Four of the main techniques used to analyze surveys are frequencies, crosstabs, means, and graphs. The techniques and their advantages and disadvantages can be described as follows.
Frequencies involve making a count of the number of instances of the categories for each variable and finding the percentages for each category selected based on the total number of people in the survey or of those answering that question, if the missing responses are eliminated. Frequencies can be used for individual or multiple variables and for both descriptive and evaluative research. For example, in looking at gender, one might look at the percentage of the sample that are males and the percentage that are females; in looking at age, one might look at the percentage of people in each of the age groups. Another example of using frequencies is determining the percentage of people choosing an action in a forced choice question.
The advantages of using frequencies is that this is a simple way to provide an overview of responses to a questionnaire. Also, the frequencies for the categories for a variable can be combined to create a cumulative per cent for certain types of variables, where the categories can be grouped together, such as age or the amount someone has spent on something.
A disadvantage of this approach is that if there are multiple choices for different categories for a variable, the percentages will add up to more than 100% which might make it difficult to compare responses to that variable across samples. Another disadvantage is when there are multiple questions since there will be multiple frequency and percentage and cumulative percentage charts, which can be unwieldy for presenting the data. Also, the frequency procedure doesn’t work well when there are numerous categories for ordinal or Likert-type variables.
Crosstabs, involves conducting a cross-tabulation of two or more variables to look at the relationship between those variables. These are used in explanatory and evaluative research. For example, one might do a cross-tabulation between a demographic variable, like age or gender, and the response to a question to see if there is any difference between the groups in their response to that question, such as whether different movies appeal more to younger or older age groups or to men or women.
The choice of which total to use as a row or column percentage depends on the data, based on which comparison one wants to make (i.e. whether one wants to compare the demographics for a particular movie, or whether one wants to compare the movie preferences for members of a demographic group). Besides two-way cross-tabulations, one can use a three-way cross-tabulation or more, if the sample size is large enough. For example, one can look at the sex and age breakdown for different movies.
The advantage of using crosstabs is that one can compare differences between different groups, and the results can be used to help explain these differences. Crosstabs can also be used to compare different user and customer groups in evaluative research.
The disadvantage of crosstabs is that it can lead to a very large number of tables when there are multiple responses, because of the many different ways the variables can be cross-tabulated with each other. Also, not all of the crosstabs may be meaningful, although it may not be clear which ones are meaningful or not until one has done the cross-tabulations. Another disadvantage is the number of items that can be cross-tabulated with each other can be limited if there is a small sample size.
Means, involves finding the means or averages for certain types of variables, and this method of analysis is used for all types of research – descriptive, explanatory, and evaluative. However, means can only be used if there are scales or ordinal data. It is not meaningful to use means if one has used numerical codes for nominal variables.
The advantage of using a mean is that it can provide a single statistic that can be used in comparing different responses, rather than trying to look at a frequency table showing the percentage of responses for different categories in ranking or rating something.
However, a disadvantage of using means could occur if the mean has resulted from widely different responses, such as when a large percentage of the respondents strongly agree with something and a large percentage of the respondents strongly disagree. This would be a bimodal distribution, and the average of the two results would make it seem like there is little opinion, because it averages the very different results. A mean is also a disadvantage when there are a few extreme cases, such as in a few people with a very high income which skews the whole distribution, so the average income is much higher for everyone. In such cases, a median might be a more accurate statistic to use since it more accurately reflects the middle point of the data.
Graphs are a way to present the results of an analysis in graphic form, such as a bar graph, stacked bar graph, pie chart, line graph, or scatterplot. The bar graph, which is also called a histogram, is the most common form used in leisure and tourism research, and it shows the number or percent of cases on one axis of the graph and the category measured on the other.
If two variables are cross-tabulated with each other, these results can be shown on a stacked bar graph, in which one variable is shown in one color or pattern and the other variable is shown in the other, so together they make up the total stack for each of the categories into which a variable is divided. An additional variable might be shown by one stack next to each other, such as for a study conducted in two cities or in two different years.
The advantage of using a graph is that it shows visually the count or percentage differences in the results for different variables, rather than just looking at the count or percentages in a table. A disadvantage in using a graph is that the graph could be misleading based on how it is drawn to shown the differences between groups. For example, if there is a great difference between groups, but the percentage categories on the side are close together, this might underplay the differences; or conversely, if there are only small differences, spacing the percentages categories far apart could make it seem like the differences are greater than they are. Then, too, it might be hard to know what the actual percentages are unless they are written in or on top of the bars.
Pie charts are a type of graph which divide up the number or percentages of categories or responses for a variable into the sections of a pie. The advantage of a pie chart is that it is helpful to show the relative size of the different responses when there is a meaningful total, such as 100%. However, a pie chart doesn’t work well when there are multiple responses, so that the total is greater than 100%.