NURS 8211 Week 7 Level-Matched Data Displays Example

Four Measures, Four Levels, Four Displays: Descriptive Summaries Matched to Level of Measurement for the Heart Failure Teach-Back Project

Student Name

College of Nursing, Walden University

NURS 8211: Data Collection and Analysis for Evidence-Based Practice

Instructor Name

Month Day, Year

Four Measures, Four Levels, Four Displays: Descriptive Summaries Matched to Level of Measurement for the Heart Failure Teach-Back Project

What This Paper Does

Four measures at four levels of measurement cannot share one chart type or one summary statistic. This paper summarizes and displays four measures drawn from the composite unit's teach-back work, one at each of the levels Stevens (1946) defined: nominal, ordinal, interval and ratio. For each, it gives the summary statistic suited to the level, a display suited to the data and a caption that states what the display shows and what it leaves unsaid.

All data in this paper are illustrative. They were constructed to resemble six months of discharges from a unit of this size, 214 eligible discharges home, and they do not describe any real unit or patient. They are used to show method, not to support conclusions.

Levels of Measurement and Why They Matter

Stevens (1946) argued that the operations permitted on a set of numbers depend on how those numbers were assigned. Nominal data are categories with no order; the permitted summaries are counts, percentages and the mode. Ordinal data have an order but no fixed distance between categories; medians and percentiles are permitted, but means are not meaningful. Interval data have equal distances but no true zero, so differences can be compared but ratios cannot. Ratio data have a true zero, so all arithmetic, including ratios, is meaningful. Choosing the wrong summary, such as averaging ordinal codes, produces numbers that look precise and mean little.

Measure 1 (Nominal): Who Received the Teaching

Field 7 of the abstraction form records whether teaching was given to the patient, a caregiver, both or not recorded. The categories have no order. The summary is counts and percentages, and the mode is patient alone.

Table 1

Recipient of Teach-Back Teaching, Illustrative Data (n = 214)

Figure 1 (described). A horizontal bar chart with one bar per category, sorted from largest to smallest, with the not recorded bar shown last in gray. Caption: The chart shows who was taught; it does not show whether the person taught was the one who manages medications at home, which the project did not record.

Measure 2 (Ordinal): The Patient's Response on the Weight Rule

Field 4 codes the response on the daily weight rule as correct the first time, correct after re-teaching or not correct at discharge. These categories are ordered, but the distance between them is not fixed; there is no sense in which correct after re-teaching is exactly halfway between the others. The summary is frequencies in order and the median category. Among the 206 discharges with a recorded response, the median category is correct the first time, since more than half fall there.

Table 2

Response on the Daily Weight Rule, Illustrative Data (n = 206 with a recorded response)

Figure 2 (described). A single stacked horizontal bar divided into the three categories in their natural order, with the percentage in each segment. Caption: The chart shows that about nine in ten patients could state the rule by the time they left; it does not show whether they could still state it a week later, or whether stating it led to daily weighing.

Measure 3 (Interval): Clock Time of Discharge

Clock time is interval data. The difference between 2 p.m. and 4 p.m. is the same two hours as between 6 p.m. and 8 p.m., but midnight is not a true zero, so it makes no sense to say that an 8 p.m. discharge is twice as late as a 4 p.m. one. Means and medians are both permitted; the median is reported because discharge times are bunched in the afternoon and trail into the evening. In the illustrative data, the median discharge time is 4:40 p.m., the interquartile range runs from 2:10 p.m. to 6:55 p.m. and 38% of discharges occur after 5 p.m.

Figure 3 (described). A histogram of discharges by clock hour, from 8 a.m. to 11 p.m., with a vertical line at 5 p.m. marking the usual end of the day shift. Caption: The histogram shows when patients leave; it does not show when teaching took place, which may have been hours earlier, so it cannot by itself explain differences in teach-back documentation by time of day.

Measure 4 (Ratio): Minutes From Discharge Order to Departure

Minutes from discharge order to departure are ratio data, with a true zero. All summaries are permitted, including statements such as one wait being twice as long as another. The data are skewed, since a few patients wait many hours for transport or medications, so the median and interquartile range are reported alongside the mean. In the illustrative data, the median is 142 minutes, the interquartile range 96 to 215 minutes and the mean 168 minutes, pulled upward by a small number of waits longer than eight hours.

Table 3

Minutes From Discharge Order to Departure, Illustrative Data (n = 214)

Figure 4 (described). A dot plot showing every discharge as a point, with a box marking the interquartile range and a line at the median. Caption: The plot shows the full spread of waits, including the long tail; it does not show why patients waited, and it cannot say how much of the wait, if any, was due to teach-back.

Why These Display Choices

The displays were chosen to show the data rather than hide them. Bar charts of means are common in reports, but they conceal the distribution behind each bar, and a review of published figures argued for displays such as dot plots that show individual values, especially for small samples (Weissgerber et al., 2015). The dot plot for waiting time shows the long waits that a mean alone would blur. For categorical measures, bar charts remain appropriate, since each bar is a count, not a summary of a hidden distribution.

Every caption states a limit because descriptive displays invite causal readings. A reader who sees that 38% of discharges happen after 5 p.m. and that documentation is weaker in the evening may conclude that late discharges cause poor teaching. The data in this paper cannot support that conclusion; they only describe. Analysis of change over time, in the next stage of the project, will use run and control charts, which display data over time and have rules for telling signals from random variation (Perla et al., 2011).

References

Perla, R. J., Provost, L. P., & Murray, S. K. (2011). The run chart: A simple analytical tool for learning from variation in healthcare processes. BMJ Quality & Safety, 20(1), 46-51. https://doi.org/10.1136/bmjqs.2009.037895

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

Weissgerber, T. L., Milic, N. M., Winham, S. J., & Garovic, V. D. (2015). Beyond bar and line graphs: Time for a new data presentation paradigm. PLOS Biology, 13(4), Article e1002128. https://doi.org/10.1371/journal.pbio.1002128