Hypothesised explanatory variables included in the regression models and hypotheses regarding their influence on output power levels
| Variable | Level of aggregation of information | Categories | Hypothesis |
| Network/study area characteristics | |||
| Study centre | Individual | 12 centres (including one repeat study) | Characteristics of the study location influence output power levels |
| Network operator | Individual | 36 operators | Technical characteristics of a network influence the efficiency of APC |
| Time period of SMP study | Individual | 2001–2002, 2003, 2004–2005 | Use of APC may have changed over time |
| Phone characteristics | |||
| SMP model | Individual | 4 SMP models | Downregulation of output power under APC may depend on handset design |
| Call characteristics | |||
| Duration of call | Phone call | <5, 5–10, 10–30, 30–60, ⩾60 s | Very short calls may have higher average power due to the initial connection at high power level |
| Time of day | Phone call | Day 8:00–19:00, evening 19:00–23:00, night 23:00–8:00 | Calls during busy time of the day may have higher power levels as more handovers are expected when base stations are full |
| Phone use circumstances | |||
| Frequency of use in urban/rural areas | Individual | Mainly urban, mainly rural, both | Power levels may be higher in areas with lower densities of base stations (ie, rural areas) |
| Frequency of use in city centre/suburban areas (if urban) | Individual | Mainly centre, mainly suburb, both | Idem |
| Frequency of use in a moving vehicle | Individual | Never, less than half time, about half, more than half, nearly always | Power levels may be higher if phone is used often in moving vehicle due to frequent handovers |
| Frequency of use inside buildings | Individual | Never, less than half time, about half, more than half, nearly always | Power levels may be higher inside buildings due to shielding of walls |
| User characteristics | |||
| Age | Individual | ⩽30, 31–40, 41–50, ⩾51 years | Power levels may vary between age groups due to other, unmeasured factors |
| Gender | Individual | Male, female | Power levels may vary between genders due to other, unmeasured factors |
APC, adaptive power control; SMP, software-modified phone.