Any organisation which provides access to data over a long period of time should be fully trusted only with a public statement describing the practices they follow and the provenance of data they provide. Standards of trust are critical.

A statement setting out an organisation's practices in handling data should be a given. How otherwise can we tell the extent to which the data may have been altered during processing before they arrive on the user's desktop?

We implicitly trust national statistical organisations to publish correct and valid data, but without some reassurances we cannot be certain that the data published by others on one day are the same as the data published the day before.

There are three key standards which digital repositories like the UK Data Archive can use to assess themselves and to support public statements about their level of trustworthiness:

• Data Seal of Approval (DSA)
• Audit and certification of trustworthy digital repositories (ISO 16363)
• Criteria for Trustworthy Digital Archives (DIN 31644)

The standards provide the basis of a framework by which different levels of trust of digital repositories can be demonstrated. They do this by providing a methodology to justify and objectively prove adherence to certain criteria which provides the right levels of trust.

None of the three standards have been designed for the specific demands of social science research data, and all need some local interpretation in order to apply them.  However, we believe that it is necessary to provide both a method of assessment for more mature archives and, for less established organisations, a method of understanding the demands and constraints of inculcating trust.

It is a feather in the cap for organisations to be successfully audited and registered against a standard, but this mark of approval alone is not usually sufficient to meet all the trust requirements between two parties. Two reasons for this stand out:

• the body which places its trust in another may not always be aware of the small print of the standard itself, and thus may not have full confidence that the standard meets its levels of trust.
• this body may also have specialised requirements which cannot be addressed by a broad-based standard.

Both need to be considered when working within this emerging and less than completely objective and measurable environment.