Doctoral thesis

6.3Population and Sample

While scholars have linked high response rates to accuracy (Olson 2010), other researchers have found that lower response rates do not significantly affect the accuracy of results (Curtin et al 2000). Robert Gray (2012) writes "It is generally understood that, with reasonably rigorous sampling procedures, distributions are reflective of the attitudes held by the population at large". However, based on the initial estimates from existing research and pilot interviews, it was questionable whether the methodology could generate data sets large enough to facilitate valid statistical inference to the total population of participants in either country. In addition, the equal proportionality of respondents generated from email lists was not guaranteed, due to lack of direct control over the entire process. In view of the possibility of low sample size and uneven distribution of the sample, the use of inferential statistical methods did not appear to be attainable. Processes to overcome these limitations could be deployed, such as weighting the data, but in consultation with advisors at the Faculty of Social Studies of Masaryk University, this option was not recommended. Therefore, the project adopted a non-probability logic of sampling, whereby the results represent only the respondents of the surveys, and not statistically inferred to represent any larger populations.

In Austria, the 340 valid responses out of a total estimated population of 2,600 participants produced an estimated 13% sample (see table 5.1).

Table 5. Estimate of Participants and Survey Respondents in Austria

The response of participants in Austria was somewhat less than expected. The survey statistics (table 5.2) reveal a dropout rate of 41%, which for a target population who are known to be active participants, suggests some issues within the survey may have influenced the number of valid responses.

Table 5. Survey Statistics (Austria)

Survey Statistics (Austria)
Completed	289
Partially completed	51
Total Valid	340	59%
Completely empty	4
Partially empty	7
Entered first page	37
Entered introduction	218
Total Invalid	266	41%
Total Surveyed	609

One issue contributing to the response could be the duplicative nature of the estimates of participants in Salzburg and Linz, where the community televisions grew out of the radio organizations, creating non-distinct populations of participants. This overlap may have caused an over-estimation of the total participants, thereby causing a commensurate reduction of the response rate estimate. Additionally, in the case of Radio Salzkammergut, a quite active and successful radio, post-study correspondence indicated some organizational communication errors in the email distribution process, causing a zero response.

Beyond the estimates of total population and operationalization issues, the low response rate in Austria may also have been related to language factors. Multilingualism is an accepted reality in Austrian community broadcasting, as research projects have confirmed its presence (Purkarthofer et al 2008). In addition, pilot interviews with stakeholders and regulators also revealed strong views on the importance of foreign languages for community broadcasting, and advocated for the use of multiple languages in the research (Moser 2013, Grinschgl and König 2014). Thus, the Austrian survey was offered in Turkish and Serbo-Croatian languages, resulting in a 4% rate of respondents selecting a non-German language option.

In the Czech Republic, the target list of online non-commercial alternative broadcasters currently operating around the country was compiled from the author’s connections and experience as a practitioner, advocate, and researcher. The same processes and methodology used in Austria were subsequently deployed in the Czech Republic, facilitated by Czech stakeholders. That included email requests and survey links sent to the broadcasting organizations shown in table 5.3.

Table 5. Estimate of Participants in Czech Republic

The response rate for the Czech survey was higher than Austria both among the organizations, and also in the survey performance statistics (table 5.4). While the surveys deployed in both countries were identical (aside from language translations), perhaps the language issue was less of a factor in the Czech case, resulting in a slightly lower dropout rate.