Considering the Socio-Technical Model (STM) as a basis for ongoing research has revealed unexpected theoretical results. The research in question is an examination of the impact of technology on user trust and subsequent intent to use election systems. In the process of conducting this research, a new survey instrument was created based on a newly created theoretical foundation built on STM. Our interim findings have exposed an unexpected theoretical finding that suggests further consideration. The ifndings were a result of data validation done on the new survey instrument. These findings indicated that the principle components of STM may not be valid when tested. At the least, the components may be relevant in a non-equal way. Another possibility is that the model should be re-examined to determine if the components accurately reflect the nature of people and technology and their interactions with large scale systems.
Research is currently underway to investigate the American electorate’s perceptions of security
in election systems and resultant intention to participate in elections [
To facilitate the research, a theoretical framework was developed to provide a framework
and basis for the creation of a survey instrument. As the Socio-Technical Model (STM) deals
with the interaction between technology and the users of the technology [
The first component of STM, structure, describes the social structures that support the system.
In the context of US Presidential Elections, the structure of the system is complex. Presidential
elections are a federal position but the election itself is managed at the state-level. Furthermore,
within the state, there are separate localities that run the actual election. Therefore, this provides
for the potential of weak policies and procedures for administering the election [
The second component of STM, people, describes how individuals interact with the system under study. For election systems, people play a critical role as humans interact with all aspects of the voting system from the campaigns to counting the ballots. Of most importance, it is the individual conducting the primary transaction: voting. If a voter does not trust the system, their intention to use is very likely to be afected. If they believe that the systems used to administer the elections are insecure, they are unlikely to use the system.
The third component of the socio-technical model is technology [
The final component of the socio-technical model is tasks. Tasks deal with the interactions of the humans within the system. In the context of elections, this includes campaign ads, casting ballots, and ballot counting. These are just a few of the tasks that occur during the election and have critical efects on the overall election and the election results. If the processes involved in these tasks are tampered with in any way, it could potentially result in flawed results.
Given the basis of STM, a new survey instrument was created to facilitate the research into voter perceptions’ of trust in election systems. A theoretical framework was created whereby each component of STM was applied within the context of elections. Specific topic areas were created for each component that applied to election systems. The list below presents the results of this:
A survey utilizing the election system elements was created. After building the survey, we conducted a pilot study to validate the survey instrument. A pilot study was conducted using google forms which collected responses from eligible voters in the United States. This group generally consists of all adults over the age of 18. There are however exceptions to this in some states. For example, in 11 states felons lose their voting rights indefinitely for some crimes.
In the pilot study, we collected 205 responses. This number of responses gives us suficient
data to be able to perform the validation on the survey [
The purpose of this pilot study is to validate the survey instrument since it is a new survey
and has not been validated previously. Validation of the survey allows us to determine how
well the data we collected covers the areas and topics that we are researching in this study [
An analytical tool used for validating a new survey instrument is the Principal Component
Analysis (PCA) test. The PCA test facilitates the determination of the correlation structure
between the diferent variables [
The Kaiser-Meyer-Olkin (KMO) value determines how adequate the sample size was from
the data [
The next value within the PCA analysis would be the p-value. A low p-value shows that the results are replicable and that the efect is large or that the result is of major theoretical, clinical or practical importance. This value should be close to 0 and determined to be acceptable if less than 0.1. The p-value was calculated to be less than 0.001. Therefore, this demonstrates that the likelihood that the data we collected from the pilot study was less than a 0.001% chance that it happened by chance.
After reviewing the results of the PCA analysis, it was determined that there are 2 main components within the survey instrument questions. The first main component found in the questions is the structure of elections and the technology used in elections. The other primary components in the questions are people and tasks. These components line up with the STM component but appear to be combined into a smaller number of components.
The PCA analysis revealed that some of the questions that we grouped together should be grouped diferently based on the interpretation of the questions by the voters. For the most part, the PCA analysis showed a general theme between the two components, with only a few questions that did not seem to fit the general theme of the component.
As demonstrated in Table 2 above, most of the questions from the ST component, which is component 1, were strongly correlated with that component. However, some of them correlated with both components using the threshold of 0.3 and -0.3, the question ST1 was strongly correlated with both components. Most of the first component questions had negative loading which indicates that it is a negative correlation. ST7 from the first component was not a strong correlation with either of the components as the highest loading was in the first component as -0.148.
Within the second component, most of the item loadings were strong as most of them were higher than 0.4, however questions PT2 and PT6 were both lower than 0.3, which indicates that it was not as strong of a correlation with the component as well as it did not show a correlation with component 1 as well.
Even though the PCA did not group the questions together as we had them grouped, they still demonstrated a general theme in the questions and some of the STM components merged together. The reasoning behind this could have been due to how respondents interpreted the questions as well as some of the questions do have some crossover between the four STM components.
While the intent of the study was to examine the impact of technology on user trust and subsequent intent to use election systems, our interim findings have exposed an unexpected theoretical finding that suggests further consideration. STM has long informed scholars on the interaction between people and technology and is well established in the literature. Despite this well established application, our process of validating the STM based instrument did not support the validity of the theory itself.
The question loadings within the PCA test did not load well into the four core constructs of STM. This validation testing shows that structure and technology have an outsized impact on the system while people and tasks are much more minimally significant. Feedback from subject matter experts in quantitative data validation discussed the possibility that the way the questions were asked may have impacted the question loading. The survey is in the process of being adjusted to account for this possibility.
If the subsequent validation testing shows similar results, it opens the possibility that the theoretical foundation of STM should be revisited. The implication that the components of STM are equally impactful is fairly unlikely and should be investigated further. Perhaps refinement of the components themselves is called for as well.