E.g. My Aunty Mary damaged her car while parking (actually, someone drove into her while she was parking, but I don't mention that). Steps One and Two combined should give you a good number of anecdotes - at least twenty or so.
STEP THREE: Compile a list of these 20+ anecdotes, moving seamlessly from one to the next without giving your audience pause to think "Er, hang on, is that really a case of an X being Y?". The sheer quantity of anecdotes you pile up should be enough to quash any such doubts.
STEP FOUR: Ignore evidence to the contrary. Don’t mention examples of X that aren't Y. e.g. women who are professional and reliable drivers, Jews who are extraordinarily generous. Or, even better, mention some examples to give the impression that you really are striving to be 'balanced', but imply these examples are the exception to the rule.
STEP FIVE: Step back and wait for your audience to start thinking of their own examples of Xs being Ys. "Actually, my Jewish neighbour is a bit of a miser, isn't he?", "The fact is, my Mum really can't drive for toffees." Now you have set up the expectation that Xs are Ys, confirmation bias will likely set in and they will have no trouble finding more examples of their own (including many genuine examples, no doubt). It won't occur to them to look for counter-examples. Or, if counter-examples are spotted, they'll be dismissed as anomalies: exceptions 'that prove the rule', etc.
STEP SIX: Once this way of thinking about Xs being Ys has really set in, it may be seem so obvious to your audience that X's are Y's that they'll be astonished that others can't see this too. “Are you blind?" they'll ask incredulously when someone questions whether Xs are Ys. "Of course they are!", they'll add, pointing to their now vast collection of anecdotes. They will likely think there's something wrong with those who can't see that Xs are Ys when the evidence is right in front of them. Indeed, they will likely conclude that those who question whether X's are Y's are weirdly insensitive to the truth - they are the ones who are biased.
No doubt most of us are familiar with how this works. What are the warning signs this is going on? Heavy reliance on anecdotal evidence. Distinct lack of other more reliable forms of evidence. No mention of anything but a few tokenistic counter-examples to establish 'balance'.
Now go read any Nick Cohen piece on 'The Left" in which he argues e.g.:
'Corbyn and his comrades bring their support for the .... the women-, Jew- and gay- haters of radical Islam...' Source.
‘Corbyn and his supporters do not want us to think about Paris because they cannot accept that privileged westerners can be victims. If Isis kills them, it is their own or their governments’ fault.' Source.
How much of Cohen's evidence for supposing Corbyn supporters are West-hating Islamist-apologists is anecdotal? Surely most of it. Then read this report providing non-anecdotal evidence of what Corbyn supporters on twitter actually think about Islam, Western foreign policy, etc.
Similarly, read any recent newspaper article by Cohen etc. denouncing ‘The Left’ as having a major anti-semitism problem. Replace X with ‘The Left’ and Y with ‘are (more-than-averagely) anti-semitic’. Notice anything? By all means go ahead and try to make the case for the Left having a major problem with anti-semitism. But you won't succeed by trotting out lists of anecdotes.