Surveys Flashcards
Which are the three types of surveys’ design?
We can have cross-sectional, repeated cross-sectional and Panel
Which are the cons of postal surveys?
High costs, time delay of respondents and low repsonse rate of participants
Which are the pros and cons of telephone surveys?
They allow for conducting interviews without travel costs, there is no tight control of interviewers and target specific geographical areas. They also allow for open answers. Among the cons we find the labor intensity of this method.
Why would someone use face to face surveys?
They are expensive in terms of travel and time costs but they allow for deeper information and interviewer’s mediation
Which are the pros and cons of internet surveys?
They are very fast and cheap. Their reliability is the same as the other modes of administration. One of the disadvantages is that there is a lack of a representative sample.
Which are the main issues regarding surveys?
- Open vs closed answer questions (where there is a trade off between cost and precision)
- Ranking vs scaling
- Rating can be bipolar with 7 points or unipolar out of 5 points
- Testing can be conventional, behavioral or cognitive
Which are the 4 characteristics that a sampling frame should have?
a. Relevant to the study
b. Complete
c. Precise
d. Up to date
What are the differences between probability and non probability sampling?
In the first type every unit in the population has a non-zero probability of being selected and it can be determined. In a non-probability sample for some units this probability is either zero or can’t be determined.
What are the 3 types of probability sampling?
- Simple random sampling
- Stratified sampling
- Cluster sampling
What are the characteristics of simple random sampling?
a. For each sample size each unit and each possible combination of units in the population have equal and known probability of being selected
b. Known population
c. Sampling frame
d. Random selection
What are the characteristics of stratified sampling?
a. Division of the population into subgroups which have to be mutually exclusive and collectively exhaustive
b. Random sampling from each stratum in proportion to the population
c. More precise estimates than with simple random sampling
What are the characteristics of cluster sampling?
a. Random sampling of clusters of units
b. Inclusion of some or all the other units within each cluster
c. The clusters must be pre-existing and naturally occurring groups
d. No independence of observations
What are the 2 types of non-probability methods that we have seen?
Quota sampling and Snowball sampling
What are the 4 characteristics of Quota sampling?
a. Division of the population into meaningful strata
b. For each stratum reasearchers set a quota of units in proportion to the population
c. Nonrandom selection of units to fill the quota
d. Often used in market research with face-to-face interviewing
What are the 4 characteristics of snowball sampling?
a. Multi-stage sampling
b. The sample snowballs in size through a process of reference from one respondent to the next
c. Among the advantages we have that the accumulation is quick, nominators reference the researchers
d. Primarily used in small scale exploratory samples or when units are difficult to contact
What are the main response rate determianants?
- Characteristics of the respondent
- Topic/subject of research
- Researcher identity
- Way/means of contact
- Social context
- Good survey practices
How is the adequacy of the response rate determined?
- No systematic differences between respondents and non respondents
- In line with expectations and similar surveys
- All good practices used
- Still high risk that non response biases the results
- Two type of non response (refusal and non contact)
Non response is a major validity threat (think about voting Berlusconi surveys)
Which are the main 3 approaches to manage missing items?
- Use only non missing item data
- Input mean values
- Estimate missed data from other covariates
