Chart 3: Index of Similarity

Men Women All
1984 69.09 40.79 57.80
1985 65.26 38.17 54.58
1986 65.11 40.37 55.38
1987 64.98 41.15 55.63
1988 61.93 38.66 52.74
1989 52.39 30.03 43.61
1990 53.59 27.66 43.45
1991 57.68 32.25 48.74
1992 59.08 33.66 51.04
1993 56.28 35.29 49.83
1994 53.93 32.10 47.03
1995 52.67 28.33 44.63


The Index of Similarity used to compare the coverage of the LFS unemployment and Count of Claimant series is defined formally as:

Index of Similarity = 100 * A/(B + C - A)

Where A is the population who are classified as unemployed by both series, B is the total number classified as unemployed by the Count of Claimants, and C is the total number classified as unemployed by the LFS unemployment series. If the two definitions series cover exactly the same population A=B=C and the Index has its maximum theoretical value of 100. If the two definitions have no population in common then the Index has its minimum theoretical value of zero.

The numerator A of the Index of Similarity is the category labeled in the statistics as LFS claimants which is are included in both the Count of Claimants and the LFS unemployment definition. In 1995, for example, there were just over a million LFS claimants. But more than 800 thousand Economically inactive Claimants and Claimants in employment (i.e. B minus A in the Index) were included in the Count of Claimants but not the LFS. And there were nearly a million LFS non-claimants (i.e. C minus A in the Index) were included in the LFS definition but not in the Count of Claimants. The average Index of Similarity works out at 46% as shown in Chart 3. In other words, it could be said that the two series by 1995 had less in common than they had in difference.

The year-to-year variations in the size of these indexes shown in Chart 3 are associated with year-to-year changes in the conditions of the labour market (see Thomas, Integrating measures of unemployment, forthcoming).