Overview and details of the sessions of this conference. Please select a date or room to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).
B4: New Challenges in Applied Research
Using mobile phone data for statistics
Centraal Bureau voor de Statistiek (CBS), Netherlands, The
Relevance & Research Question:
(Keywords: mobile phones, geospatial activity, movement, tessellation, signatures)
In this research we aim to gain insight on the geospatial activity of mobile phone users. We explore the possibility of using this data for statistics on day-time population density and the mobility within certain areas. Another aim of this research is exploring the relation between (the signature of) call activity and economic properties of an area. A derived research question is deducing a method to obtain a tessellation of cell serving areas from a cell plan and combining different tessellations.
Methods & Data:
(Keywords: call-events, location, cell plan tessellation)
For our research we obtained a dataset from a telecommunication company containing records of all call-events on their network in the Netherlands for a time period of two weeks. Each record contains information about the time and serving antenna of a call-event and an identification key of the phone. The dataset contains around 600 million records which introduced difficulties in the (pre)processing.
Also a cell plan containing information on the geo-locations of 20.000 antennas was provided. We devised a method to transform this cell plan into an appropriate tessellation needed for geospatial analysis. We experimented with different data normalisations to detect deviating day patterns.
(Keywords: animation, population density, Voronoi, clustering)
We produced a geospatial animation from which it is clearly visible that high call intensity coincides with high population density. We used the Voronoi algorithm to create tessellations from the cell plans and applied rastering methods to combine them. We were able to locate anomalies by deviations from the “normal” call activities. By analysing the call activity signatures of the cells through time we found indications for possible clustering relating, for instance, to the economic properties of that area. From the dataset and research we obtained insights in the movement of mobile phones.
(Keywords: whereabouts, movement, mobile phones, clustering, classification)
With this dataset we were able to obtain detailed information on call density and movement of mobile phones. Clustering areas based on the signature of call-activity is a novel way to classify areas economically.
Improved cost-effectiveness in mobile surveys using HLR-Lookup
GESIS - Leibniz Institute for the Social Sciences, Germany
Relevance & Research Question: The growing mobile-only population is a concern for telephone survey research as a possible cause of non-coverage error. Though the dual sample frame approach includes mobile numbers (e.g., the probability sampling method by Gabler and Häder), that option raises survey expenses. Another problem may arise when mobile phone numbers and landline numbers are treated equally in response rate calculation. Due to the difficulties in identifying the non-existent numbers by means of standard interviewer contact, proportions of non-existent numbers may be biased causing inaccurate estimation of the response rates. One of the reasons is that German mobile operators differ in reporting the status of a number being reached. The study illustrates an application of using a service available in the GSM-network (HLR-Lookup), which may allow a precise classification of numbers as non-eligible instead of numbers with unknown eligibility along with reducing the survey costs.
Methods & Data: 30000 mobile phone numbers were fielded as part of a telephone survey. The outcome of the contact attempts such as successful contacts, non-contacts, as well as reasons for non-contacts are compared with the results of the HLR-Lookup service which allows a real-time network query of a mobile subscriber’s status (active, absent, unknown, roaming, etc.).
Results: Preliminary results of the numbers‘ check allow to sort out about 43,5% of numbers as being non-eligible. Further analysis will classify and match the outcomes of the survey and the HLR-lookup into eligible, non-eligible and of unknown eligibility more precisely. Most importantly, the results will show the accuracy and efficiency of the HLR-lookup compared to a telephone survey which made call-attempts to all numbers.
Added Value: HLR-lookup promises substantial lower costs for samples with non-valid numbers. Especially in cases when interviewers have to make all call attempts manually and an automated dialer is unavailable due to legal or technical reasons. For survey quality the technology could help to achieve higher accuracy in response rate calculation.
Combining Quantitative and Qualitative Approaches
intanges interviewtechnik gunnar harde, Germany
Relevance & Research Question:
Classical questionnaires support a systematic questioning process and can be efficiently and automatically analyzed, while qualitative surveys allow an exploration of individual insights. This abstract outlines the Intanges Method, which combines the strengths of both approaches: in contrast to qualitative surveys, the researcher interrogates up to several thousand participants and receives the results of multivariate statistics. At the same time, in contrast to quantitative questionnaires, the researcher gains an insight into the participants’ individual rating criteria.
Methods & Data:
The Intanges Method is based on the Repertory-Grid-Technique. The researcher selects the elements (e.g., brands or products) which constitute the survey subject. Via an online-interview-platform, the participant decides if two of these elements are rather similar or different. If the elements are similar, the participant defines the similarity and its opposite; otherwise, she defines the difference. Then she rates the remaining elements due to this construct. The participant performs this process several times. Finally, she has built up her own multi-dimensional space, in which she has positioned each element. In this way the interview platform collects constructs and element positions of every participant.
The Intanges-Analysis-Software generates multi-grid-plots; i.e., graphics which show the relevant factors and their associated attributes stated by the participants. Multi-grid-plots display the condensed psychological cosmos of all participants. Tag clouds for each element are also generated, containing the important and significant attributes associated with the element. Furthermore, the participants are clustered according to their ratings. In combination with additional socio-demographic data, the researcher can perform socio-demographic segmentation. The researcher examines these results with the intanges software iPlotX.
Instead of making assumptions about the mindset of the participants, the researcher exploits the mindset of each participant individually without bias. Qualitative statements are generated, which enable the researcher to perform a lively presentation of the results. Completed cases have shown that this hybrid approach of multivariate analysis and participant-defined constructs supplies both, statistically significant results and the psychological and linguistic framework of the participants.