Category Archives: Biostatistics

We need more biostatistician

I feel like Malaysia is lacking biostatistician. We do have a lot qualified biostatistician but not enough practicing biostatistician. In Malaysia, in most universities especially UKM, UM & USM, biostatistics is taught together with epidemiology. Eventhough there are similarities between epid & biostat but they are different. I am not practicing epid that much now so my knowledge on epid is disappearing. But talking about biostatistician, for medical & health research in Malaysia, we need to produce more and more biostatistician.

The issue now is whether we produce non-doctor biostatistician or doctor biostatistician? For some, they said it is a waste for medical doctor to do anything else other than medicine. Well….then Dr Mahathir is the example of major waste in the country. So do few of other leaders, oppositions alike.

I feel that medical doctor should take up biostatistic with epidemiology as elective (rather than core subjects) if they want to be a biostatistician or medical statistics. We know the lingo. We can easily understand what clients want. For non-doctors, they need to practice longer time to understand medical & health terms.

Secondly, being statistician, even though we need to learn mathematics & know the basics of everything but when talking to clients, we should speak their ‘language’ i.e. medical & health language and not statistical jargons. Just hide certain details like covariance, estimate, likelihood etc. Convert these terms to miningful meaningful terms for the clients.

Thirdly, educate the clients. Tell them when to use standard deviation and when to use standard error for example. But tell them slowly & must provide example.

Lastly, don’t print all the statistical output to them. Select relevant tables & graphs only. Guide them on how to report those findings.

I want to change to SAS!

I plan to switch from using SPSS to SAS & Stata because SPSS is too expensive & because many people are using it, it becomes the standard statistical software at least in medical & health data analysis in Malaysia. When demand is too high, the price will go up. Even though I teach SPSS, I think it is high time for me to change to SAS to give alternative to SPSS.

SPSS BI Malaysia…if you read this…start lowering your price…..offer us unlimited license for complete module please. Please don’t use that hardkey to get the license. It is too tedious.

SAS…..if you read this, please contact me. I want to conduct a workshop using medical & health data. You need to sponsor the participants tho’…. :)

Is there a free software? Yes! It is R. Is it easy? No…not at all. :(

Update: Please read this :D

Linear Mixed Model

The more I read statistics, the more I don’t know!

During my time as a medical student (in early 90′s, t-test & chi-square are already the most sophisticated statistical tests on earth. People don’t even bother to run One-way ANOVA to compare variances of 3 or more means. Then I learn about regression. My MPH thesis was using survival analysis. Yup, those with cox-regression. That time I already know about GLM (General Linear Model), which is a group of procedures that utilise t-test, ANOVA, regression and the like) where observations are independent and fixed. What if the dependent variable still continuous but the there are correlation between independent variables? That’s when Linear Mixed Model (LMM) come into the picture.

The most important thing to know is the hierarchical model. What is hierarchical effect? If you understand linear regression, it will be easy for you to understand hierarchical modelling or multi-level analysis. Linear regression analyses effects at one single level. What is level?

To be continue….