Survey Work and Design

1. overall survey design and preliminary planning

2. questionnaire design and pre-testing

3. final survey design and planning

4. sample selection and data collection

5. data coding, data file construction, analysis, and final report

What are the 5 stages of survey work?

sampling and non-sampling

What are the two types of error in survey work?

sampling error

• directly attributable to the fact that not all the population units are observed

• extent depends on sample size and variability of the population

sampling error

only asking selected college students re: online learning readiness

non-sampling error

All error sources unrelated to the sampling of the respondents.

non-sampling error

coverage bias, selection bias, nonresponse bias, interviewer error, response error, coding error

sampling error + non-sampling error

What is the formula for the total survey error?

• Determine appropriate sample size

• define target population

• attempt to maximize population coverage

• select sample that fairly represents entire population

• obtain data from as much of the selected sample as possible

• properly train and supervise interviewers

• use good questionnaire design

• exercise decent control over coding and data entry

How do you minimize errors?

target population

Population from which information is ideally acquired from.

survey population (sampled population)

Population from which a sample is actually taken.

• List of elements covering target population

• e.g. list of all registered voters in the province

• each unit must be counted only once, and must be distinguishable from other units

• provides up-to-date information

A sampling frame must posses the following characteristics, these are?

Sampling Unit

Unit which is part of the sampling process.

Observation unit

Object on which the measurement is actually taken (element)

list of HEIs, 25 HEIs in Western Visayas, and deans/chancellors

It is argued by some that the overall quality of graduates in higher education institutions or HEIs (colleges, universities, and the like) in the Philippines, particularly in Western Visayas, is declining. A survey of these HEIs will be undertaken to investigate this claim. A random sample of 25 higher education institutions is taken from the list of all HEIs in the region. Then, the chancellor/deans of each selected university is interviewed.

What is the sampling frame, sampling unit, and observation unit in this situation?

more confidence can be placed in the conclusions drawn

In a sampling plan, the larger sample means that?

less reliability

In a sampling plan, greater diversity means?

True

TRUE or FALSE: In a sampling plan, bias must be avoided when selecting a sample.

False

TRUE or FALSE: In a sampling plan, there must be a selective factor.

True

TRUE or FALSE: In a sampling plan, the researcher must consider the efficiency of the scheme (reliability in relation to unit cost.)

False

TRUE or FALSE: In a sampling plan, mere size assure representativeness in a sample.

Sampling methods

• process of selecting a sample

• done to estimate and test a claim about an unknown parameter from a well-defined population

1. no need to conduct a full census to obtain information

2. less cost

3. timeliness

4. accuracy

5. detailed information

6. destructive testing

What are the reasons for conducting a sampling method?

Destructive sampling

Carried out to the specimen's failure in order to understand its performance or material behavior under different loads (e.g. light bulbsarried out to the specimen's failure in order to understand its performance or material behavior under different loads (e.g. light bulbs)

1. procedure should be easily implemented and practical

2. for statistical analysis, a sample must represent the population and reliability must be measurable

What are the 2 criteria for acceptability of a sampling method?

Probability of selection (P₁)

Chance, ranging from 0 to 1, that each unit in the population is included in the sample, usually determined from the elements of the sampling frame, where N - populations size and n - sample size.

Non-probability sampling

• Not all the elements in the population have the chance to be included in the sample

• Pi not specified

• Researcher cannot assert that the sample is representative of the larger population; statistics must be descriptive only

• extreme difficulties in locating or identifying subjects

• only few are willing to be interviewed

• probability sampling is more expensive to implement

What are the useful cases to apply non-probability sampling?

purposive, convenience, quota, snowball

What are the 4 types of non-probability sampling?

Purposive

• judgment sampling

• sets out to make sample agree with the population with regards to certain characteristics

Purposive

Which type of non-probability sampling is this?

Conducting pilot testing to gather feedback from the market before launching a new product; researcher chooses expert wine tasters to do this.

Convenience

• Haphazard/accidental sampling

• Choosing units which are at hand or are convenient to interview

Convenience

Which type of non-probability sampling is this?

Standing at a grocery store and asking customers if they are willing to answer questions re: satisfaction

Quota

Sample has the same proportions of individuals as the entire population with respect to known characteristics.

Quota

Which type of non-probability sampling is this?

10% of the population is made up of indigenous peoples → 10% of sample are IPs

Quota

Sample has the same proportions of individuals as the entire population with respect to known characteristics

Snowball

Which type of non-probability sampling is this?

Investigating lifestyle of those engaged in illegal prostitution → asking prostitutes to refer other people

Probability Sampling

• Every element of the population is given a known nonzero chance of being selected in the sample

• Used if main objective is to make inferences about a population

• specifies rules for both sample selection and estimation

1. simple random sampling (SRS)

2. stratified random sampling

3. systematic sampling

4. cluster sampling

5. multi-stage samp[ling

What are the 5 types of probability sampling?

simple random sampling

• Selecting a sample size of n, giving each unit an equal chance of being included in the sample

• Can be one with replacement (SRSWR) or without replacement (SRSWOR)

• Selection of units uses some random process,

• theory involved is easy to understand

• estimation methods are simple and easy

What are the 2 advantages of simple random sampling?

• sampling frame is necessary

• sample chosen may be widely spread → higher transportation costs

• sample chosen may not be truly typical of the population if it is heterogeneous

What are the 3 disadvantages of simple random sampling?

Stratified Sampling

• Population is stratified/divided into more or less homogeneous subpopulations (strata) before sampling is performed

• SRS is performed in each of the stratum

equally and proportionally

Allocating n among the strata can be done in 2 way, thee are?

equally

This is the formula for stratified sampling in which way?

proportionally

This is the formula for stratified sampling in which way?

• may increase precision of the estimates of the characteristics of the population

• more comprehensive data analysis

• administratively convenient to implement

What are the 3 advantages in stratified sampling?

• listing of the population for each stratum is needed

• stratification of the population may require additional prior information about the population and its strata

What are the 2 disadvantages in stratified sampling?

Systematic Sampling

• Selection of samples involves taking every kth unit from an ordered population.

• First unit (r) is selected at random

1 and k

In systematic sampling, if k is an integer, r is between…

1 and N (circular systematic selection)

In systematic sampling, if k is not an integer, r is between…

• drawing sample is administratively easy to implement

• possible to select a sample from the field without a frame

What are the 2 advantages in stratified sampling?

If periodic irregularities are found in the list, a systematic sample may consist of similar unit,

What is one disadvantage of systematic sampling?

clusters

Mutually exclusive subpopulations which together comprise the entire population

Cluster Sampling

• Involves selecting a sample of distinct groups • sample clusters may be chosen by SRS or systematic sampling

• Difference: clusters are preferably formed with heterogeneous units.

Cluster Sampling

What type of probability sampling is this?

From a hospital's 12 departments, six were randomly selected; every intern in these six sampled departments was given a questionnaire (+followed-up) until all of them successfully completed the questionnaires.

• No need for the population list of elements

• Administratively convenient to implement

What are the 2 advantages of cluster sampling?

Difficulty of estimation procedures

What is one disadvantage of cluster sampling?

Multi—stage sampling

• Selection of sample is accomplished in two or more steps

  • 1. population is divided into a number of primary/first-stage units from which a sample is drawn

  • 2. within the sampled first-stage units. a sample of secondary/second-stage units is drawn

  • 3. further stages may be added if desired

Multi-stage sampling

What type of probability sampling is this?

From the 12 departments in the hospital, 4 were randomly selected; from each sampled departments, 10 interns were given questionnaire

Simple two-stage sampling

1. Units are grouped into N primary sampling units (PSUs) and a simple random sample of n PSUs are selected

2. Each of the n PSUs selected with Mi elements, a simple random sample of mi units called secondary sampling units (SSUs) will be obtained.

reduced listing and transportation costs

What is one advantage of multi-stage sampling?

• difficulty of estimation procedures

• sampling procedure entails more planning before actual selection is done

What are the two disadvantages of multi-stage sampling?

True

TRUE or FALSE: A stratified sample yields more reliable results than a SRS of same size.

False

TRUE or FALSE: Cluster sampling may be expected to yield a more reliable result over a SRS of the same size

True

TRUE or FALSE: A small random or stratified sample is much superior to a larger but badly selected sample