Quantitative Marketing Research Methods
Qualitative research uncovers the culture, history, and behaviors of a population. Culture need not be that of an entire society; it may be limited to a profession, organization, or family.
History need not cover an entire era; it may be restricted to a few months, weeks, or days. Behavior need not include all activity; it may simply be a workflow, function, or gesture.
One of the questions researchers face when developing a research plan is who will be chosen to be research participants. The word ‘population’ is commonly used to define everyone of interest who could be possibly included in a research study.
Researchers may define a population by geographic area. In addition, they may also define a population using such demographic data as age, gender, income or ethnicity.
Because marketing promotion is often targeted based on psychographic segmentation, researchers may also define a population-based on interests, values or lifestyles. Product usage can also be a means for defining a population, such as nonusers, occasional users and frequent users.
These variables can also be used in combination. The resulting population may be very large, such as daily users of toothpaste who live in Germany, or very small, such as people over the age of 70 in the UK who are attending university as full-time students.
A standard process should be followed when selecting a sample. First, the population needs to be defined along with the frame, which are the people in the sample. A method for sampling from people in the frame will be the next issue, along with determining how many people need to be research subjects. Finally, the research subjects will be chosen.
Decide on the population to be studied
Develop sampling frame
Choose the sampling method
Determine sample size
Choose sample participants
Using a census
When conducting quantitative research, researchers are attempting to support a fact or hypothesis. This fact might be how many consumers would buy a product, what type of new promotion would work best, the effect of pricing on purchase behavior, or the best store in which to distribute a product.
If a 100 percent accurate answer is needed, researchers must ask every person who is included in a population.
Asking everyone in a population is called a census. Conducting a census is possible if the number of people from whom information is needed is small and all of the members can be reached.
For example, a professor might wonder if the students in his class study each evening. A census could be used for surveying all the members of this class about their plans for the evening.
However, if one student is absent due to illness the census would be incomplete. Even if everyone is present, if one student refuses to answer this question, again, the census would be incomplete. Obviously, there are problems with trying to conduct a census, so instead a sample is used.
Using a sample
In fact, most research will involve the sampling of a population rather than a census. This is because the cost of trying to reach every possible research subject is simply too high. For example, researchers might be developing a new toothpaste flavor that they plan to introduce in France.
Asking everyone in the population will simply take too much time and money. In addition, the company commissioning the research does not need to know with 100 percent accuracy that the new flavor will be acceptable.
Less accuracy will still provide the information as to whether the flavor will be acceptable to a significant percentage of adults in France. Thus, asking a sample of the population will still provide an accurate enough estimate of consumer preference and this will allow the company to proceed with product development plans.
In this type of research situation, researchers will save money and time by surveying a sample of the total population.
The data obtained from asking a sample of a population can never provide as accurate an answer as a census of everyone. In the example mentioned above, the professor who wants to learn how many students study at night could ask all of his students.
However, if the same professor only asks 10 out of the 30 students enrolled in his class, there is the possibility that the answers provided are not representative of everyone. This is called sampling error.
This error might result if the professor asked those 10 students who always attend class. They would respond positively that they study nightly. The professor would then believe that all his students worked hard.
On the other hand, he could also include too many of those students who didn’t attend class regularly, get the response that they didn’t study and, as a result, will believe that all his students are irresponsible. No sample can perfectly represent a total population.
Other types of errors that result from using a sample are called nonsampling errors. These errors do not result from the fact that a sample was used instead of a census.
Instead, those designing and conducting the research cause these errors. Because they are caused by human error, nonsampling errors can be controlled. Types of nonsampling errors include specification errors, design errors and selection errors.
Specification errors result when the wrong population is specified in a research design. A university admissions office may wish to study students’ views on what type of art classes should be offered.
For convenience, the admissions office might choose to survey students on the campus where the admission office is located. However, if most of the students who attend classes on the campus are business students, the survey results will be flawed.
Business students may have little interest in art classes. They may not even know what type of art classes could potentially be offered. Their opinions will have no comparability to what students interested in the art might want. Therefore, the wrong choice of the population will result in data that are not useable.
Design errors occur because of human failure and include data recording, data entry, data analysis, and nonresponse error. When the same researcher conducts many surveys, it is possible that some responses may be recorded incorrectly.
With self-administered surveys, respondents may also make mistakes in recording data. To minimize data recording errors, survey takers should be properly trained and also motivated to do their job correctly. Survey forms should also be designed so that respondents will be able to easily find the response they wish to indicate.
The answer space should be designed so that once an answer is indicated, it will not be confused with other possible responses. For example, if the question requires a yes or no answer, whether a respondent is to make a tick, cross out, circle or underline should be clearly stated.
In addition, the blanks for completion of open-ended questions should be far enough apart to ensure there is no confusion when an answer is indicated.
Selection errors occur when the correct population has been chosen, but the sample taken from the population is not representative of the entire population.
For example, a university may decide to ask students who are enrolled in art classes what additional classes they should offer. A selection error occurs when the students who are asked are still not representative of the whole student population.
For example, a young male student employed to conduct the survey might view the task as a good opportunity to chat up lots of young women. As a result, a much larger population of women than men will be included in the sample.
If women want different types of classes than men, this could result in the survey returning inaccurate information due to selection error.
Determining the Target Population and the Sample Frame
One of the most critical steps in the quantitative research process is determining the target population to be researched. It is not unusual early in the research process for both marketing researchers and managers to speak in generalities. They may discuss the need to research the attitudes of current customers.
They may also discuss wanting to research potential customers who are older, retired couples, for example. At first these might appear to be reasonable research requests. However, marketing researchers will understand that both of these definitions of a population are too vague.
A target population always needs to be clearly defined so that the correct individuals are included in the sample frame from which the final participants will be chosen.
This is especially true of the population and the sampling frame for online surveys. Because geography does not need to be considered, when defining a population, it is easy to do so too broadly resulting in a higher nonresponse rate.
An example of this process would be the management of a football team researching how a rise in ticket prices would affect attendance at games. The first decision that must be made here in defining the population is whether to research the effect on fans who currently attend games or potential fans.
If the decision is to research fans who are currently buying tickets, the next step in defining the population is to discuss whether the population includes everyone in this category or a smaller group. Management, with the assistance of market researchers, will need to define what they mean by ‘current’, ‘fans’ and ‘buy tickets’.
Defining current ticket buyers at first seems self-explanatory. However, it is important to define the time period meant by ‘current’. Does this refer to just the most recent game, or to the past month, the current season or the last one or two years?
The answer to the question will depend on whether attendance is consistent or varies over time. If management is aware that the same group of fans attends each game, then a short time frame is acceptable. However, if attendance varies for a longer time period will be needed.
The next question that needs to be addressed is if a ‘fan’ includes those people who attend occasionally or those who attend frequently. Management may decide it is more important to learn the attitudes of those people who attend frequently.
Finally, does it make a difference if those who attend frequently buy season passes or purchase tickets for each game individually? A final definition of the population could be those people who attended at least 75 percent of the games during the last season and bought season tickets.
Once a population has been defined, marketing researchers will be able to decide whether they can do a census of the population or whether they will need to choose a sample.
If the population is small and the research question demands 100 percent accuracy, a census could be conducted. However, in most situations, researchers will decide to use a sample of a population to participate in the survey.
In order for a sample of participants to be chosen, researchers must first have potential access to everyone in that population.
For the football team mentioned previously this list of potential participants in the population, or sample frame, might be developed using data from the football team’s ticket office database.
Using this database, researchers will be able to construct a list of ticket buyers and the frequency of ticket purchases for the past season. Researchers can use this list to look for fans who have purchased tickets for 75 percent or more of the previous year’s games. This method is still not foolproof, however.
There may be ticket buyers who paid cash, or who refused to provide their names when they bought tickets. However, the frame should include a high percentage of people, if not everyone, who make up a population.
Probability vs. nonprobability sampling
Probability sampling uses techniques that result in an ability to calculate exactly the probability of a single person in a sampling frame being chosen to participate.
This probability is based on the number of total people in that sample divided by the number of total people in the population. If the population is a known number, this is quite easy to calculate.
A survey that includes 250 people, out of a population of 1,000, means that every individual in the population has a 25 percent probability of being included. The methods of conducting probability sampling include simple, stratified, systematic and cluster.
Of course, for most research studies the total number of people who are in a population at any given moment is unknown.
For example, even at a university, it is impossible to know exactly how many students are attending classes based on the last registration list. Since that list was compiled, students may have withdrawn or new students may have transferred in.
There may also be students who have stopped attending classes but have not yet notified the university. Probability sampling can still be conducted with the probability calculated based on a reasonable estimate of the entire population.
The methods of probability sampling from which researchers can choose include simple, stratified, systemic and cluster. These vary in the randomness of the resulting sample.
They also vary in their complexity and the time and effort it will take to construct a sample. Sometimes researchers may decide to use the services of a commercial provider.
Simple random sampling
The most easily understood method is simple random sampling. In this method, each sampling unit has an equal chance of being chosen to participate in the research study. The probability can be calculated by dividing the number of people in that sample by the total number of people in the population.
For example, a university’s administration may wish to conduct a study of all of its students. They decide to survey 300 students out of a total population of 15,000. Thus, the probability of being chosen is 2 percent or 300 divided by 15,000.
While it is easy to calculate the probability of being chosen, what is difficult is choosing the 300 participants so that this is done randomly. The university’s administration could go to the entrance to the main building and ask the first 300 people to come in the door to participate.
However, this would not be random because participation would be determined by who happens to have a class right after that time period. Even just going down the list of student names choosing every fiftieth name will not result in a truly random list because someone has to choose a starting point, so this first name is not random.
The problem is easy to understand if a low-tech solution to the problem is described. One way to achieve a random sample of student names is to write each student’s name on a piece of paper and put all of these in a hat. Someone who is not looking then draws out the names.
Each name will have the same 2 percent chance of being drawn. After a name has been drawn, a blank piece of paper is put back into the hat to represent that name, so that each time there is still one chance in 15,000 of being drawn.
Of course, this would be somewhat cumbersome. Most researchers will use a random number generator or table.
A random number table consists of rows and columns of numbers that have no order or sequence. The first step in using the table is to give each unit, or name, in the population a number. Students could be numbered from 1 to 15,000.
Then the researcher uses the list to choose the number of the first participant. The researcher will use the next 299 numbers in the random list to determine who is included in the sample. Other methods to ensure that a sample is truly random are using automatic dialing, a software program or a calculator that generates random numbers.
Even though it is a much simpler method to use, systematic sampling will result in a sample that is almost random. In systematic sampling, after the population has been determined, all units in the population are listed and counted.
A skip interval is then calculated by dividing the total population by the sample size and this interval is used to choose who will be included in the sample.
Using the example described above, the skip interval is 50, or 15,000 divided by 300. A random start point is chosen and then the skip interval is used to count off every fiftieth name on the list which is then included in the sample.
Calculating the skip interval
At first glance, this may seem to be as random as simple sampling, but this is not so. The starting letter of family names is not random but often determined by ethnic and national origin.
For example, if a university has a large number of students of Korean ethnicity, Korean student names would not be scattered randomly throughout a list of student names.
A very common Korean name is Kim and students named Kim would not have an equal chance of being chosen because their names would be grouped together on the list. Using student numbers would also be a problem as these are implemented as students enter a university, resulting in a bias by time period.
One solution to this problem is to randomize the list before using the skip interval. If the list is computerized it can be sorted by a more random factor than a name or student number. Another method is restarting the skip interval and counting once or twice when choosing the sample. Each time this happens a new random start would be chosen.
Stratified sampling is used when researchers believe that answers will vary depending on the demographic, psychographic, geographic or usage characteristics of each person in a population.
There are two main reasons for choosing a stratified sample. First, the population may be skewed in such a way that it is difficult to obtain a random sample using either a random or systematic method. In this case, stratified sampling is used to increase the randomness of the sample.
Second, the research study may call for comparing results between specific groups within a population. The main reason for using stratified sampling is to ensure that any differences are diminished by the sampling procedure. Another reason is when the research study is designed to learn more about the differences between groups.
Most populations can be divided into smaller groups based on shared characteristics. These characteristics can be based on demographic factors such as gender, age, income, and education level. They can also be grouped by psychographic characteristics such as interests and lifestyles.
Geographic grouping may be an important option to consider if the answer to a research question might vary depending on where participants live. Finally, product usage status (such as nonusers, occasional users and frequent users) might be of interest to researchers.
The example of using stratified sampling when conducting a survey of university students can be used to clarify this method. Perhaps a university’s administration is interested in learning the views of students from all ethnic backgrounds.
The administration, therefore, wants to ensure that their sample includes at least some students from all the ethnic groups represented on the student body.
Using a simple random or systematic sampling procedure might include students from all the ethnic groups or might not, because each student at the university has a random chance of being chosen.
Using a stratified sample allows marketing researchers to examine each of the strata separately. Initially, researchers must decide how many participants will be chosen from the population as a whole. The researchers must then decide on how to distribute the number of total participants among the different strata.
Using proportionate sampling, each stratum will be given the same proportion of sample participants as the stratum’s proportion of the entire population.
For example, the strata may have been developed using ethnic groups of students for a university in the USA that would include Americans of European descent, Asian-Americans, African-Americans, and those from a Hispanic ethnic background.
The university also possesses census data that give the proportion of each group in the total population. This information is used to calculate how many participants will be chosen for each group if the total population is 15,000 and the sample is 300.
Mean of population
To implement proportionate sampling, a separate list of units or names must be developed for each group and then the names must be chosen using systematic sampling.
However, the university, in this case, may choose to use disproportionate sampling if they decide that they need much more information about groups other than the majority.
Cluster sampling also divides populations into groups. However, cluster sampling differs by reason of why a population is divided and the number of resulting groups. In stratified sampling, groups are based on a few demographic, psychographic, geographic or usage characteristics.
The purpose here is to ensure that at least a few individuals from each group will be included. In cluster sampling, a population is divided into many groups. Rather than being different, the aim of the resulting groups is that each will include individuals who represent the total population.
The purpose of cluster sampling is to obtain economically and efficiently a sample from a large population that provides an accurate representation of an entire population. Such a method could also be combined with other sampling methods to form subgroups.
Cluster sampling is often used when it is impossible to determine the exact number of individuals in a population. One example would be if a university’s administration wanted to conduct a telephone survey of potential students living in the city.
While the university could determine the exact number of current students, they would never know how many potential students existed. Another problem is that a telephone survey of the entire geographic area from which the university draws students would be too expensive. As a result, they may decide to cluster sample.
One common means of clustering the population is to randomly divide a geographic area. If the same university believes that potential students are evenly dispersed throughout the city, they can subdivide it into several geographic areas. The first step would be to choose how the city was to be subdivided, such as by postal code.
Assuming that there was an even chance that potential students were living in any of the postal codes, the next step would be to randomly choose which of the postal code areas would be included in the cluster sample.
An additional step can be added here to the cluster sampling technique. Once the postal codes that are to be included in the study have been chosen, another probability technique could be used to sample individuals within each of the chosen areas.
For example, the university may know that potential students are scattered evenly throughout the metropolitan area. However, they may also know that not all of the residents are likely to be potential students.
They may wish to include in the sample in each cluster only those people within the population who are members of a certain age range.
Determining a Sample Size
Once a sampling procedure has been chosen, researchers must then decide upon the number of subjects that should be included in the research to ensure that the results are representative of the entire population.
The management of the company commissioning the research should understand that more is not always better when determining the number of research subjects to be included in the study.
Research costs money and the more research subjects that are involved, the more the study will cost. If carefully chosen, a small sample from this population can be reasonably representative of the whole.
There are a few concepts that must be considered when the sample size is determined. First, the more variation there is in a population as a whole, the larger a sample will need to be.
For example, when planning the design of a new student center a university will need to know the recreational preferences of its students, such as whether they are sports, arts-related or video gaming. The university may have a fairly homogeneous student population of the same nationality and age.
Therefore, it can be assumed that their interests might be similar and it will take a smaller sample to determine their preferences. However, another university might have the same number of students but that comprise both traditional college-age students as well as older adults who come from different countries.
In this case, it can be assumed that a larger sample will be needed as it would be expected that the interests of the students will vary widely.
A second factor that must be considered in this example is the precision in the range between the survey answer results and the reality of the population as a whole. A survey question might ask how much each student would be willing to pay to use the new center. From these responses, an average fee can be calculated.
The university must then decide on how large a range or interval they are willing to tolerate between the average tabulated from the survey responses and the true answer that would be known only if everyone in the population was asked this same question.
A third factor that must be considered when determining sample size is the need for confidence that the research findings reflect the reality of the total population.
Total accuracy requires a census of all participants. Since this is not possible, the university must decide what level of confidence they need that the survey data accurately reflect the whole. The larger a sample is, the higher will be the confidence that this is so.
Factors to consider when determining sample size
Variation within the population
More variation means a larger sample
The precision needed between a range of given answer and true answer
A smaller range means a larger sample
Confidence that research findings represent the population as a whole Higher confidence means a larger sample
Normal distribution and variation
To understand the variability of data it is not necessary to have an advanced knowledge of a subject. However, when discussing variability, the terms ‘parameter’ and ‘statistic’ need to be understood. If a market researcher could conduct a census on how much students spent on entertainment last weekend, a true mean or average could be calculated.
This number is called the parameter. Of course, conducting a census is too expensive and time-consuming so the researcher would survey a sample of students. From these data is calculated a number, or statistic, which is used as an estimate of the entire population.
Perhaps the researcher wants to further make sure that this statistic is an accurate estimate, so the survey is conducted a second time with a new sample. The same sampling procedure is chosen so the same number of subjects are asked the research question. However, the sample will now consist of different individuals.
The researcher will again need to calculate the statistic. If the researcher were able to survey every possible sample in the population, this would result in a series of means. If all the resulting means were examined, it would be found that some answers will occur more frequently than others.
This similarity results from the fact that members of a population will have characteristics in common, which in turn will mean that many will behave in similar ways. However, there will also be some whose behavior will vary.
When these similarities and differences are shown visually they result in a graph where the most commonly resulting mean is in the middle with diminishing returns on either side for numbers that are higher and lower. This is the classic ‘normal’ or bell distribution curve.
Fortunately, most data will fit a normal distribution curve and, therefore, the variability within a population can be estimated. Normal variability is considered to be plus or minus three standard deviations away from the true value. Therefore, a researcher can take an estimated range of variability and calculate the variation.
1. Including everyone in the population in a research study is called a census. But to include everyone in the population is expensive and impractical for research purposes.
Sampling is used when only certain people in the population are chosen to be part of a research study. When sampling is conducted, it is inevitable that there will be errors.
Errors result when the chosen sample is not absolutely representative of the whole population. Mistakes can also result due to human error when recording or enter information.
2. The first step in developing a sample is to determine the targeted population. It is important to narrow the definition of a population so that only those participants who can answer the research question are included.
A list of names from this population is called a ‘sampling frame’. The next step in the process is to select a sampling method, which can be based on nonprobability or probability.
3. Probability sampling includes simple, stratified, systematic and cluster. The methods vary in the randomness of the resulting sample. The probability of someone being chosen during sampling can be calculated by dividing the sample size by the total number of people in the population. Simple random sampling provides the most random sample.
Systematic sampling is easier and results in almost as random a sample. Stratified sampling divides a population into smaller groups based on shared characteristics and with a certain number of subjects chosen from each group.
In cluster sampling, the population is divided into many groups that are similar, rather than dissimilar as in stratified sampling. Cluster sampling is an economical means of sampling a large population.
4. Factors to consider when determining a sample size include the acceptable range of error of the estimated answer from the true answer, the level of confidence needed that the answer is within that range and the variation within the population.
The sample size can be calculated using these three numbers. The sample size for estimating a proportion will be higher than the sample size for estimating a mean.
Survey Research Methodology
Surveys are a quantitative research methodology that consists of a set of questions with predetermined answer choices from which participants must choose.
In addition, some survey questions may be open-ended, where participants are able to respond in their own words. Surveys can be administered by a researcher, either in person or over the phone.
Surveys can also be self-administered whereby they are sent via the postal service or put online. Whatever the form of survey administration, the survey design methodology and the questionnaire development process remain basically the same.
Uses of survey research
Surveys are most effective when obtaining information if marketing researchers already know something about the research problem at hand. This information will be needed as researchers must anticipate participants’ responses when they write possible answers to questions.
This knowledge can be the result of secondary research on the external environment or already conducted qualitative research. There needs to be some initial knowledge of the research problem so that there is enough information to write the correct survey questions and answers.
Surveys can be used to gather information on any aspect of the marketing mix including the customer, product, promotion, price and place, satisfied with its products, or it may be used to determine the likelihood of a new target market segment being interested in purchasing.
The effectiveness of a promotional campaign can be discovered by asking if a consumer has purchased a product as the result of seeing or hearing an advertisement or sales incentive offer.
Technology has increased the number of ways that marketing messages are being delivered and a researcher might be interested in how consumers are now receiving marketing messages.
Of course, this researcher must first be aware of all the new forms of social media that are being used to communicate such messages, otherwise, these won’t be added to the possible answers to a question.
Survey questions can also be written to ask about the effect of price changes on purchase likelihood. In addition, where consumers could most conveniently purchase a product could be discovered through a survey question.
Questionnaire development needs
When designing a questionnaire for a survey a researcher must keep in mind the needs of a number of groups that may have an interest in what questions will be on the survey form. Each group will view the completed questionnaire from their own unique perspective.
First, the management of the company that has commissioned the research will be concerned that the survey questions will provide the information needed to answer the research question.
In addition, they will also want questions on demographic and geographic characteristics that will allow for comparisons and contrasts between groups of consumers.
The research participants will want the questions to be ones they can answer and that these are worded in a way that is easily understood. Participants will also want the form designed so that it is simple to record their responses.
Those administering the survey (if it is not self-administered) will want the questions to be written clearly so that participants do not need to ask for clarification.
Management – that it asks the right questions
Participants – that it includes questions that can be easily answered Surveyor administrator – that the questions are clearly written
The Questionnaire Design Process
Once management has determined the research question, the questionnaire design process will begin with a meeting between the researchers and management.
At this meeting, researchers and management will discuss the topic areas that need to be covered in the survey to provide the data needed to answer all aspects of the research question. The researchers will then write a draft survey form. This draft will be reviewed by management for the appropriateness of the questions.
Another issue that needs to be addressed is how the process should be adapted if the survey study is to be conducted in a different country.
This process will be repeated until everyone is satisfied with both the questions and the answers. The questionnaire will be laid out in its final form. The last step will be to test the survey with potential participants.
Questionnaire design process
1 Meet with management to determine question topic areas
2. Write a draft survey
3.Review draft with management
4.Layout questionnaire design
Question topic areas
Before the design process can begin, management must have determined the research question. It may be that researchers will simply be given the research question and told to write a survey form. However, it would be preferable for management and the marketing researchers to work together during the process.
Marketing researchers should meet with the management of the company commissioning the research to discuss those topic areas for which questions will need to be written to answer the research question. In addition, any questions needed to qualify participants for the study and any demographic questions need to be discussed.
For example, a survey designed to answer the question of what factors motivate young people when purchasing a car might include question topic areas on preferred price, design and power.
Several sample questions such as ‘How important is the price when making a purchase decision?’ will be written for each question topic area to help management visualize the final questionnaire. At this meeting, the management of the company may suggest questions to the researchers.
However, they will only be viewing the question from a consideration of their own needs. Sometimes management will propose questions that are too broad and which would only result in general data that cannot be turned into useful information. It will be the researchers’ responsibility to explain that the question is too broad or too complex to be answered.
Writing the draft and management review
Once the topic outline and sample questions have been approved a draft survey needs to be created. Researchers will then review the form with the management of the company to determine the appropriateness of the questions and any additional questions that should be asked. This process might be repeated any number of times.
While it is management’s responsibility to decide on the topic areas, each time the form is reviewed it is the marketing researchers’ responsibility to explain why certain questions should be included and why some should not. The exact wording of the questions also needs to be reviewed.
This process must be repeated until everyone is comfortable with the survey questions. Trying to save time at this step in the process may result in wrong or badly worded questions being asked and unusable data. In addition, the answers to the questions must also be developed.
Other issues that will need to be discussed at this stage in the process will include translation issues that must be addressed even for single country surveys. In addition, the organization and design of the physical layout of the form need to be discussed.
Coding the question answers
Surveys are a quantitative research methodology. This means that statistics such as averages and percentages will need to be calculated during the analysis stage.
To do so, answers in words will need to be converted into numbers. These numbers are referred to as ‘coding’ and should be designed as part of the survey form and not added later.
This will save time and effort when the data are entered into a computer software program. For example, if a question calls for a research subject to decide if they agree or disagree with an answer, the form can be written so that a number is circled that corresponds to an answer.
It is the number and not the word that is entered into the software program, as numbers can be manipulated statistically to find if there is a relationship between the data. If the survey is being administered online, the coding will be automatically added by the software used to develop the questionnaire.
Testing the survey form
All survey forms should be tested. When testing the survey form, both the content of questions and their wording should be examined. Each question should be one that the participant can and is willing to answer.
In addition, the wording should be checked to see if a participant understands the question in the way that the researchers intended.
Besides the questions, the answers should also be examined. The available responses for a question should contain the answers that most participants would provide.
If the written answer does not do so, ‘other’ will be ticked too often, or participants will choose a response even though it does not accurately reflect their opinions.
In addition to the content and wording, the form of each question should be examined. If an open-ended question leaves participants confused, it could be rewritten as a close-ended question so that potential answers can be provided.
The words used in the scale that is used to answer a question, such as ‘strongly agree’ to ‘strongly disagree’, should have the same meaning for all participants. Lastly, the instructions should be tested to ensure that they are written in an easily understood format.
The survey questionnaire should be tested with participants who are similar to the research subject sample. If the sample consists of a variety of subgroups, at least some of the test participants should be from each group.
During the first step in the testing process, a researcher should be present while a participant completes the questionnaire.
This way the researcher can make a note of any direction, question or answer that causes difficulty. If there are major problems with the survey form, the researcher should address these through changes and the testing should begin again.
If there only a few minor changes that result from testing the questionnaire, the testing can proceed to the next step.
The questionnaire should now be tested using the planned methodology, whether administered in person, over the phone, by mail or online. The sample for the test can be small, with as few as five to ten individuals.
The reason for testing the methodology besides the questions and answers is to determine the length of time that it takes for the survey to be completed. The amount of time it will take to complete will have an effect on the completion rate.
Data entry testing
Once the survey delivery methodology testing is completed, a final test should be conducted of the data analysis or entry system. The data analysis section of the online survey software should be checked to ensure that all responses are being recorded.
Testing of the questionnaire form
Administer the survey personally to discover problems
Administer with the delivery methodology
Test data analysis
Writing the Questions and Answers
Most survey questions are close-ended, with possible answers provided from which a participant must choose. These answers are determined on the basis of previous qualitative research, such as focus groups or interviews.
For example, the answers that focus group participants provide most frequently to a similar question will determine the responses used on the survey form. In addition, data discovered through observation and interviews will also be used to construct the answers to close-ended questions.
Open-ended questions allow survey participants to provide their own answers that they will write on the form or type online. Since the purpose of surveys, which are a quantitative research methodology, is to support a fact or hypothesis, open-ended questions must be kept to a minimum.
This is because the answers to open-ended questions may vary so much that there aren’t enough similar responses to allow for statistical calculations.
General guidelines for question writing
There are general guidelines that researchers should remember when writing any survey question. First, the questions should not be hypothetical as some people will have difficulty imagining such situations.
For example, questions that ask for imagined responses such as ‘How would you feel if you found out that you had bought defective merchandise?’ are difficult to answer on a survey form.
Such questions are best left to focus groups or interviews, where researchers have the time to draw out responses. Survey questions should only deal with what participants already know or have experienced.
Problem Question: How should manufacturers change cars so that they get better gas mileage?
Rewritten Question: How important is high gas mileage to you when you purchase a car?
Use simple terms
Researchers are often very familiar with the terminology used by the industry commissioning the research. If not initially, they will have certainly become familiar with the industry terminology while conducting secondary research.
It is important for researchers to remember that participants in a survey might not have this knowledge. Researchers should always write the question using words that are commonly understood.
Problem Question: What means of transport did you use to visit this establishment?
Rewritten Question: How did you travel to the store today?
Besides the issue of terminology, respondents may not read at the same academic level as researchers. Reading level involves the number of words in a sentence, the sentence’s grammatical structure and the length of the words used.
It is very important to have questions written at the correct level. If researchers are unsure they should always write at a lower level, which will make a sentence easier and quicker to read for everyone.
Problem Question: In terms of motivational desires, what would you consider of primary importance?
Rewritten Question: What was the most important reason why you bought our product?
Ask one question at a time
It is important to keep a survey form short so that participants are motivated to complete it. However, the researchers should not be tempted to shorten the number of questions by combining more than one at a time.
Asking two questions simultaneously will only confuse participants. If a participant is unsure of the question being asked, the answer will be meaningless.
Problem Question: How did you travel to the store today and did you encounter any difficulty?
Rewritten Question: How did you travel to the store today? How convenient was this method of transportation?
Ask only what can be answered
Questions that ask when ‘a person’ or ‘someone’ has done something will leave respondents confused. They will find it easiest to respond to questions that ask about their own activities.
For example, a question such as ‘Do people buy products that they have seen on infomercials?’ cannot be answered as participants will not have such knowledge. A question that asks ‘Do you purchase products that you have seen on infomercials?’ is easy to understand.
Problem Question: How enjoyable is a visit to the theatre?
Rewritten Question: Did you enjoy your visit to the theatre this evening?
Writing questions in more than one language
It is important that any common words or phrases are comparable when asked in more than one language. This translation may be necessary even when a questionnaire is being used in only one country.
Even simple words such as ‘miserable’, ‘disgusted’, or ‘thrilled’ will be difficult to translate with the exact same level of meaning.
Writing the answers
There are two components to a survey answer – the wording of the answer and the means of response. The same rules that apply to write the questions apply to write the answers, but even more so. Because each answer consists of so few words, it is vitally important that these are the right words.
There are several ways that participants can be instructed to answer a question. An answer to the question of ‘What motivated you to enroll in this marketing course?’ could be written using fill-in-the-blank, dichotomous, forced choice, multiple choice, checklist, ranking or rating responses.
Open-ended questions, where respondents are only provided with a blank line, are used when researchers do not wish to lead participants. An example of an open-ended question in a survey would be ‘Why did you enroll in this marketing course?’
The advantage of this type of question is that it allows any type of response, including those which researchers might not have thought of.
One surprising response might be ‘A person I’m attracted to signed up’. After all, potential romance might not have been among the responses considered by researchers.
The disadvantage of this type of fill-in-the-blank response is that every questionnaire will have to be read and the answer will need to be coded.
A dichotomous choice answer allows respondents to choose one of two responses that are usually the opposite. Examples would be answers that allow respondents to tick ‘yes’ or ‘no’.
A dichotomous choice answer directly addresses a research issue and forces participants to make a choice. Perhaps researchers are interested in discovering whether the love of learning motivates students to enroll in a marketing course. However, this response may not be one that occurs to a student.
This would leave researchers with the conclusion that love of learning plays no role in motivations for enrolling in the class.
A dichotomous choice question would ask ‘Is love of learning one of the reasons why you enrolled in this course?’ The student will tick either ‘yes’ or ‘no’. This type of answer forces students to reveal whether the love of learning had any role in motivating their decision.
A forced choice question asks respondents to choose between two responses. However, the responses do not need to be opposites. In fact, they can have no direct relationship with each other.
Researchers would use this type of question when they want to determine which of the two responses is more important. It might be that researchers notice that mid-afternoon classes have the most enrollments. Is this because of a popular professor or the time of day?
In this case, the forced choice answers would be written as ‘Like to take mid-afternoon classes so I can work later’ and ‘Heard the professor was friendly’. Therefore a student cannot choose both.
If most of the students tick the first answer, mid-afternoon might be the time to schedule professors who have lower enrollments because students do not perceive them as friendly!
When researchers have a number of variables that might affect choice, they may wish to write a multiple choice question. The answers would then list the motivations that were uncovered during earlier exploratory research.
Researchers must decide how many reasons to list. If too many are provided, participants may find it difficult to weigh them mentally and come to a conclusion about which are the most important.
Usually, four or five possible answers are listed. Of course, the listed reasons might not include any of the reasons why an individual student might have enrolled in that class.
This situation can be handled by adding ‘none of the above’ as a response, or by allowing a fill-in-the-blank line for respondents to write in their own responses.
The problem with a multiple choice answer is that more than one of the answers may be true. A student might have been strongly motivated by two or even three of the reasons.
A checklist solves this problem by allowing participants to choose as many variables as apply. Because participants do not have to weigh one possible answer against another, a checklist can include many more possible answers.
A variation on the checklist is the ranking question. This type of question and answer assumes that the listed responses will include a number of variables that will apply. It allows participants to indicate not only which of the reasons apply, but also the relative importance of each reason.
Theoretically, researchers could ask participants to rank all of the answers that apply. However, this may prove to be too difficult for participants, so usually, a question will ask participants to rank their top three to five answers.
The importance of asking a ranking question instead of just providing a checklist will become apparent when data analysis is considered. In data analysis, when calculating the frequency of responses, they will not just be counted but will also be weighted by choice.
For example, the number one choice for students might vary widely while the second choice is almost always ‘it will help me get a job’. When the variables are weighted, the choice that was ranked second may be the most common.
A rating question will allow respondents to choose more than one answer and also to rank the importance of each. The answer can consist of three, five or seven possible rankings. One of the rankings will be that the answer had no effect on choice. This is always the middle of the rating.
A researcher can then write the rating to allow a respondent to indicate whether the answer had either a positive or negative effect on choice, such as ‘agree’ or ‘disagree’.
In addition, the ratings can allow participants to show how strongly the answer positively or negatively affected their choice. Even complex questions can be answered in this way. A researcher may obtain this information by adding more possible answers, such as ‘strongly agree’ and ‘strongly disagree’.
Along with the response of ‘no effect’, the answer now has five possibilities. A researcher can continue to expand the answer to add the responses of ‘very strongly agree’ and ‘very strongly disagree’. Of course, it is still difficult to predict how each participant will evaluate the choice of rating.
For self-administered surveys, the way that the questions and answers are visually presented is critical. This is true for both paper and online survey forms. A poorly laid out questionnaire may confuse participants and result in unanswered questions.
When laying out a form, researchers must remember the visual impact that results from the use of margins, spacing and font size. Proper use of these elements will result in a survey form that is visually appealing and easy for the participants to read.
Researchers will also want to keep a survey form to as few pages as possible, as fewer pages will keep reproduction costs down. Whether on paper or online, a survey with many pages will discourage potential subjects from participating.
However, these concerns should not lead researchers to print the survey in a font that is too small and therefore difficult to read. The use of white space, such as larger margins and extra lines between the questions, will result in a longer survey form but will also make the survey more attractive.
Routing is an issue that needs to be considered when a questionnaire layout is designed. For example, there might be a follow-up question designed only for those who drive an automobile to go shopping.
A question that asks where participants park their car when they travel to the store will not be relevant for those participants who took the bus. Therefore, these latter participants should be directed to skip the question and go on to the next question.
The instructions for routing must be very clearly and simply stated on the survey form as most people will assume they will need to answer every question. One advantage of an online survey is that it can be programmed to automatically bring the subject to the next appropriate question.
The questions should not be listed on the form in a random manner. The first questions should be for the purpose of qualifying a participant.
The types of qualification questions asked would depend on the research question and the purpose of the survey. They may ask a participant about the frequency of purchase of a specific consumer product or whether they rent or own their own home. The questions may also ask about age or family status.
Even if basic demographic data are not part of the qualifications or research question, they should still be included. These demographic questions should come first as they are the easiest to answer.
For example, while the research question might not have asked about a gender difference in purchase habits or attitudes when the data are analyzed it may be found that there is a gender difference in consumption. The same may hold true for age, education level, and even geographic location.
The next questions should be general in nature, with more in-depth questions to follow. For example, a question that asks what brand of sneakers a participant prefers is easy to answer.
A more specific question, as to why the participant chooses to purchase this specific brand, will take more thought.
If difficult questions are listed on a survey form too early on, they may discourage participants from continuing and completing the form. For this reason also, any sensitive questions, such as ‘Do you feel you are overweight?’, should be saved for last.
Easy to answer questions
Online Survey Forms
The development of an online survey should be no different from that of a traditional survey. There are software packages now available that make the creation of a survey much easier. However, this does not make the decision of what topics to address and how to write the questions and potential answers any easier.
While some programs provide pre-written questions on a variety of subjects, the researcher should be careful to ensure the questions used are specific to the current research project.
There are a number of reasons for the increasing popularity of putting survey forms online. One is the difficulty in motivating individuals to participate in traditional survey research.
An online form can be completed at a time that is convenient for a participant, unlike a phone or personal survey. Also, the completed form is automatically returned – unlike a mail survey.
In addition, there are design advantages to using an electronic form. With an electronic form, there is less concern to keep a questionnaire to as few pages as possible.
Participants only see one or two questions at a time and, therefore, will not be intimidated by the overall length. In addition, the form can be laid out with colorful graphics that make it visually appealing.
Coding of responses is not needed because the results from an electronic form are calculated automatically without data entry. In addition, the routing of questions can be handled automatically.
The form can be designed so that the next question that needs to be answered will appear on the screen based on the answer to the previous question.
On an electronic survey form, when a participant responds to a question on how they travel to the store with the answer ‘by car’ the next question will automatically ask about the difficulty of parking. If the response to the question is ‘by bus’, the next question will ask about the convenience of the bus stop.
Another advantage to laying out an electronic survey form is the ability to use drop down boxes for answers to questions. In written surveys, researchers face the temptation of limiting the number of responses provided to anyone question so that the survey form does not become too lengthy.
Using drop-down boxes – where participants use their cursors to pull up a menu of answers and then make their choice – solves this problem.
Sending out the survey
If the organization has a list of email addresses, text numbers or Twitter followers, they can be used to find participants. If possible the message sent should be personalized rather than being the same standard message sent to everyone. Social media can now be used to disseminate survey forms.
To do so a link to the form is simply posted on the organization’s social media site. This method will only be successful if there is already a social networking site that is actively used by current and potential customers. If the site does not have sufficient traffic, not enough participants will be found.
Information posted on the site needs to include the reason for the survey, the participants being targeted and the length of time the survey will take to complete. To ensure the right participants are being included in the survey, screening questions should be used.
Using technology to design new types of responses
Besides convenience, using an online format allows creativity in designing a questionnaire. This is particularly true in writing the answer format.
For example, when product or brand preferences need to be indicated, they can be shown with pictures or logos along with their names. Participants may more easily recognize a package or logo than the name of a brand. Beside each picture would be a ‘radio’ button that a participant would click to indicate his or her choice.
Survey questions often ask participants to rank their choices of products or desired benefits. In a paper survey, such choices are ranked in the same order.
There may be a bias here toward the early responses as research subjects may find a likely answer and stop before they read the complete list. With an online survey, the order can be randomly generated so as to eliminate this bias.
When it Comes to Surveys, How Long is Too Long?
Business has become more complicated. Companies may sell online and in stores, may have multiple product lines, and several target markets in more than one country.
To try to gain an understanding of their customers’ motivation and behavior, surveys are getting longer. At the same time, people have become less interested in responding to surveys and have less patience when they do.
Overall response rates for surveys have fallen: according to the Pew Research Center, from 36 percent in 1996 to only 9 percent in 2012. There is every expectation that the rate will continue to fall.
Many companies offer incentives to encourage completion. It has been proven that this can lead respondents to ‘click through’ a survey. They simply click any answer to get to the end to get the reward. The longer the survey the more likely this behavior is to occur.
Finally, more people are responding on their smartphones. It is estimated that around 50 percent of surveys are completed on mobile devices. Unfortunately, the small screen size makes long surveys seem even longer so people are less likely to respond.
Question: If your job was to design surveys how would you approach these problems?
1. Surveys can be used to research any aspect of the marketing mix. The development of a survey should be a joint undertaking between management and marketing researchers. Both participants and data entry clerks should also be considered when designing a questionnaire.
2. The survey development process starts with the research question and proceeds through meetings between the marketing researchers and management where question topic areas will be discussed. Drafts of the survey form are created and reviewed until everyone is satisfied that the questionnaire will provide the needed information.
The questionnaire is then tested on potential participants and should be first tested verbally to check for understanding and possible confusion. After this, it will be tested with a few survey participants using the planned delivery method.
3. Survey questions can be either closed or open-ended. The general rules for writing effective questions include using simple terms, writing at the correct reading level, asking only one question at a time and writing in the active voice. It should be remembered that questions may need to be translated into other languages.
There are a number of different ways to structure the answers. These include fill-in-the-blank for open-ended questions. Close-ended questions can be answered using dichotomous, forced choice, multiple choice, checklist, rating or ranking responses.
4. The questionnaire’s layout should ensure the form is easy to use and to read. It should look attractive so that people will be motivated to read the questions.
The routing of the participant through the questionnaire should be clearly explained. The questionnaire should start with qualifying questions and then move on to demographic questions, easy to answer questions, in-depth, and finally sensitive questions.
5. If a form is to be administered online, the same development process must take place. One advantage of an online survey form is that it can be designed so that only one question can be seen at a time and routing can be automatic.
Another advantage is that answers can be constructed using graphics and drop down boxes. In addition, radio buttons can be used for answers and photos of products and logos can be shown instead of words.
Methods of Conducting Surveys
After deciding upon a research question, researchers must choose the research methodology that will be used. The most common quantitative technique for collecting primary data is the research survey, which asks all participants the same predetermined questions.
If researchers decide to conduct a survey, they must ascertain the appropriate population to be surveyed, the size of the sample and the sampling method.
Researchers must also decide upon a survey method. Methods for conducting surveys can be divided into researcher-administered surveys and self-administered surveys. When researcher-administered surveys are used, participants are assisted in completing the questionnaire.
With this method participants are helped with any questions they find difficult to answer and are encouraged, if necessary, to complete the survey. With self-administered surveys, participants are provided with a survey form, which they then complete on their own.
When conducting researcher-administered surveys researchers or their assistants ask the questions and complete the questionnaire forms. The primary methods of conducting researcher-administered surveys are personal face-to-face surveys and phone surveys.
There are three advantages to conducting researcher-administered surveys. First, if a research participant does not understand the meaning of a question, the researcher can help to clarify it.
This will ensure that the question is understood correctly and will, therefore, result in more accurate research data. For example, a college student taking a survey might wonder if a question on yearly income refers to their own or their family’s income.
The second advantage of this survey method is that a researcher can encourage a participant to answer all the questions and complete the questionnaire.
Research participants may initially agree to take a survey because they do not want to seem rude when asked to participate. However, once they have the questionnaire they may remember that they are in a hurry to get dinner on the table or to meet friends to see a movie.
Because they are in a hurry, participants may skip any confusing questions or just neglect to turn over the survey to finish the second page. Using a researcher-administered method a researcher will notice these omissions and prompt the participants to answer all the questions on the form.
Choosing helpful and encouraging administrators can increase response rates. There are some groups of people who may distrust any form of marketing research.
This would include those groups who may feel disenfranchised from mainstream society. As a result of this feeling of disenfranchisement and the distrust that follows, members of these groups are least likely to participate in surveys.
By having someone who is of a similar background personally conducting a survey, members of these groups are much more likely to participate.
Advantages of researcher-administered surveys
Clarify confusions and misunderstandings
Prompt participants to complete all the questions in a survey Establish a rapport with difficult to reach groups of participants
Some survey methods use paper questionnaires or online survey forms that are self-administered. With these methods, there is no researcher physically present during the surveying process. One of the major advantages is that self-administration is much cheaper, as there is no need to pay for a researcher’s or assistant’s time.
Another advantage to using a self-administration surveying format is that a researcher cannot lead a participant in any way to respond with a specific answer to a question.
Of course, a good researcher is not going to tell a participant how to answer a question. However, when explaining a question to a participant, a researcher may still unintentionally influence their answer.
A third advantage of a self-administered survey is that participants can take their time to complete the form. If they are interrupted during the process of completing the survey, they can even leave the form and come back and complete it at another time.
An additional advantage of having a self-administered survey is that the method allows participant privacy. When a researcher administers the survey, a participant may be concerned about that researcher’s response to their answer.
This would be particularly true of answers to questions about personal behavior. However, even with a routine question, participants may want to provide a more positive answer when personally responding to a researcher.
Advantages of self-administered surveys
Less cost because there is no need to pay for a researcher’s time
A researcher cannot lead a participant to provide a specific response A participant can complete the form at their own pace
There is no concern about a researcher’s reaction to the answers
Researcher-Administered Survey Methods
Researcher-administered surveys can be conducted personally or by using the telephone. In addition, personal surveys can be conducted using computer technology rather than paper questionnaire forms.
When in the field conducting interviews, the success of researcher-administered surveys will depend on being organized and treating everyone with respect. In fact, even the tone of voice used can affect whether people are willing to cooperate.
Personal surveying has important advantages over telephone surveying as a methodology. With personal surveying researchers are able to use prompt cards or other visual stimuli that will assist participants with their answers. For example, if a question asks if participants have purchased a specific brand of shampoo, a picture of the brand can be shown.
Many people can more easily remember a product visually, by remembering what the bottle looked like than by remembering a specific brand name. If a survey is about consumer preference in the color or style of a product, researchers can have the product ready to view.
For food products, participants could be given a sample to taste before they respond to the survey questions. If a research question involves consumer preferences for types of promotion, examples of ads or web pages can be shown.
Location of personal surveying
There are a number of different locations where personal surveys can be administered. The natural location for research that involves current customers would be at a business or organization itself.
For example, if the organization commissioning the research is a sports team wanting to know more about their fans, the survey could be administered at a game. If the organization is a retailer wanting to know about product preferences, researchers could survey shoppers at the store. A problem is that using these locations would only reach current and not potential customers.
Another location for conducting surveys would be at a place with a number of retail businesses. This type of research is often referred to as ‘mall-intercept surveys’.
For businesses selling consumer products, large regional shopping malls or areas is a favorite location for conducting personal surveys, as such an area will generally attract a demographically diverse group of shoppers.
A large mall will also attract customers from a wide geographic area who will provide a mix of urban, small town and rural potential participants.
A large mall will also have a variety of different types of retail, dining and even entertainment options that should draw a psychographically diverse population. Researchers should choose a number of different times to administer the survey, as different age groups may shop at different times of the day.
In addition, they should station themselves beside different businesses that attract a variety of types of shopper. Using personal surveying, researchers stop potential participants and survey on the spot or direct them to a separate area with comfortable seating.
A final location for conducting personal surveys is at a public place where potential participants are gathered. This might be a community park or another civic amenity such as a swimming pool or zoo.
Surveys could also be administered on a street corner where a targeted sample of potential participants passes by. Another idea is to conduct a personal survey at a community festival. The choice of location would depend on what type of event attracts the required population.
Advantages of personally-administered surveys
Motivate responses through interaction
Use visual aids to remind subjects of products or brands
Can be done at the location where groups of participants are found
Disadvantages of personally-administered surveys
Time-consuming, which costs money
Must train survey takers in both technical and interpersonal skills Research subjects may hesitate to commit to a personal encounter
Computer-aided personal surveys
Technology has provided a means to discard the paper questionnaire form traditionally used when conducting personal surveys. Instead, a computerized tablet is used by researchers to read questions and record answers. A researcher or assistant would still be present but the answers would be recorded electronically.
There are advantages for researchers in using this form of electronic data collection. Researchers will save money as there will be no data entry costs and the recorded answers can be downloaded directly into a computer program for analysis. Besides the cost savings, direct downloading increases reliability by excluding data entry errors.
Complex survey forms will often be written with directions for a participant to skip ahead to a specific question based on their response to a previous question. With an electronic form, this will happen automatically, minimizing confusion. In addition, the sequencing of questions can be randomized.
This is an important consideration if there is any concern that question sequencing might be influencing the resulting answers. Of course, there will be the cost of buying the technology and training people in its use.
Advantages of computer-aided surveys
The software will automatically enter and calculate answers Saves money on data entry and copying forms
Disadvantages of computer-aided surveys
Cost of technology
Training of survey takers
Researcher-administered telephone surveys
The most significant advantage of a telephone survey over personal surveying is that it allows participants to be geographically dispersed. A phone survey can reach potential subjects wherever they live at little extra cost.
In addition, participants can be reached that researchers may not be able to survey at a business or in a public place. For example, the elderly may be easier to reach by surveying them at home over the phone rather than in public.
Another advantage of telephone surveys over personal interviewing is that the telephone allows a participant to remain anonymous. As a result, participants may provide information that they may not wish to provide to someone personally.
They may also be more willing to provide negative feedback, which they may feel would be rude if delivered in person.
Problems with conducting telephone surveys
There are, however, significant challenges that researchers must overcome when conducting phone surveys. Contacting potential participants is an increasing problem due to the growing number of individuals who rely on their cell phones rather than having a land line. Researchers can obtain phone lists for landlines, but not cell phones.
In addition, when calling a home phone on a landline market researchers at least know that individuals are at home and could potentially participate in a survey.
Even if market researchers had access to cell phone numbers, using them for a cell phone survey would probably result in a poor response rate. People may be answering their cell phones in a store, or while driving, or may even be at work. Individuals thus engaged would usually not be receptive to the idea of taking time to participate in a survey.
Another difficulty is the number of people who have caller ID. Such households may not even answer the phone if a number is not recognized. Another issue that researchers must consider is privacy laws. The popularity of Do Not Call lists means that many households cannot be called for commercial marketing research.
Advantages of telephone surveys
Can have larger geographic reach economically
A participant can remain anonymous
A participant may be more likely to give negative feedback
Disadvantages of telephone surveys
Increasing use of cell phones makes it more difficult to contact People do not answer cell phones when they may not have time to participate, even if interested
Methods of Conducting Self-Administered Surveys
Surveys can be self-administered by providing participants with a form using the postal service, email or social media sites. The mail survey has been the traditional means of delivery for self-administered surveys. However, this method has been largely replaced by using technology to deliver questionnaires.
A short survey form can be sent to participants in the body of an email which is then returned via email when completed. Another means of delivering the questionnaire electronically is to send potential participants an email with a link that directs them to a website to take the survey. Even text messaging on cell phones is now being used to ask survey questions.
One of the traditional means of delivering a self-administered survey form to participants is the postal service. There are significant advantages to using mail surveys, which is why they continue to be used by some researchers despite the growing popularity of electronic delivery of questionnaires.
A mail survey will reach designated households as everyone has a physical address, while not everyone has an email address. There is no problem obtaining a list as there are commercial companies whose business is to provide such information to researchers. Marketing researchers could also use a publicly available source such as a voter registration list.
Problems with conducting mail surveys
The success of a mail survey depends on the quality of the mailing list that is being used. A poor quality list that does not target the right households or has old addresses will not save money, as the responses will not provide the needed information or even arrive at households.
While one advantage of using a mail survey means there is no researcher to bias responses, the lack of personal communication can also be a disadvantage. When using a mail survey researchers have no means to encourage completion or to explain questions that respondents might find unclear.
Another issue with mail surveys is that researchers must wait for responses to be returned in the mail. The research will take more time to complete as researchers must wait while not knowing if participants are going to respond – or if they have thrown the form away and more participants will need to be sought.
Advantages of mail surveys
Able to reach every household as all have address Survey taker can’t introduce bias
Mailing lists are easily available in some countries
Disadvantages of mail surveys
Nonresponse rate is high as no one to motivate response
People may not respond promptly
No one available to explain questions
Emails are generally not widely used for delivering questionnaire forms because of practical issues. An email attachment that is large enough to contain all the text and formatting for a questionnaire may be too large for a potential participant’s computer to handle. In fact, such a large attachment may be screened out as SPAM.
However, electronic delivery using email is appropriate for short questionnaires targeted at groups that are familiar with the organization and therefore likely to respond. Most electronic survey forms though will be on separate websites and email will only be used to send the relevant website link that provides access to the form.
Most common are commercial online survey software packages that will assist in the design of a survey, collect the responses and tabulate the results. There are products which are free or relatively inexpensive and easy to use.
While the fact that the form is electronic does not change the survey questions, it can change the way the questions and answers are presented. Visuals can be incorporated and the sequence of questions can be easily varied.
The major difference in using an online survey versus a mail survey is that the survey results appear immediately. This allows researchers to track the number of responses and send email reminders to those who have not responded.
It also allows researchers to change any of the questions that seem to be causing confusion or to reconsider any of the questions that are not being answered.
Difficulties with conducting online surveys
There are also disadvantages to online surveys. First, respondents must have computer access. They must also be computer literate and feel comfortable responding online. It should be remembered that not all types of jobs require people to be in front of a computer.
In addition, not everyone enjoys being online. For those individuals who are not online every day because of their job or personal interests, completing an online survey will mean that they must go online specifically for that purpose, which is not as easy as picking up a pen or answering the phone.
Another disadvantage is that the potential participants who wish to respond to online surveys may be skewed toward younger, better-educated individuals. This is fine if a research question addresses a product that is of interest to younger people.
However, if a research sample calls for responses from a quota of specific demographic and psychographic types, online may not be the best methodology to get the responses needed.
Advantages of online surveys
Results appear immediately
Survey software widely available
Electronic survey questions can be customized
Disadvantages of online surveys
Subjects may not have computer access
Challenging to reach some segments
Posting survey links on social media
Disseminating surveys to research subjects using email allows the marketing researcher to individually choose each potential participant. Another means of obtaining research participants is to send out text messages to the research sample with a link to the survey. Social media is a third technology that can also be used as a means to attract research participants.
The difference with this method is that the marketing research does not control whom will respond. However, if the research is to gather opinions and not to provide a quantitative answer to a research question, then posting is an option.
The organization should only use social media that it is already using. The followers have demonstrated that they are interested in what the company has to offer.
If the survey is going to be posted or tweeted it should be short. Even if the link brings the research participant to the survey software, people on social media are rarely interested in spending time answering questions. It may be necessary to offer an incentive for completing the survey form.
As the people who are on the organization’s social media sites are interested in the product offered, providing a discount should motivate completion. Another idea is to enter all those who complete the survey into a contest where the prize has a higher monetary value than a discount.
The response rate is the number of people who respond to a survey compared to the number of respondents who are eligible to participate. Low response rates should be expected. In some research studies, the sample can be as low as 10 percent.
This is the reason why more sample participants are selected that are needed to meet the sample size requirement.
Busy people walking down the street will usually not be pleased when they are stopped and asked to complete a survey form. The first impulse of anyone answering the phone, only to discover there is a researcher on the line, will be to hang up.
Imagine people with too little time picking up their mail or clicking on to their email – they will quickly sort through, looking for personal correspondence and work-related items. What is left is usually termed ‘junk’ mail or SPAM and will be quickly discarded.
However, there are reasons why individuals will participate, including a desire to assist others, a general interest in giving their opinions, or for personal gain. If the purpose of the research is to help in finding a solution to a problem, those directly concerned with that problem may be willing to help for altruistic reasons.
For example, if a survey is to determine people’s attitudes toward global warming, those interested in environmental issues are likely to respond.
Sometimes research for commercial products may even elicit responses because the use of a specific product is of personal interest to a participant. Subjects such as the use of technology and a choice of entertainment options may motivate participation because the topic is associated with an interesting activity.
In addition, some people are always interested in giving their opinion. They may find that the attention they receive fulfills an inner need to be noticed.
However, these reasons will not always be present and, therefore, researchers must consider the need to motivate participants by using a financial or product incentive in order to complete both a researcher-administered and self-administered survey.
Uber in India: Will it Succeed?
It might be surprising that the second-ranked country for the number of rides hailed by using the Uber app in India, second only to the United States. Uber has spent $1 billion in entering this marketplace of 1.2 billion people.
While there are a lot of potential customers, finding and hiring drivers is a challenge as only 5 percent of Indian households own cars.
Uber also faces local competition, a company called Ola, which operates in 102 cities and has over 500,000 drivers employed. At first, Uber decided to use the service model that worked so well in other countries.
Drivers would be hailed using the app, which would handle the payment. However, when Uber started operating in India, they found they needed to change the model in the following ways:
Most Indians are not familiar with apps. In addition, if they do use a smartphone they may have very small data plans and be reluctant to use any for hailing an Uber. Uber developed a special website where rides could be booked.
Most Indians don’t use credit cards. Uber started to accept cash payments.
Cars are expensive for most Indians. Much cheaper are motorbikes. Uber now hires motorbike drivers to take fares.
Many rural areas in India are not mapped and even if they are, lack addresses. Uber is hiring local engineers to develop maps.
Question: What research should Uber have conducted so that they could have made these adjustments before they entered the Indian market?
Providing information to potential participants
Some people may choose to participate in a survey because of personal interest in a product or because they believe the results of the survey will help others.
Therefore, it is of critical importance that any information about who is sponsoring the research and how the results will be used is communicated to potential participants.
For personally-administered surveys, this information should be provided verbally. For both mail and online surveys, researchers should enclose or attach a letter that includes this information.
If a covering letter is sent along with a survey form to provide credibility, it should be on a letterhead from either the research firm or the business or organization sponsoring the research. An email request should provide contact information, to allow someone who wants to verify the identity of the sending or sponsoring organization to do so.
Both the letter and the email should explain why that particular person or household has been chosen. People are more likely to respond positively if they can understand why they have been chosen to participate.
For example, the letter or email might explain that the survey is being sent to people who work in education or the medical profession. Even if researchers are using a random sample, there will still be a reason why a specific population was chosen.
The letter or email should explain the purpose of the research in terms that someone unfamiliar with such research terminology can understand. This language issue is so important that the covering letter or email should be as carefully tested as the survey form.
The explanation should also include information on the research methodology in order that participants will know how the data are being gathered. If the research is of a sensitive nature, information on confidentiality should be given.
This should include a reassurance that no participant will be identified by name, that only tabulated total results will be released, and that all forms will be destroyed on the conclusion of the study.
The next issue that should be addressed is how participants will benefit from completing the survey form. Some research on social or political issues, while not benefiting individuals directly, will benefit society as a whole.
In this case, the letter or email should appeal to the altruistic nature of participants. If this is not the case, the letter or email should describe any financial or alternative incentive that is being offered to those completing the survey.
To encourage participation, the letter or email should discuss the length of the survey and the projected time it should take to complete. Also, to encourage completion of an addressed and stamped envelope must be included for mail surveys. For online surveys, the link for the survey’s website should be easily located.
To generate interest, even the envelope that contains the survey or the subject line in the email should communicate the purpose of the research. After all, no one specifically picks up their mail or reads their emails hoping for the opportunity to participate in a survey.
Finally, the letter or email must be brief, must use simple everyday language, should look attractive and must be visually designed to be read easily.
Covering letter or email components
Place on an official letterhead or provide contact information in an email
Explain who is being sent the letters (sample) and how they were selected
Explain the purpose of the research and the methodology used
Communicate how society or participant will benefit
Describe the length of the form and the length of time to complete it
Include mailing instructions and an envelope or a website link
Place motivating information on the envelope or email subject line
The use of incentives
If the subject of the research is something rather mundane, such as the frequency of auto maintenance or a soft drink consumption preference, most recipients will probably not wish to participate in the survey. In this case, researchers must consider the use of an incentive to motivate participation.
This incentive could be indirect, such as a respondent who completes the survey being entered in a prize draw. Because most people will understand that they will probably not win, the prize must be so exciting that the chance of winning will motivate participation.
The incentive chosen should be of particular interest to the population sample being targeted by the research survey.
For example, a survey targeting frequent travelers could use as an incentive the possibility of a free trip, as most people would find this an attractive enough incentive to motivate their participation even if they only had a slim chance of winning.
The incentive could also be direct. A free product, such as digital music which can be delivered online, or a coupon good for redeeming a free product at a store, could be given to anyone who completes the survey. As this can be expensive, a discount coupon for a product that the company supplies could be offered.
For example, for a survey on auto maintenance preferences, researchers might offer a coupon from the sponsoring company for a half-price oil change if the form is completed. This incentive has the advantage of motivating completion and also providing promotion for the company.
If researchers do not wish to tie the coupon to a specific company as it may affect participants’ answers, a voucher that could be used at one of several companies could be offered instead.
The Survey Process
The survey process is similar for both researcher-administered and self-administered surveys. First, the research question must be written. As with all forms of research, this will be a team effort between researchers and management.
The research question will help to determine the profile of the sample that will be needed to participate in the research.
The next step will be to decide upon the method of conducting the survey, whether the method is a researcher- or self-administered. The survey will then be written and tested.
The times and locations for researcher-administered surveys must be chosen to meet the needs of participants. For self-administered surveys, the dates for mailing or emailing a survey will need to be decided.
For researcher-administered surveys, the next step will be to identify staffing needs and hire and train the personnel who will conduct the survey. After the personnel is hired and trained, the survey can be conducted.
For self-administered surveys, the survey is now mailed or the website is opened, after which a researcher will wait for the responses to be returned. Once the data have been returned – either by the survey taker, through the mail or electronically – the results are then analyzed and the report is written
Training survey takers
The success of researcher-administered surveys depends on the skill of survey takers. These survey takers do not need to be professional researchers. Large corporations can use the services of a commercial call center.
Small businesses or research companies that only occasionally conduct surveys can hire employees by the hour. This type of part-time job could be attractive to college students or the retired.
However, because these employees will not have experience in conducting surveys, training should be conducted before the survey process begins.
Training should inform the survey takers of the purpose of the research as they will be more motivated to have participants complete a survey if they understand the importance of the research question.
The second step in the training is to explain the structure of the survey and the reasons for the questions. The role of the supervisor or lead researcher should be explained, namely that they are there to assist with difficult participants and to provide general support.
The survey takers should also be taught the general rules for conducting successful surveys.
Once training has been completed, the survey takers should first administer the survey on each other. In addition, the supervisor should role-play uncooperative or rude potential participants so that survey takers learn how to diffuse such situations. Once survey takers are on the phone or out in the field, the first surveys should be supervised and monitored.
Training survey takers
Describe the research topic
Explain the survey structure and the reasons for the questions
Define the role of the survey taker and the supervisor or lead researcher
Practice role-playing survey taking
Supervise and monitor first surveys
Conducting a survey
The survey process for both phone and personal surveys should proceed through the same steps. Survey takers should first be trained to introduce themselves and their organization.
They then should use a script written by the researchers to explain the purpose of the research and the sample selection process. The survey takers should next determine eligibility with one or two screening questions.
These can be as simple as asking if the person speaking is the head of a household, or in a specific age range, or if they have children living at home.
After eligibility has been determined, survey takers should explain the length of the survey in terms of the time requirement, not the number of questions or pages. They must then obtain consent to proceed with the survey.
While the survey is being conducted, a survey taker should be prepared to explain confusing questions without leading a participant to provide any specific response. Finally, a survey taker should always sincerely thank a participant for their assistance.
Customers value transparency in an organization. Posting the survey results on social media is an option that should be considered. Knowing that they will be able to see the results will provide additional motivation for participation.
Any steps that are taken as a result of the survey, such as improvements to existing products or the introduction of new products, should also be shared on social media. Customers want to know that the organization listened and responded to what they had to say.
The process of conducting researcher-administered surveys
1.Make the introduction
2.Explain the purpose and sample
3.Determine eligibility with a screening question
4.Explain the time requirement
6.Conduct the survey, clarifying questions when needed
7.Thank the person for their participation
There will be occasions when a survey taker contacts a potential participant who reacts to the request to participate in the survey with rudeness or even hostility. A survey taker should be trained to never respond in kind, as it may damage the reputation of the organization conducting the research.
Instead, a survey taker should be trained to not take the response personally and simply move on to the next call. At times, some of these people may wish to speak to a supervisor, realizing that the survey taker is not the one responsible for the survey call. Supervisors should always be available to handle these calls and soothe any irate callers.
1. A researcher or an assistant can administer survey forms or these can be self-administered. The methods for administered surveys include in-person, computer-aided personal surveying and by telephone.
The advantages of these methods include the ability to clarify misunderstandings, to prompt completion and to establish rapport.
Self-administered surveying includes mail and web-based forms. The advantages to self-administered surveys include less cost, an inability for researchers to bias the response, and participants’ ability to complete at their own pace and convenience.
2. Personal surveying has the advantage of allowing the use of visual prompts. A researcher can also demonstrate product use. Computer-aided personal surveys use a hand-held device to collect answers.
This reduces the cost of data entry while also making complex question sequencing easy to understand. Telephone surveys suffer from low response rates but they still have the advantage of reaching certain groups and providing anonymity.
3. Mail and email surveys are inexpensive and with this methodology, there is no danger of an interviewer biasing the response. The success of both mail and email surveys depends on the quality of the mailing or email list used to reach participants.
Online surveys can be inexpensive to create and should provide researchers with the ability to see results immediately and to track response rates. In addition, results can be analyzed immediately. The disadvantage is that a sample may be skewed toward young people.
4.To motivate completion, information that is provided to participants should explain the reason for the research and why a certain subject has been chosen.
It should also assure confidentiality and provide contact information for those who wish to assure themselves regarding the legitimacy of the research. Information should also be provided on any financial or product incentive that will be given to the research subject.
5. The survey process starts with writing the research question, selecting the sample, and the method of surveying. If the survey is to be administered, the process must include the training of survey takers.
Survey takers should train by practicing giving the survey to each other. The trainers should monitor the first surveys conducted by the survey takers.