Correlational Research Question Example in Nursing: Staffing and Missed Care

This correlational research question example comes from adult medical nursing. It shows how a concern raised by ward managers becomes a question about the relationship between two measurable variables, and how that one question then shapes the sub-questions, hypotheses, instruments, sample size and statistical analysis of a small quantitative study.

The example is presented as a worked plan for a master’s or final-year dissertation. Each step explains the decision that was made and why, so that you can follow the same route with your own variables.

From topic to correlational research question

The nurse began with a broad topic and narrowed it in four steps. Writing each step down made it easier to explain the final question to the ethics committee and to the dissertation supervisor.

The main correlational research question

Sub-questions under the correlational research question

A correlational dissertation usually needs a descriptive starting point before the relationship can be tested, and then a step that checks whether the relationship holds once other factors are taken into account. Each sub-question below is linked to the data and test that will answer it, which keeps the analysis chapter focused.

Variables and operational definitions

A correlational research question stands or falls on how the variables are measured. “Staffing” and “missed care” are everyday words, so each one needs an operational definition that tells the reader exactly what number will be collected.

The conceptual framework behind the question

A correlational research question is stronger when the dissertation explains why the two variables might be linked. In this example the reasoning is simple and can be shown in a short framework: when each registered nurse has more patients, the total amount of care required on a shift rises while the time available stays the same, so nurses prioritise urgent tasks such as medicines and observations and postpone or omit tasks such as mouth care, walking patients and patient teaching. Other factors can also influence how much care is missed, so they are measured and controlled rather than ignored.

The middle box is not measured in this study. It is the explanation offered for any relationship that is found, and it is the part that a later qualitative or mixed-methods study could explore directly by asking nurses how they decide what to leave undone.

Hypotheses

Each analytic sub-question of the correlational research question has a pair of hypotheses. Writing them before data collection protects the study from “fishing” for significant results afterwards.

H1 (sub-question 3)

Higher numbers of patients per registered nurse are associated with higher missed nursing care scores.

H0 (sub-question 3)

There is no association between patients per registered nurse and missed nursing care scores.

H2 (sub-question 4)

Patients per registered nurse remain a significant predictor of missed care after controlling for shift type, ward and experience.

H0 (sub-question 4)

Once shift type, ward and experience are controlled, patients per registered nurse do not predict missed care.

Why this is a correlational research question

• Two measured variables. Staffing and missed care are both recorded as they occur.
• No manipulation. The researcher does not change staffing levels, which would be neither ethical nor practical.
• Relationship, not cause. The wording asks how the variables move together, not whether staffing causes missed care.
• Quantifiable. Both variables can be scored, so a correlation coefficient and a regression model can be calculated.

What the published evidence already suggests

Before finalising the correlational research question, the nurse checked that it was worth asking and not already fully answered for this setting. Ball and colleagues (2014) surveyed registered nurses in English hospitals and found that nurses frequently reported leaving necessary care undone, and that care left undone was strongly associated with higher numbers of patients per nurse. On a larger scale, the RN4CAST study by Aiken and colleagues (2014) linked nurse survey data on workload to patient mortality across 300 European hospitals. See our cross-sectional study example for how that study was carried out.

These studies justify the question but do not make it redundant. Local managers need local data, and the hospital in this example has no measure of missed care at all. The literature also supplies tested instruments, which saves the student from writing an untested questionnaire.

Design that fits the correlational research question

A cross-sectional survey of registered nurses across the medical wards suits this correlational research question. Each nurse reports staffing on the last shift they worked and completes the MISSCARE Survey at the same time. Because both variables are measured once, the design can show whether they are associated, but not which came first.

Design

Cross-sectional, correlational survey

Population

Registered nurses working on adult medical wards in one hospital

Sampling

All eligible nurses invited (total population sampling); agency staff excluded because they may not know ward routines

Instrument

MISSCARE Survey Part A, plus questions on patients and nurses on the last shift, shift type, ward and experience

Data collection

Anonymous online or paper questionnaire over four weeks, with two reminders

Analysis

Descriptive statistics, correlation and multiple linear regression in SPSS

Sample size

Sample size is calculated for the most demanding analysis, which is the regression model in sub-question 4. Using G*Power for a multiple regression with four predictors, a medium effect size (f² = 0.15), a significance level of 0.05 and power of 0.80 gives a minimum of 85 completed questionnaires. The medical wards employ more registered nurses than this, so inviting all of them and allowing for non-response makes the target realistic. If fewer responses are received, the regression is reported with a note on reduced power, and the correlation in sub-question 3 still stands.

Analysis plan

1. Clean the dataCheck for missing items and impossible values, such as more nurses than patients, and record how each was handled.
2. Describe the sampleReport shift type, ward and experience to show who responded.
3. Answer sub-questions 1 and 2Calculate missed care scores and rank the elements most often missed.
4. Check assumptionsInspect the distribution of both variables; use Spearman’s rho instead of Pearson’s r if they are clearly skewed.
5. Answer sub-question 3Report the correlation coefficient, its direction, its size and its p value.
6. Answer sub-question 4Enter patients per RN with the control variables in a regression model and report the coefficient for staffing with its confidence interval.

Piloting the questionnaire

Before the main survey, the questionnaire is piloted with a handful of registered nurses from a ward outside the study, for example a surgical ward. They are asked how long it took, whether any questions were unclear and whether the question about their last shift was easy to answer accurately. In this example, the pilot might show that nurses on long day shifts are unsure whether to count patients admitted late in the shift. A short instruction, such as “count all patients you were responsible for at any point during the shift”, would then be added. Changes made after piloting are recorded in the methods chapter, and pilot responses are not included in the main analysis.

Ethics and practical points

Because nurses are reporting care that was not done, the survey must be anonymous and must not collect information that could identify an individual or a specific shift. Ward results are only reported where enough nurses responded to protect anonymity. The participant information sheet explains that the aim is to understand pressures on the system, not to judge individual nurses, and that taking part is voluntary. Approval is needed from the university ethics committee and the hospital’s research and development office.

How the result would be reported

Writing the result sentence in advance keeps the answer to the correlational research question in correlational language. The template below has blanks where the real figures would go.

Writing the correlational research question into Chapter 1

The correlational research question is the anchor of the introduction chapter. The problem statement leads to it, and the aim and objectives restate it in practical terms. The extracts below show how the pieces fit together for this study.

Aim

To examine the relationship between nurse-reported staffing adequacy and missed nursing care on adult medical wards.

Objective 1

To measure the level and type of missed nursing care reported by registered nurses.

Objective 2

To measure staffing adequacy as patients per registered nurse on the last shift.

Objective 3

To test the association between staffing adequacy and missed nursing care, controlling for shift type, ward and experience.

Objective 4

To make recommendations to ward and hospital managers.

Limitations to acknowledge

Stating limitations in advance shows the examiner that the correlational research question has been thought through. For this design, the main limitations follow directly from the wording of the question and the choice of a single-hospital survey.

Interpreting the possible answers

A correlational research question can produce three broad answers, and each one is useful. A clear positive association would support managers in reviewing staffing on the wards with the highest workloads and would justify a later intervention study. A weak or absent association would suggest that other factors, such as skill mix, ward layout or the number of admissions and discharges, deserve attention, and these could be added in a follow-up study. An association that disappears once ward is controlled would suggest that differences between wards, rather than shift-to-shift staffing, explain most of the variation. Thinking through these possibilities before collecting data helps to write a balanced discussion chapter whatever the result.

Weak and strong versions of the correlational research question

How the same topic changes with other question types

Seeing the topic rewritten as other question types makes it clear what makes this a correlational research question, and shows which designs each version would need.

Related examples and reviews

Staffing questions appear across nursing research. Our literature review on burnout and staff wellbeing covers the workforce evidence behind this topic, and the review on medication administration errors shows how workload is linked to safety. If your question is about an intervention rather than a relationship, the PICO example shows how to frame it, and the explanatory sequential design example shows how a survey like this one can be followed by interviews.

Common mistakes this correlational research question avoids

• Causal language. Words such as “effect” and “impact” are kept out of the question, the hypotheses and the conclusions.
• Undefined variables. Both variables have operational definitions before any data are collected.
• No link between questions and tests. Every sub-question is matched to a specific analysis.
• Ignoring other explanations. Shift type, ward and experience are measured so their influence can be controlled.
• Unrealistic sample size. The target is calculated, not guessed, and checked against the number of nurses available.

Use this correlational research question example as a template: keep the structure of question, sub-questions, variables, hypotheses and analysis plan, and replace the variables with your own.

References

1. Aiken, L.H., Sloane, D.M., Bruyneel, L. et al. (2014) Nurse staffing and education and hospital mortality in nine European countries: a retrospective observational study. The Lancet, 383(9931), pp. 1824–1830. https://doi.org/10.1016/S0140-6736(13)62631-8
2. Ball, J.E., Murrells, T., Rafferty, A.M., Morrow, E. and Griffiths, P. (2014) ‘Care left undone’ during nursing shifts: associations with workload and perceived quality of care. BMJ Quality & Safety, 23(2), pp. 116–125. https://doi.org/10.1136/bmjqs-2012-001767
3. Kalisch, B.J. and Williams, R.A. (2009) Development and psychometric testing of a tool to measure missed nursing care. Journal of Nursing Administration, 39(5), pp. 211–219. https://doi.org/10.1097/NNA.0b013e3181a23cf5