Due to the escalating usage of cellphones, cellphone distracted driving remains one of the fastest-growing traffic safety problems. Two basic types of cellphone distractions are – talking (handheld or hands-free) and manipulating (texting, internet browsing, using apps for social networking, navigation, entertainment, etc.). To explore the association between the cellphone distraction type and characteristics of a driver, roadway, and vehicle, we used Fisher's exact test and association rule mining to analyze the data from ten different hour-long observation sessions performed both at intersections and on straight continuous segments. Among 3727 drivers observed, 22.1% were found distracted by cellphones. More drivers were involved in manipulating (12.8%) than talking on the phone (9.3%). The Fisher's exact test results show that intersections and the presence of passengers may be significantly associated with driver distraction type. This study makes a novel attempt to utilize association rule mining in cellphone distraction observation. A set of generated rules and their network visualization demonstrated how a specific group of characteristics can be related to cellphone conversation or manipulation through the measures of support, confidence, and lift. 'Peak hour', 'Absence of passengers', and 'Urban setting' are more likely to be associated with distracted driving than other characteristics. The results from association rules revealed that male drivers were found to be predominantly involved in cellphone manipulation during peak hours on urban segments than female drivers, and they were also largely involved in cellphone conversation during peak hours on the rural segment. Female drivers engaged in manipulating cellphones at the urban intersection during both peak and off-peak hours. In the rural area during peak hours, female drivers were manipulating at the intersection and were talking while driving on the continuous segment. Drivers aged '30–60 years' were found to be more associated with cellphone use compared to '<30 years' and '>60 years' age group. 'Passenger car drivers during peak hours' were more associated with manipulating cellphones than expected compared to any other collective group, whereas 'female passenger car drivers driving on continuous segment' were strongly associated with cellphone conversation. The detailed discussion of the findings from association rule mining will be useful for targeting a specific group of relevant characteristics in potential cellphone distraction reduction strategies.