Chapter 5 Notes from The Human side of statistical consultancy
Two models
- Problem solving
- Go away
- Solve
- Present
- Client achieving learning
- Work with client to clarify goals
- Develop proposals consistent with client situation
5.1 Minimum Required Knowledge (their thoughts not mine)
- Regression Analysis
- Basic Statistical Models
- ANOVA
- Plots and summarisation
Design of experiments (power law etc)
- Variance- erimation
- Bio stats (bioassay)
- Categorical data analysis
- Quilaty controil & acceptance sample
- Non-linera estimation
- Non-parametic metjhos
- Ranked & paired comparisons
Multvariate analusis
- Probability models
- Simulation
- Reliability & life data danalysis
- Numerical analysis
- Time series analusis
Survey exampling
5.2 Expections
Clients will expect that: * You’re knowldegable * You have contacts for allied services * Good at managing time & money * Loyalty * Comparssionate and human * Enthusiastic and interested
5.3 The effective industrial statistician
- Trained in theory & practice of statistics
- An effective problem solver
- Good communication skills (oral & written)
- Can work within contraints of the real world
- Good programming skills
- Is familiar with the statistical literature (current)
- Understands the realities of statistical practice
- Pleasing personality - works well with others
- Gets involved with the solution of company problems
- Can extend and develop statstical methodology
- Can adapt quickly to new problems & challenges
Produces high quality qork in a timely fashion.
- The more knowdlege the better
- Application not just theory
Attiude & business
5.4 Chapter 5 - Choosing a statistical technique
- Get clients to clarify objectives
- What do you hope to answer by running that analysis
- How will you know if you’ve answered your question
- It’s one this to know how to do a t-test, another to know when do do it and what it means
- When you have a choice, use the methods you understand well
- Don’t offer colutions that are beyond the comprehension of the client (or his ability to describe them)
- The simple graph is a major consulting virtue (Sometimes all you need!)
- Simple techniques are easier to give good time & money estimates
- Confirm who is doing the analysis - client or consultant
- How much analysis does the client need to understand - and at what level
- Use techniques already accepted in the (client’s) field
- Choose the simplest technique that will do the job, not the fanciest technique you can handle
- If the client’s solution is decent, don’t push yours unless necessary
- Help with the data collection and questioning
- Push for client understanding
- Agree work division before deciding a technique
Writing * Write sooner than you think * Avoid technical jargon * Keep tone friendly & helpful * Short, simple and concise
Remember the target if your client’s audience. What will persuade them. Get to know the subcultural - which arguments are the most convincing?
5.5 Chater 7 Business aspects of consulting
- What tasks need to be done
- Who will do them
- Who gets the credit
- Project time
- Project funding
- How will you get paid, and when?
- Who decides when the work is done?
- Who does the quality control of variables?
- What sort of consultant-client relationship
- Determine any other involved parties and roles
- Who signs off on the final report, who writes it?
- How will my analysis be presented - do I have any control over this?
- Will I prepare a written report of my activities?
- Stakeholder mapping?
Check in with your feelings before writing this to ensure correct tone (friendly & helpful)
- Be kind, firm, fair and empathetic
- Time estimates
- Break into chunks
- Review data before the final time estimate
- It doesn’t matter how good it is if it isn;t on time
- Listen to everyone who asks for a consult, but triage
Are you talking to the real client or a messenger?
- Discuss time & money early on
- Communicate any delays or unexpected costs
- You’re responsible for time/money of services you employ
Allocate responsbilities without coming across as legalistic