Differences

This shows you the differences between two versions of the page.

--- level_of_variables [2016/03/04 18:17] – hkimscil
+++ level_of_variables [2016/03/04 18:23] – hkimscil
@@ Line 65: / Line 65: @@
 </code>
-위의 글을 읽으면서 독자가 느꼈듯이 쌍대비교법의 단점은 (1) 비교할 항목 수가 많아지면, 쌍으로 비교한 문항이 더 많아진다는 점이다
+위의 글을 읽으면서 독자가 느꼈듯이 쌍대비교법의 단점은 (1) 비교할 항목 수가 많아지면, 쌍으로 비교한 문항이 더 많아진다는 점이다 $ (n(n-1))/2 $. In this case, 문항수 $ 162(162-1))/2 $
-{
-{{#!asciimathml
-`# ====== (n(n-1))/2` <br />
-In this case, 문항수 `# ====== (162(162-1))/2`
-}}}
-(2)
 ====== Lickert ======
@@ Line 84: / Line 78: @@
 ====== Level of measurement - English ======
 Measurement is the process of assigning numbers to objects in such a way that properties of the objects are reflected in the numbers themselves. There are four different measures.
- * NOMINAL -- Nominal measures have only the characteristics of exhaustivenss and exclusiveness. Examples include gender, religions, affiliation, political party affiliation, birthplace, college major, and hair color (Babbie,  - 98, p.142).
+  * NOMINAL -- Nominal measures have only the characteristics of exhaustivenss and exclusiveness. Examples include gender, religions, affiliation, political party affiliation, birthplace, college major, and hair color (Babbie,  - 98, p.142).
- * ORDINAL -- Ordinal measures, on the other hand, represent relatively more or less of the variable. Examples might be social class, conservatism, alienation, prejudice, intellectual sophistication, and so on (Babbie,  - 98, p.142). Therefore, such measures imply some sort of order. Each category, however, DOES NOT have measurable distance. In other words, we can arrange the attributes in order. But, we cannot assume that one is twice much as former.
+  * ORDINAL -- Ordinal measures, on the other hand, represent relatively more or less of the variable. Examples might be social class, conservatism, alienation, prejudice, intellectual sophistication, and so on (Babbie,  - 98, p.142). Therefore, such measures imply some sort of order. Each category, however, DOES NOT have measurable distance. In other words, we can arrange the attributes in order. But, we cannot assume that one is twice much as former.
- * INTERVAL -- Interval measures, as you guess, provides the distances among attributes. The distances are meaningful, which means the distance between the attributes can be expressed and understood as the unit of the attributes.
+  * INTERVAL -- Interval measures, as you guess, provides the distances among attributes. The distances are meaningful, which means the distance between the attributes can be expressed and understood as the unit of the attributes.
- * RATIO -- Ratio measures have the same characteristics as Interval measures do. In addition to that it has a meaningful zero point (or absolute zero), which represent "nothing-ness" of the attribute. Fahrenheit or Celsius scales are interval measures because the zero point does not represent the absence of temperature (no heat). Kelvin temperature scale, on the other hand, is a ratio measure because its zero point really represent the absence of heat (I am not able to imagine what it means (or what the reality is), though :) ).
+  * RATIO -- Ratio measures have the same characteristics as Interval measures do. In addition to that it has a meaningful zero point (or absolute zero), which represent "nothing-ness" of the attribute. Fahrenheit or Celsius scales are interval measures because the zero point does not represent the absence of temperature (no heat). Kelvin temperature scale, on the other hand, is a ratio measure because its zero point really represent the absence of heat (I am not able to imagine what it means (or what the reality is), though :) ).
 Why such distinctions? That is because "certain quantitative analysis techniques require variables that meet certain minimum levels of measurement. To the extent that the variables to be examined in your research project are limited to a particular level of measurement -- say, ordinal -- you should plan your analytical techniques accordingly" (Babbie,  - 98, p.143). One interesting point is that "in moving sequentially from nominal to ratio level data, the measurement scale contains the same information as the previous scale(s) while simultaneously adding a new piece of information (Weiss & Leets,  - 98, p.18).
@@ Line 95: / Line 89: @@
 ^ Level of Measurement  ^ Properties  ^
 | Nominal  | Indicates differences among the attributes of the variable  |
-| Ordinal  | Indicates differences among the attributes of the variable [[br]]Indicates the direction of the differences (e.g., more than or less than)  |
+| Ordinal  | Indicates differences among the attributes of the variable \\ Indicates the direction of the differences (e.g., more than or less than)  |
-| Interval  | Indicates differences among the attributes of the variable[[br]]Indicates the direction of the differences (e.g., more than or less than)[[br]]Indicates the amount of difference in equal intervals of the variable   |
+| Interval  | Indicates differences among the attributes of the variable \\ Indicates the direction of the differences (e.g., more than or less than) \\ Indicates the amount of difference in equal intervals of the variable   |
-| Ratio   |  Indicates differences among the attributes of the variable[[br]]Indicates the direction of the differences (e.g., more than or less than)[[br]]Indicates the amount of difference in equal intervals of the variable)[[br]]Contains an absolute zero[[br]]   |
+| Ratio   | Indicates differences among the attributes of the variable \\ Indicates the direction of the differences (e.g., more than or less than) \\ Indicates the amount of difference in equal intervals of the variable) \\ Contains an absolute zero   |
 Source: Bartz, A. (1999). Basic statistical concepts. NJ: Prentice-Hall. (p.11)
@@ Line 103: / Line 97: @@
 As you can see, more information is added into the higher level of measurement. We may think that it would be always better if we use the highest level of measurement -- when we measure a variable. However, many variables used in social research are in fact ordinal, or even nominal -- and sometimes interval. The only convenience for the higher level of measurement is that it can go down. That is, you can use interval variable as nominal variable. You cannot do this in the reversed way, however. The main reason for the distinction is that, as mentioned before, it helps us incorporate statistical methods such as chi-square test, t-test, regression analysis, etc.
-----
-CategoryResearchMethods