SD and SE in a quitting report answer different questions

Read the label attached to the exact result

A paper may give a mean with SD in one table and an effect estimate with SE elsewhere. Locate the result behind the claim you are reading and copy its label accurately. A plus-or-minus sign or a short vertical line does not reveal the statistic by itself. Check the caption and methods.

Cochrane’s effect-measures chapter explicitly warns about confusing SD and SE. That is a reporting and analysis distinction, not evidence about a quitting treatment. This guide does not supply a normal range, determine whether a person’s score is concerning, or calculate an individual clinical outcome.

Cochrane: Cochrane Handbook, Chapter 6: Choosing effect measures

Keep participant variation separate from estimation uncertainty

An SD can describe how scattered the observed scores are around their mean under the stated calculation. An SE of that mean addresses uncertainty in estimating a population mean using the data and assumptions. An SE of a group difference concerns a different estimate again; do not detach the letters from that target.

For an original fictional reading example, one report labels bars “SD of observed scores,” while another labels them “SE of the group mean.” The same-looking bars do not describe the same quantity. You cannot conclude that participants were more similar in the second report merely because its bars are shorter.

Avoid a conversion that leaves the methods behind

In a simple independent-observation setting, an SE of a mean relates to the SD and observation count. Actual study designs and estimates can require other calculations. Repeated observations, grouping, and adjusted analyses should not be reduced to a formula copied from an unrelated example.

Cochrane discusses obtaining SD from SE only with the correct target and conditions, including distinguishing a group mean from a difference between groups. For an introductory reading task, the safer output is the reported label and any unresolved definition. Ask the authors or a qualified statistical source for clarification rather than reverse-engineering bars from a screenshot.

Cochrane: Cochrane Handbook, Chapter 6: Choosing effect measures

Describe what the statistic does not establish

An SD is not a treatment-effect estimate or a safe individual threshold. An SE is not protection against bias, a test of whether the outcome matters, or your chance of quitting. Keep the comparison, measurement definition, population, and follow-up visible alongside either statistic.

A bounded summary says: “The table reports this mean with this labeled statistic; it concerns observation spread or estimate uncertainty as specified.” If the label is missing, say that you cannot identify the bars. This guide cannot validate a scale, reanalyze a named study, or turn informal craving notes into the reported research measure.