Last month's JNLR radio listenership figures were accurate and within the normal statistical margin of error, an investigation has found.

The 2002 figures had been queried by the chief executive of FM 104, Mr Dermot Hanrahan, who said the sample sizes used in the survey did not reflect the breakdown of the population generally. The JNLR figures are produced by MRBI.

However, an investigation by a JNLR technical committee made up of senior advertising and radio figures found the survey results to be "sound" and "robust".

However, the investigation concluded there was an "unacceptable slippage in the sample size for 25-34 year olds".

"As a consequence of this, the technical committee are recommending that MRBI provide a monthly sampling structure report to ensure that such slippage will not occur again," said a statement from the JNLR technical committee.

Earlier this month, the chief executive of the Irish Advertising Practitioners Institute, Mr Steve Shanahan, issued what he called a "health warning" about the Dublin section of the survey.

Mr Hanrahan also raised concerns, particularly about the accuracy of figures for two key groups: housewives and housewives with children. Mr Hanrahan declined to comment on the outcome of the investigation last night.

The JNLRs are the main way advertisers judge the listenership of individual stations. A sharp movement in the figures for a particular station can have an impact on how much advertising an agency places with that station. If the set of figures is wrong for one station, the whole survey could theoretically unravel.

The investigation agreed with Mr Hanrahan that upon first viewing there seemed to be "unusual variations" in the figures for housewives and housewives with children.

However, it said after comparing 2002 figures with 2001, the figures were accurate.

While the controversy over FM 104's figures may end, the statement from the JNLR technical committee said figures for newer stations should be "treated with caution".

"To help the newer stations being measured, the JNLR management committee had requested that six month share data be produced, for the first time, in this report.

"Some of this data has proved not to be statistically valid in certain sub-cells as a result of small sample sizes and should be treated with caution," said the JNLR technical committee statement.

Mr Hanrahan's original queries surprised the JNLR committee because over the years the figures have been accepted at face value by the whole radio and advertising communities.

Mr Hanrahan was particularly concerned that there appeared to be difficulties about the 25-34 age group, the most sensitive audience for virtually all radio stations.