Just to clear up some things that are mentioned above which indicate a slight misunderstanding of data acquisition. The Flight Data Acquisition Unit (FDAU) captures data from the aircraft sensors. ICAO Annexe 10 is the source book for the rules, but different regulatory bodies may interpret these rules differently and several possible data frames are possible while remaining within ICAO Annexe 10 (E.g. FAA and UK CAA dataframes are significantly different with respect to capture rates for certain parameters.). Also, the FDAU monitors the mandated data to a dataframe setup that depends on when the aircraft was originally certified and what regulations are in place. Older aircraft are not usually required to upgrade the Flight Data monitoring system when new rules are brought in.
After signal processing and verification, FDAU output is passed to the DFDR for recording. (Verification simply means that the output data is checked for "reasonableness".)
Flight Data Recorder (DFDR) data from a representative flight sector is downloaded from each aircraft regularly by engineering and checked for capture error rate (must be <2% generally) The data is analysed and stored for reference - should there be an accident, the stored data is used as a data reference, as each aircraft will have slight variations in the parameter reference levels. In addition, the aircraft maintenance schedule calls for regular checks of the DFDR system and each of the recorded parameters is checked right across its range. (e.g. each aileron position measured at centre, full up and full down) using a data monitoring test set. Discretes are checked in both possible states (e.g. "On" or "Off") Any irregularities are corrected before a Certificate of Release to Service (CRS) is issued by the certifying engineer.
In most Quick Access Recorder systems - but not all! - the FDAU data is captured in parallel with the DFDR data and is therefore exactly the same data.
In interpreting data, experience and objectivity are required. The data doesn't tell you the reasons for any particualr measurement or change in value. It is important to bear in mind that the raw data is providing information on what the sensor is measuring, not what was really happening. The sensor may be faulty or completely detached!
I have personally listened to inexperienced personnel looking at data that I had downloaded for investigation say things like "Look, he put in excessive nose down pitch and the speed began to run away within ten seconds."
No!
The data says that "pitch increased to 9 degrees nose down and the speed increased steadily over the next ten seconds to reach 275 knots". The data doesn't say why - it is just dumb data.
One reason why there may be a perception that DFDR data is more reliable for legal purposes than QAR data is that it is subjected to airworthiness engineering surveillance. Obviously that is true, but the same data is being recorded by the QAR. If there is any doubt about the veracity of information found on the QAR, the parameter in question can be verified against the DFDR recorded parameter. In the extreme an engineering check can be performed against the documented values.
But be aware that there is always a 2% possibility of data error for any individual parameter in any particular record.