Studies of the reliability of eyewitness identifications show little correlation between the accuracy of a witness’s account and the confidence the witness has in the account.
Conclusion
We shouldn't do police lineups of suspects if the eyewitnesses can hear each other's assessments.
Evidence
There's little correlation between the accuracy of someone's account and the confidence they have in it. And certain factors can affect a witness's confidence without affecting their accuracy.
Evaluate
This argument kind of makes a curious jump from the first two sentences to the last idea. What does "confidence and accuracy are two very different things" have to do with ?
Since the second sentence is talking about "certain factors" affecting the confidence of eyewitness accounts, I'm presuming that the author is worried that hearing someone else's assessment of a police lineup might influence the confidence of that person's own assessment of the police lineup.
I'm not really sure why the author cares if someone's confidence is impacted by hearing another witness's assessment; maybe she thinks that someone who wouldn't have sounded confident in their inaccurate pick will now be biased into sounding more confident about their inaccurate pick.
Or vice versa -- the person would have sounded confident in their accurate pick, but based on hearing other eyewitnesses, they become less confident about their accurate pick.
Goal
Find something that feels like it takes us from and gets us closer to "we shouldn't allow witnesses to hear each other's answers to the police lineup.