In: Statistics and Probability
Online surveys have become common due to the ease of distribution. Discuss how random Internet-based surveys really are, explaining your rationale. Is it really possible to report a margin of error for Internet-based surveys? Why or why not? Please provide a link to an outside source that supports your position. Outside sources include materials outside of the textbook, such as a website.
1.
This is in trend due to
(a) rise in internet penetration
(b) relatively low cost in web surveys.
Advantages:-
(a) convenient for respondents in terms of time and pace.
(b) less social desirability bias due to lack of an interviewer.
(c) Allow to use multimedia elements (videos or audio clips).
Drawbacks:-
(a) Dependence on the internet So leads to significant biases like undercoverage and nonresponse.
(b) Access to the internet based on demographic differences based on age, location, education level.
(c) No systematic way to tap people via mail list, phone to complete survey online.
2.
In random samples, there should be equal chance of each member in the population to be selected. But internet sample is generally considered as convenience sample because generally sample consists of people who are filling surveys based on their convenience.
(a) Nonrandom methods are used to recruit potential respondents
(b) Also, pollsters generally use weighting and modeling techniques.
So reporting margin of error for surveys is not possible without random sampling. The "theoretical basis" required to generalize views of sample to large population is absent.
References:-
http://www.people-press.org/methodology/collecting-survey-data/internet-surveys/
https://www.huffingtonpost.in/entry/margin-of-error-debate_n_6565788?ec_carp=689804291912634746