Skip to content

Summary of the impact and implications of revisions to mid-year population estimates

2002 mid-year population estimates (MYEs) for England and Wales were published on 26 September 2003 together with revised 2001 MYEs for England and Wales. This document describes the revision to the 2001 MYEs, the effects the revision has on the estimates, and other implications. Details of the revisions methodology can be found in the Downloads section.

Estimates for Scotland and Northern Ireland are the responsibility of the General Register Office for Scotland (GROS) and NISRA respectively. GROS and NISRA have published their 2002 MYEs and are content that the MYEs in Scotland and Northern Ireland do not require revision either for 2002 or earlier years. If new information emerges which indicates a need for a change to the MYE, the information will be considered in the next round of MYEs for Scotland and Northern Ireland.

National Size of the Revisions

The overall effect of the revision to the 2001 MYE is to add 192,600 people to the population of England and Wales. Most of the addition is to young adult men, 83 per cent (159,700) of the total additional population is males aged 25 to 34, with males aged 35 to 49 making up another 14 per cent (26,800).

This addition represents an increase of 0.36 per cent, when compared to the earlier estimates for the 2001 MYEs, males increasing by 0.74 per cent. The revision is less in Wales at 0.15 per cent than it is in England at 0.38 per cent.

MYEs are subject to a statistical 'margin of error', which, because the size of revision for 2001 is relatively small, will be close to that calculated for the previous 2001 estimate, +/- 0.2 per cent.

Make up of the Revisions

Revisions are being made for four separate factors:

1. To add men who were not measured as usually resident by the Census
2. A small addition to the base for unprocessed Census forms
3. Updating of Census base between September 2002 and February 2003, and
4. Previous double counting between E&W and NI of some armed forces personnel

Of these the first factor is the most significant adding in 187,100 males of whom 160,000 are aged 25 to 34 and the remaining 27,100 are aged 35 to 49. This revision affects 68 local authorities.

The addition for unprocessed forms also only affects a small number of local authorities, 16 in all. However, this revision is split across all ages and both males and females, with 3,600 additional females and 3,300 males. Further information on this revision can be found in the Related links section.

Updating the Census base, from the one that was published in September 2002, to the one published in February 2003, has a negligible effect on the national total. It does however improve the quality of single year of age data. Further details on the differences between the two bases can be found in the Related links section.

Finally there has been a reduction of some 1,300 people due to double counting of some armed forces personnel. These were personnel stationed in Northern Ireland, with a 'home address' in England and Wales. These personnel had originally been included in both Northern Ireland and England and Wales. Since their usual residence is the place where they are stationed in Northern Ireland they have now been removed from the England and Wales estimates. These are armed forces personnel and therefore most of this reduction relates to young adult males.

Geographical Impact of the Revisions

The revisions range in size across local authorities. Lambeth with an increase of 2.8 per cent, 7,400 people, and Birmingham with an increase of 9,400 - though with a bigger population this amounts to an increase of just under one per cent, have the biggest increases. Harlow with a decrease of just over 100 people, or -0.2 per cent, is the only authority to have a decrease of over 100. The decrease is due to the change in the Census base between September and February. The Isles of Scilly sees a bigger percentage decrease, but because of the small size of the authority this amounts to a decrease in single figures only.

The revisions have most impact in London with an increase of 119,900 or 1.7 per cent. In fact all authorities with an increase of over two per cent and all bar one with an increase of over 1.5 per cent are in London.

The relatively large change for Anglesey of 1,000 or 1.5 per cent is due to a number of unprocessed forms for one ward, Amlwch Port, in this authority.

Outside of London the biggest increases are mainly in large urban areas, for example Manchester, Bradford, Liverpool, and Birmingham. This is reflected in the regional increases, with the South West having the smallest increase of 0.06 per cent closely followed by Yorkshire and the Humber, 0.07 per cent and the North East 0.09 per cent. Outside of London, the West Midlands has the highest increase at 0.30 percent, followed by the North West with 0.23 per cent.

Effects by Age

The increase to young adult men accounts for the largest change, and is described in general terms above. This increase is split relatively evenly over the 25 to 34 age group, and also, albeit at a lot lower level, relatively smoothly over the 35 to 49 age group. However, the updating of the Census base does mean there are large changes for some individual ages. This is most notable in a decrease of 8,200, -1.2 per cent, in the number of 15 year olds, which is almost balanced by an increase of 6,400, 1.0 per cent, in the number of 19 year olds.

Sex Ratios

The effect of the revised 2001 estimates on sex ratios can be seen in a graph from the Downloads section.

The changes in the population have not been made by targeting sex ratios, so the resulting sex ratios provide an extra insight into the impact of the changes. The changes made were disproportionately to particular age and sex groups and so they have produced a change in the shape of the sex ratio curve as compared to the original 2001 estimates. As in the original 2001 estimates, at ages 20-24 there are more women than men. This is consistent with the female bias in the flows of international in-migrants aged 15-24. However at ages 25-34 there are now more men than women, with sex ratios a little higher than those estimated for 1991. This is consistent with the male bias in the flows between 1998 and 2001 of international in-migrants aged 25-44. There is a drop in the sex ratio at ages 35-39 even though a small number of men have been added at these ages. It should be noted that it would only take a difference in the number of males and females at ages 35 to 39 of 1 per cent (or +/- 0.5 per cent in each) for the 35-39 sex ratio to form a smooth mid-point between the 30-34 and 40-44 data points. Further research will be carried out as part of the programme of future demographic analyses to give us better understanding of the sex ratios.

Effects on 2002

The above revisions are all applied to 2001 which is the base for the MYEs for England and Wales for the rest of the decade. Thus all the changes also apply to 2002 and are aged on a year.

Unattributed Change

The revision to the 2001 MYEs explains a further 193 thousand of the difference between the previous MYEs and the 2001 Census leaving some 290,000 as unattributed change.

It is thought that a large part of this remaining amount may be due to changing residence status. This is a complex issue which is likely to cause difficulty for estimating migration accurately during the intercensal period. In addition a small part of the difference is thought to be due to including some births to non resident mothers as resident in England and Wales.

Thus the best evidence available is that MYEs remain subject to a small statistical 'bias'. This bias will mean that, unless corrected for, MYEs will gradually 'drift' from their true value over the forthcoming decade.

Consequently for 2002 MYEs a factor has been applied, calculated for each local authority by age and sex, to correct for unattributed population change. This is calculated as the revised 2001 MYEs less the 2001 'rolled forward MYEs' (after allowing for revised migration, the 1991 rebasing and the 2001 revision) divided by 10. Thus the adjustment is fairly small at a national level of 29,100.

This is essentially the approach that has been taken by Scotland and Northern Ireland. Other agencies also apply similar factors to their population estimates, including Statistics Canada and the Australian Bureau of Statistics.

There are several assumptions implicit in this calculation. The biggest assumptions are that future bias will be the same as that which occurred during the 1990s, and that this bias was spread equally across the decade. Without additional information the first assumption has to be the best estimate of what will happen over the next decade; as this is only the first year of the new decade the impact of any error in this assumption is relatively small. On the second assumption the increase in migration levels at the end of the 1990s might lead to a conclusion that the effect was stronger during this period. Therefore, if anything, the allowance for unattributable population change might be an underestimate.

Since the adjustment is calculated separately for each local authority the relative size of the adjustment does vary. The size of the adjustment depends directly on how much the authority 'drifted' over the 1990s and the extent to which that over or under estimate has not been explained by the revision to the 2001 MYEs.

As ONS implement improved methods and data sources the continued need for the adjustment for unattributable population change will need to be reassessed. This reassessment will be carried out annually.

Content from the Office for National Statistics.
© Crown Copyright applies unless otherwise stated.