Variables in a research work are widely used to obtain correct inferences. This is reached when the variables are altered, measured and manipulated. Accordingly, variables are classified into two sets namely dependent and independent variables (Cohen, Manion & Morrison, 2000). The latter can be altered and changed by a researcher. It is imperative to note that the state of an independent variable determines the manner in which a dependent variable reacts. For instance, in a study on the use of a limb sock to reduce edematous residual limbs, the limb sock is an independent variable while the dependent variable is the residual limb. A measurement of the volume or circumference of the socks is made. In this case, a researcher can change the application of the independent variable because he/she has control over it and not the reaction of the volume of the limb after pressure has been applied.
In a study to investigate security threats on e-commerce, there are a number of independent variables and a single dependent variable that are investigated. Security of online transactions is the dependent variable. Independent variables like technology, viruses, inside man, legacy systems, Trojan horses, denial of service attacks and poor system designs all affect this dependent variable (Bryman, 2008).
The letters X and Y are used to denote variables. X is used to identify independent variable while Y is used to measure dependent variable. X is plotted on the abscissa of the graph and Y on the ordinate or Y axis.
Variables are classified as either continuous or discreet (Frankfort-Nacimas & Nachimas, 2007). The later uses discreet values. The discrete variables such as race and blood groups use numerical integers or alpha characters such as 0, 1,4,6 or A,B,C,D respectively. By using numerical figures, a dependent variable like famine can be graded as unbearable-4, severe-3, mild-2 and slight-1.
Continuous variables use infinitely divisible items such as force, blood pressure height and age (Trochim & William, 2001). They are measured on a real scale with numeric figures. Poor classification or choice of a variable to use results into judgment error. In agreement, Parahoo (2006) suggests the use of a statistical test to determine the best variable to apply.
There are four levels of measurements of variables in a study. Cohen et al. (2000) indicate that the type of measurement a researcher uses is critical in determining the nature of analyses to be made. The measurements include nominal, ordinal, interval and ratio level.
Nominal measurement level uses numeric figures to classify data. According to Robson (2002), it may also include letters. In this case, entities or items that are being measured can have similar properties but differ with others in the other categories such as gender or sex. Hence, male gender is classified as M while the female gender categorized as F.
Bridger (2003) indicates that the classification of observations using ordinal measurement involves the use of symbols. The numbers of items are placed in an ordered relationship. Ordinal measuring provides a reason for such ranking or measurement. The ratios along the scale for different observations are not the same.
This level of measurement applies to numbers. Robson (2002) notes that the latter specifies distance from a lower to a higher interval along a scale. For instance, the measurement of temperature interval in centigrade is the same between 83 to 85 Celsius degrees and 92-94 degrees Celsius.
The measurement in this level includes a zero. Ranking of observations are tailored along size and intervals on scales that are equal. Examples of ratio measurements are speed, weight, velocity and area.