The main types of scale identified by using the search engine are measurement scales (nominal scale and ordinal scale, interval scale and ratio scale). For the measurement of attitudes, researchers use physiological measures of effect, indirect qualitative methods and self-report. In self-report, researchers distinguish single item scales and multi-item scales, numerical and non-numerical scales, pictorial and graphical scales (Google Search 2008). Also, researchers distinguish comparative scales: rank-order scales, paired comparison and constant sum. Taking into account studies and researches of theoretical literature, no new and innovative approach have been developed. Thus, the researched scales allow choosing the best one for further research (Huberman and Miles 2002).
The weaving and sifting of categories of variables to formulate the relationships among them allow for, at least, a claim of subjectivity on the part of the researcher and, at most, a gross misinterpretation of actual facts. The grounded theorists accuse the empiricists of imposing a priori rating scale values and codes on the subjects’ responses, while it may be that the grounded theorists’ own processual analysis is even more firmly based on researcher bias (Flick 1998).
In place of immersion in the sources followed by a sudden insight, quantitative research demands the testing of a suggested relationship with statistical procedures. While statistics cannot “prove” the metaphysical “truth” of any statement, they can help examine the nature of the connections between variables assumed in a given hypothesis. Basically, this procedure involves rejecting the null hypothesis, which states that chance might have produced the same result. The kind of mathematical techniques to be used in testing a given relationship between variables are determined by the four levels of measurement:
- Nominal measures assume only that categories are exclusive and exhaustive. For instance, Nazi members can be either male or female, but not both.
- Ordinal measures are rank-ordered categories, implying some hierarchy of value. One might, for example, assume that college professors and industrialists were members of the elite, store clerks and craftsmen members of the middle class, and industrial workers part of the proletariat and rank them accordingly on a descending scale.
- Interval measures demand a standard unit of measurement which remains constant in size.
Until the last decade or so, it was generally true that the higher the level of measurement, the more powerful were the statistical procedures that could be applied, but more recently, statisticians have developed sophisticated analytical techniques for categorical variables as well (Gubrium and Holstein 2001).
In contrast to a general sense of meaning, quantitative research necessitates the stringent interpretation of the statistical results. The importance of this last step cannot be overemphasized. All too often, scholars drown in a mass of confusing printouts without coming to grips with the original hypothesis. In principle, there are three ascending levels of formality: Even the most simple procedure will already yield descriptive statements about the data set. Especially in cases where the research is path-breaking, such basic information about the structure or development of masses of data will already constitute considerable scholarly progress.
More analytically promising is the testing of the initial hypothesis. The confirmation, rejection, or more likely modification of the original statement about the relationship among variables ought to be the normal aspiration of a quantitative historian since it uses the powers of the method more adequately. Possible only in a minority of cases where the data are strong enough to support theory is model building. The combination of several hypotheses into one model is likely to do the complexity of historical processes greater justice than the testing of a single hypothesis. However, it can only be achieved in fortunate circumstances where the database includes enough factors, and there is already a preexisting theoretical framework (Denzin and Lincoln 1995).
The differences between these categories and to be able to identify which scale each variable in a data set reflects. The descriptive and inferential statistics that are appropriate for answering particular research questions are largely determined by the scale of measurement of the variable or variables involved. The values of nominal scale variables represent a set of exhaustive and mutually exclusive categories of the characteristic or quality being measured.
Hence they are also sometimes called categorical variables. Such variables, in effect, derive from simple classification schemes. The only information imparted by such variables is that cases assigned to one category are deemed equivalent with respect to some characteristic being measured and different with respect to that characteristic from cases placed in other categories (Denzin and Lincoln 1995).
Though it is customary to do so, it is not necessary to assign numbers to the categories of such variables since letters or any other set of unique symbols would do as well. But it is important to remember that any numeric values employed are merely labels applied without regard to their ordering or magnitude. In the Fort Moultrie data, birthplace, occupation, and marital status would be good examples of nominally scaled variables. Other common categorical variables in historical research include sex, religious preference, and province or state (Denzin and Lincoln 1995).
Ordinal scale variables satisfy the exhaustive and mutually exclusive categorization rule for nominal scale variables. In addition, their categories must be orderable to indicate which cases have more or less of some property. While the numeric values assigned reflect their categorical ranking, they do not indicate the magnitude of any differences between them in the amount of the property (Denzin and Lincoln 1995).
Interval scale variables have all the ordinal properties with the addition that the intervals between the ordered categories have equal values. This means that differences between values can be measured by subtraction and that the zero points on such scales are arbitrary. The most familiar illustrations of this measurement type are the Fahrenheit and centigrade scales for temperature. While each has a different zero point with respect to water freezing, it is possible to calculate differences in temperatures in constant units on each scale and to compare these differences (Denzin and Lincoln 1995).
The measurement for ratio scale variables is even more stringent than for interval variables due to the requirement of an absolute zero point. Thus, not only can differences be compared as ratios, but also ratios of the values themselves. Age and value of the real estate in the Fort Moultrie data are examples of ratio scale variables. For most practical statistical purposes, the subtle distinction between interval and ratio scales can be ignored, and the two types of variables are thought of together as quantitative or continuous variables (Denzin and Lincoln 1995).
Bibliography
Denzin, N.K. and Lincoln, Y.S. 1995, Handbook of Qualitative Research, Thousand Oaks CA: Sage.
Google Search: “Attitude Scales and Marketing research”. Web.
Gubrium, J. F. and Holstein, J. A. (eds) 2001, Handbook of Interview Research, Thousand Oaks CA: Sage.
Flick, U. 1998, An Introduction to Qualitative Research, London: Sage.
Huberman, A. M. and Miles M. B 2002, The Qualitative Researcher’s Companion, Thousand Oaks CA: Sage.