After the concept noun is clear, the next step is variable design, including three contents: operation variable design and variable attribute design scale selection.
Variable is a measurable noun. A scientific research, especially empirical research, needs quantitative data as the analysis basis, and inevitably has to deal with many variables. Some variables, such as temperature and daily output, can be measured directly. Other variables have clear connotations, but they are difficult to measure directly. For example, the term labor productivity is conceptually divided by the total number of employees, but there are different understandings when collecting data and calculating, which need to be specified. For example, does the total number of employees refer to the number of registered employees, or does it include temporary workers and contract workers? "Total number of employees" is a nominal variable, while the operational variable may be "number of registered employees in enterprises" or "total number of registered employees plus contract workers".
Transforming nominal variables into operational variables is an important content of variable design. For example, the Teachers' Law of People's Republic of China (PRC) promulgated by 1993 stipulates that "the average salary level of teachers is not lower than or higher than that of national civil servants, and will be gradually increased", but so far, there is no report on the implementation of this provision. Compared with the civil servants in recent years, the average salary level of teachers is high or low, and no one can say clearly how big the gap is. The reason is that it is difficult to measure and test according to the proposition expressed in this regulation. "Average wage level" is a nominal variable. If it is to be calculated, it must be converted into reasonable operational variables, and the meaning of each variable should be clear. For example, does the average wage level refer to all teachers and civil servants, or does it refer to the comparison between teachers in various schools and corresponding types of civil servants? Salary refers to basic salary or actual salary including performance salary. These details can't be counted unless they are explained clearly.
Variables must be measurable. This means that there are quantitative differences in some attributes of nouns (concepts). For example, the variable "number of employees" refers to the number of employees, and its attribute is the number of employees. The attribute of the variable "worker sex" is only male or female. The attribute of the "worker's age" variable can be set to youth, middle age, old age, or a number between 18 and 60.
A variable is a collection of attributes, and different attributes should be measured by different scales. The number of employees property set is a number greater than 1, and the proportion used is fixed. For example, "1000 people" is an attribute of "number of employees". The attribute set of "workers' gender" is only male and female, which belongs to the classification scale variable, and it is classified according to the attributes of workers. "Workers' age" belongs to the classification scale if its attributes are youth, middle age and old age. If it is set to 18 to 60 years old, it belongs to the scale. If the variable "employee education" is set, it can be divided into undergraduate, master and doctor. If it is necessary to prioritize various attributes, a ranking scale can be used. For example, in the recruitment of employees, they are ranked according to academic qualifications, which are set as undergraduate, high school, master, doctor and junior high school, and the ranking scale is marked as first to fifth accordingly.
Research work is always inseparable from the relationship between variables. Variables are nouns and concepts that can be measured by numerical values. Some variables have only two values, namely 0- 1 variables. For example, "gender" as a variable has only two attributes of "male" or "female", and the state of shells is only explosive and non-explosive. Of course, attributes can also be added. For example, the nationality of an individual can be "1, 2, 3, 4, 5, ...". . If represented by automobile brands, Changan is 1, Geely is 2, and Jetta is 3. These variables are all discrete, and generally cannot be expressed by decimals such as 3.2. Another kind of variables is continuous, such as annual income, test scores, age and so on. , can be expressed in decimal.
The measurement of variables and attributes such as the total number of employees, age and education is relatively intuitive and can be completed by a single indicator. In some cases, variables need to be measured by multiple indicators, involving multi-dimensional attributes. Management research often encounters such variables, such as satisfaction, cohesion, execution and so on. Unlike length, age, weight and other variables can be measured by a single indicator, researchers often have to design a set of multiple indicators to indirectly measure such variables, which is the difficulty of management research, but it also provides a unique research space for management researchers. Designing an effective measurement indicator is the result of a research work.
Second, the process of variable manipulation.
From hypothesis to variable design, there are a series of transformation and refinement links, which constitute the entity research content with personal characteristics in the paper work. Graduate students can't ignore and despise this process of transformation and refining, and it is not easy to complete all the work correctly. Here is an example to illustrate this process.
There is a folk proverb "beautiful women are unlucky", which is actually a hypothesis. Some people put forward this argument according to their own observations and feelings, while others found it reasonable and profound, so it gradually spread, but if it is to become a scientific conclusion, it needs to be demonstrated. "Beautiful women are unlucky" can be literally understood as "the fate of beautiful women is not good", and if expressed in hypothetical language, it means "all' beautiful' women have bad fate". Or to put it another way: "a woman's face value is negatively related to her fate." No matter what the explanation, the research object is "woman", and this hypothesis involves two variables: "face value" and "fate". The attributes of these two variables can be set as discrete, for example, the attribute of Yan value can be "beautiful, beautiful, average and ugly"; The attribute of fate can be "good luck, average, bad luck".
If the attribute is set to continuous, it can be expressed in numerical values according to the beauty of face value and the beauty of fate, such as 1...5. Five of them are the most beautiful and the fate is the best. In order to prove and meet the requirements of data collection, this nominal variable must also be transformed into a measurable operational variable. Although there is no scientific instrument to measure face value and fate in reality, as a scientific research, measurable problems must be solved.
In this case, there are two ways to solve the problem. One is the method of logical reasoning, and the other is the method of direct feeling judgment.
The method of logical reasoning is that we can't find a direct way to measure "face value" or "fate", so we should design several indicators to indirectly measure variables according to the extension of "face value" or "fate". This leads to the term "indicator". As mentioned above, the independent variables derived from the independent variable tree to the operational level are called operational independent variables, and the variables are operational variables. Some of these operational variables can be measured directly, while others can't, so we need to find a group of variables that can be measured directly. This variable, which can be directly used to collect data, is usually called "indicator" in practice, and multiple or multi-level indicators form an "indicator system".
Imagine that "face value" can be transformed into three variables: beauty of appearance, beauty of form and beauty of charm, which is one step closer to the requirement of operability, but it can't be measured directly, so it can be decomposed into the next level variables, such as height, weight-height ratio, leg length-height ratio and so on. Variables such as height can be directly measured, which can be called indicators, and the quantitative value of physical beauty can be indirectly measured by using this group of indicators. Variable design is basically completed in this step, and the follow-up work includes setting the attributes and scales of manipulated variables. Fate is a similar situation, and an operable index system needs to be designed.
Intuitive judgment is to find some experts and judge the face value and the beauty of life subjectively by intuition. As will be discussed later in the questionnaire method section, even subjective judgment is very important for experts to answer any questions. They can't directly ask, "Is this a good life?" And "Is this person beautiful?" . Because the experts who answer questions have different understandings of the concepts of a better life and a better life, these direct answers lack comparability and consistency, which is of little statistical significance. Like a mature "IQ" test questionnaire, instead of asking the client "How is your IQ? Please choose from seven grades. "A well-designed IQ questionnaire should make the testee unaware that he is testing IQ. The direct sense judgment method also needs to design a set of questionnaires similar to the operation index system.
As can be seen from the above discussion, it is not easy to prove the general assumption that "beauty is unlucky" according to scientific methodology. If we really take the above example as a research work, it is a valuable research work to design an operational index system of the two concepts of "face value" and "fate". In management research, we often encounter abstract concepts such as "cohesion" and "openness". Therefore, there is still a lot of work to be done from the hypothesis statement to the design of operational variables and measurement indicators, and the transformation and refinement of management dissertations.
In the process of transforming nominal variables into measurable operational variables and indicators, there are two problems worthy of attention.
First, don't confuse variables with attributes. Attributes represent the differences in the types or degrees of variables, and there are always related comparable concepts, while variables are relatively independent concepts. For example, gender is a variable with the attributes of "male" and related "female". In a paper, the same concept cannot be both a variable and an attribute. For example, in the front of the article, "Yan value" is set as a variable and "beautiful" or "very beautiful" as an attribute, but "beautiful" cannot be regarded as a variable and given a set of attributes about beauty. However, this confusion between variables and attributes often occurs in dissertations.
Second, from nominal variables to directly measurable indicators, it is necessary to demonstrate the effectiveness of each link. Some papers involve nominal variables such as "enterprise innovation" and "enterprise performance". When empirically measuring this variable, they simply rely on a cognitive question in the questionnaire: "Do you think the innovation (performance) of this enterprise belongs to: very strong (very good), strong (good), average, poor and very good? Faced with such questions and options, employees can only give answers based on personal impressions. The validity of the data collected by these answers is not convincing.
Where did the article come from? Research and writing guidance for MBA dissertations
Author | Li Huaizu