Evaluation of Two New Assessments for Selecting Telephone Customer Service Representatives

The Phonemin Company is a distributor of men’s and women’s casual clothing. It sells exclusively through its merchandise catalog, which is published four times per year to coincide with seasonal changes in customers’ apparel tastes. Customers may order merchandise from the catalog via mail or over the phone. Currently, 70% of orders are phone orders, and the organization expects this to increase to 85% within the next few years.

The success of the organization is obviously very dependent on the success of the telephone ordering system and the customer service representatives (CSRs) who staff the system. There are currently 185 CSRs; that number should increase to about 225 CSRs to handle the anticipated growth in phone order sales. Though the CSRs are trained to use standardized methods and procedures for handling phone orders, there are still seemingly large differences among them in their job performance. CSR performance is routinely measured in terms of error rate, speed of order taking, and customer complaints. The top 25% and lowest 25% of performers on each of these measures differ by a factor of at least three (i.e., the error rate of the bottom group is three times as high as that of the top group). Strategically, the organization knows that it could substantially enhance CSR per­for­mance (and ultimately sales) if it could improve its staffing “batting average” by more accurately identifying and hiring new CSRs who are likely to be top performers.

The current staffing system for CSRs is straightforward. Applicants are recruited through a combination of employee referrals and newspaper ads. Because turnover among CSRs is so high (50% annually), recruitment is a continuous process at page 357the organization. Applicants complete a standard application blank, which asks for information about education and previous work experience. The information is reviewed by the staffing specialist in the HR department. Only obvious misfits are rejected at this point; the others (95%) are asked to have an interview with the specialist. The interview lasts 20–30 minutes, and at the conclusion the applicant is either rejected or offered a job. Due to the tightness of the labor market and the constant presence of vacancies to be filled, 90% of the interviewees receive job offers. Most of those offers (95%) are accepted, and the new hires attend a one-­week training program before being placed on the job.

The organization has decided to investigate the possibilities of increasing CSR effectiveness through sounder staffing practices. It is not pleased with its current methods of assessing job applicants; it feels that neither the application blank nor the interview provides an accurate and in-­depth assessment of the applicant KSAOs that are truly needed to be an effective CSR. Consequently, it engaged the services of a consulting firm that offers various methods of KSAO assessment, along with validation and installation services. In cooperation with the HR staffing specialist, the consulting firm conducted the following study for the organization.

A special job analysis led to the identification of several specific KSAOs likely to be necessary for successful performance as a CSR. Three of these (clerical speed, clerical accuracy, and interpersonal skills) were singled out for further consideration because of their seemingly high impact on job performance. Two new methods of assessment provided by the consulting firm were chosen for experimentation. The first is a paper-­and-pencil clerical test assessing clerical speed and accuracy. It contains 50 items and has a 30-minute time limit. The second is a brief work sample that could be administered as part of the interview process. In the work sample, the applicant must respond to four different phone calls: a customer who is irate about an out-­of-stock item, a customer who wants more product information about an item than was provided in the catalog, a customer who wants to change an order placed yesterday, and a customer who has a routine order to place. Using a 1–5 rating scale, the interviewer rates the applicant on tactfulness (T) and concern for customers (C). The interviewer is provided with a rating manual containing examples of exceptional (5), average (3), and unacceptable (1) responses by the applicant.

A random sample of 50 current CSRs were chosen to participate in the study. At Time 1 they were administered the clerical test and the work sample; performance data were also gathered from company records for error rate (number of errors per 100 orders), speed (number of orders filled per hour), and customer complaints (number of complaints per week). At Time 2, one week later, the clerical test and the work sample were re-­administered to the CSRs. A member of the consulting firm sat in on all the interviews and served as a second rater of performance on the work sample at Time 1 and Time 2. It is expected that the clerical test and work sample will have positive correlations with speed and negative correlations with error rate and customer complaints.

After reading the description of the study and observing the results above,

1. How do you interpret the reliability results for the clerical test and work sample? Are they favorable enough for Phonemin to consider using them “for keeps” in selecting new job applicants?

2. How do you interpret the validity results for the clerical test and work sample? Are they favorable enough for Phonemin to consider using them “for keeps” in selecting new job applicants?

3. What limitations in the above study should be kept in mind when interpreting the results and deciding whether to use the clerical test and work sample?

How do you interpret the reliability results for the clerical test and work sample? Are they favorable enough for Phonemin to consider using them “for keeps” in selecting new job applicants?

Introduction

The reliability of a test or job sample tells you how accurate the results are. If there is too much variation in your test score, for example, then it might not be useful for predicting performance at all levels of the organization. A high coefficient means that your results are more consistent than those from other tests or work samples; thus, they would be more reliable candidates for hiring purposes. However, since there are no agreed-upon standards for interpreting reliability coefficients—only guidelines from organizations like Phonemin—it’s up to each employer to decide which metric best reflects their needs and criteria when evaluating potential employees for new jobs within their company or organizational structure.

What are reliability results?

When you take a test, such as the clerical test and work sample, the reliability results are your score. If the scores are high, then it means that the scores were consistent over time. For example, if you took two tests on different days and got identical scores for each one (e.g., 90%), then this would be considered “reliable” since there was no way for your performance to change between tests due to fatigue or boredom from sitting in front of a computer screen all day long!

Reliability is important because it helps us determine whether our current system is effective at selecting staff members who will perform well at their jobs. If we don’t have reliable data on how well people do their jobs after they’ve been hired by Phonemin Corporation Ltd., then we won’t know whether our hiring practices are working or not; therefore we wouldn’t know whether or not our current system needs improvement/changes before moving forward with new policies around recruitment/hiring strategies etc…

What is a reliability coefficient?

A reliability coefficient is a measure of how consistent and stable you are when answering questions on a test. It’s calculated by dividing the average standard error by the average standard deviation (a measure of variation). The higher your coefficient, the more reliable your results are!

The Cronbach Alpha is another common reliability coefficient used in psychology research. It measures how strongly correlated different participants’ answers are with one another—in other words, it represents how consistent results tend to be across different people who’ve taken tests together at different times/locations/etc…

How to interpret clerical test and work sample reliability coefficients.

A reliability coefficient of .70 is considered to be good.

A reliability coefficient of .80 is considered to be excellent.

A reliability coefficient of .90 is considered to be perfect.

A reliability coefficient of 1.0 or more is impossible

There is no way to know how much weight to give to a particular test or job sample until you know the source of their reliability.

There is no way to know how much weight to give to a particular test or job sample until you know the source of their reliability.

The clerical test and work sample were designed to measure clerical aptitude, so their reliability is good. However, they do have limitations:

• The clerical exam was designed for specific jobs in specific organizations; people who have applied unsuccessfully may have different abilities from those who succeeded in getting hired as an assistant manager at Wal-Mart (or wherever). This could affect how well the results predict future performance on other tests or assignments.
• The job sample was chosen based on its similarity with what Phonemin wants employees at some point in time—but it does not represent all possible types of tasks that might be assigned by managers over time because there are many different roles within any given organization’s hierarchy (e.g., salesperson vs customer service agent vs customer service representative). If you’re interested in becoming someone else’s assistant manager someday and want a better idea of what kind of person would be best suited for that position, consider taking our “Assessing Potential” test instead!

Conclusion

In order to use test and work sample reliability coefficients, you must be able to interpret them. This is not always as easy as it sounds, because the way that people report their results can be different from what they actually mean by their responses. However, there are some general guidelines that will help you figure out if a test or sample is reliable enough for you to consider using it “for keeps” in selecting new job applicants.