Statistical analysis challenges in validity and reliability for business research

Wal Mart Stats

Wal-Mart Survey Analysis

In order to begin addressing two key research questions, how can Wal-Mart ensure decent quality products and simultaneously keep consumers happy with prices and how would Wal-Mart keep the costs of its products down, a survey of Wal-Mart customers was taken. Attitudes towards prices, quality of products, and overall shopping experiences were collected and analyzed via in-person questionnaires/surveys to consumers exiting Wal-Mart stores after shopping. All responses were recorded on a Likert scale of 1-5, with 5 representing high satisfaction or strong agreement and 1 representing extreme dissatisfaction or strong disagreement. The following paragraphs assess the quality of this survey and its results in terms of addressing the research question.

Validity and Reliability

The research questions themselves as detailed above are very valid questions to be asked of Wal-Mart, a company that built its business model on the ability to offer low-priced goods to consumers while of course still satisfying consumer demands for quality and performance (Wal-Mart, 2012). Understanding how a balance is struck between the two competing elements of low price and high quality in order to achieve customer satisfaction, not only in products but with the shopping experience as a whole, is quite a valid approach to analyzing and perhaps improving Wal-Mart's business. While the specific data collected in this study should prove useful in furthering the stated ends of the research questions, though, the data collected does not in and of itself represent a valid or necessarily a reliable manner for addressing these specific research questions. The sample size should be large enough to ensure statistical reliability and therefore yield reliable analysis, yet the wide variance in responses and the high numbers for overall satisfaction ratings of 5, 4, and 3 as well as the imbalance in other responses vs. overall satisfaction ratings all suggest that the survey does not reliably measure this overall satisfaction, as consistency of meaning might not exist from respondent to respondent (Cooper & Schindler, 2011; McClave et al., 2011). The survey questions also do not directly address the research questions, only measuring consumer experience and not identifying any operation-end actions Wal-Mart can take or is taking to meet the goals of low prices and high quality/consumer satisfaction; though consumers seem satisfied this research does not identify why, and combine with the above-described issues this means it lacks internal and external validity (Cooper & Schindler, 2011).

Steps to Minimize Challenges

Though there are certain inherent and foundational validity issues in the research, there are steps that can be taken to minimize the degree to which these erode the usefulness of the research. Short of changing the research questions there is nothing that can be done to increase external validity and force the survey to address the specific questions identified, but it might not be a bad idea to adjust these research questions to fit the data collected, making the investigation still useful if not in the manner originally intended (Cooper & Schindler, 2011). The survey is a valid measurement of Wal-Mart's success in blending low prices with a demanded quality level, so shifting the question from "what can Wal-Mart do" to "is Wal-Mart doing things right" would greatly increase the external validity of the survey (Cooper & Schindler, 2011). Acknowledging the variance in responses and attempting to account for it through an identification of linguistic ambiguities in the survey will demonstrate a commitment to maintaining the reliability of the research and will make the analysis more readily accepted by others, and focusing the analysis on the large-scale and reliable trend of overall satisfaction with Wal-Mart rather than trying to break the results into more discrete levels of satisfaction will also improve the reliability of the results and analysis (Cooper & Schindler, 2011; McClave et al., 2011).

Classification of Findings