In the first article in this series, I introduced readers to the four phases of testing and, hopefully, showed just how pointless it is buying into a biotech company's price rises too early in the testing cycle.
After that article was published, Sirtex Medical Limited (ASX: SRX) provided a perfect example of how long clinical trials can take.
It took Sirtex more than three years just to recruit enough patients for its Phase III liver cancer trials – which will take an additional eighteen months to conduct.
That's obviously not counting the time spent in Phase I, Phase II and preliminary/animal trials.
So if you're getting excited about price spikes after pre-phase I work at companies like Cynata Therapeutics Ltd (ASX: CYP) or Novogen Limited (ASX:NRT), you're basically just gambling.
Today's article is going to cover a few more important pieces of information, namely; sample size, statistical significance, and correlation.
Sample Size
'Sample' refers to the group being tested, while 'size' refers to the number of people in the group.
A sample should be simultaneously 'representative' (variables can include severity/type of symptoms, age, weight, , etc) of the population being tested, and large enough to give meaningful statistical results.
Generally speaking scientists have a large say in many ASX research companies, and very little investor oversight is necessary.
One warning sign to watch out for is if the company appears to be run more by businessmen, like Phytotech Medical Ltd (ASX: PYL), as there can often be a disconnect between what business interests are trying to do versus what the research says can be done.
Another warning sign is a company hyping initial success in animal/lab testing or trials with small sample sizes.
These trials are a LONG way from the final word on the topic, and face the additional hurdle of having to be proven better than existing 'best practice' treatments for a given disease before they hit mainstream success.
Statistical Significance/ Probability
This refers to the likelihood of a given result being due to chance, and is generally expressed in the form p = x.xx. Sometimes < and > are used to indicate that probability is less than (<) or greater than (>) a given figure.
For instance if we toss a coin, aiming for heads (50% chance of success), this would be expressed as p = 0.50.
In scientific research probability refers to the likelihood of results being due to chance, and the aim is to get the figure as low as possible.
In the medical arena, generally only p< (less than) 0.01 is acceptable, as this means there is a less than 1% possibility of results being due to chance.
HOWEVER, you need to be careful what you are reading, because the numbers can get confusing.
Taking a real-life example from Sirtex's book, proprietary SIR-Spheres combined with Hepatic Artery Chemotherapy (HAC) were found to deliver a significantly longer (and thus better) Time To Progression of colorectal cancer than HAC alone (p=0.001).
Now, using a made-up example, imagine that the same study found that Quality of Life (QoL) for patients with SIR-Spheres + HAC was moderately better than those on HAC alone (p=0.07).
Because probability of 0.07 was higher than the cut off of 0.01 and 0.05 (another common cut off level), the QoL results are NOT significant, i.e. there is no difference in patient QoL between the two treatments.
Logically if quality of life is the same, you are naturally going to choose the treatment that is more effective, making the non-significant QoL results probably irrelevant in this situation.
Watch out for companies that publish results showing that a treatment is not effective, yet the announcement spins the results by saying something like 'we are committed to delivering on marketable outcomes for our' (ineffective) '_____ product.'
I won't name any names, but rest assured that it does happen and is a potential risk to owning a research company.
Correlation
Last and least is correlation, which means the relationship between two variables.
A correlation is expressed like probability, except using the letter r, as a fraction of 1, e.g. r=0.8. Correlations can be either positive or negative.
You won't see too many of these in the ASX because they're kind of worthless as far as research goes.
To give a made-up example; there is a significant positive correlation (r=0.96, p=0.0001) between being a white, obese male under the age of 25 and having all four limbs.
The shrewd reader might realise that in fact being white, male, obese, or under 25 has very little to do with having two arms and two legs and thus, the correlation is worthless.
So you have to be wary of what is being correlated.
Trends involving consumer spending, product preferences, and payment patterns are where correlations can really come into their own, especially for companies like Woolworths Limited (ASX: WOW).
Given Woolies' recent parcel delivery partnership with eBay, you can bet your bottom dollar that researchers will be investigating if there's a significant positive correlation between picking up a parcel and making a purchase.
In fact as businesses continue to develop, I expect to see more and more correlations being employed to measure various developments – but whether their usage is reported is anyone's guess.
So there you have it, three more pieces of information essential to identifying whether a stock is really going to deliver big – or whether the market's just buying into the hype.
Part 3 of this series will focus on how you can get more in-depth with a research stock, and really break down the public announcements to see precisely how things are going.
What did you think of the article? Was there anything I missed you'd like to know more about in a future version?
Until then, invest Foolishly and check out The Motley Fool's free report to find out how you can invest more like Warren Buffett, one of the world's richest men and someone whose investing style everyone should aspire to emulate.
