University of Waterloo STAT 220 - W. H. Cherry - Planned Data Collecting: Experiments and Sample Surveys - Program 13

Sep 17, 2026

Uploaded by Graffin

Planned data collecting focuses on experimental design using blocking and on sample survey design. Blocking is explained as grouping similar units so randomization can occur within blocks, illustrated through strawberry plot treatments. The survey section contrasts samples with censuses, discussing speed, cost, and sources of inaccuracy like non-response and undercount. Simple random sampling is defined via random digits, emphasizing bias avoidance. The material includes guidance on comparison error, lurking variates, and key strawberry plant response variables.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages2

Reading Time5 min

AccessFree to download

Planned Data Collecting

Blocking and randomized block design

Sample surveys vs. censuses

Sampling design and simple random sampling

Clarifications: comparison error, non-response, and measurement error

\#8.11 University of Waterloo STAT 220 –W. H. Cherry Figure 8.3. PLANNED DATA COLLECTING: Experiments and Sample Surveys Program 13 in: Against All Odds: Inside Statistics This program has two main purposes. First, we look at some more complex designs for experiments, especially the use of blocking. Blocks are groups of experimental units or subjects that are similar in some way important to the response. In a randomized block design, randomization is carried out separately within each block. Forming blocks is another type of con-trolin an experiment. The video shows an agricultural experiment in which the experimental units are small plots of land and the treatments are different varieties of strawberries. Because land in different locations varies in fertility, drainage and other ways that affect plant growth, it is helpful to divide the land into blocks containing a few plots in the same location. The differ-ent varieties are then assigned at random to plots separately within each block. In addition to blocking, the program also mentions experiments with more than one factor. In such experiments, a treatment combines a particular value of each factor. The second purpose of this program is to introduce the design of sample surveys. In a sample survey, a sample is selected from the population of all people or things about which we desire information. Conclusions about the population are based on examination of the sample. The video contrasts a sample with a census which attempts to examine every member of the population. Even a census is usually not completely accurate, as a discussion of the U. S. population census shows. Some people are missed by the census, and this undercount is concentrated in the minority neighbourhoods of large cities. Only a census can give detailed information about every part of a population, but for overall information a sample is usually faster, less expensive, and just as accurate. The varied uses of sampling are shown by a look at the manufacture of potato chips: sampling is used at every stage, from the arrival of potatoes at the plant to checking the ﬁnal product on supermarket shelves. The design of a sample refers to the method used to select the sample from the population. The most important statistical principle in sampling design is to select the sample at random from the population. The basic probability sample is a simple random sample, which gives every possible sample of a given size the same chance to be the sample selected. Simple random samples are chosen by labelling the members of the population and using random digits to select the sample. Random sampling avoids the systematic favoritism or bias that often results when a sample is formed by human choice. Copyright © 1985 by Consortium for Mathematics and Its Applications (COMAP), Inc. Reprinted with permission of W. H. Freeman and Company. 1 Comment critically on the last two sentences of the video summary given above. • Simple random samples are chosen by labelling the members of the population and using random digits to select the sample. • Random sampling avoids the systematic favoritism or bias that often results when a sample is selected by human choice. When viewing this Program 13, keep in mind the following clariﬁcations or details of the discussion. Blocking in an experimental Plan means forming groups of units (the blocks) with the same values of one or more non-focal explanatory variates; units within a block are then assigned different values of the focal variate. When answering a Question with a causative aspect, statistical best practice to manage comparison error (to reduce the limitation it imposes on the Answer) is to: * block (to the extent that is feasible in the Question context) on known and measured lurking variates,* use equiprobable assigning (EPA) to manage unblocked, unmeasured and unknown lurking variates, Thus, blocking is used to manage the magnitude of comparison erro

FAQs

What is blocking in experimental design, and why is it used?

Blocking forms groups (blocks) of experimental units with similar values of one or more non-focal explanatory variates, then assigns different values of the focal variate within each block. It helps manage comparison error from known sources of variation.

How does a sample survey differ from a census?

A sample survey selects a sample from the population and draws population conclusions from that sample. A census attempts to examine every member, but it can still be inaccurate due to missed individuals and undercount, often concentrated in minority neighborhoods.

How are simple random samples selected, and what bias does this avoid?

Simple random samples label population members and use random digits to select the sample. Random sampling avoids systematic favoritism or bias that can occur when humans choose the sample.

Sep 17, 2026

Uploaded by Graffin