What are variables - Essay Example

? Variables A variable is anything whose value can vary. For instance age is a variable as age varies for different persons. Variables need not be numerical always. Gender is a variable as it consists of two text values ‘male’ and ‘female’. Every variable has some attributes. An attribute refers to specific values of the variable. For example, the variable gender has two attributes- male and female. The variable agreement may have five attributes- (a) strongly agree, (b) disagree (c) neutral (d) agree and (e) strongly agree (Trochim, 2006). For studying cause- effect relationship, the distinction between independent variable and dependent variable must be clear. The independent variable is what the researcher manipulates. The dependent variable is that which is affected by the independent variable. The independent variable is the cause or treatment and the dependent variable is the effect or outcome. For example, while studying the effects of traffic rule awareness campaigns to reduce accidents, the program is the independent variable and the achievements are dependent variables. A variable should be exhaustive and mutually exclusive. Exhaustive means that a variable should include all possible responses. For this purpose, list all the important attributes and use a general category like “other” to represent remaining ones. Also, variables should be mutually exclusive. It means that no one should be able to have two attributes simultaneously. Operationalization The meaning of a research study depends on how objectively the phenomena under consideration are observed. So developing a reliable and valid set of procedures for measuring the variables is crucial for the validity of research study (Operational Definitions, 2005). The first step for an operational definition is to specify the constructs. How observations are made, what is observed, how observations are to be recorded have to be specified accurately. Sample Sampling is the process of selecting units from a population so that by studying the sample, it would be able to generalise results back to the population. Related to generalisation is the concept of external validity. Validity refers to the approximate truth of conclusions (Trochim, 2006). External validity refers to the degree to which conclusions in the study would hold for other persons in other places at other times (Trochim, 2006). There are two major methods for selecting sample for generalisation. One is called sampling model. In it, the population to be generalised is identified and a fair sample is drawn to conduct the research. Since the sample is representative of the population, the generalisations are applicable to the population. The second approach is called proximal similarity model. The term proximal similarity was suggested by Donald T. Campbell. In this approach, several generizability contexts are considered and theories are developed as to which situation or context is more similar to the study and which are less similar. By doing so, it is possible to generalise the results of the study to other persons, places or times that are more like the study (Trochim, 2006). The population to be generalised is called theoretical population. The population that is accessible to the study is called accessible population. The list of accessible population from which sample is drawn is known as sampling frame. For example, if a phone survey is to be done with a phone book, the phone book is the sampling frame. The sample is the group of people, who are selected for the study. But, the group that actually completes the study is the sub sample of the sample only. There is a distinction between random selection and random assignment. Random selection refers to how sample is selected. Random assignment refers to how different treatments are assigned to the sample selected for study. In sampling, the units are usually people and they supply responses. A response is a specific measurement value. A statistic is used to look at the responses of the entire sample. Mean, median, mode etc. can be used for this purpose. Scale Scaling is the assignment of objects to numbers according to a rule (Trochim, 2006). It involves construction of an instrument that associates qualitative constructs with quantitative metric units (Trochim, 2006). It attempts to measure abstract concepts. Scales may be unidirectional or multi directional. Height, thirst, self esteem etc are unidirectional. Height can be either shorter or taller only. Thirst, can be more or less only. Self esteem can also be more or less. If the concept under study is multi directional, multidimensional scaling is required. For example, measuring academic achievement in a unidirectional scale is not appropriate. Some may be good in maths, but verbally terrible. So, only a multidirectional scale can capture the type of achievement. The three unidiamentional methods of scaling are thurstone or equal-appearing interval scaling, likert or summative scaling, and guttman or cumulative scaling. Measurement Measurement is the process of observing and recording the observations that are collected as part of a research effort (Trochim, 2006). Here, it is necessary to understand the various levels of measurement. The levels of measurement used to measure the relationship among the values that are assigned to the attributes for a variable vary. For example, consider party affiliation as a variable. Assume that the attributes are Republican, Democrat and independent. For the purposes of analysis, the values 1, 2, 3 respectively may be assigned to the attributes. Here the role of values is to distinguish the attributes only. This is nominal scale of measurement. In ordinal scale of measurement, the attributes can be ordered. For example, in a survey code, using 1 for some High school, 2 for high school, 3 for some degree, and 4 for degree denotes that higher numbers mean more education. In interval scale, the distance between attributes has a meaning. For example while measuring temperature; the interval between two values in a thermometer is interpretable. In ratio scale, there is an absolute zero. Weight is a ratio variable as it can be zero. The number of students in a class is a ratio variable. Measurement Error Measurement errors are of two types- random error and systematic error. Random error is caused by any factors that randomly affect measurement of the variable across the sample (Trochim, 2006). For example, a person’s mood may influence his performance. If mood influences performance on the measure, it artificially inflates or deflates the observed scores. An important property of random error is that it does not affect the average performance of the group. Systematic error is caused by any factors that systematically affect measurement of the variable across the sample (Trochim, 2006). For example, if there is loud noise near the class room where students are taking tests, it will affect the scores of all the students. Systematic error may be positive or negative. Systematic error is also known as bias in measurement. One method to reduce measurement error is to test the instruments and getting feedback from the respondents regarding testing environment etc. Training the observer and checking the data carefully is very important. All data entry for computer analysis should be "double-punched" and verified (Trochim, 2006). Causation A cause is an explanation for some characteristic, attitude, or behaviour of groups, individuals, or other entities (Causation and Research Design, 2006). Most social scientist tries to make causal explanations for the phenomena under study. Usually the independent variable is the presumed cause, and the dependent variable is the potential effect (Causation and Research Design, 2006). It is to be noted that finding a relationship and causation are different. Causation cannot be proven with statistics. Statistical methods help in finding a relationship. It should be clearly analysed with the help of theories and logic for making causation Causal explanations are of two type- ideographic and nomothetic. A nomothetic causal explanation is one involving the belief that variation in an independent variable will be followed by variation in the dependent variable, when all other things are equal (Causation and Research Design, 2006). An ideographic causal explanation is the concrete, individual sequence of events, thoughts, or actions that resulted in a particular outcome for a particular individual or that led to a particular event (Causation and Research Design, 2006). Three criteria must be fulfilled for establishing causal relationship. (1) The cause must occur before the effect, (2) there should be some type of relationship, and (3) there should not be any other plausible alternative explanations. Plausible rival explanations A relationship does not always mean that it is a causal one. It is possible that some other variables may cause the outcome (Trochim, 2006).This is called the third variable or missing variable problem. Dealing with it is very important and it is at the heart of internal validity. A research design should rule out plausible rival explanations. This is possible by introducing a control group in experimental designs. Hypothesis A hypothesis is a statement of relationship between two variables. It is a statement of prediction (Trochim, 2006). It tells clearly what is expected in the study. Exploratory studies need not have hypothesis. In a study, there may be more than on hypotheses. The way to set up hypothesis test is to formulate two hypothesis statements, one that describes the prediction and one that describes all other possible outcomes with respect to the hypothesized relationship (Trochim, 2006). If the prediction is that variable A and variable B are related, the only other possible outcome would be that A and B are not related. The hypothesis that supports the prediction is called alternate hypothesis and the hypothesis that describes other possible outcomes is the null hypothesis. The notation Ha or H1 represents the alternative hypothesis and H0 or Ho represents the null one. Reliability The word reliable means “dependable.” In research, the term reliability means "repeatability" or "consistency". (Trochim, 2006). A measure is considered reliable if it gives the same result when repeated. There are four general classes of reliability estimates. They are: inter-rater or inter-bserver reliability, test-retest reliability, parallel-forms reliability and internal consistency reliability. Unit of Analysis The unit of analysis refers to the major entity that is to be analyzed in the study. It can be individuals, books, artefacts (books, photos, newspapers), geographical units (town, census tract, state) or social interactions (dyadic relations, divorces, arrests) (Trochim, 2006).When comparing the academic achievements of children of two classes, the unit of analysis is each individual child. When comparing two classes on class room climate, the unit of analysis is the class room. Main Types of Validity Validity is whether the measurement truly reflects the concept we are studying (Operational Definitions, 2005). Internal validity is relevant for studies trying to establish causal relationship only. It is the approximate truth about inferences regarding cause-effect or causal relationships (Trochim, 2006). Internal validity is relevant for studies that assess the effects of social programs or interventions. Internal validity, in simple terms, is the evidence that the outcome is the result of what is done by the researcher. External validity is related to the question of whether the results of the study can be generalised to the people outside of the study (Operational Definitions, 2005). Construct validity means that the things manipulated and measured in the study truly represent the ideas in the mind of the researcher. For a study to determine the effect of traffic awareness class among youth in reducing accidents the independent variable is awareness classes and the dependent variable is the youth who had undergone such classes. To ensure internal validity, a control group also need to be selected. The unit of analysis would be the participants. The alternate hypothesis can be that there is a relationship between traffic awareness classes and accident rate among youth. The null hypothesis can be that there is no relationship between traffic awareness classes and youth accident rate. References Causation and Research Design. (2006, 11 12). Retrieved February 28, 2010, from http://www.chnri.org/resources/1.%20Learning%20Resource%20Material/Epidemiology/Epidemiology/Association/Papers/Causal%20and%20Research%20Design.pdf Operational Definitions. (2005). Retrieved February 25, 2011, from Wodsworth Cengage Learning: http://www.wadsworth.com/psychology_d/templates/student_resources/workshops/res_methd/op_def/op_def_02.html Trochim, W. M. (2006). Research Methods Knowledge Base. Retrieved February 25, 2011, from Web Center for Social Research Methods: http://www.socialresearchmethods.net/kb/variable.php Read More