Month: April 2014

A

Assumptions of Multiple Regression

The mathematics behind regression makes certain assumptions and these assumptions must be met satisfactorily before it is possible to draw any conclusions about the population based upon the sample used for the regression. That is, the assumptions must be met in order to generate unbiased estimates of the coefficients such that on average, the coefficients derived ...

S

Screening may greatly improve survival rate but do absolutely nothing for mortality

Given a group of patients diagnosed with cancer at the same time, a 5-year survival rate would simply be calculated as the proportion of these patients still alive 5 years after diagnosis to the total number of patients diagnosed with cancer. Mortality rate is different. The mortality rate is the proportion of people within some ...

L

Left side of a rating scale is the preferred side when questions framed positively

It never ceases to amaze me to what extent seemingly small changes in a questionnaire can impact responses. Imagine you are a college student and asked to rate your agreement to the following question: My college has an excellent reputation Now rate your agreement to this question: My college has an excellent reputation The only difference ...

O

Order effects on questionnaires

A considerable body of evidence has shown that the order in which questions are asked can influence responses. For example, if a survey has a mix of open and close-ended questions, placing the closed-ended questions first can bias responses to subsequent open-ended questions toward those mentioned in the close-ended questions. More than 50 years ago researchers Herbert ...

T

The Pill Scare of 1995

In October 1995, the UK Committee of Safety of Medicine (CSM) issued a warning stating that third generation oral contraceptives doubled the risk of potentially fatal venous thromboembolism (VTE)  in the legs or lungs and recommended a switch to older drugs except where possible. The statement read, “It is well known that the pill may rarely produce thrombosis ...

E

Error bars: Expressing uncertainty

Statistics is a tool that can be used for 2 purposes: 1) to describe a sample and 2) to infer things about a population based on a sample. Error bars presented on graphs can be used for both purposes. Error bars for description: Say we want to assess how well a teacher is performing based ...