The VDO sensor is rated at 10 ohms to 180 ohms, but can vary rather widely, but 1.5 ohms seems a stretch, unless there is a sensor type that I am not aware of.
I recommend you use a 250 ohm (249 ohm 1%) metal film resistor, or go as high as 300 ohms if you wish. At 250 ohms, you will draw about .1 watt through a shorted sensor (worst case scenario), which is well within the rating of a 1/4 watt resistor.
Go here to this URL for a voltage divider to determine the output voltages at the two extremes. You are looking at a divider of 250/10 for "0psi" and and 250/180 for 30psi or 100psi depending on which VDO you are using.
If you need more resolution and/or precision, you should go with the sensors I discuss below. A 4.0V spread will give you about 820 counts to work with, and the precision and linearity of the sensor is outstanding.
http://hyperphysics.phy-astr.gsu.edu/hbase/electric/voldiv.html
For 0psi, I show an output of 0.192V, and for maxpsi, an output of 1.895V (assuming a high Z input -- I'd use a 10K resistor between the divider and the Picaxe input if I were you).
The difference is about 1.70V - so divide 1.7V by 5.0V and multiply by 1023 and you get a 348 count, which is the number of steps you have to work with over that pressure range.
I feel that both the VDO and the Stewart Warner resistive sensors are primitive and inaccurate by modern standards. If you want to do precision measuring, suggest you purchase an Autometer 2245 (15psi) or 2246 (100psi) sensor. Then, you'll have a full 4.0V spread between 0psi and maxpsi, and an overall error of <1% of FS across the entire pressure range. Or go to Digi-key and type in "pressure transducer" and you'll find SSI models available for around $65 with 24-gauge wire pigtails. This is the best buy you can get, but be aware they are not protected for reverse or over voltage and I don't like the idea of 24-gauge wire in engine compartments.
Hope this helps
PS With reference to "spikes" on the input: I would recommend that, rather than a capacitor, you write some code to average the input over a period of time. Write a loop that runs while x<20, and add the input value each time. Then, after taking twenty readings or however many you want (with a "pause 2") in between, divide the value by 20, and you've got an average count over around 50ms that will take care of any variation. Your routine should run fast enough that you will not notice the delay. You can play with the values in the routine.