Test-retest correlations of memory span are low, questioning the validity of a single memory span test¶

Claim · All claims
Evidence · 1 study · associational r1 · q3
Currency · newest study 1918 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Across 45 inter-trial correlations for 97 children tested on ten days, the average coefficient was 0.37 for Consecutive Series scores and 0.54 for Maximum Series scores, with only two exceeding 0.70. → Mitchell 1918

Evidence¶

Mitchell 1918¶

Mitchell, David. (1918). Variability in Memory Span. The Journal of Educational Psychology. https://www.google.com/search?q=%22Variability+in+Memory+Span%22+Mitchell+1918

q3 · i? · associational · r1

Correlational analysis of the ten-day study: 45 coefficients computed among trials for 97 children. The article reports "The average is 0.37" for Consecutive Series and "The average coefficient of correlation for the "Maximum Series" score is 0.54." No standardized effect size is printed for this finding.

"The average is 0.37. The average coefficient of correlation for the "Maximum Series" score is 0.54. Thirty-three come between 0.50. and 0.69; only two are greater than 0.70."

Discussion¶

Related Claims¶

• Lengthening the interval between dictated digits markedly lowered children's memory span scores on one testing day — related
• Individual children's memory span scores vary considerably between two testings five weeks apart — related
• Group average memory span scores remain nearly identical across days while individual scores vary considerably — related
• The digit series used was too short, with 73 of 970 trials reaching the maximum score of ten — related
• In one class of 13-year-old boys, memory scores were kept to a steady rank order by repeated practice (successive tests correlating +.87 to +.90) before fatigue testing began — related
• Across repeated digit tests pupils improved with practice, with large uniform increases attributed to unconscious memory of the previous day's digits, and no fatigue effect on recall was found — related
• The digit memory-span increases with age rather than with grade-promotion measures of intelligence, reaching a limit of six digits for most grammar and high school pupils — related
• Psychological methods of fatigue measurement afford useful average values and reliable information about the course and effect of fatigue (Offner) — reports the opposite
• During the final fatigue test, time per problem rose gradually, then markedly, with no obvious correlation between time and accuracy — related
• The study's conclusions rest on only sixty-five individually tested unretarded cases, limiting generalization — related
• Mental index ratings on the group test showed close correspondence over a two-year interval (r = .72, rank correlation .77) among 46 retested children — related
• Auditory memory (digit-span) tests are overrepresented in the adult scale and depress the rankings of otherwise high-performing students — related
• Longer letter series yield somewhat higher average scores because they offer opportunities for occasional high scores via mental grouping — related