June 03, 2009
Longer Tests Improve Mental Performance?
Does this seem plausible to you?
In the study, 239 freshman college students from the Atlanta area took three different versions of the SAT Reasoning Test. Under conditions simulating the actual exam, with start times of 8 A.M. on three consecutive Saturdays, the students completed tests specially constructed for three different durations: 3.5, 4.5 and 5.5 hours. (The current SAT is 3.75 hours of testing over a 4.5-hour session. In this study, the short version of the test had one less of the verbal, math and writing sections; the long version had one more of each. Otherwise, the tests were the same.) Students received a cash bonus if they beat their previous SAT scores.
Before, during and after each test, students completed a questionnaire designed to asses their mood, emotions, confidence, subjective fatigue and more. As expected, the longer they worked on a test, the more the students reported mental fatigue. At the end of 5.5 hours of testing, students reported high levels of fatigue.
However, even though students reported greater fatigue for longer tests, their average performance for both the standard and long tests was significantly higher than for the short test. In fact, the short-form average score was 1,209 out of a possible 1,600; the standard-form average score was 1,222; and the long-form average score was 1,237. Scoring was weighted to make performance comparable across the different length tests.
Does the mind become more focused as the test proceeds? Do distractions fade from consciousness the longer you are hunched over puzzling out test questions?
Here is the full text of the research paper.
As an utter guess: you begin to forget about everything else and the mind works only on the questions.
If that is true I would expect the best performance would come in the middle hours of testing. By that time you are no longer thinking about what you did before the test. And you aren't thinking of what you might do after the test, because the end seems far away.
And graphs of how many questions were answered correctly and incorrectly for each 15 minute interval. Such data would be very hard to gather unless the test were given on a computer.
And does a right answer lead to more right answers?
i.e. Does answering a question correctly change your chance of answering the next correctly? Same for incorrect of course.
I can't open the paper itself. Something about file being damaged.
I would guess this is 100 % related to anxiety.
I think the focus and anxiety possibilities are likely, and would like to add a third. Over a longer test, small differences in intelligence will become more pronounced. It is impossible to remain completely focused even on the short version uninterruptedly. (I have some background in this if it becomes relevant to the discussion.) Therefore, how well one does when focus is diminished is more dependent on "natural" abilities. As anxiety reduces, one's natural abilities are less impaired.
As a check on this, I would like to see the experiment done over, in reverse order: have them do the longest test on the first Saturday, the shortest on the third. Then see whether they score better on the longest.
It could be simply that the students were getting better at test-taking from week to week.