#Quality Assurance during Fieldwork

This use case illustrates how the quality of collected data can be ensured during the on-going field phase.

#Example from the User’s Point of View

From the very beginning of the field phase onward, collected data must be checked for irregularities. The preliminary data can be uploaded into Inspirient on a regular basis to calculate a table of quality indicators, both for review but also for integration into existing quality control data streams. Key focus is on metrics regarding irregularities and fraud indicators. Specific indications of incorrectly collected data can be checked on a case-by-case basis, and resulting evaluations can be saved for documentation purposes. Afterwards, if necessary, potential irregularities can be brought up with the interviewers of the field unit.

#Approach

Key steps in implementing this use case are as follows:

1. Uploading data – The analysis starts, like in most cases, by uploading the preliminary dataset, for example as an SPSS, Excel or CSV file
2. Guidance / Prioritisation – Since all dimensions are equally relevant for quality assurance, selecting a uniform prioritisation is recommended to get an unbiased evaluation
3. Guidance / Annotations – For quality checks to be applied correctly, the correct use of relevant annotations should be checked, especially SURVEY_DURATION and SURVEY_INTERVIEWER_ID
4. – the user is notified by e-mail upon completion of the autonomous analysis –
5. Downloading reports – Relevant metrics can now be checked in the generated quality report in order to draw conclusions about irregularities

If a dataset is spread across multiple files, these files can be evaluated either separately or as a joint analysis.

#Data Validation after Completion of Fieldwork

This use case shows how data quality can be checked and documented after the completion of the field phase and how this may support one’s own evaluation.

#Example from the User’s Point of View

A final update to the current survey dataset has arrived, and it is now important to quickly check the data quality in order to identify and correct any remaining issues in the collected data before the actual evaluation. For this purpose, Inspirient analyses the data set and creates a report on data quality, from which problems such as inappropriate distribution of responses across demographic cohorts or straight-lining can be deduced. Prominent outliers can be checked directly in the user interface, providing a good overview of the data quality aspects to be considered.

#Approach

Key steps in implementing this use case are as follows:

1. Uploading data – The analysis starts with uploading the final data set
2. Guidance / Prioritisation – As before, a uniform selection of priorities is recommended for an unbiased view of data quality
   • Individual dimensions that are irrelevant to the evaluation can be ignored using the lowest priority setting in order to obtain more focused results
3. Guidance / Annotations – The correct use of relevant annotations should be briefly checked, especially for SURVEY_DURATION and SURVEY_INTERVIEWER_ID
4. – the user is notified by e-mail upon completion of the autonomous analysis –
5. Reviewing a story – The generated data quality story contains the results of the survey quality checks that were applied to the data
6. Reviewing results – With a first review of results, an overview can now be obtained by focusing in particular on results with relevant patterns such as outliers

#Quick Insights

This use case shows how light-weight preliminary analyses can be created for short-term customer inquiries, but also for relationship building.

#Example from the User’s Point of View

For the next client update, interim results of the current survey are to be made available to a customer at short notice, incl. initial charts and tables. The current survey dataset can be analyzed with Inspirient on short notice. Navigation options, and especially slide recommendations, make it possible to quickly compile preliminary results that are relevant for this meeting. The automatic visual preparation of these results allows for the results to be exported directly as a PDF and sent to the client by email. In addition, the client can easily be provided with initial tables structured by socio-demographic dimensions for their own analyses.

#Approach

Key steps in implementing this use case are as follows:

1. Uploading data – This analysis starts, once again, with uploading the current dataset at a time
2. Guidance / Prioritisation – For prioritisation, those dimensions for which a strong interest of the client can be assumed should top the list
3. Guidance / Annotations – When checking annotations, attention should be paid to the correct use of DEMOGRAPHIC_VARIABLE, SURVEY_RESPONSE and, where appropriate, AGGREGATION_WEIGHT
4. – the user is notified by e-mail upon completion of the autonomous analysis –
5. Downloading report / files – Depending on the customer’s needs, there are two document types to choose from:
   • Stories allow the client to get a quick visual impression of the data
   • Contingency tables allow the client to see the details of the collected data
6. Reviewing results – Additional results can be selected from the analysis results and then be downloaded as a PDF or PNG image and sent to the client

#Analytical Deep-dive

This use case demonstrates how Inspirient can be used to analyse survey data in detail, prioritise findings to share with the client and support in preparing the final report.

#Example from the User’s Point of View

Both during and after the field phase, the progress of data collection must be continuously monitored and documented. During the on-going field phase, it is important for control purposes to monitor response rates of the data collection in order to ensure a sufficient number of interviews per cohort group. Once field phase has been completed, the return rate of the data collection must be documented for the client in the method report. In both cases, the automated evaluation of the respective status of the survey by Inspirient is there to help, even applying multivariate methods such as logistic regressions (OR, AME) where appropriate. And Inspirient’s graphical output also makes it possible to quickly decide which topics should be presented in detail in the client report.

#Approach

Key steps in implementing this use case are as follows:

1. Uploading data – The analysis starts with uploading the dataset
2. Guidance / Prioritisation – The prioritisation should be broad rather than narrow in this use case, in order to obtain a balanced overview of the results
3. Guidance / Annotations – When checking of Annotations, the key point is to ensure that DEMOGRAPHIC_VARIABLE, SURVEY_RESPONSE and, if applicable, AGGREGATION_WEIGHT are set correctly
4. – the user is notified by e-mail upon completion of the autonomous analysis –
5. Reviewing results – When browsing results, the Tag Cloud should be employed to…
   • Focus on tags concerning analytical methods and patterns
   • Interesting results can be recorded for later consideration using positive feedback (‘Likes’)
6. Sharing results – The ‘liked’ results can be shared with colleagues for quick feedback and coordination
   • Similarly, colleagues with whom the analysis has been shared can highlight further results via ‘likes’
7. Downloading reports – Selected results can be summarised and downloaded in dynamic reports

#Comparison over Time

This use case shows how several separately collected datasets of the same long-term study can be combined intelligently.

#Example from the User’s Point of View

For a long-term study, newly collected data must be analyzed at regular intervals. For this purpose, existing and new data sets of the same study can be evaluated via Inspirient. The system intelligently links old and new tables with each other. In the user interface, noteworthy changes since the last evaluation can be identified quickly. Consolidated data sets can be downloaded for further analysis.

#Approach

This use case follows the same approach as for the analytical deep-dives. However, in this case, several datasets are submitted for analysis to be analyzed over a longer period of time. These are merged by the system if there is a sufficiently large overlap between the columns.