LECTURE 2

GUIDELINES FOR CONSTRUCTING FREQUNCY DISTRIBUTIONS:

The purpose of constructing frequency distributions is to condense a body of data into a table that is readily appraised and comprehended.

Definition of class intervals:

    Each observation in the data set must belong to one and only one class interval.

Width of class intervals:

    Class intervals of equal widths, if possible. There are situations where unequal widths are necessary for various reasons, e.g., income distribution.

Number of Class Intervals:

    No hard-and-fast rules, but generally between 5 and 20.

Relationship between Number and Width of Class Intervals:

    Width of class interval = (largest value – smallest value)/ No. of class intervals.

Position of Class Intervals:

    Midpoints should be close to the average of the observations included in the class interval.

THE CUMULATIVE FREQUENCY DISTRIBUTION:

    It shows the number of measurements in the population that lie below or above a certain value.

    e.g., how many firms in Table 1.8 paid their chief executives less than $1.4 million in 1991?

---------------------------------------------------
Salary               No. of firms (millions $)
---------------------------------------------------
Less than 0.20                  0
Less than 0.60                  5
Less than 1.00                  8
Less than 1.40                 16                                                       Table 1.11:
Less than 1.80                 20                                                       Cumulative
Less than 2.20                 23                                                       Frequency
Less than 2.60                 24                                                       Distriution
Less than 3.00                 24
Less than 3.40                 25
---------------------------------------------------

One can plot the cumulative frequency distributions, called OGIVE. All these are less-than cumulative frequency distributions.

One can also construct greater-than cumulative frequency distributions and ogives, which show the number of frequencies that exceed particular values.

FREQUENCY DISTRIBUTIONS FOR SAMPLE DATA:

Sample vs. Entire population

Generally we do not have the frequency distribution of the entire population. We have frequency distribution based on a sample. They will be somewhat different. It is good to visualize the frequency distribution of the population, and how this is related to that based on only one sample.
 

SUMMARY MEASURES:

Measures of Central tendency and Dispersion:

Central tendency: Þ average level, typical value.

Depending on the purpose, there are a number of alternative measures of central tendency.

Dispersion: Þ degree to which the individual measurements vary about the average value.

Think of a society with two individuals. In one their yearly incomes are $50,000 each. In the other, one earns $100,000 and the other $0. In both societies the mean income is $50,000. However, the dispersion is much higher in the latter society.

PARAMETERS and STATISTICS:

Population parameters are seldom known. We are interested in drawing inference about population parameters from a sample.

Estimates of population parameters based on a sample is known as statistics.

Arithmetic mean: Sum of the numbers in the sample divided by their number.

m = (X1 + X2 + …. + XN)/N

m is the population value.

When based on a sample

X = (X1 + X2 + … + Xn)/n

    = ( S Xi )/n

Calculation of arithmetic mean from grouped data:

Use the midpoint of the class interval as the value of measurements in that class as an approximation.

Use table 2.2:

X = S (f.i XI)/n