Friday, April 13, 2012


In demographic data the most important and common error is committed in recording of ages. Ages are recorded in whole numbers in completed years. Extract date of birth are seldom asked, even if exact date of birth are known, the date is tabulated with a reference to date in incomplete year which creates biased over above ages are misstated due to conscious acts and same times unconsciously wrong ages are recorded. Studies of various ages data revealed that following type of misstatement of ages are usually recorded.

Over Statement of Ages of Children/ Infants:
 An over statement of ages before adult hood and under statement afterward. An over statement of ages of person just below 65 years an over statement of ages toward extreme of life. A heaping of ages at 21 years for male and 18 years for female. Heaping of ages of ending digits which are multiple of 5 that are 0 and 5 there after even digits or odd digits ages not studies not recorded not stated or unknown.

Methods of Studying Error in Age Reporting:
Various method of studying and measuring error in ages reporting have been develop some of them are objectives and other are subjective. First among them is to see presence of digits preference by contracting what we call the POPULATION PYRAMID  taking population on axis such that one side male and on the other side female population is taken with ages on y-axis. Ages may be taken in single year or 5 years interval. From theoretical assumption if the population is closed to migration pyramid should smooth triangle, however this is not the case as the mortality pattern every age is not same and also the sex differential. The pyramid also reveals level and pattern of fertility and mortality.

Subjective method of detecting miss-reporting of ages is based on comparisons of data. Comparison is possible with in and between the data of two series. Another comparison is possible with the last census or the survey result. A group of persons who are in age group 5 to 90 in a particular census will be 15 to 90 years of age after 10 years that is at second census with possible reduction due to mortality keeping population close to migration. A closed data examination reveals that there are consternation of persons at ages, ending at (0 and 5) and even digits. This is attributed to wrong recording of ages. Because this is not possible that births occur at such interval as to have lesser number of person of ages ending at odd digits at a particular time.
The indexes which are use in measuring it are:

1. Whipple’s Index.
2. UN Secretariat Method.
3. Myer’s Index.

Whipple’s Index reflects the age heaping on multiple five that is terminal digit (0 and 5) with in the age range from 23 to 62 years hence Whipple’s Index is defined as

Approximate 100 values of Whipple’s Index are for low preference. Well it exceed 175 the quality of age data is very rough as for UN standard. Maximum value of WI is 500 indicating that only zero and five (0-5) ending digit has been recorded.

The method is applied when we have data in 5 years age interval. In this method sex ratio and age ratio for 5 years age group are found up to 70 years. In case of sex ratio successively differences between one age group and the next age are found and their average is taken without considering algebraic signs.

In case of age ratio for sex deviation from 100 are obtained and averaged of unrespectable of algebraic sign.
U.N.S. = 3 {Mean of |SR|} + Mean of |Average Ratio Male| + Mean of |Average ratio of female|

Three times the average of sex ratio differences is then add to the two averages of the deviation of the age ratio from 100 to complete the index. The resulting index is not exact and should by regarded as order of magnitude rather than a precise measurement.

A method of blended population suggested by MYER for obtaining the number of persons at ages ending with digits 0 to 9 starting from 10 years onwards and then starting from 20 years onwards. The population at each digit in first case is given the weights as 1, 2, 3 … 10 and population in second case is given the weights as 9, 9, 7… 0. Then the weighted average of these two populations has found and termed as blended population.

The range for the two populations must be kept same. The data below age 10 and after age 80 is usually not consider for forming Blended population (BP) as the accuracy of information is highly questionable.
The blended population of digit 9 is made the unit of comparison and the ratio is found. A ratio of 1 shows low preference for the digit and the ratio of more than 1 show more preference.
Another way of finding the digital preference is to measure accuracy of the total BP of each digit. A extend of preference is indicated by the preference deviation from 10%. In order to measure the total amount of error in age reporting for a population Myer drive the preference which is simply a sum of absolute deviation from 10%. In the absence of error its index is zero and maximum value is 180.

MIDP= ∑ |column (10 – 11)|

No comments:

Post a Comment