Data quality and criteria

2.2 Understanding of Data Governance within EGIT | 17

see from this that it is the new media and technologies that mainly influence this develop­

ment. As a means of data quality assurance in the 1990s, data registration was still considered the most important task of data quality management (Baumöl & Meschke, 2009, p. 62). The opinion was that the more reliable and impeccable the information on which strategic and operational decisions are based, the less time and effort is required for subsequent and neces­

sary error corrections. There is nothing wrong with that, as long as you always know where your relevant data is and you have access to the origin of the data. However, the emergence of social media brought with it a very different way of processing data and the existing data quality frameworks were not defined for this kind of data processing. The most important paradigm shifts that must be considered for future data governance are listed in Table 2.1. It should be noted that many of these shifts were recognized much earlier and are also being addressed in science. Many of the articles published on this topic show that data management has changed significantly and that a major impact can be expected from a governance point of view.

Table 2.1 Paradigm shifts in data management

From → To

Transaction (tables) → Event (streams) (Appel et al., 2013)

Cycles → Realtime (Lu et al., 2016)

Centralized systems → Federated systems (Villegas et al., 2012) Hard to scale → Dynamic scalable (Zhang & Zhang, 2010) Human centric → Software centric (RPA) (Kehoe et al., 2015) Product centric → Customer centric (Rust & Kannan, 2003) Deployment in weeks → Continuously in seconds (Savor et al., 2016) Inflexible, tightly coupled → Agile, loosely coupled (Orton & Weick, 1990)

Costly → Inexpensive (Armbrust et al., 2010)

Not easy to share → Easy to share (Hamari et al., 2016)

Idle → Utilized (Armbrust et al., 2010)

Closed source → Open source (Ye & Kishida, 2003) Software (As is) → Service (As used) (Turner et al., 2003)

Data It is the distribution of data in the cloud which will be a major challenge in data man­

agement in the future. Regardless of where data is ultimately made available, the consolida­

tion of data serves to gain information and knowledge. It is about measuring the organization.

In order to check whether goals have been achieved, whether the company behaves in accor­

dance with the rules and whether it has adequately positioned itself against risks, managers

2.2 Understanding of Data Governance within EGIT | 19 need data from information systems. This data has to be first of all understood, secondly, interpret correctly and thirdly, to be able to initiate the appropriate actions. In order to be able to carry out all these steps, it is necessary to provide data that is unobjectionable and unambiguous for the respective purpose. The terms data and information are often used syn­

onymously in connection with the quality of data and information, whereby the two terms must be distinguished from each other. As a general rule, data are neutral facts (Y. W. Lee, 2006, p. 9). Data is distinguished between master data, inventory data, transaction data and change data due to its different characteristics (Stahlknecht & Hasenkamp, 1999, p. 160).

Master data is data about objects that change very rarely or not at all. This includes, for example, the name of a person, the address of a company, the date of birth or the marital status. In production planning, master data also includes bills of material or work sched­

ules (Stahlknecht & Hasenkamp, 1999, p. 160). Changing data show the change of data and trigger a change of master data, as for example the change of the marital status with the mar­

riage of two persons (Stahlknecht & Hasenkamp, 1999, p. 160). Balance data always shows a balance. For example, for an account, the account balance is a position, or for a warehouse, it is the warehouse stock. Stock data has the characteristic that it is updated at periodic and time intervals (Stahlknecht & Hasenkamp, 1999, p. 160).

Information Zins (2007) names no less than 130 definitions which mention the connection with data, information and knowledge. The definitions agree that data is theraw materialfor information. Data, in combination with additional context, becomes information. Informa­

tion represents knowledge about facts or persons. Depending on the situation and context the information can be relevant or irrelevant (Zins, 2007) . In business processes the information is processed and is the starting point for decisions in the enterprise (English, 1999).

Data quality criteria In order to make data quality measurable, quality criteria must be defined. The table 2.2 shows mentioned data quality criteria by the literature (Helfert, Her­

rmann, and Strauch, 2001, p. 7; Apel, Behme, Eberlein, and Merighi, 2015, p. 8). An impor­

tant aspect of data security is data integrity (Maletic & Marcus, 2000). Data integrity requires that the data is not damaged or altered during processing or transmission. Integrity stands for correct content, consistency and correctness (Shaikh & Sasikumar, 2015, p. 494). An more detailed perspective on the quality criteria for data is taken by Wang and Strong (1996, p. 11) and shown in table 3.5. The criteria presented later in table 3.5 have been taken into account with those from table 2.2 as a baseline for developing the model. There is no question that a future model must also meet these quality criteria.

Table 2.2 Data quality criteria (Helfert et al., 2001, p. 7; Apel et al., 2015, p. 8)

Actuality Universality Age

Change frequency degree of processing meaning Usability Degree of Confirmation Determination

Level of detail efficiency clarity

Faultlessness Flexibility Integrity

Validity Accuracy Credibility

Validity Manageability Integrity

Information Level Clarity Compactness

Compression Consistency Constancy

Correctness Neutrality Objectivity

Operationality Performance Portability

Precision Problem Adequacy Forecast Content

Verifiability Quantifiability Timeliness

Non­redundancy relevance correctness

Robustness Rarity Security

Significance Memory Requirements Degree of Standardization

Subject adequacy testability scope

Independence Verifiability Transferability

Validity compression ratio availability

Power of disposal Linkability Reliability

Encryption Level Comprehensibility Trustworthiness Readiness for use completeness truth content

Probability Maintainability Reusability

Effect duration time adequacy time reference

Time Optimal Accessibility Reliability

Quality According to DIN EN ISO 9000 quality is defined as follows (ISO, 2015):

Quality is the degree to which a set of inherent characteristics meets requirements.

Quality is therefore a correspondence between demands and expectations of a product and its properties. Quality can be perceived negatively or positively. According to Garvin (1988, p. 40), the concept of quality is divided into five approaches (Kamiske & Brauer, 2011, p. 167):

1. The transcendent approach describes quality as a person’s subjective experience of the specific characteristics of a product or service.

2. The product­related approach describes the quality of a product based on the fulfilment of defined requirements. Quality can be determined without subjective perception.

3. The customer­related approach describes the quality of a product via the product user.

The customer decides whether the product corresponds to the required quality.

4. In the process­oriented approach, it is assumed that the product corresponds to the quality if the process runs optimally and according to plan. All product specifications are adhered to.

2.2 Understanding of Data Governance within EGIT | 21 5. The value­based approach sees quality as the fulfilment of a service at an acceptable cost. A product is of high quality if the costs and the performance are in an acceptable relation to each other.

The above approaches show that quality can be perceived differently, depending on the per­

spective. The evaluation of the quality of an identical product can therefore be different. Fi­

nally, the term data quality can be derived from the above general definition of data and qual­

ity. According to Würthele (2003, p. 21) data quality is defined asmultidimensional measure of the suitability of data to fulfill the purpose associated with their collection/generation. This suitability may change over time as needs change. Based on the definition of Würthele (2003, p. 21) and the definition of data, quality and information, the term data quality is defined as follows in this dissertation: Data quality is the degree of matching between an individual’s demands and the properties of the data. The requirement may change over time and may take on a different meaning due to different uses.