Transcript Joint UNCTAD – ITU – UNESCAP Regional Workshop on
Joint UNCTAD – ITU – UNESCAP Regional Workshop on Information Society Measurements in Asia-Pacific
Bangkok, 26-28 July 2006 ESCAP, United Nations
Koay Hock Eng
Malaysian Communications and Multimedia Commission
“No, son, MCMC does not only stand for Markov chain Monte Carlo”
Greetings from the MCMC
1 MCMC in a nutshell
•Regulator for communications and multimedia in Malaysia. •Malaysian Communications and Multimedia Commission Act 1998 (MCMCA 1998) set up the Commission. •Communications and Multimedia Act 1998 (CMA 1998) spells out the regulatory and licensing framework on a convergence platform. •Obligated to monitor industry performance and report to the Minister of Energy, Water and Communications at the end of each financial year. •Timely and relevant data was necessary to fulfill the said reporting role.
• http://www.mcmc.gov.my
2 The Statistical & Knowledge Resource Department of the MCMC
•In charge of statistical activities from collection to dissemination. •It is staffed by three statisticians and headed by a manager •It looks into 3 areas namely Primary data, Secondary data and Methodology.
Primary Data •Modest primary data capability in the form of a 20 seat CATI Centre powered by Nipo software ( http://www.niposoftware.com
). •Web survey software by Perseus ( http://www.perseus-asia.com
) •Mail surveys •The main surveys: Annual Household Use of the Internet survey, Annual Hand Phone Users Survey and the annual Licensees Survey. •Planned surveys: Business use of ICT, e-readiness and community access. •Dissemination of primary data collected in surveys is done through a series called
Statistical Briefs
(
ISSN:1823-2523
)
Secondary data •Administrative records of licensees. •Monthly basis •Published on a quarterly basis according to an advance release calendar. •
Communications and Multimedia; Selected Facts and Figures (ISSN: 1675 6223)
•Record Keeping Rules under section 268 of the CMA 1998
Communications and Multimedia; Selected Facts and Figures (ISSN: 1675-6223)
Methodology •Mindful about statistical methodologies used •Has developed standards appropriately. •Surveys conducted or the market research bought must be on a scientific basis and that documentation of those surveys must be done according to some accepted standard. •It also embraces the Fundamental Principles of Official Statistics.
3 Quality, standards and capacity building
Statisticians •Dr T.J. Rao, visiting professor of the Indian Statistical Institute in Kolkata to train its statisticians in survey sampling. •Skills diet includes scripting in relation to the CATI solution that its CATI centre supports as well as programming skills required to create and publish web questionnaires. •The statisticians at MCMC also trained in statistical forecasting to meet increasing demands for projections.
He was not showing us how to break the da Vinci code
Interviewers •Handpicked.
•Most of them have at least the SPM (O – levels) but mature candidates without the SPM are also given an opportunity.
•The candidate must first come across well on the phone. •Required to enquire by phone.
•An initial assessment is made: pleasant voice, good diction, self-confidence and politeness. •Interviewers at MCMC’s CATI Centre are properly trained: organisation, subject matter, questionnaire and telephony skills.
The organisation Functions of the MCMC so that they may be able to answer simple queries from respondents wishing to know more about the organisation. More difficult queries are referred to their supervisors who are statisticians with the MCMC.
The subject matter For example in the case of a survey on hand phone users, interviewers will be given a roundup of the industry including some basic facts and figures.
The questionnaire Question by question through the questionnaire, terms used are explained zero knowledge assumed in all surveys, even a hand phone users survey although practically all of our interviewers are hand phone users themselves; Skips and branches in the questionnaire and the logic behind questions in particular those that need careful probing and those that needed cross checking with answers given to earlier questions.
Telephony skills •In-house training manuals for telephone interviewing skills. •Mock runs •The performance of the interviewers are reviewed. Those that do not meet MCMC’s stringent standards are dropped while those who do are invited to become part time interviewers. •Piece rates •Database of all good interviewers and draws from this pool when it launches a survey.
4 Harmonization of survey questions
•MCMC's surveys took specific data needs of the regulator and other local users into consideration. •No attempt to harmonise totally with other surveys although some of these questions would also coincide •In 2005, the MCMC joined the Asia Pacific Internet Research Alliance (APIRA) and as such will fulfill its obligations to incorporate questions agreed upon in its 2006 Household Use of the Internet Survey. • It will also take this opportunity to incorporate questions agreed upon by the Partnership on Measuring ICT for Development.
5 MCMC approach to indicators
•Indicators in household ICT normally relate a certain measurement to a hundred inhabitants or a hundred households. •An example would be the Internet penetration rate by gender. When viewed in this context then the logical thing to do to derive these indicators would be to do a household survey and make the necessary estimates from that survey. •The CATI surveys conducted cannot do that as a true household survey by CATI would require a very high fixed line penetration rate in the households.
•MCMC surveys collect descriptive statistics about its target population viz households with Internet access or individuals with hand phones. •The MCMC sees the derivation of indicators as a separate exercise which brings together 2 pieces of secondary data; one, from its surveys and another, demographic estimates from the NSO. •To illustrate, the number of household Internet users of the feminine gender can be estimated from the Household Use of the Internet survey and current estimates of the total number of females in the population of the country can be obtained from the NSO. •Dividing the first by the second will yield an estimate of the Internet penetration rate for females.
6 Users of statistics
Domestic users include: MCMC Industry players Parliament Central Government and Central Planning Agencies Ministry of Energy, Water and Communications Ministry of Science, Technology and Innovation Ministry of Finance Department of Statistics, Malaysia Economic Planning Unit, Prime Minister’s Department Bank Negara Malaysia (Central Bank of Malaysia) Malaysian Industry Development Authority Multimedia Development Corporation
6 Users of statistics
State and local authorities State Economic Planning Units State Resource Centres Consultants Local press Researchers and students International users ITU APT Foreign trade missions Foreign research agencies
7 Lessons learned and the challenges ahead
Data collection •Response rates to MCMC CATI surveys have consistently been in excess of 70 %. •A challenge faced at least in terms of household Internet usage statistics, is that as other Internet access modes grow in prominence, the users that the MCMC can reach through its CATI centre as a proportion of the target population will diminish. As such, findings will become less useful since its scope will be limited to access through dialup and ADSL. At the same time, the MCMC is also mindful that ICT statistics does not start and end with household Internet usage.
•Community access is gaining attention and will continue to gain attention and this is the next challenge that the MCMC will have to come to grips with. It has been envisaged that a team of interviewers to make observations on the ground or to canvass data on a face to face basis may be necessary.
Dissemination •Efforts need to be heightened in dissemination to make users more informed about the nature of statistics collected in MCMC CATI surveys. •There has been misconception that the MCMC CATI surveys are household surveys even though the MCMC has always tried to distinguish its surveys from household surveys.
•As mentioned elsewhere in this paper, a household survey by CATI would require a very high household telephone penetration rate. In the case of the MCMC CATI surveys, the target population is not the population of households but rather the subscriber base, be it hand phone, fixed line or Internet access. •The estimates from the MCMC surveys are then used in conjunction with demographic statistics to derive indicators in a separate exercise.
MCMC Household Use of the Internet Survey 2005
1 Background
•First ever purpose built survey of its kind conducted by the MCMC. •Objective was to address user side data gaps on core attributes and current trends in the use of Internet in Malaysian homes. •Proportions of the home subscriber base that fell into the various categorization schemes of the key variables are estimated •Proper use of these estimated proportions in conjunction with population estimates at reference date will facilitate the derivation of various Internet penetration rates.
2 Survey design
Reference Date 31 March 2005. To qualify for inclusion into the sample, a potential respondent must be able to answer “yes” to a screening question on whether he was a user at reference date. Target population The target population is the universe of Internet users regardless of age who access the Internet from a dialup or xDSL from their own homes at least once in the past month. This effectively covers 99% of home Internet users in Malaysia.
Users of wireless access were excluded from this survey.
Sampling scheme •This is not a household survey in the conventional sense and the survey was not founded on any household frame. •Two stages are identified in the selection of a respondent. •In the first stage, unique randomly generated PSTN numbers were dialled to screen out non-active and business numbers. When a household is reached, it is screened to see if it is Internet equipped (via dialup or xDSL) at reference date. •If it is, then a user from that household is selected at random. This is done by asking to speak to the Internet user in that household who has the next birthday.
Sample size •With a predetermined = 0.01 and d= 0.02 a sample of 4,925 Internet users in households were drawn to provide estimates at national level. •At the same time 2,000 non-user households were also interviewed as to why they were do not have Internet access at home.
•Sadly though, stratification was not possible since a “virtual” frame was used. It would have been possible if we had a detailed frame of Internet users in Malaysia.
Data collection •Hybrid CATI / web survey. •Trained interviewers call respondents from the MCMC CATI centre regardless of whether they have Internet access at home or not. Those contacted but had no Internet access at home at reference date fell into the “non-user” sample while those with Internet access at home at reference date fell into the “user” sample. •“Non-users” were interviewed immediately through a CATI module while “users” were required to log into a website to do a longer questionnaire.
•In both cases pains were taken to explain to respondents the purpose of the survey. •In order to garner a credible response rate to the “user” module, the Association of the Computer and Multimedia Industry of Malaysia (PIKOM), which had many times in the past collaborated with the MCMC, sponsored a contest for those who participated in the web survey. The grand prize was a PC. ADSL and dialup modems were also offered as consolation prizes. •However, considering that most users were on slow dialup access the MCMC developed a CATI version for “users” as well. “Users” would be asked if they prefer to go online or “do it now” in CATI. However for the contest they would still have to logon to the contest website.
•Fieldwork started 1 May 2005 and ended on 24 July 2005
The survey probed: Socio-economic Gender Marital status Usual state of residence Urban / rural distribution Occupational status Income Others The length of time in years that internet access had been available at home Type of access Mode of payment Intention to migrate to broadband and when, or otherwise with reasons (for those with dialup only)
Average use per week Usual time to log on to the Internet Purpose for use of the Internet Practice of consumer e-commerce Type of products or services purchased through the Internet and amount spent in the past 3 months Number of Internet users at home Number of PCs owned by members of household, including laptops and notebooks but excluding PDAs Purchase of PC in the immediate past year to replace an old/obsolete one age of PC that was replaced Concern about security on the Internet Who in the household pays for access Whether access to Internet is made elsewhere too and where
•Response rate to the survey as calculated according to AAPOR formula number 1 i.e. RR1 was 76.47 %. •It was noted that on any given day, response seems to be slower in the morning and early afternoon but pick up sharply after 5 pm.
Survey Main Findings
Main Findings
No gender divide !
50.2 % 49.8 % Malaysian population 50.9 % male, 49.1 % female
Main Findings
Age Distribution
Below 15 15 - 19 20 - 24 25 - 29 30 - 34 35 - 39 40 - 44 45 - 49 Above 50 6.5 % 18.6 % 42.3 % 17.2% 12.5 % 12.2% 9.9 % 9.6 % 5.1 % 8.4 % Single largest age group: 15 to 19 year olds 18.6 %
Schooling Status
36.5 % of all home users are students at all levels.
Among them Primary school Secondary school Tertiary institute 8% 46.8% 45.2%
None Primary Secondary Diplomas Degrees
Schooling Status
63.8 % of all home users are no longer studying. Their highest educational attainment are as follows 0.1 % 1.4 % 37.7% 25.4% 35.4%
Marital status
Unmarried Married Divorced / widowed 55.0% 44.3% 0.7%
Main Findings
Age of Internet account
New user 1 – 3 years 3 – 5 years More than 5 years 14.9% 30.9% 20.8% 33.3%
Intensity of Usage
On a weighted average basis, the Malaysian surfer tends to surf 9 hours per week.
In comparison •Korea 11.7 hours (2005) •Mainland China 14 hours (2005) •Singapore 10 hours (2002)
Time of Day and Main Use of the Internet
5 most popular activities
E-mail 73.7% Education / Research Finding information Chatting 46.8% 40.5% 25.9% Reading online newspapers 20.2%
70.2% usually logon at night
Buying on the net
Only 9.3% of home Internet users bought on the net in the last 3 months.
Main Findings
Expect slow adoption of Broadband
Of those not on broadband only 18.6% plan to migrate to broadband. Among these, 50.2% mentioned they will do so in 12 months time or later.
Main Findings
The reasons for not going broadband included: • perceived lack of need (56.5%) • lack of awareness (23.0%) • costs (14.1%)
Main Findings
Slow too for PCs*
56.4 % of non-PC owners with landlines have no intentions of buying a PC. Only 43.6 % of non-PC owners with landlines have intentions of buying a PC. In 3 months In 6 months In 9 months In 12 months > 12 months 5.73 % 5.96 % 2.41 % 18.81 % 67.09 %
Main Findings
New PCs for old?
•Among user households, as high as 62.8 percent has only one PC.
•Households typically keep their PCs for an average of 4.3 years before purchasing a replacement for whatever reason including obsolescence.
Main Findings
Security
74.2% home Internet user expressed varying degrees of concern about security on the Net.
25.8 % expressed no concern at all.
Main Findings
Access from other places
Home Internet users also access the Internet from other places : Work School/University Public (library, cybercafe, etc) Others
Percent
51.47
25.44
32.97
3.08
A comprehensive measure of Internet access will have to take this into account.
A Composite Profile of the Average Internet User from survey findings
• Young • Just as likely to be male or female • Mainly on dialup • Shares with at least another family member • Spends on the average, up to 9 hours a week on the Internet at home • Usually at night
A Composite Profile of the Average Internet User from survey findings
His main reason for going on the web is to • check for emails • some educational and research work • finding information about goods and services
A Composite Profile of the Average Internet User from survey findings
If at all he purchased anything online, it would be • air tickets and • an occasional book He is not a big spender and spends less than RM500.00 most of the time.
He is generally concerned about security on the Internet.
Conclusion
•A need for good and timely statistics and indicators has always been felt by the MCMC. This was the main motivating factor that prompted it to embark on a primary data capability in 2004. •Surveys have been successful in filling data gaps, much remains to be done in terms of reviewing and maintaining statistical standards and quality assurance. •The Statistical and Knowledge Resources Department of the MCMC has much to learn and benefit from harmonization of survey concepts and data items across jurisdictions. •Mutual exchanges could also be useful such as the visits by the telecommunications regulators from Botswana and Brunei to the MCMC CATI Centre in 2005.
For more…
MCMC Statistical Brief Number 2 …