# "Successful projects share the same 5 keys qualities. In each quality our project achieves over average score". What can we conclude about this?

I'm not sure if this is a statistical problem, but I suppose it's a frequent problem involving using statistics, so I ask here anyway. Hope it's on-topic.

Although music is a very risky business, we found that Radiohead’s In Rainbows campaign shares the same five key qualities as any successful “pay as you wish” pricing program:

1. A product with a low marginal cost
2. A fair-minded customer
3. A product that can be sold credibly at a wide range of prices
4. A strong relationship between buyer and seller
5. A very competitive marketplace

In each quality, we assess our project has a higher than average score like this:

Quality no. Score (over 10)
1 6
2 7
3 8
4 7
5 7

There is no measurement on each quality. We just reason with what we have and rate our confidence. Basically we just "feel the number".

What can we conclude from this? What techniques we can do to confirm or deny any assumption we have? And if we think "our project is more likely to success than fail", do we have any fallacy?

Related: What stops the pay-what-you-want pricing strategy from being more popular? on Economics SE

• Right now, I don't see a statistical question here. If you say what data you have, then ... maybe? It would also need ways to measure the five characteristics, and an operationalzation of success. Maybe someone who has read the book could fill some of that in? Dec 31, 2023 at 13:02
• That's why I say "I'm not sure if this is a statistical problem, but I suppose it's a frequent problem involving using statistics". To explain more on why I decide to ask here, at first I thought I had felt in to the illusory correlation. Then I read more about the famous "correlation does not imply causation", and decide that I need help from folks with strong statistics background. Is this line of thinking reasonable? Dec 31, 2023 at 13:19

do we have any fallacy?

$$P(Success\vert Traits)$$ is what you want to know. You have $$P(Traits\vert Success)=1$$. However, flipping the conditioning leads to a different probability. You are not (nearly) assured of success by having those traits. You have to look at how many times those traits exist yet success is not achieved.
• Yes, conditional probability is the term I'm looking for. As for the assessment of how well our project achieves the rate, what is the term of that? What is the relation between the scores, the probability of the success of the project, and $P(Success)$ in the sample space? Dec 31, 2023 at 16:28
• @Ooker Then you have the model as $P(Success)=\sum_{j=1}^5 0.2 x_j$ for the $x_j$ indicators of whether characteristic $j$ is present, but then you’re not doing any machine learning or regression modeling. You’re not learning from the data. You’re just assuming that each factor has that influence, and I’ll venture a guess that, even if every successful case has those five factors, not every time those five factors are observed is there success.