← All work Statistics · English Longitudinal Study of Ageing

Staying active is worth about seven years of memory in later life

Data ELSA Wave 7, 9,670 adultsMethod Mann-Whitney, Kruskal-Wallis, Spearman, chi-square, regressionTools SPSS
TL;DR

Age, sex, health, mood, physical activity and social participation together explain 21% of the variation in a 0–20 word-recall score. Measured in "years of ageing", high activity is worth about seven years and each social group joined about three. The activity gap holds in every age group, so it is not just younger people being more active, though a snapshot cannot prove cause.

Open the full report (PDF, 35 pages) ↗Methods, five tests with hypotheses, regression and full SPSS output

The question

How much of the difference in memory between two people of the same age comes from things they can change, such as activity, mood and social life? ELSA measures memory directly, with a face-to-face word-recall test, rather than asking people to rate it.

Approach

  • Memory = immediate + delayed recall of ten words (0 to 20). Scores were not normally distributed (Kolmogorov-Smirnov p < 0.001), so tests were non-parametric.
  • Five tests, each with its own hypotheses: Mann-Whitney U (sex), Kruskal-Wallis (self-rated health), Spearman (age), chi-square with odds ratio (activity and low mood), then multiple regression with collinearity checks.
  • Effect sizes translated into "years of ageing" (each coefficient ÷ the age coefficient), because with 9,000 people almost everything is significant.

Findings

≈ 7 yearsmemory difference linked to high physical activity
4.66×odds of low mood, least vs most active (95% CI 3.85 to 5.63)
21%of variation in memory explained by the seven factors
The activity gap holds at every ageThe activity gap holds at every age
Average memory score by activity level for each age group: about two points higher for the most active in every group.
Ageing hits holding on to information more than taking it inAgeing hits holding on to information more than taking it in
Immediate recall falls about 30% from the youngest to oldest group; delayed recall falls about 44%.
Low mood rises as activity fallsLow mood rises as activity falls
Low mood reported by 5.9% of the most active, 9.7% of the middle group and 22.5% of the least active.

Limitations

  • Cross-sectional: memory decline could reduce activity, not the other way round.
  • Low mood is a single question; activity is frequency, not intensity.
  • People aged 90+ have no exact age, so the age effect is likely understated.
  • Next step: follow the same people across several ELSA waves.

All figures

Select a figure to enlarge it.