From the department of Surgery and also Perioperative Care, Dell medical School in ~ the college of Texas in ~ Austin, Austin, Texas.

You are watching: Inappropriate use of descriptive statistics

Accepted for publication August 4, 2017.

Published ahead of publish September 7, 2017.

Funding: None.

The writer declares no conflicts of interest.

Reprints will not be obtainable from the author.

Address post to cutting board R. Vetter, MD, MPH, department of Surgery and also Perioperative Care, Dell clinical School at the university of Texas in ~ Austin, Health exploration Bldg, Room 6.812, 1701 Trinity St, Austin, TX 78712. Resolve e-mail come .


Anesthesia & Analgesia: November 2017 - Volume 125 - worry 5 - ns 1797-1802
doi: 10.1213/ANE.0000000000002471
free Metrics
AbstractIn quick

Descriptive statistics are particular methods basically provided to calculate, describe, and also summarize accumulated research data in a logical, meaningful, and also efficient way. Descriptive statistics room reported numerically in the manuscript text and/or in that is tables, or graphically in its figures. This simple statistical accuse discusses a series of fundamental concepts about descriptive statistics and also their reporting. The mean, median, and mode are 3 actions of the center or central tendency the a set of data. In enhancement to a measure up of its main tendency (mean, median, or mode), one more important properties of a research data set is that variability or dispersion (ie, spread). In simplest terms, variability is how much the individual recorded scores or observed values differ native one another. The range, traditional deviation, and interquartile selection are 3 actions of variability or dispersion. The traditional deviation is commonly reported for a mean, and the interquartile variety for a median. Trial and error for statistical significance, together with calculating the it was observed treatment effect (or the toughness of the association between an exposure and an outcome), and also generating a matching confidence interval space 3 tools commonly used by researchers (and your collaborating biostatistician or epidemiologist) come validly make inferences and an ext generalized conclusions indigenous their accumulated data and descriptive statistics. A number of brickandmortarphilly.com, including Anesthesia & Analgesia, strong encourage or call for the report of pertinent to trust intervals. A confidence interval can be calculated for virtually any variable or outcome measure in one experimental, quasi-experimental, or observational study study design. Typically speaking, in a clinical trial, the trust interval is the range of values within i beg your pardon the true treatment effect in the populace likely resides. In an observational study, the to trust interval is the range of worths within i beg your pardon the true stamin of the association in between the exposure and also the result (eg, the risk ratio or odds ratio) in the population likely resides. There room many feasible ways to graphically display or show different types of data. While there is frequently latitude regarding the an option of format, ultimately, the simplest and also most comprehensible layout is preferred. Usual examples incorporate a histogram, bar chart, heat chart or heat graph, pie chart, scatterplot, and box-and-whisker plot. Valid and also reliable descriptive statistics deserve to answer straightforward yet important questions about a study data set, namely: “Who, What, Why, When, Where, How, how Much?”


I store six moral serving-men;

(They taught me all i knew)

Their names space What and Why and When

And How and Where and also Who.

Rudyard Kipling (1902), “The Elephant’s Child” from Just therefore Stories

Descriptive statistics are particular methods basically provided to calculate, describe, and also summarize collected research data in a logical, meaningful, and also efficient way. Descriptive statistics space reported numerically in the manuscript message and/or in that tables, or graphically in the figures.1–3

As insightfully observed by Grimes and also Schulz,4 “Descriptive studies frequently represent the first scientific toes in the water in brand-new areas the inquiry. A an essential element of descriptive reporting is a clear, specific, and also measurable an interpretation of the disease or condition in question. Like newspapers, an excellent descriptive report answers the 5 simple W questions: who, what, why, when, where…and a sixth: therefore what?” This simple statistical accuse discusses the following basic concepts around descriptive statistics and their reporting:

complete study sample dimension versus study team sizes research sample allude estimate Frequency, percentage, ratio, and also proportion procedures of the central tendency of data actions of the variability or dispersion of data to trust interval (CI) as measure the the precision that a allude estimate Graphically displaying different types of data

TOTAL research SAMPLE dimension VERSUS STUDY team SIZES

An understanding of statistics is predicated on the key distinction between a research sample and also the population from which the is selected, in that traditional inferential statistics permits one to extrapolate indigenous the descriptive statistics the a sample come infer other or to make conclusions around the research population.3,5–7

For example, in a clinical study choose a randomized controlled trial, the statistical evaluation of the descriptive data native a random sample of patient is supplied to make inferences and conclusions around the larger populace and similar other, future patients.5

Prospectively determining the ideal study sample size and also the study group or subsample sizes is vital in conducting a valid and also ethical quantitative research study.8–10 The mechanics of doing so will be the subject of a future statistical tutorial.

When report the descriptive findings of a study study, that is very important to provide (a) the full sample dimension (ie, the total number of subjects sampled) and (b) the study team sizes (ie, the number of subjects sampled in every study team or subsamples) in the text, tables, and also figures.11,12

SAMPLE point ESTIMATE

Based ~ above a sample that data the is randomly accumulated at a given allude in time, an approximated value have the right to be generated. Together a sample suggest estimate is a single value that have the right to be offered to validly estimate the corresponding population parameter.3,13–15 A collection of repeated, arbitrarily (unbiased) samples of data built up from the very same underlying populace would be supposed to generate various yet same valid point estimates the the corresponding populace parameter.13–15

FREQUENCY, PERCENTAGE, RATIO, and also PROPORTION

As debated in the previous statistics tutorial, some demographic and clinical features can it is in parsed into and described using separate, discrete categories. Such categorical data deserve to be one of two people dichotomous (2 categories) or polytomous (more 보다 2 categories).2,16–19

Dichotomous and also polytomous categorical data can be explained as (a) the life counts or absolute frequencies (eg, “50”) of the category or (b) the percentages or loved one frequencies (eg, “50%”) of the categories. Family member frequencies can also be reported together ratios (eg, “50:50”), proportions (eg, “50/100”), or decimal (eg, “0.50”).

Provide Numerators and Denominators for all Reported Percentages or Proportions

When reporting any kind of observed portion or relationship in the abstract, text, and/or tables that a manuscript, the authors should administer the matching numerator and also denominator.

See more: My Sister The Heroine, And I The Villainess (Novel) Series, My Sister The Heroine, And I The Villainess

MEASURES of THE main TENDENCY the DATA

*
Figure 1.:
A hypothetical instance of a histogram displaying a discordant mean, median, and mode values for a skewed, nonnormally (non-Gaussian) dispersed data set.