The Cq values reflect the amount of RNA you have extracted, which of course depends on the amount of sample you use, the extraction method and final resuspension volume, amongst other things. So there is not definitive answer to your question. However, when we extract RNA from 0.5 mg colorectal tissue biopsies using Qiagen's RNeasy and suspend in 50 ul, we observe Cqs in the region of 12 or so. If the same individual uses the same amount of starting material and the same extraction method  then the sample at Cq 16 suggests a problem with the extraction process. This could be inhibition, so you should cary out a 1:10 dilution an compare Cqs. If they differ by 3.2, then there is no inhibition, just less RNA in that sample. "A bit higher" is too vague to comment on, since a bit might be 0.5 Cqs or 2, with the latter indicating a significant 4-fold difference.

No, the CT values do not reflect the amount of RNA extracted.  They reflect the cDNA input to the Taqman assay itself.  I always quant and normalized the RNA before cDNA conversion and then quant and normalized the cDNA before running the taqman RT-PCR.  I always got very low (5-6) values for 18s but it is dependent on the cDNA input to the RT-PCR reaction.  You can't really compare CT values without knowing the cDNA input to each reaction.  The "16" value is probably a bad sample or not properly diluted.  I'd re-quantify your 16 samples and then normalize and repeat.  You do want to make sure that you do at least a 1:10 dilution of the cDNA before running the assay.  Even though you normalize to a housekeeping gene, it is always good practice to normalize your RNA/cDNA input before running the assay and make sure the cDNA is diluted at least 1:10.  You should avoid such large CT variations this way.  

I agree with Todd. Quantify total RNA of your samples using Nanodrop if you can do it, this will give you some useful information: total amount of RNA, the integrity of this RNA and if there are any contaminants in your samples that can interfere in cDNA conversion reaction or taqman RT-PCR assay (like fenol contamination). I have obtained 18s values between 9 and 15 depending on the experiment, but the most important thing is to minimize 18s variations in the same extraction and in taqman RT-PCR assay . I have experience with Qiagen columns and with TRipure RNA extraction, and it is very difficult to obtain similar values with Tripure than with the columns extraction kit because of fenol contaminations. Is that the reason why we use a housekeeping gene, but if the Ct values of your samples are too different, it can create artifacts in your final data.

Although it is of course correct to say that the Cq reflects the cDNA input into your qPCR reaction, the cDNA input is dependent on the amount of RNA you have reverse transcribed in the first place. Hence, if you start off with  the same amount of RNA between samples, your Cqs will be a reflection of the original RNA concentration. However, the efficiency of conversion from RNA to cDNA is RT , sample and target dependent, and can vary considerably (see Stahlberg's two papers in Clin Chem. In 2094). But assuming you are using the same amount of RNA  from each sample to start off with and the same RT you can quantify your RNA with reasonable accuracy, and if you are using relative quantification this san be within 5-10 fold, depending on how much RNA you are analysing.

In most cases the levels of 18S was quite high as compared to the target you are testing. Ct value of 18S varies with samples and with extraction procedures. Sometimes, higher abundance of 18S transcripts may mislead the study conclusion if target transcripts were very low than the 18S which is going reach plateau early and this cannot be comparable with target. Hence using of competing primer you can get linear amplification rate of 18S and target. This will provide better comparison of target expression levels. For details refer this paper "Modified competing polymerase chain reaction primer for single tube quantitative PCR" published in Analytical Biochemistry 427 (2012) 175–177.