Statistical & Financial Consulting by Stanford PhD
Home Page
DISSERTATION ASSISTANCE


DISSERTATION TIPS

As a future expert in your area, you have spent years on learning topics which are dear to you. You would prefer to keep going and focus on the issues that got you in the field in the first place. The catch is the empirical analysis you need to perform. It has to be formal, it has to rely on the statistical methods which are most efficient for your problem. Moreover, once you have performed the analysis, you have to be prepared to talk about all kinds of diagnostics, statistical assumptions and limitations which would justify the chosen modeling approach in the eyes of the thesis committee. You may have an idea about the statistical model and the estimation approach, or you may be completely lost. Either way you need quality control by a statistics expert before the committee members hear anything important from you.

The truth is the following: even if you think that you've got a portfolio of reasonable statistical ideas, you may be slightly off the track. Most likely, you have taken only a handful of statistics courses in your life. Most likely, those were the ones required for people in your field. I have taught such courses and know that often (not always) they are limited to the main and most popular statistical techniques. This may not be enough for a very specific problem you've got. You may need a more advanced version of a presented method, because of serial correlation, censored observations, nonlinear interactions, hidden factors or any other issues reflected in your data set. You cannot know all the relevant techniques but a statistics expert can. Once the optimal solution is identified, it can be explained to you reasonably fast. It's not like you are not fit to understand the intuition and some of the proofs behind the optimal procedure. It is the small size of your scope that may not allow you to see this optimal procedure yourself at the beginning of the research.

And the decisions made at the beginning are most important. One wrong choice and the consequences grow as an avalanche. Not knowing what your capabilities are, you submit ambitious research questions to the committee. They approve them and now you are stuck. You cannot address those questions with the data at hand. Sooner or later you will have to go back and ask for restructuring of your responsibilities... Another situation: you phrase the research questions correctly but propose wrong statistical methods. Your committee does not see the data or have not studied the details of the problem. The proposal looks good to them at first, they approve the methods but sooner or later the truth comes out and you look bad. Get statistics help before you have begun the empirical research.

In 90% of the cases, the members of your committee understand that you cannot become knowledgeable in statistics overnight. They will encourage your getting statistical assistance from outside because that means they will have less work to do. Oftentimes they think they have better things to do and view the advisory responsibilities as necessary but burdensome. In the least case, they will approve somebody else teaching statistics to you. In the intermediate case, they will approve an expert statistician helping you out with the data analysis. In the best case, they will pay for statistical consulting services out of their grants. Get in touch with your advisers now and get a sense of how much freedom they are willing to give you.

If you are stuck and not sure how to do the analysis, you spend weeks trying to read more statistics, learn statistical software and identify potential pitfalls in your data to see if more data are necessary. You spend days waiting for another meeting with an advisor each time you’ve got several questions. All that may ultimately prolong your dissertation phase and increase your tuition. Even if it doesn’t, it distracts you from the courses you could take or audit as well as other activities you are spending your tuition on. Think how much tuition you would save by hiring a statistical consultant for a substantially smaller fee.



HOW I CAN HELP YOU

I am able to consult you in every statistical aspect of your dissertation from the beginning to the end.

1] I will look at your data, objectives and time-frame and formulate the statistical research questions.

2] I will coach you in going back and forth between the committee members.

3] Most importantly, I will develop a statistical model based on your qualitative insights, choose the most accurate estimation procedure and make sure that all the important diagnostics is performed after the estimation.

4] I will help you get the modeling approach approved by the committee. I will help you incorporate any their suggestions into the analysis.

5] I will teach you everything about the techniques we've employed. Moreover, I will teach you everything about the 2-3 competing techniques. Some of the most frequent questions at the defense, preliminary talks or seminars are "What do you get if you try this alternative approach?", "What rationale did you have to forego method A and concentrate on method B?", "How sensitive are your findings to the modeling assumptions?", "Are there outliers or high leverage points in the data set?" You should be prepared to answer these questions. Oftentimes, I will train you to understand selected segments of statistics better than your advisors.

6] I will answer any statistical question and perform any follow-up analysis. I will make sure to be available to you on a short notice in the days preceding your defense. I have been through the process myself and understand all the stress you may be experiencing.


Please read the detailed description of the services offered in the area of statistical consulting: home page, experience, case studies and payment options. You may also find the following pages useful: statistics resources, finance resources, statistical software and statistics tutoring.