4. COMPARATIVE EVALUATION OF EACH SOFTWARE PACKAGE
4.10 Cost effectiveness
All power comes at a cost. However, the question is whether that cost is appropriate for the software services provided (i.e. whether the CAPI costs lead to value in data collection).
The CAPI products under consideration roughly follow three pricing models: software as services, where purchase of the product grants the user unlimited technical assistance; software as software, where the package costs a per-unit fee and assistance is extra; and software as freeware, where the package is free to all users.
The CAPI products include several software products, as well as some service ones. Cost effectiveness must therefore consider the price of each component cost and what it means for survey operations. Each CAPI package considered here includes two to three programs that are typically priced separately. The first piece, responsible for creating and modifying a CAPI program, is the designer software. This is typically purchased in relatively few units for a small team of CAPI programmers. This can typically be considered as a fixed cost of CAPI survey operations.
The second piece, responsible for running the CAPI program in the field, is the implementer package. This is purchased for each field interviewer, and can thus be considered a marginal cost. Projects of larger scale will typically require more licenses.
The third piece, responsible for moving data from the field to the survey headquarters, is the server software. Like the designer license, this can be considered a fixed cost of CAPI survey implementation. For some software, this is a separately priced software product; for others, this is integrally integrated into the core CAPI program.
44 Alongside software is technical support. There are a few different modes of support on offer. One is one-off technical training (e.g. classes on how to program) that is incurred as a set-up cost of the CAPI implementation. Another is on-going technical assistance that can either be a fixed cost (e.g. as-needed technical assistance that acts as an insurance policy against catastrophic CAPI failure) or a flexible cost that increases with the scale and complexity of the CAPI survey (e.g. a technical adviser that assists with design and management).
What is rated here is how well the costs for these items translate into value for the CAPI implementer. Value: Although the final calculus of cost effectiveness depends on the scale of deployment and the technical competence of the implementing agency, a few cost trends merit emphasis. First, software as service (i.e. software provision with unlimited remote technical assistance) may seem expensive but could pay dividends for organizations with lower technical capacity.
For example, Entryware has high unit costs for designer and implementer programs, but paying these fees entitles users to unlimited remote technical assistance via phone and e-mail. The rate seems high until one factors in the price of technical assistance. However, this type of pricing model may be too costly for organizations with higher technical capacity. These organizations could benefit from more modestly priced software whose technical assistance is priced separately and only paid for on an as- needed basis.
CASES and Pendragon also largely follow this model. Both provide software but also offer unlimited remote technical assistance as part of their product. This support must implicitly be factored into the product price.
Second, software as software (i.e. software provision not inherently inclusive of any technical assistance) can be an inexpensive pricing model as long as users do not require technical assistance. Technically competent organizations may benefit, all other things being equal, from software whose price is divorced from any substantial technical support. They can simply pay for the price of the
software and rely on internal human resources for absorbing and implementing the technical content of CAPI programs. Less technically competent organizations, on the other hand, may find this pricing model expensive. The software packages that offer technical support as a separate item typically charge dearly for that support.
Blaise, CSProX, and Surveybe adopt this model. Software is priced and sold separately from support services. End users may use software without paying for support, but will have to pay for support if it is needed.
Third, free software, though appealing, often has hidden costs. MMIC and ODK illustrate this point well. MMIC is a free software package, but using it effectively would likely require training and costly
technical assistance. These are services that RAND provides for a fee alongside its free software. ODK is also free, but programming with it could require considerable learning costs, as ODK does not offer much in the way of technical assistance. Therefore ODK adopters would likely need to hire an outside software firm, find competent programming resources, or task staff with the arduous task of learning all of the components of programming an ODK questionnaire (i.e. XML, JavaRosa, etc.). Free, then, is not always entirely free.
45 While cost effectiveness may be an organization-specific concept, there still seem to be a few CAPI packages considered here that offer an interesting value proposition. In the software as software group, two packages offer real value. Blaise offers competitive fixed costs for developer software and
reasonable variable costs for interviewer software. Surveybe may provide an attractive pricing model for some – where fees match usage, as measured by data points collected, rather than the number of machines – but its appropriateness is highly dependent on scale. Perhaps a deal for small surveys, Surveybe can become increasingly expensive for either large sample sizes or large questionnaires. In the software as freeware group, both MMIC and ODK could be strong value propositions, but on the condition that users have a high level of programming expertise, particularly with web programming languages. MMIC may require some up-front training costs in order to get users acclimated. ODK is much more approachable, but also more limited in functionality. The value proposition is also large if organizations are willing and able to incur relatively higher adoption costs, whether through dedicated staff time or hiring external organizations to train. Once the fixed cost is incurred, these free software packages have no marginal cost.
In determining the best priced package, organizations should consider the features that are strictly necessary for its survey deployments, the level of existing internal programming expertise, and the degree of technical assistance required to get a CAPI system operational, given the quality of a software’s documentation and the organization’s technical absorptive capacity.
Additionally, it is worth mentioning that most of the featured CAPI software providers also offer bulk pricing. The pricing and cost effectiveness analysis here relies on publicly available pricing information that is more relevant for smaller scale surveys. Those interested in larger scale implementations should contact CAPI software providers directly to inquire about larger purchase agreements.