The table shows the call outcomes for all calls dialed for projects A and B.

These results can be depicted as follows:

Connected calls correspond to calls that have a call outcome of local hangup, remote hangup, or extension on hook. For Project A this was 188414 calls out of the 383863 dialed, or 49.1%. For Project B this was 68135 of the 101803 dialed, or 66.9%. These connect rates correspond to failure rates of 50.9% and 33.1% respectively. The most common reason for calls failing to connect is no answer, which is higher for residential Project A than for business Project B. Project A also had 7.4% of calls resulting in "Unobtainable number" and 2.4% as "Call failure (cause unknown)", while Project B had virtually none of these outcomes. This is probably a result of the Project A sample including a percentage of random digit dialing.

Of the other calls that failed to connect, a small percentage (around 3% to 4% of calls) was to busy numbers and about 3% were fax machines. The other possibilities for call failures were relatively insignificant at less than 1% of calls dialed.

For silent calls (no extension available), 631 were made for Project A, corresponding to 0.33% of connected calls, and 181 were made for Project B, corresponding to 0.27% of connected calls. Both of these are well below the Ofcom limit of 3% of connected calls.

The next two charts are simplified versions of the previous ones that combine local hang-up, remote hang-up, and extension on-hook into a single Connected segment and show only the top five call failures. Project A has a higher no answer percentage and more call failures which result in the percentage of connected calls made for that project being lower. Calls to fax/modems, busy numbers, and other failures are relatively consistent between the two projects and are relatively low (around 4%, 3%, and 1% respectively).

The call outcomes for Project A were also examined with regard to time. Two periods were considered: 3pm to 4pm (near the start of a shift) and 8pm to 9pm (near the end of a shift), both on 6 November. Around 2000 calls were made in both periods. The number of connected calls was constant at 65%, but the number of no answer calls decreased from 32% to 19%, presumably because this was a residential survey and more respondents were home at night. Strangely, there were more failed calls (unobtainable number, unknown call failure) and fax/modems during the later period. Examination of the data for other shifts showed that the number of failed calls varied greatly from hour to hour which suggests that the random phone numbers may not have been distributed evenly throughout the sample.

Because of the potential uneven distribution of the sample, a similar analysis was performed on the Project B call logs. Following are charts for two periods (10am to 11am and 3pm to 4pm) for a single project. The call outcome distribution is virtually constant throughout the day; the only difference is in the silent call rates, 0.4% in the first period and 3.1% in the second period.