Creativity is a about freely generating new ideas and culling 99% of them. The end product of creativity is thus new, useful ideas.
One’s quality as a scientist is a product of these two abilities.
Some folks are so hyper-critical of any new idea that nothing leaves their notebook. Even writing a paragraph becomes an excruciating ordeal, as any abstraction is ruthlessly worried and excised.
Some folks fill their notebooks with idea after idea, but end up swallowed up in the thicket unable to commit to any one idea for long (sort of an intellectual ADHD). Consider this post a dose of analytical Ritalin, prescribed by the good Dr. Tufte.
Edward Tufte is someone every beginning scientist should get to know. His Visual Display of Quantitative Information is the best introduction to the theory and practice of effective graphics. His website includes a bulletin board on all topics analytical, graphical, aesthetic, and concrete. The site’s only downside is that it can swallow Friday afternoons whole if you let it.
In short, Tufte is the paragon of GTDA’s guiding principle:
Quality = great content*great design.
Here is the opening salvo (plus my commentary) from his “Advice for effective analytical reasoning.”
1. “Be approximately right rather than exactly wrong.” John W. Tukey
Unless you are designing a bridge it is not necessary to be both accurate and precise the first time around. Great science is about discovering new territory, not running the subsequent surveyor team. There is a reason that introductory chemistry courses begin with the principles of significant decimal places. Know the limits of your measurements, and that two imprecise measures, multiplied together, increase the imprecision.
2. “The first principle is that you must not fool yourself–and you are the easiest person to fool.” Richard Feynman
It is so easy to fall in love with an hypothesis, especially if you thought it up. But we protect the ones we love. Fight this tendency. The best way is to generate multiple hypotheses and let them compete to explain your data. (A more pernicious example of Feynman’s principle, is collecting and massaging data just to get a probability value <0.05, the magical 1 in 20 cutoff that is the gold standard of most scientists.)
3. Ask questions.
Scientists are characterized by one property above all others. They are curious. And ultimately all good science starts with a good question. Yet beginning grad students are often reluctant to ask questions, particularly in public. Swallow your pride and wallow in. Sometimes you will get a cursory answer. Sometimes you will get a quizzical look and a request to “rephrase the question”. This is great, as now you have been given an opportunity to hone your skills at questioning. I’ve never heard a grad student criticized because she “Asks too many questions.”. I have heard professors wonder out loud, “Is she even listening?”.
4. Develop and fine-tune a sense of the relevant, both for identifying the key leverage points in any problem and also for examining large amounts of information to find the rare diamonds in the sand.
This is where your reading comes in. The more you read and the more you dissect papers in study groups, the more you will recognize areas of bedrock certainty, the mushy areas of things we think we know, and the crevices of ignorance and illogic. When planning your research, focus on the latter two. Also, remember that every hypothesis has three weak points: its assumptions, its logic, and its predictions. Each is fair game for attack.
5. Nearly all serious analysis requires multivariate-thinking, comparison-thinking, and causal-thinking. Develop such thinking.
Most things worthy of study are complex–representing multiple processes linked together in networks. Search first for the skeleton of that network. Identify the two or three most important processes, and explore with pencil and paper all the ways these processes can interact. Then design your experiments to test for these interactions. Develop skills in multivariate statistics, computer programming, calculus, and/or structural equation modeling. All develop your analytical muscle, and all are useful tools in making the complex accessible.