 |
 |
Looking at the design, the pattern of rotation seems similar to that used in the Labour Force Survey (similarly eurostat-mandated) but with annual rather than quarterly waves. Therefore, 75% of the sample are common to any two adjacent waves. New samples (25%) are drawn each year and replace those retired. Therefore the idea of a discontinuity after the fourth year is unlikely to occur, as the fifth year sample is composed of the new sample members selected in the second, third, fourth, and as well as the new additions in the fifth year. Unlike cross-sectional designs, this design permits longitudinal analysis. Therefore analysis of moves into poverty or out of poverty can be analysed, which is not possible with cross-sectional designs. However, the analysis of persistent poverty is limited by the design. Once the survey is operational over a medium-term horizon, then each sample member would be in the panel for five annual waves. Initially, some sample members might be present for only one wave, etc. Analysis methods would be similar to those used in longitudinal surveys - i.e. pooling data over several waves for transitions of interest, while cross-sectional data can be analysed as in a cross-sectional survey. The methods seem to be relatively well understood and to pose few problems, but may be unfamiliar to those who have not used longitudinal studies. --------------------------------------------------------- Paul Bivand Head of Analysis and Statistics Direct Line: 020 7840 8335 Inclusion 3rd floor, 89 Albert Embankment, London SE1 7TP Tel: 020 7582 7221 Fax: 020 7582 6391 Inclusion website: www.cesi.org.uk The contents of this e-mail and any attachment(s) are intended solely for the use of the addressee. If you are not the intended recipient please return the e-mail to the sender and delete from your mailbox. The contents of this e-mail and any attachment(s) are intended solely for the use of the addressee. -----Original Message----- From: jay ginn [mailto:[log in to unmask]] Sent: 12 April 2005 10:10 To: [log in to unmask] Subject: FW: EU-SILC - concerns[Scanned] Dear listmembers, Just a note re the planned EU-SILC survey (Income and Living Conditions) which will be used to generate Eurostat reports on eg poverty rates. As I understand it, in Britain it is planned that the General Household Survey will be adapted to provide the data required by Eurostat, to a) include the questions prescribed by the EU b) make it a 4-year rotational panel (panel survey over 4-year periods, with 25% of the panel being replaced each year and a new sample drawn in the 5th year) Although the initial sample will be relatively large (eg 13,000 adults in private households in UK), clearly by the 4th year only 25% of the original population will be present. So analysis will probably be on a mixed sample of original respondents plus mostly new ones. Its not clear to me how the 25% to be replaced will be selected nor how the replacement group will be selected. Of course the statisticians involved will have thought about a suitable method. Nevertheless, my concern is that this complex design leaves lots of room for error compared with a repeated cross-sectional design. I would not be surprised if the results showed a trend over the 4 years followed by a discontinuity. And that trend would be due to the sample design not to changes in society. Official statisticians may then try to correct for this using weighting and imputation techniques, both of which lend themselves to manipulation for political purposes. It is true that the BHPS design, on which ECHPS was based, also has weaknesses, mainly the attrition (which tends to be selective, ie biased) that occurs in a panel study. Also its sample size was small and I found cases where its results differed markedly from the larger cross-sectional surveys. But replacing the BHPS with the new design seems like jumping from frying pan to fire. I believe this design is being required for all member states and I would appreciate comments on it from Radstats statisticians. Am I worrying unnecessarily? Best wishes Jay Jay Ginn Visiting Professor Surrey University Guildford, GU2 7XH, UK 01737 559341 ****************************************************** Please note that if you press the 'Reply' button your message will go only to the sender of this message. If you want to reply to the whole list, use your mailer's 'Reply-to-All' button to send your message automatically to [log in to unmask] ******************************************************* ****************************************************** Please note that if you press the 'Reply' button your message will go only to the sender of this message. If you want to reply to the whole list, use your mailer's 'Reply-to-All' button to send your message automatically to [log in to unmask] *******************************************************