According to recent world-wide conferences atteneded by qPCR afficianados since 2009, a mixture of randomers and Oligo dT primers is best. If just using Oligo dT for priming RT reactions, you get "3' bias." But, when using the mixture of randomers and dT's, you get better overall representation of the transcriptome including fairer representation of 5' regions. BIO-RAD sells this mixture - but I'm not sure you can find out what ratio the mix is. I've heard it is a 1:5 mix of Oligo dT to randomers. Or it might be the other way around. If only using dTs for priming, it is then suggested to design your primers nearest the 3' end of all targets since the 3' bias introduced by poly A/dT priming preferentially reverse transcribes the 3' ends/regions more completely than the 5' regions.