Page 285 - Special Topic Session (STS) - Volume 2
P. 285
STS493 Sofie d.B. et al.
The three criteria refer to the survey topic properties that are prone to
measurement error and/or nonresponse. To provide examples: Examples that
satisfy the first criterion are topics that require keeping a diary for a specified
time period, say a week or a month, and provide details about all time periods.
Other examples are surveys that require consultation of administration and
archives, for example about assets, investments and finances. The second
criterion is satisfied, for instance, for travel surveys where respondents need
to provide exact coordinates of locations they have visited, for health surveys
where respondents need to describe sleeping patterns and STS and SPS
surveys where businesses need to provide detailed information about their
activities and output. The third criterion applies to complex socio-economic or
psychological concepts such as happiness, health or wealth, or research and
development where many questions are needed to derive latent constructs.
From the sensor point of view, the main criteria are:
Omnipresence: The sensor(s) are available to most population units.
i.e. as sensors in contemporary devices and wearables or as sensors in
IoT systems;
Data access: Data generated by the sensor(s), as well as metadata
about the properties and accuracy of the sensor data, can be accessed
and processed;
Quality: The sensor data is comparable, reproducible and accurate;
Costs: Any costs associated with the sensor(s) or implementation of
the sensor measurement process are affordable for an NSI;
The four criteria all link to the utility of the resulting sensor data. The
omnipresence criterion refers to the coverage. In theory, tailored instruments
can be developed that record complex phenomena and behaviours, but these
are until now used only in lab settings. Smartphone sensors are examples of
omnipresent sensors, whereas sensors in wearables have a much lower
population coverage and pose challenges with regard to data access. The data
access criterion means that sensor data can be stored, manipulated,
processed, evaluated and interpreted. For instance, location data can be
stored and processed, but it is not always clear what sensor, GSM, Wi-Fi or
GPS, produced the data and how accurate the data are. For businesses there
is the issues that so far, only few innovative companies have installed sensors
because of costs issues. In order to produce business statistics, a
representative number of companies should have sensors installed. The
quality criterion originates from the statistical objective to derive accuracy of
statistics and to be able to compare statistics between persons or companies
and in time. Sensor accuracy is the equivalent of survey accuracy and should
be evaluated on measurement and missing data properties. For instance,
location data can be used to estimate travel distances but are subject to
missing data, measurement errors, sensor data drift and potentially also device
274 | I S I W S C 2 0 1 9