RES 710 Week 1 Variables and Measurement Example

What Exactly Are We Measuring? Variables and Levels of Measurement in a Credit Union Service Study

[Student Name]

University of Phoenix

RES/710: Statistical Research Methods and Design I

Week 1 Assignment

[Instructor Name]

[Date]

The learner, the credit union, its branches and all data are composites written for a model paper.

The learner in this paper is a composite operations director at a member-owned credit union in Grand Rapids, Michigan, with 24 branches, a phone center and a mobile app serving about 180,000 members. Member satisfaction scores fell four points on a 100-point index over two years, and branch managers blame longer lines since staffing was cut, while the digital team suspects members who move to the app are less attached. The director wants to study these relationships carefully for her DBA research. Before collecting any data, she needs to decide exactly what she will measure, how and at what level.

From Concepts to Variables

A concept, such as satisfaction or waiting, is an abstract idea; a variable is a measurable characteristic that can take different values; an operational definition states exactly how the variable will be measured in this study. The concept of waiting, for example, could be measured as minutes from arrival to service, as members' perception of whether the wait was long or as the number of people ahead in line. Each choice measures something different, and the study's conclusions will apply only to the measure chosen.

Roles of Variables

Field (2018) describes variables by their role in a research design. An independent or predictor variable is the one whose effect or association is studied; a dependent or outcome variable is the one expected to vary with it; control variables are measured to account for alternative explanations; moderating variables change the strength of a relationship; and mediating variables carry it.

In the credit union study, wait time is the main predictor; member satisfaction is the main outcome; service channel may moderate the relationship, since waiting on the phone may feel different from waiting in a lobby; member tenure, age and branch size are controls.

Levels of Measurement

Stevens (1946) proposed that measurement scales fall into four types, nominal, ordinal, interval and ratio, each defined by the operations that are meaningful on it, and that the type determines which statistics are permissible. Nominal scales name categories, such as branch, phone or app, and allow counts and modes. Ordinal scales rank categories, such as education level, and allow medians and rank-based tests. Interval scales have equal distances without a true zero, such as temperature in Fahrenheit, and allow means and standard deviations. Ratio scales add a meaningful zero point, as with minutes waited or dollars deposited, and allow all arithmetic operations, including ratios.

Classifying the Study's Variables

Wait time, recorded by the branch queue system in minutes, is ratio. Service channel is nominal with three categories. Member tenure in years is ratio. Branch size, measured as daily transactions, is ratio. Member age is ratio but may be grouped into categories for reporting, which would make it ordinal.

Satisfaction will be measured with a four-item scale, each item rated from 1, very dissatisfied, to 7, very satisfied. Each single item is technically ordinal, but the average of several items is often treated as approximately interval, a common practice that the learner will justify by checking the scale's distribution and reliability. Net promoter likelihood, a single 0-to-10 item, is ordinal but often analyzed as interval; she will run both rank-based and mean-based analyses to check whether conclusions differ.

Minutes waited and satisfaction ratings look equally numeric on a spreadsheet, but only one of them has a true zero.

Choosing Among Possible Measures

For several concepts, the learner had to choose among measures. Satisfaction could be captured by a single overall rating, which is quick, or a multi-item scale, which is more reliable; she chose the scale. Waiting could be measured objectively from the queue system or by asking members how long they felt they waited; she will collect both, since perceived wait may matter more for satisfaction than actual minutes, a possibility worth testing. Channel use could be the channel of the most recent transaction or the share of a member's transactions in each channel over a year; she chose the latter as a second variable, digital share, measured on a ratio scale from 0 to 100 percent.

Validity Concerns Specific to This Study

Two threats stand out. Members who respond to surveys may be more engaged or more upset than others, so satisfaction scores may not represent all members. And satisfaction asked right after a transaction may reflect that visit more than the overall relationship. The learner will compare respondents with nonrespondents on tenure and channel use and will add one item asking about overall satisfaction with the credit union, to distinguish visit satisfaction from relationship satisfaction.

Measurement Quality

Reliability is the consistency of a measure; validity is whether it measures what it is meant to measure. Hinkin (1998) described steps for developing sound survey measures: generating items from theory, reviewing content with experts, pretesting, assessing internal consistency and examining factor structure and relationships with other variables. The satisfaction items come from an established banking service scale, which has reported internal consistency above .85 in prior studies; the learner will check reliability in her sample. Queue-system wait times are objective but may miss time spent before members take a ticket; she will observe 200 visits at four branches to estimate that gap and adjust if it is large.

The Variable Table

Wait time: minutes from ticket to service; queue system; ratio; predictor; correlation, regression.

Service channel: branch, phone or app; transaction record; nominal; moderator; group comparisons.

Satisfaction: mean of four 7-point items; survey; treated as interval; outcome; means, t tests, regression.

Net promoter likelihood: 0 to 10; survey; ordinal or interval; secondary outcome; rank and mean tests.

Tenure, age, branch size: records; ratio; controls.

Why Levels Matter for Later Weeks

The levels chosen now constrain every later analysis. Because wait time is ratio, the learner can compute means, correlations and regression coefficients and say, for example, that each additional five minutes of waiting is associated with a certain change in satisfaction. Because channel is nominal, she can compare groups but cannot compute an average channel. If she had recorded wait time only as short, medium or long, she would have lost information and been limited to rank-based tests. Recording variables at the most precise level available keeps options open.

Data Collection Notes

Survey invitations will go to members within 48 hours of a transaction, linking each response to that transaction's wait time and channel. This timing reduces recall error but means members with no recent transactions are excluded, a limit to note when interpreting results for less active members.

Conclusion

Translating the director's concerns into operationally defined variables with clear roles and levels sets up every analysis that follows. Ratio wait times, nominal channels and scaled satisfaction ratings each permit different statistics, and planned checks of reliability and validity will show whether the measures deserve trust. Week 2 will describe the data with appropriate summary statistics.

References

Field, A. (2018). Discovering statistics using IBM SPSS statistics (5th ed.). Sage.

Hinkin, T. R. (1998). A brief tutorial on the development of measures for use in survey questionnaires. Organizational Research Methods, 1(1), 104-121. https://doi.org/10.1177/109442819800100106

Stevens, S. S. (1946). On the theory of scales of measurement. Science, 103(2684), 677-680. https://doi.org/10.1126/science.103.2684.677