Confidence intervals in repeated-measures designs: The number of observations principle.

Jarmasz, Jerzy; Hollands, Justin G.

doi:10.1037/a0014164

Cited by 183 publications

(217 citation statements)

References 16 publications

Supporting

Mentioning

215

Contrasting

Order By: Relevance

“…2 Addition response time transfer relative to control addition problems as a function of problem size, with multiplication-practice and addition-test language the same or different. Error bars are the 95 % within-subjects confidence intervals based on the mean square error for the language×size interaction (Jarmasz & Hollands, 2009). Large problems revealed a response time cost for same, relative to different, practice and test languages Note.…”

Section: Discussionmentioning

confidence: 99%

Interoperation transfer in Chinese–English bilinguals’ arithmetic

Campbell

Dowd

2012

Psychon Bull Rev

View full text Add to dashboard Cite

We examined interoperation transfer of practice in adult Chinese-English bilinguals' memory for simple multiplication (6 × 8 0 48) and addition (6 + 8 0 14) facts. The purpose was to determine whether they possessed distinct number-fact representations in both Chinese (L1) and English (L2). Participants repeatedly practiced multiplication problems (e.g., 4 × 5 0 ?), answering a subset in L1 and another subset in L2. Then separate groups answered corresponding addition problems (4 + 5 0 ?) and control addition problems in either L1 (N 0 24) or L2 (N 0 24). The results demonstrated language-specific negative transfer of multiplication practice to corresponding addition problems. Specifically, large simple addition problems (sum > 10) presented a significant response time cost (i.e., retrieval-induced forgetting) after their multiplication counterparts were practiced in the same language, relative to practice in the other language. The results indicate that our Chinese-English bilinguals had multiplication and addition facts represented in distinct language-specific memory stores.

show abstract

Section: Discussionmentioning

confidence: 99%

Interoperation transfer in Chinese–English bilinguals’ arithmetic

Campbell

Dowd

2012

Psychon Bull Rev

View full text Add to dashboard Cite

show abstract

“…They advocate that graphically displaying a set of means along with a representation of statistical error (CIs) helps the researcher identify how the means in a set relate to each other, which is more meaningful than making binary decisions about particular pairs of means within the set. We cannot resolve this debate here but note that the CIs proposed in Jarmasz and Hollands (2009) can simply be multiplied by -2 / 2, which should yield results similar to those obtained with the approaches favored by Tryon or Blouin and Riopelle and allow direct comparison of pairs of means. In a sense, this is a graphical representation of the rule of thumb described earlier.…”

Section: Caveatsmentioning

confidence: 99%

“…

Confidence intervals (CIs) can be used to draw statistical inferences about differences between conditions in an experiment.

…”

mentioning

confidence: 99%

“…The relation of the effect to the larger design is important because it is this relation that affects the size of the MS e term computed in the ANOVA. The interested reader is referred to Jarmasz and Hollands (2009), where we have provided a more detailed treatment.…”

mentioning

confidence: 99%

“…In a recent article (Jarmasz & Hollands, 2009), we have specified the procedures for computing CIs across the range of effects found in factorial RM and IM-RM designs, following the Loftus and Masson (1994;Masson & Loftus, 2003) approach. We reviewed the use of CIs for inferential purposes, explained the difficulties in correctly obtaining the number of observations, and developed a general method for obtaining the number of observations for various effects, which we coined the number-ofobservations principle.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Revisiting confidence intervals for repeated measures designs

Hollands

Jarmasz

2010

Psychonomic Bulletin & Review

View full text Add to dashboard Cite

Loftus and Masson (1994) proposed a method for computing confidence intervals (CIs) in repeated measures (RM) designs and later proposed that RM CIs for factorial designs should be based on number of observations rather than number of participants (Masson & Loftus, 2003 (Jarmasz & Hollands, 2009). In this note, we provide a brief summary of our approach.Confidence intervals (CIs) can be used to draw statistical inferences about differences between conditions in an experiment. Indeed, CIs bear a systematic relationship to significance tests. The difference between the means of two conditions is significant if it exceeds half the total length of the CI (called the margin of error) multiplied by a factor of -2 (Loftus & Masson, 1994). A corresponding visual rule of thumb states that when the margins of error for two means overlap by less than half, the difference is significant. Thus, graphing CIs serves as a highly useful technique for determining whether the various conditions involved in an effect are different from one another.The procedures for computing CIs for independent measures (IMs; also called between-subjects) designs are well known (e.g., Kirk, 1982;Loftus & Loftus, 1988). In contrast, methods for computing CIs in repeated measures (RMs; also called within-subjects) designs have a shorter history. For a long time, there were no published methods for computing RM CIs. However, in 1994, Loftus and Masson published a landmark article in Psychonomic Bulletin & Review, showing how RM CIs could be computed, how they are related to RM significance tests, and how they can be used for making inferences about RM effects. This article has served as an important reference for the large number of experimental psychologists who use RM designs on a frequent basis.In IM designs, a CI can be computed around the mean for each condition, and the size of the CI is affected by the number of participants contributing to each mean. This assumes one observation per participant. For RM designs, a participant serves in multiple conditions, thus generating multiple observations. How does one determine the number of observations for a particular effect in an RM design? The literature on RM CIs has not provided a comprehensive guide to computing the number of observations for factorial designs. Indeed, as Cumming and Finch (2005) noted, it is unclear how well CIs can be effectively used by researchers to understand effects within complex experimental designs. In RM designs, the number of observations is affected by the number of participants, the nature of the effect, and the overall design of the experiment. Thus, there has been a need to clarify the procedure for computing the number of observations for a variety of designs and effects.In a recent article (Jarmasz & Hollands, 2009), we have specified the procedures for computing CIs across the range of effects found in factorial RM and IM-RM designs, following the Loftus and Masson (1994;Masson & Loftus, 2003) approach. We reviewed the use of CIs for inferential purposes, expla...

show abstract

The Effects of Feedback and Reflection on the Questioning Style of Untrained Interviewers in Simulated Child Sexual Abuse Interviews

Krause

Pompedda

Antfolk

et al. 2017

Applied Cognitive Psychology

View full text Add to dashboard Cite

We provided immediate and detailed feedback in a training paradigm in which simulated interviews with computergenerated avatars were used to improve interviewers' questioning style. Fifty-nine untrained student/interviewers conducted eight interviews each and were randomly assigned to a control, feedback or feedback and reflection group. Compared to the control group, the groups receiving feedback used a higher percentage of recommended questions and retrieved more relevant details while using a lower percentage of not recommended questions and retrieved less wrong details. Only the groups that received feedback reached a reliable change in the proportion of recommended questions. The reflection intervention proposed in the present study did not enhance training effects above and beyond feedback in the present sample. The present study replicated previous findings regarding the role of feedback in improving the quality of investigative interviews, however, failing to show an effect of reflection. Further studies on different reflection tasks are suggested.

show abstract

Confidence intervals in repeated-measures designs: The number of observations principle.

Cited by 183 publications

References 16 publications

Interoperation transfer in Chinese–English bilinguals’ arithmetic

Interoperation transfer in Chinese–English bilinguals’ arithmetic

Revisiting confidence intervals for repeated measures designs

The Effects of Feedback and Reflection on the Questioning Style of Untrained Interviewers in Simulated Child Sexual Abuse Interviews

Contact Info

Product

Resources

About