2. What is stratification?
The stratification is a technique used in combination with
other data analysis tools. When the data, from a variety of
sources or categories, have been grouped its meaning may
be impossible to interpret. This technique separates the data
so that the patterns of distribution of two or more groups
can be distinguished.
3. Objective
The stratification aims to understand a problem or analyze their
causes observing factors or possible and understandable elements.
The data collected from a population is divided into layers or levels.
The aim is to isolate the cause of a problem, identifying the degree
of influence of certain factors on the result of a process.
• Allows you to isolate the cause of a problem, identifying the
degree of influence of certain factors on the result of a process.
• You can support and be the basis on quality tools.
4. Strata
The strata are grouping data with similar characteristics. The
most frequent are performed in lathe a:
• Commodity raw material (dates, suppliers, composition).
• (Sex, age, seniority, experience, career) labor.
• Machinery & equipment (model, antique, technology, hours
of use).
• Time (shift, stations, hours, month, week).
• Location or areas of the company (casino, systems, parking,
manufacturing, administrative, etc.)
• Processes (procurement, technology, projects, etc.)
5. Use in other quality tools
• Stratification can rely on quality tools, the histogram is the
more usual present it.
6. For what do you use it
• Identify the causes that have greater influence on the
variation.
• To understand in detail the structure of a group of data,
allowing you to identify the causes of the problem and
carry out suitable corrective actions.
• Examine the differences between the mean values and the
variation between different strata, and take action against
the difference that may exist.
7. Processing Steps
1. Define the target population.
2. Then we need to organize or arrange the data we have so that there is
consistency.
3. Use a tool (histogram, Pareto, or cake) that allows you to represent the data
collected (this is optional).
4. Select the factors of stratification on the basis of same characteristics (strata).
5. Classify data into homogeneous groups according to selected stratification
factors.
6. Calculate the phenomenon being measured in each category. Graphically
represent each homogeneous group of data. With the help of other tools
(histogram or Pareto).
7. Show the results by homogenous groups of data from each stratification criteria
comparison.
8. Example of Stratification (Pareto graph)
In this example we have a few pieces rejected for defects. We collect
data and build a Pareto chart, but analyzing it we find that the
problems are more or less equally important and none stands out.
9. However, to stratify the data base in the week in which the
parts were produced we clearly see a difference.
10.
11. Stratified Sampling
Stratified sampling is a technique of probability sampling
where researcher divides the population into different
subgroups or strata. Then, it randomly selects the final
subjects of the different strata in proportion.
12. Types of Stratified Sampling
• Proportional stratified sampling:
In this technique, the size of the sample from each stratum is
proportional to the size of the population of the stratum when
compared with the total population. This means that each layer
has the same fraction of sampling.
13. • Example: Suppose you have 3 layers with 100, 200 and 300
population sizes, respectively. The investigator chose a fraction of ½
sampling. Then, the researcher must prove random 50, 100 and 150
subjects from each stratum, respectively.
Estrato A B C
Tamaño de la población 100 200 300
Fracción de muestreo ½ ½ ½
Tamaño final de la muestra 50 100 150
14. • Disproportionate stratified sampling:
The only difference between stratified random sampling
provided and the disproportionate sampling fractions thereof
are. In disproportionate sampling, the different layers have
different sampling fractions. The precision of this design is
highly dependent on the allocation of fraction of sampling of
the researcher.
16. Using the hypothetical example described in the table
above, if you want to carry out a detailed analysis of the
zone 2, one could on sample elements of that area; for
example, instead of a sampling of only 12 items, it shows
130 elements. In order to carry out a more meaningful
analysis, analyzes in detail the zone 2, the size of the sample
for that district should be greater than 12 elements. The
results of the distribution of elements in the sample area.