Page 70 - Anthropometry, Apparel Sizing and Design

Developing apparel sizing system using anthropometric data 99

n f ¼ N h∗ n (1.6)

where n m is the sample size for the male stratum, n f is the sample size for the female
stratum, N h is the population size for stratum h (age), and n is the total sample size.

4.2.3 Step 3: Anthropometric survey—Manual method
A preliminary study is conducted before the main anthropometric survey to check the
feasibility of the research approach and to improve the design of the research. ISO
standard 8559:1989 (garment construction and anthropometric surveys—body dimen-
sion) can be used as a guideline for taking body measurements. In the traditional man-
ual technique, measurement tools to be used include calibrated nonstretchable plastic
measuring tapes, height scale with movable head piece, long ruler, elastic 5-meter
tapes, and digital weight scale. Since measuring a single subject can take from
20 to 40min, provision for refreshments for measurers and the subjects should be
made to incentivize them. The survey data collected in the form of categorical (demo-
graphic data) and continuous data were screened and stored in a standard format.

4.2.3.1 Data entry
All the collected data are keyed into a software such as SPSS or MS Excel. The usual
format is to key in the subject’s name and data into a row, which is known as a case.
The body variables are keyed into the columns. The demographic information (cate-
gorical data), gender, ethnic group, age, and geographical area (urban or rural), comes
first followed by columns containing numeric body measurements (continuous data).

4.2.3.2 Data screening
Data screening consists of examination for data entry errors, missing data, or outliers.
The entire data set is filtered to ensure that there are no errors or missing data. Errors
can creep in due to mistakes in keying in the data; these can be rectified by cross-
checking with the raw data. The distribution of data can be tested using graphical
and numerical methods. The graphical method makes use of histograms, while the
numerical assessment is based on values of mean, median, skewness, and kurtosis.
Histograms provide a useful graphical representation of the data. Data are normally
distributed if the histogram shows a Gaussian distribution. This involves evaluating
the bell shape of the data distribution. When tabulating common key dimensions like
height, chest girth, bust girth, waist girth, and hip girth, the mean and median values
should be the same, while the skewness and kurtosis should show values of 0 and 3,
respectively; this indicates that the data are normal (Tabachnick and Fidell, 2007).
Skewness refers to the asymmetry of the distribution. If the skew has a negative value,
this means the data are skewed to the left; if positive the skew is to the right. Kurtosis
refers to the peakness or the flatness of the graph (Hutcheson and Sofroniou, 1999).