EditTypes of Survey Design
The
cross-sectional design is the most commonly used form of survey design. These surveys ask a group of respondents a set of questions at one point in time.
[1] As the name suggests, cross-sectional surveys cross-analyze factors within the data. Popular examples include comparing results among those of different ages, gender, or ethnicity.
[2]It is important to administer cross-sectional surveys for a limited amount of time, as external changes (social, political, cultural) may affect the attitudes of respondents as time progresses. Since, these surveys attempt to capture the attitudes of a sample in a specific timeframe, this shorter timeline is vital in gleaning relevant results.
[3]For example, it would not be advisable for a college or university to administer a survey assessing its campus climate for longer than a few weeks. If a major campus event or social shift were to occur, the respondents who had taken the survey after the event may have a shift in opinion. Then, the data collected before the event will no longer be temporally relevant. It is prudent to think of cross-sectional surveys as snap-shots at a particular point in time.
While cross-sectional designed surveys are perhaps the most popularly used, some argue that
longitudinal surveys are less fallible and more suitable for measuring possible correlations within data. Longitudinal surveys are administered at least two times, and these studies either track change within a sample or strengthen the reliability of the results within multiple samples.
There are three types of longitudinal studies:
- A trend study is a type of survey that is administered to two samples within a population at different points in time. As the name suggests, these samples track trends and concepts and how they change over time.[4]
- A cohort study is essentially a trend study, which tracks changes among a specific cohort. A cohort can be defined as a group or organization, which belong to a similar time period or location and tend to have similar experiences. Often cohort studies involve those born at or around the same time.[5]
- A panel study aims to track change within a particular sample over a period of time.[6]
EditUniversal Survey Weaknesses
Surveys are undoubtedly a useful assessment tool, especially when carefully designed. And yet, even with a large sample size and carefully designed questions, surveys remain invariably less scientific than other more experimental research. Some of the undisputed survey design and implementation weaknesses include:
- Non-response rate is a lack of survey participation. Just about any survey, even the most mandatory government census, is inherently voluntary. Respondents can easily drop out at any time.
- Target population or sample is a portion of the population that is selected as a representative piece of the entire population. While it is statistically possible to find a large enough survey sample to produce a normal distribution of data, there will always be some error in the representation of the population. Since large scale surveys are administered to, sometimes, thousands of people, human individuality and other characterizing aspects of the population will often be ignored. For more info, please see Types of Survey Sampling.
- Socially desirable responses refers to the psychological tendency of respondents to provide socially acceptable survey responses rather than ones that reflect their own true opinions. Some research has shown that this phenomenon stems from two separate human impulses, one being the human propensity to maintain a positive self-image through self-deception and illusion, and the other being a wish to uphold the positive impressions of others.[7]
Besides the more universal survey weaknesses, smaller mistakes in survey design can also be detrimental to the logic or strength of the survey. Some design issues include:
- A leading question is a poorly designed survey question, which insinuates the type of response the researcher expects. This dilemma can go hand in hand with socially desirable responding, as both deal with the issue of authenticity within survey responses.
- A double negative is, in general, an overly complicated way of communicating and should be avoided in survey design. Typically, survey questions that are clear and straightforward will render the most useful and valid results.
- A double-barreled question is a question that contains two attitudinal objects in the same question. For example, “Please rate your overall level of satisfaction with the staff and the environment.” “Staff” and “environment” are both attitudinal objects, and with both in the question, a survey respondent cannot rate either definitively.[8]
References