Levels of Measurements in Research
When we design a research problem its's the idiosycratic of a person as how it will define it. But the definition in the beginning will define the shape and solution it will require to be producible. Therefore, our selection of variable plays a significant role in our research.
Variable as seen is categorised predominantly in four ways.
Variable as seen is categorised predominantly in four ways.
- Nominal
- Ordinal
- Interval
- Ratio
1. Nominal (Discrete)
These variables are common in nature such as persons belonging to various cities, different caste, race, various holidays, celebration days, perception of sports and so on.
The summation of the Nominal category is that we cannot make a Ranking order of our choices all of them are important and equal. Also, this category comes in light when the quality of each category is important that the quantity. Therefore, no arithmetic property is associated with it. We will deal in repetition of a a category so Mode will be calculated. Also when we validate the relation then we will use testing method of dependency Chi-Square Method.
NOTE: For nominal data we use Contingency Table for data description
For Example - for a study of a employee of a company their commuting pattern was seen it was found they were commuting by
- Bus
- Metro
- Bicycle
- Train
- Personal Vehicle
we can calculate Frequency, Percentage
It is for categories
2. Ordinal (Discrete)
Ordinal itself says "ORDER" This involves all the property of Nominal Category along with the ease of Ranking. All the option that will be used can be ranked in a order. For Example. How your subject teacher can be rated 1. Average, 2. Good 3. Very Good 4. Excellent.
Median, quartile and percentile can be determined in this category. Various test can be performed like Spearman's Ranked Correlation Coefficient and Kendall's Coefficient of concordance .
Note: For Ordinal Data again we use Contingency Table
We can Calculate Median,
Percentile is an ordinal data
3. Interval (Continuous)
This category is used where we have to define something in a particular interval for example 3 PM to 6 PM. Interval variables contain lots of information. It does not elaborate the "0" . Also, Ratio can not be defined in this variable form For Example 20 degree Celsius is not as hot as 10 degree Celsius.
In Interval "0" does not mean that there is no temperature but the temp is there
Some more example includes - The year you went to school, 2002, 2004
Note: For Interval Data we use use ScatterPlot
No absolute Zero. One cannot compare between the data
Rating data like rating at the rate of 10 for a car service
4. Ratio (Continuous)
This Category is richest in nature. Almost all the arithmetical calculation can be done on it. The "0" can be defined in this category. Comparisons like twice as much as high, or half of something can be mentioned and is followed. Very suitable for all statistical operations.
For example- Reaction time taken to catch a ball, for player A it is 3 second for player B it is 4.5 sec so I can say player A is 50% faster than player B.
More example- The marks you got in a statistics test
NOTE: For Ratio data again we use ScatterPlot
It has absolute Zero.
Data in currency, height ,weight.
All the basis of statistics can be done with Ratio Data
LIKERT SCALE
(1) Strongly disagree
(2) Disgree
(3) Neither agree nor disagree
(4) Agree
(5) Strongly agree
for most of the questionnaire this scale is used, now the point lies that in which class does this Likert scale falls- yes it falls in Ordinal
CONFOUNDS- If a researcher is looking for data of the people who smoke and it causes cancer than there will some participants who got cancer but not due to smoking but due to other factor of diet & environment. These other factors that affects the research are known as"confounds".
PLACEBO EFFECT - Call a patient and give him a chemically inert drug for his illness he will have it and became better than when he will not be given any drug :)