Millisecond Forums » Millisecond Forums » Inquisit 5 » Adaptive N-back: Scoring formula seems taken from normal n-back and therefore unreliable?

Adaptive N-back: Scoring formula seems taken from normal n-back and therefore unreliable?

aquinox — Wed, 24 May 2017 18:42:49 GMT

Hi,

I have used the N-back for a while, assuming that scoring correction is taking place for getting in higher levels of N, because obviously they are harder. If someone enters a higher level of N but scores not so good on that N level, he could end up scoring lower than someone scoring quite well in a lower N level.

Lets say there are
30 target trials in total, with each time 0 FAs, someone scores
Person A : N=2: 9/10 hitsN=3: 8/10 + 7/10= 15/20 hits (less than 90% correct, so no N=4)Final score will be 24
Person B:
N=2: 9/10 hitsN=3: 9/10 hits (pass to N=4)N=4: 4/10 hitsHis final score is 22. While person B reached N=4, his final score will still be worse.

Or am i missing something?

Thanks!

RE: Adaptive N-back: Scoring formula seems taken from normal n-back and therefore unreliable?

Dave — Wed, 24 May 2017 18:42:49 GMT

[quote]

[b]aquinox - Wednesday, May 24, 2017[/b]

[quote]

[b]Dave - Wednesday, May 24, 2017[/b]

[quote]

[b]aquinox - Wednesday, May 24, 2017[/b]

Hi,

I have used the N-back for a while, assuming that scoring correction is taking place for getting in higher levels of N, because obviously they are harder. If someone enters a higher level of N but scores not so good on that N level, he could end up scoring lower than someone scoring quite well in a lower N level.

Lets say there are
30 target trials in total, with each time 0 FAs, someone scores
Person A : N=2: 9/10 hits N=3: 8/10 + 7/10= 15/20 hits (less than 90% correct, so no N=4) Final score will be 24
Person B:
N=2: 9/10 hitsN=3: 9/10 hits (pass to N=4)N=4: 4/10 hitsHis final score is 22. While person B reached N=4, his final score will still be worse.

Or am i missing something?

Thanks!

[/quote]

I'm confused: Are you referring to the adaptive n-back or the normal (non-adaptive n-back)? I don't quite get how you came up with the numbers in the example and how they relate to the adaptive n-back. I'm also not sure what exactly you are referring to as "final score." If you could clarify, that would be great.

[/quote]

I'm referring to the adaptive N-back.

They are just some round numbers to illustrate the problem I think exists in how the adaptive n-back has been set up.

So what I mean with final score is the way the test results are scored. Participant B reached N=4 but didn't score so well in this N level, and will get a lower final score than Participant A because he scored higher in lower N levels and there's no differentiation between N levels. (it appears as if the scoring system has been taken over from the normal Nback)

[/quote]

Okay, thanks. Cases where a participant who reaches a higher n-level ends up with a slightly worse score due to bad performance at that level than a subject with not-great but more consistent performance at a lower n-level are theoretically possible.To the best of my knowledge, the scoring in the Inquisit script is the way also Jaeggi et al score the adaptive n-Back.

aquinox — Wed, 24 May 2017 18:10:25 GMT

[quote]

[b]Dave - Wednesday, May 24, 2017[/b]

[quote]

[b]aquinox - Wednesday, May 24, 2017[/b]

Hi,

I have used the N-back for a while, assuming that scoring correction is taking place for getting in higher levels of N, because obviously they are harder. If someone enters a higher level of N but scores not so good on that N level, he could end up scoring lower than someone scoring quite well in a lower N level.

Lets say there are
30 target trials in total, with each time 0 FAs, someone scores
Person A : N=2: 9/10 hits N=3: 8/10 + 7/10= 15/20 hits (less than 90% correct, so no N=4) Final score will be 24
Person B:
N=2: 9/10 hitsN=3: 9/10 hits (pass to N=4)N=4: 4/10 hitsHis final score is 22. While person B reached N=4, his final score will still be worse.

Or am i missing something?

Thanks!

[/quote]

RE: Adaptive N-back: Scoring formula seems taken from normal n-back and therefore unreliable?

Dave — Wed, 24 May 2017 17:52:34 GMT

[quote]

[b]aquinox - Wednesday, May 24, 2017[/b]

Hi,

I have used the N-back for a while, assuming that scoring correction is taking place for getting in higher levels of N, because obviously they are harder. If someone enters a higher level of N but scores not so good on that N level, he could end up scoring lower than someone scoring quite well in a lower N level.

Lets say there are
30 target trials in total, with each time 0 FAs, someone scores
Person A : N=2: 9/10 hits N=3: 8/10 + 7/10= 15/20 hits (less than 90% correct, so no N=4) Final score will be 24
Person B:
N=2: 9/10 hitsN=3: 9/10 hits (pass to N=4)N=4: 4/10 hitsHis final score is 22. While person B reached N=4, his final score will still be worse.

Or am i missing something?

Thanks!

[/quote]