• PVC pipe (especially white PVC) is not a preferred material for antenna construction for the following reasons:
• PVC of the sort used for pipes has poor mechanical strength.
• It has rather poor longevity when exposed to direct sunlight.  Being intended for use underground or inside buildings, prolonged exposure to direct sunlight will cause it to discolor rather quickly, and it will become brittle.
• Under stress and when exposed to heat (e.g. hot sun) it will often sag.  This can skew element alignment for longer antennas.
• If you must use plastic pipe for a boom, you can paint it - to protect it from UV, or you can use UV-stabilized PVC.  Also consider ABS pipe as an option as it is somewhat more resistant to the effects of prolonged UV exposure.  (It's still plastic pipe and rather flimsy, though.)
• Soft-drawn copper tubing is, electrically speaking, quite good but it is not a very good choice for use in elements, mechanically speaking:
• The material is very soft and is easy to bend.
• Being hollow, it tends to accumulate insects (especially spiders.)  This is more of an annoyance rather than a performance issue.
• In northern climates, if left uncapped, they tend to accumulate moisture and ice - especially if they are full of spiders/webs.  This can cause them to become heavy (as they fill) or even split.  (A plug of silicone in each end of the element will help prevent this problem.)
• When building any antenna - especially a yagi - the diameter of the elements has an effect on how long the element should be.  For example, consider two antennas that are identical except for the diameter of their elements:  The antenna with skinny elements will have shorter elements than the one with rather fat elements.  This means that if an antenna designed with, say, 1/8" diameter elements is built using 1/4" diameter elements, it will be high in frequency.
• An antenna really does need to exhibit a reasonable match.  If the return loss of an antenna is too low (that is, the SWR is very high) then one cannot efficiently transfer power into and out of the antenna.  The design of any antenna should really be checked for consistency and repeatability.  For instance, a return loss of 5 db means that over 56% of the power sent to the antenna gets reflected back - and this correlates with an SWR of at least 3.5:1.  This sort of mismatch not only yields poor performance, but it has the potential to damage a transmitter connected to it.
• As mentioned above, a matching section should be effective and repeatable.  Long experience has shown that there are several basic types of matches that are simple and effective in yagi construction.  One of the things in common with all of these matches is that they have some property of impedance transformation - which makes sense because a matching section would not be necessary if there were no need for it!
• A gamma match.  This is a single-ended matching network that is quite simple, mechanically, and can be adjusted to accommodate a wide range of matching conditions.  It's complicating factor is that it typically requires a series capacitance.
• A "T" match.  This is a balanced matching network that feeds each half of the element equally.  Its main complication is that it typically uses a 1/2 wave delay line to feed the other half of the element out of phase.
• A "J-pole" type match.  This is, essentially, using a J-pole 1/2 wave element as the driven element.  The J-pole has the attraction that it's simple to construct as it has a good range in its impedance matching capabilities.  While the initial adjustment of the two tap points for proper match may be tricky, it can be repeatable when accurately reproduced.  As a driven element in a vertical yagi, it can also be used as the mechanical support if the yagi is fairly short.  It potential problem is that it can have a pattern symmetry problem - but that's usually not important for small-ish yagis.  It also isn't well-suited for horizontal polarization.
• The matching network should be fairly rugged.  The electrical nature of the matching network can change dramatically if it is mechanically flimsy.  Also, the accumulation of rain/snow can greatly affect the performance, as can degradation of the materials used upon exposure to the elements.
• Connection to the feedpoint.  Even though small-diameter coax can be lossy, such a short length of it is used that its contribution to the losses is usually negligible.  Generally, open-wire line is impractical on yagi antenna (with the possible exceptions of some log periodics and collinear arrays) and when used, open-wire line will require some form of impedance transformation to make it useable at 50 ohms, anyway.
• When mounting the coaxial connector, one must be wary of mounting it in such a fashion that there are too few threads exposed to allow the connector (especially a UHF connector) to be fully tightened.  This can be a particular problem when single-hole chassis-mount UHF connectors are used.  Typically, these are intended to be used on a fairly thin metal chassis - but when used on a thick material (especially plastic - which must be rather thick for adequate strength) too much thread is taken up and the body of the cable's connector cannot be cinched into place by the ferrule, resulting in a loose ground connection.

It should be mentioned that even the poorest of UHF antenna may easily outperform an indoor one.  Typical construction materials used in a house can easily result in 10-15db (or more) of additional attenuation.  Added onto that, a typical rubber duck antenna can have less gain that a dipole.

It should come as no surprise, then, when even a very poorly performing antenna is placed outside - even when fed with a lossy coaxial cable - it is likely to outperform nearly any indoor antenna to which it may be compared.  It should also be noted that at higher frequencies - such as 70 cm - even a few 10's of feet of a small cable such as RG-58 can have very high losses.  These losses not only "eat" both transmit and receive signals, but they can cause a false sense of security:  Even an extremely high SWR on the far end of the cable can, when losses are involved, appear to be perfectly acceptable, thus masking problems.