I think the float data is 4 bytes 32 bits.
If you are sending 4 float numbers,
#define dataNum 16 // roll, pitch, yaw, Vbatt (4 float data 16 bytes)
union unionData { // Union for bit convertion 32 <--> 8
float dataBuff32[dataNum/4]; // float : 32bit
uint8_t dataBuff8[dataNum];
};
You can confirm this by sending and receiving an easy-to-understand number.