PolyU

The Statistical Advisory Unit (StatAU) in the Department of Applied Mathematics runs both internal and external statistical consultancy and advisory services. Staff in the Department possess the experience and expertise to provide assistance and guidance on statistical matters, including:

• survey methods and sample design
• questionnaire design
• opinion poll
• quality control
• reliability
• survival analysis
• non-parametric inference
• stochastic scheduling
• simulation
• model fitting and data analysis
• meta-analysis
• medical statistics
• experimental design
• mortality analysis
• forecasting by neural network
• financial time series analysis
• time series with changing conditional variance (ARCH and GARCH)
• econometric modelling

Some of the past and existing consultancy projects:

• Household energy end use consumption estimation
• Energy efficiency factorization
• Energy consumption forecast modelling
• Transport energy consumption study
• Thematic Household Survey
• Coverage analysis of satellite broadcasting via Direct-To-Home (DTH) System
• Housing recurrent survey
• Survey on porpensity to use private cars to cross the Boundary
• FEHD Public Toilet customer satisfaction survey
• Public transport services and facilities study at boundary control points

We provide free advisory service to full time staff and research students of PolyU.

Enquiries in relation to statistical consultative and advisory services should be addressed to Dr. He Daihai, co-ordinator of StatAU in the Department of Applied Mathematics, or simply fill in a form for request of service (MS Word, PDF file) and send it to the Unit either by fax or email.

• Room TU804, Yip Kit Chuen Building, The Hong Kong Polytechnic University
• 2766 7864
• 2362 9045
• statau.ama@polyu.edu.hk

Enquiry

The importance of good quality data

Scientific research is used by academics in a wide scope of academic disciplines such as social sciences, public health, biostatistics, education, social work, public administration, and business administration; and by practitioners engaged in marketing, commerce, and industry. Data are the basis for all scientific research. Collecting good quality data plays a vital role in supplying objective information for the problems under study so that some analytical understanding of the problems and hence solutions can be obtained. Making decision on the basis of poor quality data is risky and may lead to disastrous results, as the situation may be distorted and hence all subsequent analyses and decision making will rest on a shaky ground.

Statistical sampling techniques

Sampling plays a vital role in data collection. Drawing a representative sample saves time, money and efforts in research and helps achieve desirable degree of reliability on collected data. There are many types of sample design and choosing the appropriate design can improve the accuracy and reliability of sample statistics while keeping the cost of survey low.

Designing a good questionnaires

A lot of expertise is required in designing a good questionnaire. Factual questions requiring memory may produce answers with big memory errors. Opinion questions improperly asked may lead to biased results. Questions with unspecific, complex, ambiguous wordings may produce poor quality data. Question order also may affect the answers obtained, especially when one is concerned with opinions that are unstable.

Conducting surveys

To conduct a survey, the aims must be clearly stated. The accuracy required, and the cost, time and manpower available need to be estimated. The population to be studied should be clearly defined. Whether personal interview, telephone interview or mailed questionnaire is to be used needs to be considered. A good questionnaire is to be designed. The likely sources of error of the survey data and the precautions that can be taken to minimize them should be considered. A pilot test should be performed so that the survey procedures can be finalized. Before the fieldwork is conducted, training may be given to interviewers to guarantee their quality. The method of field supervision has to be decided. Methods for processing and analyzing the survey data should also be considered.