09/01/05
			GDIGIT - Graduate Module		     cor862.txt

OVERVIEW

  The 2003 - 2005 round of data collection is the first time that the graduates
  have completed the digit ordering cognitive task.  The purpose of this task
  is to help provide a measure for the cognitive ability of the respondent.
  The instructions to the respondent are as follows:

  	"Now we are going to do a task that's quite a bit different.

  	I'd like you to make sure that you're away from any noises or other
  	distractions, and I'd like you to close your eyes for this part after I
  	explain the instructions.

  	This time I'm going to read some numbers to you, and when I am done I
  	would like you to rearrange the numbers from the lowest number I say to
  	the highest number I say and then say them back to me.  For example, if
  	I said "4 zero 2", you would say, "zero 2 4".  I can only say the
  	numbers once, so that's why it's important that you are able to hear me
  	clearly.

  	This is designed to be a little hard.  We don't expect everyone to get
  	everything right, but we really want you to try your best, OK?

  	Is this clear?

  	Okay, the first set of numbers is...".

  The interviewer then reads a series of three one-digit numbers, which appear
  on the computer screen.  The respondent attempts to say the numbers back in
  ascending order.  If the respondent answers correctly, he or she continues
  on to the next level, which uses four numbers instead of three.  If the
  respondent answers incorrectly, the interviewer reads another set of three
  numbers for the respondent's second attempt.  If the respondent answers
  incorrectly on the second attempt, the module ends.  This pattern continues
  until the respondent either fails at one level or completes the last level,
  which uses eight numbers.

BRIEF VARIABLE DESCRIPTIONS

  gi501re		Indicates if GR is in 80% digit ordering sample.

  gi502re		Highest level GR successfully completed.

  gi503re		Alternative scoring method for digit ordering task.

  gi504re		Additional information about low performance on digit
  			ordering task.

  gi531re		Tabulation of results: Level 3, Item 1.

  gi532re		Tabulation of results: Level 3, Item 2.

  gi535re		List of GRs' answers: Level 3, Item 1.

  gi536re		List of GRs' answers: Level 3, Item 2.

  gi541re		Tabulation of results: Level 4, Item 1.

  gi542re		Tabulation of results: Level 4, Item 2.

  gi545re		List of GRs' answers: Level 4, Item 1.

  gi546re		List of GRs' answers: Level 4, Item 2.

  gi551re		Tabulation of results: Level 5, Item 1.

  gi552re		Tabulation of results: Level 5, Item 2.

  gi555re		List of GRs' answers: Level 5, Item 1.

  gi556re		List of GRs' answers: Level 5, Item 2.

  gi561re		Tabulation of results: Level 6, Item 1.

  gi562re		Tabulation of results: Level 6, Item 2.

  gi565re		List of GRs' answers: Level 6, Item 1.

  gi566re		List of GRs' answers: Level 6, Item 2.

  gi571re		Tabulation of results: Level 7, Item 1.

  gi572re		Tabulation of results: Level 7, Item 2.

  gi575re		List of GRs' answers: Level 7, Item 1.

  gi576re		List of GRs' answers: Level 7, Item 2.

  gi581re		Tabulation of results: Level 8, Item 1.

  gi582re		Tabulation of results: Level 8, Item 2.

  gi585re		List of GRs' answers: Level 8, Item 1.

  gi586re		List of GRs' answers: Level 8, Item 2.

CODING

  gi501re - gi586re

    All refusals become REFUSED for remaining variables.

    Once GR gets two consecutive questions wrong, task ends and remaining
    variables are INAP.  Some INAPS for even-numbered variables are from
    respondents who answer correctly on the first attempt.

PROBLEMS

  In some instances, interviewers made mistakes while administering the module,
  such as saying a number other than the ones that appeared on the screen or
  missing a number completely.  In cases where this error occurred, and the
  respondent provided a correct answer for the wrong digits read to them, the
  answer was changed to give the respondent credit for providing a correct
  answer.

  For a few cases, interviewer error prevented the respondent from getting as
  far as the respondent's answers would warrant.  Such cases were set to 'NOT
  ASCERTAINED' since the interviewer error rather than the respondent caused
  the score.

PEOPLE

  Justin Resnick - Compilation of variable list, checking of notes, correction
  of raw data, creation of analysis variables, and writing of the COR.

  Jeremy Freese - Writing of code for creation of analysis variables and
  oversight/supervision of the module.

NOTES

  The variables for lists of answers, which are those variables ending in 5 or
  6, are string variables.  While the negative values do not have labels, they
  have the same meaning as for the other variables.

  Refusals and not ascertained cases do not receive scores for the score and
  level variables because they did not complete the task.

  Cases with values of 'not ascertained' include problems due to respondent
  error, interviewer error, instrument error, spoiled cases, and cases in which
  respondents may have written down the numbers.  The values of 'not
  ascertained' may disappear from attempt one to two at each level if the
  respondent gets the second attempt correct.