
examples of nonrandom sampling
 quota sampling
 panelling
 postal and telephone surveys

continuous data
data that can take any value whatsoever

discrete data
data that can only take on specific values, such as money, which changes value in whole units

categorical data
data classified into a number of distinct categories

ordinal data
data that has been classified into a number of distinct ranked categories

frequency distribution & relative f d
 group the data into bands of specific values and display the frequency of occurrence of each band
 relative show proportions or percentages rather than frequencies

cumulative f d & relative cum f d
 can be used to show the number of items with a value less than or equal to a given figure.
 relative show the proportion or % with a value less than or equal to a given figure

lorenz curve
visual comparison of the 2 cumulative frequency distributions

histogram
displays the number or % of items falling within a given band through the AREA of a bar

 arithmetic mean
 not necessarily an observed value
 greatly affected by extremes

 Population
 for raw data fi = 1 for each item
 for grouped data, use the midpoints
 for sample standard deviation, devide by (n1) rather than n

 Geometric mean
 useful for COMPOUNDING relationships
 UNDERSTATES the mean compared to the arithmetic mean

formulas for median, range and intertquartile range

= median
UNAFFECTED by extremes

range =
highest  lowest
considers ONLY EXTREMES

interquartile range =
UNAFFECTED by extremes

mode
 most frequently occurring item
 must be observed value
 UNAFFECTED by extremes

perfect symmetrical population
mean = median = mode

positively skewed distribution
 mode<median<mean

negatively skewed distribution
 mean<median<mode

regression is used to
establish if a relationship exists between two factors, a regression calculation provides a l ine of best fit

y = a + bx
 linear regression
 a = the intercept i.e. the height at which the line cuts the yaxis
 b = slope

interpolation
where we use the regression line to estimate future results from data within an already observed range

extrapolation
where we try to estimate results from beyond our existing range of experiences and data

APR
annual percentage rate

(1+APR) = (1 + r/n)^n
 interest rate
 r = annual flat rate
 n = number of compounding periods pa

R = e^rt 1
continously compounded interest rate

(1+r)^n
basic discounted cash flow compound factor with n years to run

Dn = Do (1 + r)^n
depreciation where r is negative

1/(1+r)^n
basic discount factor

1/r (11/(1+r)^n)
discount annuity

1/r
discount in perpetuity

1/e^rt
continous compounded basic discount factor

PV of borrowing/ ADF
regular repayment of mortgage

IRR
 the rate of interest that discounts the investment flows to a net present value of zero
 NPV is a better decision tool than IRR due to IRR limitations
 assuming reinvestments can be made at the IRR
 multiple yields (where cash flows reverse twice)

