Statistical Analysis

When using quantitative data, a researcher collects information in the form of numbers. Statistical analysis is used to make meaning of those numbers. Statistical analysis can be divided into two broad categories: descriptive statistics and inferential statistics. Descriptive statistics is used to summarize and simplify the data. Formative research frequently uses descriptive statistics to understand the current public relations situation by reporting what has been observed. Means, percentages, and frequencies (how many people responded a certain way) are commonly used in descriptive statistics. For instance, a survey can be used to collect data about the current community perceptions of the organization in terms of environmental responsibility, involvement in the community, and contributions to the community. The survey asks people to rate each of the three factors on a five-point scale using the options very unfavorable, unfavorable, neutral, favorable, and very favorable. Descriptive statistics would be used to identify how the community members perceive the organization. You could determine (a) what percentage of the community held a favorable, a neutral, or an unfavorable perception of the organization on these three factors; (b) the mean score of the community perceptions for each of the three factors; or (c) how often people rated the organization as (1) unfavorable, (2) unfavorable, (3) neutral, (4) favorable, and (5) very favorable (frequencies) on each of the three factors. Public relations practitioners are much more likely to use descriptive statistics than inferential statistics.

Inferential statistics allows a researcher to examine possible relationships between two or more variables. Researchers look at analyses of differences and analyses of relationships with inferential statistics. Evaluative research makes use of inferential statistics. Analyses of difference seek to determine if there is a statistically significant difference between two or more sets of data. Statistically significance is the degree of confidence you have that your results can be found in the larger population and not just in your sample. You need to establish that your results are not a result of chance/accident but more likely a true difference between variables. An example would be an organization testing two versions of a message designed to promote a new product. Two separate groups are exposed to each message and their recall of the message assessed. Inferential statistics, such as “t-test” or “chi-square,” could be used to see if there is a significant difference between recall scores for the two messages. If one message scores significantly higher on recall, you would want to use that message. If the two messages scored the same, you could use either one or both of the messages. Correlation is used to uncover relationships between variables. A correlation indicates if two variables change together in a predicable manner. There is a positive correlation when the value of one variable increases as the value of the other variable increases; for example, ice cream sales tend to increase as the temperature rises. There is a negative correlation when the value of one variable increases while the value of the other variable decreases, and vice versa; for example, sales of new homes tend to drop as the lending interest rate rises. If an organization launches a reputation campaign, it expects a positive correlation between exposure to the reputation messages and stakeholders' evaluations of the organization's reputation. You could collect data to determine if there is a positive [Page 814]correlation between exposure to a reputational message and favorable perceptions of the organization. Again, you would need to determine if the relationship is statistically significant.

...