The word survey comes from the Middle English surveyen or the French surveeir, “to look or to see”. In research it means the collection of data from multiple respondents. Survey research refers to asking questions and the collection of data. Survey research is versatile. Although it does not make laboratory experimentation completely obsolete, survey research is very useful and beneficial.
There are three main survey designs: cross-sectional, panel and trend. The cross sectional survey design samples a portion or small percentage of the larger population. The panel survey design is looking for research or answers from a specific group of people that have been placed together for that purpose. Lastly, the trend research design studies popular trends, and the people that are most likely to fall into that category.
In my research experiment on whether or not students at Howard Universitys experiences with financial aid and registration directly effects their rate of financial giving as alumni the survey design that I used was the cross- sectional design. I interviewed 20 different students just to get a small sampling of the general population of Howard University.
In addition there are various methods of collecting research data, such as mail, telephone, and in-person. Each method has its positives and negatives. Collecting research data via mail makes it easier for those conducting the research to put their survey out, however that does not always promise that research participants will enthusiastically return the surveys, and you never really know for certain who your respondents are, which can lead to not fully knowing if your results are truthful and accurate as well as possible fraud. Telephones as a method of collecting data research can look easy on the surface since all was has to do is make phone calls, but most people tend not to like being interrupted in their homes by surveys, and still you can never be truly sure of your results. In-person definitely prevails over all in collecting research data since you have to actually see the person and talk to them, and get a better feel of who it is your interviewing and are they being truthful? Also it is less evasive since the person can either say “yes” or “no” to your face, and does not involve mailing back a survey or an interruptive phone call.
In my research project the method that was most effective for me was in-person, since it was a small research experiment I found it easiest and most feasible to just talk to people one on one.
Conducting a survey with respect to data collection, analysis, training, callbacks, unreached respondents, underage respondents and proportionality for sex and age respondents has the potential to cause some problems. First of course is establishing what it is you want research on and the best manner in which to get it from, one of the aforementioned methods. Then an analysis takes place to group, interpret, and examine the data that was collected. Next training takes place to ensure that all parties involved in the research effort understand the intended goals, and that everyone is informed on the procedures and methods, so people wont do things differently. Then of course there are callbacks to get information on any answers that were not clear. Also depending on what type of research is being conducted and whet the research team decided prior, a method must be established to handle underage respondents, and there should be a proportionate number or ratio of men to women, depending on what is being observed.
A probability sample suggest that the sample that the sample is suited to statistical inference. A non-probability sample suggests a type of street corner or random sample. As it relates to probability samples, there are seven types: person-in-the street: telephone call-ins; convenice sample; popular samples; focus groups; guest samples: and socio-samples.
In reference to non- probability samples there are EPSEM, Equal Probability of Selection Sample; Simple Random Samples, SRS, in which no replacements are used, there is the lottery samples, which selects people at random, everyone has an equal chance; Systematic Interval Method, which uses sampling frames, usually taken at random from an alphabetical list: Stratified Sampling, selects randomly from layers; A cluster sample, which is used instead of a stratified sample when a proper sampling frame is not available; Multi-Level- Area sample is used when something like all 50 states are being sampled, however the percentage of the people must reflect the population of that state. Exit Polling is also an example of non- probability sample, which we have seen a lot in the midst of this current election, where people are interviewed generally after they vote at the polls. Weighting refers to response that are skewed and researchers have too many or not enough of a certain ethnic group. Oversampling, very much like weighting is what happens when a specific group is over or underrepresented.
In my research paper the type of sample used that best describes my methods would be a probability sample, even more specifically it was a conveince sample, because people filled out the survey at their own conveince. On the surface it could appear that the studies when conducted in class would fall under the non- probability category since it would appear that people in our surveys were selected in random, but actually all fit a very specific category. College aged, Howard University students, which is really a very specific group of people.
Measurement is the essence and the strength of the quantitative method. Examples of measurement are around us everyday, taking the form of opinion polls, television ratings, public attitudes, or standardized test scores.
Two types of measurement are semantic and numeric. Semantic measurement cannot be touched or seen, nut must be an inferred or imagined construct. Construct describes an idea formed from a combination of concepts. A numeric measurement is dubbed as bring more precise, since the numbers being involved provides an assignment value to the research and a way it can be prepared and studies.
There are four sources of error in measurement: instrumentation, application, sampling, and random sources. Error in instrumentation occurs when questions by the surveyor are poorly written or applied. Errors in application occur when there are errors in the way the study is applied. Sampling errors occur in two different ways (a) an incorrect drawing of a sample, or (b) when a sample is properly drawn, but the cases fall outside of the margin error. Random source errors occur when carelessness happens on the part of respondents.
In measurements there are internal validity, as well as external validity. Internal validity asks the question, “Have you measured what you thought you measured?” External validity ask the question, “Does the survey help generalize the population from which it was drawn?”