Recent social security developments in Austria.

usmg the approprnte standard error table Square these standard errors to get vanances and sum the variances Then take the square root of the sum to get the standard error of the d&vence 7Vlth the not&on shown above, If /DJ > (2 x %D), there 1s a s~gmficant dlfforence ,n estunates A and B at the 95.percent level Assummg the cO”aPl*nce 1s equal to zero WI11 result ,n accurate estnnates of standard errors of differences for most compansons prestmted m the report, except for those between 1973 and 1974 The 1973 and 1974 e&mates of the same charactenstxs, b&cause this 1s ri panel survey, are not uncorrelated If the covarmnce 1s assumed to be zero when computmg standard errors for year-to-year differences the result 1s an over&m&e and a more strmgent test of significance ‘J