The open-ended responses to these (and other) questions appear in a data set that contains five variables: RESPID (the unique respondent identification number); RTYPE
(whether the respondent is from the ERC mixed-gender group of those involved in start-up activity, the NSF over-sample of females involved in start-up activity, or an ERC-sponsored comparison group); ITEMNO (the interview question number; LINENO (the line number); and RESPONSE (the respondent's answer to the question).
For example, the following two commands creat views of ENTITY and RELSHIP specific to the TRANSP model: CREATE VIEW TRANSP_ENTITY AS (SELECT * FROM ENTITY WHERE (ENAME='TRANSP' AND ETYPE='model') OR (ENAME, ETYPE) IN (SELECT E2NAME, E2TYPE FROM RELSHIP WHERE RTYPE
= 'CONTAINS' AND E1NAME = 'TRANSP' AND E1TYPE " 'model')) CREATE VIEW TRANSP_RELSHIP AS (SELECT * FROM RELSHIP WHERE (E1NAME, E1TYPE) IN (SELECT ENAME, ETYPE FROM TRANSP_ENTITY) AND (E2NAME, E2TYPE) IN (SELECT ENAME, ETYPE FROM TRANSP_ENTITY))
The current view-oriented approach, on the other hand, can satisfy the impact of change in one command (assume x is the element ZIP CODE): SELECT RTYPE
,E1NAME,E1TYPE,E2NAME, E2TYPE FROM RELSHIP WHERE (E2NAME='ZIP CODE' AND E2TYPE='ELEMENT') OR (E1NAME='ZIP CODE' AND E1TYPE='ELEMENT')